TY - JOUR
AB - A decision of the Federal Joint Committee Germany states that negative pressure wound therapy is not accepted as a standard therapy with full reimbursement by the health insurance companies in Germany. This decision is based on the rapid report and the final report of the Institute for Quality and Efficiency in Health Care, which demonstrated through systematic reviews and meta-analysis of previous studies projects that an insufficient state of evidence regarding the use of negative pressure wound therapy (NPWT) for treatment of acute and chronic wounds exists. The Institute for Research in Operative Medicine (IFOM) as part of the University of Witten / Herdecke gGmbH is an independent scientific institute that is responsible for the planning, implementation, analysis and publication of trial projects regarding the efficacy and effectiveness of negative pressure wound therapy for acute and chronic wounds in both medical sectors (in- and outpatient care) in Germany. The study projects are designed and conducted with the aim to provide solid evidence regarding the efficacy of NPWT. The trials evaluate the treatment outcome of the application of a technical medical device which is based on the principle of negative pressure wound therapy (Intervention Group) in comparison to standard wound therapy (Control group) in the treatment of chronic foot wounds and acute subcutaneous abdominal wounds after surgery. All used treatment systems bear the CE mark and will be used within normal conditions of clinical routine and according to manufacturer’s instructions. The aim of the trial projects is to compare the clinical, safety and economic results of both treatment arms. Study results will be provided until the end of 2014 to contribute to the final decision of the Federal Joint Committee Germany regarding the general admission of negative pressure wound therapy as a standard of performance within both medical sectors. Introduction: Adequate patient recruitment is a key condition determining the validity, duration and costs of RCTs, yet remains challenging. The expanding interface between therapeutic endoscopy and minimally invasive surgery for the treatment of benign gastrointestinal diseases demands RCTs to compare safety and cost-effectiveness of comparable interventions. We aimed to identify patient motives and barriers for (non)participation in the TREND study. Methods: Patients with large rectal adenomas counseled between January and July 2011 for participation in an RCT comparing endoscopic mucosal resection (EMR) with transanal endoscopic microsurgery (TEM) were invited for a semi-structured interview (14 participants, 12 non- participants). Interviewees were asked to discuss their main motive for (non-) participation in an open fashion. Subsequently, potential other barriers and motivations as previously described in the literature were presented by the interviewer. Interviews were coded and analyzed by 2 independent researchers. Results: We interviewed 10 participants and 7 non-participants, aged 54-84. Key motive for trial participation was contribution to medical science and future clinical practice (all participants). Other motives included satisfactory counseling, sufficient time to reflect and the sense that participation was completely voluntary. The vast majority of non-participants had previous in-hospital experience, however more than half had not been introduced to research prior. Although most non-participants felt contributing to research was important, key barriers to participation included a distinct preference for either EMR (n=4) or TEM (n=3) or intervention-specific characteristics like type of sedation (all non-participants). Consultation of family members affected half of all decisions. Study characteristics such as randomization, blinding, insurance and ethical approval hardly influenced the decision making. Conclusion: In RCTs comparing similar endoscopic and surgical strategies, thorough, tailored and timely counseling is crucial to clarify the study purpose and to emphasize the presumed equality of allocation arms. Avoidance of common barriers to participation will improve trial inclusion. The establishment of clinical trial centers creates a demand for both infrastructure and skilled staff. While establishment and operation of clinical trial centers in surgical specialties have in recent years often been conducted by medical staff in addition to clinical practice, the intensification of legal regulations and the increased number of clinical trials in surgery requires rethinking. The ability to fulfill these more specialized demands can only be given by allocating sufficient funds. These funds can be used to employ trained paramedical staff and furthermore to release medical staff from clinical work and thereby enable full-time scientific work. On the other hand, a demand for funding is risen to implement infrastructural requirements such as IT hardware and premises. Once a clinical trial center has been established, further operation has to be ensured by generating additional funds. While the demand for start-up financing of a clinical trial unit is most meaningful arranged by governmental or institutional fundraising, the operation relies on generating sufficient funds through the conduction of industry-initiated trials. Own experience and further possibilities are demonstrated by reporting on different strategies for funding of clinical trial centers and the general demand of funds during establishment and operation of clinical trial centers. Therefore, different governmental, industrial as well as institutional supports are relevant factors. In Europe safety management requirements in clinical trials with medical devices were changed fundamentally by the council directive 2007/47/EC in 2007 and internationally with the ISO 14155:2011 in 2011. Previously adverse incidents had to be reported to the competent authority. Since implementation of the directive in Germany each serious adverse event (SAE) which happened or might have happened to a patient, user, or third party, must be reported immediately. Sponsor and investigators must report SAEs via an electronic form centrally provided by the Federal Institute for Drugs and Medical Devices (BfArM). Besides profound knowledge adequate IT equipment at all investigator trial sites are needed. The latter being a challenge for academic trials with their low budgets. To develop operational competence the quality management (QM) working group of the Clinical Trials Centers (CTC)-network developed a harmonized SOP describing thecomplex procedures of safety management for sponsor and investigator trial sites. It includes appendices providing among others definitions, a template for an SAE-manual, and instructions for completing the electronic form. This document particularly suggests solutions regarding more general regulatory requirements,e.g. follow-up information for the sponsor or for the competent authority. The implemented safety management procedures consider experiences derived from clinical trials with medicinal products and the special characteristics of academic trials. As increased numbers of clinical trials with medical devices are to be expected according to changed regulations CTCs will become more experienced with these complex procedures. Thus they will be competent partners to industry in conducting national and international clinical trials with medical devices. Our report will show the challenges of coping with this novel electronic reporting process, and provide our experiences in fulfilling the updated changed German national regulations. Due to the ever increasing global activity in clinical research, knowledge transfer between researchers and all stakeholders in clinical health care (including patients) worldwide is an indispensable requirement. A prerequisite for knowledge transfer in clinical research is an unbiased and complete view on clinical trial results. How is the arising number of clinical trials registries and results databases being utilized by different user groups including ethics committees, clinicians, patients, researchers, reimbursement and funding institutions? What are the different requirements defined by the parties involved? How can variable needs be met? What seem to be the arising issues concerning database content, language barriers, retrieval problems, implementation of knowledge into daily work, evaluation of existing evidence? What efforts are made to overcome the obstacles? Meanwhile it is daily experience that conduct of a clinical quality assurance audit involves some kind of computer system to be considered and assessed. The following fundamental question comes immediately to mind: how valid are data and information kept and handled by the systems employed? This question will be subdivided in five parts for the purpose of this session: Where does the basic computer system used come from, how has it been implemented and validated and how is it maintained? How has the computer system application used in a specific trial been developed, validated and deployed and how is it operated and maintained? Which regulations, guidelines and state of the art standards are to be expected to be observed for development, validation and operation? How big are the risks on data quality and acceptability of the study outcome in case of malfunction of the computer system application? How high are chances that system weaknesses can be detected in a standard audit setting and how can auditors optimise the preparation and conduct of their review to detect hidden system bugs? The session presents an overview over the current situation, naming specific risks and gives some answers to the questions posed afore, mainly by giving practical examples how to proceed in the most frequent audit situations, namely investigator site audits (eClinical and electronic source documents), CRO system audits and vendor audits. A flexible and simple Bayesian decision- theoretic design for dose finding trials is proposed in this paper. In order to reduce the computational complexity, we adopt a working model which produces analytic posterior distributions. In addition, this working model is sufficiently flexible to fit all monotonic dose-toxicity curves. We also discuss how to use a proper utility function to reflect the interest of the trial. Patients are allocated based on not only the utility function but also the chosen dose selection rule. The most popular dose selection rule is the one-step-look-ahead (OSLA), which selects the best so far dose. More complicated rules such as two-step-look- ahead (TSLA) are surely more efficient than OSLA only when the required distributional assumptions are met which is however often not the case in practice. We carry out extensive simulation studies to evaluate a variety of dose selection rules and have found that OSLA is often more efficient than TSLA under our proposed method. Moreover, our simulation results show that the proposed method performs superior to several popular Bayesian methods and the negative impact of prior mis-specification could be considered in the design stage. Isotonic Design using Normalized Equivalent Toxicity Score (ID-NETS) is a novel Phase I design originally proposed by Chen et al. by integrating the novel toxicity scoring system proposed by Chen et al. and the original Isotonic Design proposed by Leung et al. ID-NETS has substantially improved the accuracy of maximum tolerated dose (MTD) estimation and efficiency of trial by fully utilizing all toxicities of each patient and treating toxicity response as a quasi-continuous variable instead of a binary indicator of dose limiting toxicity (DLT) in a Phase I clinical trial. To facilitate designing and conducting a Phase I clinical trial with ID-NETS, we have developed a user-friendly software ID-NETS©TM, which has two functions: 1) Calculating the recommended next dose level from completed data; and 2) Performing simulation to obtain the operating characteristics of a trial. Currently, ID-NETS©TM v1.0 is available for download at http://winshipbbisr.emory.edu/IDNETS.html. Building on the toxicity probability interval (TPI) design in Ji et al. (2007), we present a modified TPI design (mTPI) that is calibration-free for phase I trials. Our goal is to further simplify the design and improve the trial conduct, and provide more effective and safer methods while maintaining the simplicity of the original TPI design. Like the TPI method, mTPI consists of a practical dose-finding scheme guided by the posterior inference using a simple Bayesian model. However, the new method benefits from improved dose-finding decision rules based on a new statistic, the unit probability mass (UPM). The improvement through the use of the UPM for dose finding is significant. We will present extensive simulation results comparing the mTPI design to the 3+3 and CRM methods, and provide convincing evidence why the mTPI is a practically superior method. Among the most recent advances in escalation with overdose control (EWOC) based Bayesian adaptive phase I clinical trial designs, Tighiouart et al (2011) developed EWOC based on the proportional hazards model (EWOC-PH) using time to toxicity to estimate the MTD. The method has been shown to be more efficient to estimate MTD comparing to the original EWOC and a version of time to event EWOC (TITE-EWOC) proposed by Mauguen et al (2010). In this study, we will further extent EWOC-PH to take into account patients’ baseline covariates. The extension of EWOC to accommodate baseline covariates has been developed by Babb and Rogatko (2001) and Tighiouart and Rogatko (2010). We expect that the new design will provide better safety protection by assigning a personalized dose to the next available patient based on his/her own baseline characteristics and lead to an estimation for covariate specific phase II dose. We assess the operating characteristics for the design via extensive simulations including three scenarios: (1) design using a covariate; (2) Design ignoring the covariate; (3) Design using separate trials. The efficiency of estimating the conditional MTD and safety of the trial are compared with original EWOC with covariate using DLT as a binary indicator of toxicity. Escalation with overdose control (EWOC) is a successful Phase I design and has been widely used. Chen et al. further extended EWOC to utilize fully all toxicities of patients instead of a binary indicator of dose limiting toxicity (DLT) by incorporating the normalized toxicity scoring system (NETS) proposed by Chen et al. into EWOC with a quasi-Bernoulli likelihood approach. The new design is called EWOC-NETS which has been demonstrated by simulation studies to have good operating characteristics and improve the accuracy and efficiency of the maximum tolerated dose (MTD) estimation relative to common Phase I designs. In this study, we have developed an interactive software called EWOC-NETS©TM which is very user friendly to facilitate designing and conducting a Phase I clinical trial with EWOC-NETS and available for free download at the website of Emory University http://winshipbbisr.emory.edu/Software.html. We would like to introduce this novel statistical software and translate EWOC-NETS into Phase I cancer clinical trials in the future. Some clinical predictors may or may not have monotonic properties. They may have increasing or decreasing pattern or they may remain steady during a period of observation time. A clinical question is raised on how such increasing or decreasing or constant pattern of variations may have effect on survival or on clinical outcomes independently during a period of time. Generally we ignore such point of transitions and independent pattern of variations while investigating their overall effect. Isotonicity is called the simple or partial ordered monotonicity mathematically. Isotonic regression is performed under ordered monotonic restrictions criteria. Either increasing or decreasing or constant variation of a clinical predicator or of a prognostic variable generates a change-point problem at the transition from increasing order to decreasing order or vice versa or from the steady condition to either increasing or decreasing pattern of clinical predictors. The primary thrust of this paper is to address such clinical questions using a Cox Regression Model with isotonic regressors with such change-point problems. In a case study with multiple myeloma the impact of such change- point patterns of clinically significant predictors such as hemoglobin, blood urea nitrogen, white blood corpuscles, platelets on patients median survival time have been explored. Policy makers in Denmark are starting to realize that RCT designs are needed if they want to know the impact of social interventions. Therefore the Ministry of Social Affair granted funding to do three RCTs of family programs (Multidimension Treatment Foster Care-MTFC, Parent Management Training Oregon -PMTO and Multisystemic Therapy-MST). During the last year SFI has been planning the studies including all relevant parts in the process. The process has been long and challenging. Problems faced include: 1: A pronounced resistance to randomization within all levels of social services. 2: An economic crisis in the municipalities causing both unwillingness to participate in the research study and a significant lower caseload in the MST and MTFC teams. 3: Fear of the study going to steal half of the families from the programs. 4: Underfunding of the studies. 5: Resistance against using standardized tests because this is not common in social work, and 6: Difficulties in getting data for power calculations. Recurring issues that we have had to deal with are: “do we really have to randomize? Can’t we just do some kind of matching?” and “it would be much better to spend the money on a wider implementation of the program and practice oriented qualitative research.” Recently the MST study was cancelled since all five MST teams declined to participate. Funding was transferred to the other two studies making them look more promising. A lot of resources have been used to prepare the studies and expectations are high. Hopefully by May 2012 the MTFC study will be recruiting families and the PMTO study will be ready to start recruiting in September 2012. If all goes well the two studies will be among the first RCTs carried out on social interventions in Denmark. Regulatory goals for a clinical study can be different from research and exploratory goals. For the Food and Drug Administration (FDA), Phase 3 Investigational New Drug (IND) studies (i.e., adequate and well- controlled studies) will provide the primary clinical evidence to support a marketing application such as a Biologic License Application (BLA). To support approval and initiation of these studies, the protocol should cover all the statistical issues considered by the reviewers when assessing the study design and its ability to adequately address the study objectives. As an aid to statistical reviewers, as well as to improve efficiency and consistency in our reviews of these protocols, the Division of Biostatistics in the Center for Biologics Evaluation and Research (CBER) has developed a checklist outlining the important elements of a Phase 3 protocol. Some elements included in this checklist are consistency between the study objectives and endpoints; number and regions of study sites; statistical hypotheses; planned interim analyses; sample size assumptions and calculations; randomization description; blinding techniques; analysis populations; multiplicity considerations; missing data considerations; study success criterion; and study conduct. Original IND submissions with protocols with such elements may also lead to less correspondence with the sponsor to clarify items in the protocol. We note that the elements identified in this checklist may also be helpful in designing earlier phase studies as well as non-IND studies. This presentation will focus on the motivation, contents, and use of the checklist. In non-inferiority trials, a pre-specified margin of non-inferiority must be defined a priori. However, what happens when an acceptable risk varies according to physician risk tolerance or patient circumstances, so a pre-specified margin of non-inferiority cannot be easily defined? We will discuss a double-blind clinical trial in which SLE patients on long-term MMF therapy with stable disease are randomized to continue therapy or withdraw. It is expected that withdrawal from MMF will result in increased risk of disease reactivation; however, long- term MMF treatment is not without risk of side effects. The goal of the trial is to determine if the increase in risk of disease reactivation outweighs the benefits of withdrawal. One could frame this goal as a test of the non-inferiority hypothesis that the increased risk of disease reactivation is greater than a pre-specified margin, determined by potential benefit. However, how the risks and benefits should be weighed is highly individual, even for experienced rheumatologists. Rather than design the study to test a hypothesis of non- inferiority using an arbitrary risk margin that may or may not be relevant for any particular patient, the results of this trial will be reported as effect estimates for change in risk of disease reactivation in addition to other disease activity, safety, quality of life and medication use endpoints. These effect estimates and confidence intervals will describe the risks and benefits of withdrawal from study therapy and may be used to guide physicians and patients in making decisions based on individualized assessments of acceptable risk. As clinicians are likely to find interpretation of results difficult in the absence of hypothesis testing (and p-values), we will discuss the approach planned for disseminating clinical trial results. Cluster randomised trials (CRTs), compared to individually randomised trials, require specific analysis; statistical models need to take account of between-cluster variability. This abstract focuses on two main approaches to analysis, cluster-level (CL) and individual-level (IL), and examines their efficiency for the Training Caregivers After Stroke trial (TRACS). TRACS is a CRT evaluating a complex intervention in stroke rehabilitation with 928 dyads of patients and caregivers within 36 stoke rehabilitation units. The primary outcome is NEADL score at 6 months. CL analysis is attractive due to its simplicity. We compare performance of three techniques; simple summaries (using appropriate two- sample weighted t-tests), CL analysis adjusted for covariates and nonparametric methods. The CL approach is a two-stage method; in unadjusted analysis, a summary measure for each cluster is calculated and two sets of cluster-specific measures are compared using the weighted t-test. When performing CL analysis adjusted for covariates; we carried out an IL regression with all but intervention effect covariates ignoring the clustering effect. Secondly, we compared the residuals for each cluster between treatment arms using the weighted two-sample t-test. Wilcoxon’s rank sum test was used as a nonparametric substitute to the two-sample t- test. IL analysis, using a random effects linear mixed model with stroke rehabilitation units as level 2, was the preferred method, because of its greater computational convenience and it is possible to analyse the intervention effect and other covariates simultaneously. The CL approach is robust, particularly for a smaller number of clusters. However, it may not be statistically the most efficient, especially if clusters are of varying size. Such weighting is incorporated in IL analysis. Analysis of cluster randomised trials can be undertaken in a number of ways. We will present and discuss the advantages and disadvantages of each of these analytical methods. Simple randomisation is the easiest method for allocating participants to treatment groups in clinical trials. In the long run it balances all features of participants across the groups but may not be suitable for small to medium sized trials, or for larger trials at the time of interim analysis. If important prognostic factors are identified at the design stage then minimisation can help to balance these features. There is uncertainty as to whether treatment groups need to be balanced at baseline, but many researchers believe this to be advantageous, allowing comparison between groups without the need for sophisticated analysis. Balance between treatment groups can be desirable in a range of scenarios: for small trials, interim analyses, early termination, analysis of subgroups or where the credibility of an unbalanced trial is considered problematic (e.g. in the case of a small treatment effect). Three completed trials of varying size, originally using minimisation as method of treatment allocation, were “re-randomised”, in the order that participants joined the study, using simple randomisation. Datasets frozen at the time of either interim or full analysis were used. For all three trials treatment allocation was well balanced across prognostic variables and between treatment arms when using minimisation at all time points, i.e. for any number of participants recruited at that time. With simple randomisation, at each timepoint imbalances were identified that could have made analysis more difficult. In some cases the potential imbalance across treatment groups within a factor reached 100% (where all participants with a given characteristic were in the same treatment allocation group) and no amount of sophisticated analysis could compensate for this. The simulations demonstrated the need for incorporating minimisation into the randomisation algorithm of trials of any size in order to achieve treatment balance which cannot be achieved by using simple randomisation alone. A futility analysis is commonly utilized in clinical trials for early read of trial efficacy data. An investigational drug will be declared futile if the prespecified futility boundary is met, and the trial will be stopped. In a conventional group sequential design, a futility boundary is selected based on power preservation or sufficient conditional power, which protects the probability of rejecting the null hypothesis at the end of a trial. However, the observed treatment effect size is of increasing importance beyond rejecting the null hypothesis and is directly associated with the eventual success of drug development. To ensure a high probability of observing a desired effect size, or in another words a high probability of success, we propose a new futility analysis design approach where the futility boundary is selected based on preserving the probability of success. We define the relative preservation of this probability as “pseudo-power”, using which we propose the boundary selection criteria. Via a case study, we evaluate various operational characteristic of this approach in term of the probability of correct and incorrect stopping, with respect to the futility boundary, the underlying true effect size and the timing of the futility analysis. Increasingly, the NIH and Industry sponsors are establishing large multi-center collaborative research efforts that implement multiple studies simultaneously through a network of investigators. This presents a challenge for the sponsors and leadership teams who are responsible for monitoring activities across studies. The business world has addressed this need for complex information monitoring systems with a high- level summarized overview called a dashboard. We created a web-based dashboard for the Inner City Asthma Consortium (ICAC), a large, federally-funded, multi-center research consortium. The consortium has eight active studies that are being conducted across ten clinical centers with total projected enrollment of over 3500 participants. Many research programs provide web-based resources to aid stakeholders with study supervision. Unfortunately, traditional study portals can be hindered by an overabundance of information and tedious top-down navigation. These characteristics make them an inefficient resource for large multi-center programs. Our dashboard avoids the typical information- dense, hierarchical construction of traditional websites in favor of a comprehensive single- webpage display that is easier for stakeholders to access and use. Dashboards summarize and display critical information on a single page, with an emphasis on graphical, rather than text-heavy displays. Our dashboard has a main page for the consortium, which provides comparative enrollment graphs for the active studies, and displays pertinent deadlines and announcements. Each study also has a dedicated page that follows this same “dashboard” approach. We update the dashboard data weekly. Sponsors have responded positively to the ICAC dashboard, voicing appreciation for the system’s efficient presentation of data and ease of use. Given the success of the dashboard, we plan to expand the system to other federal and commercial programs. Our experience suggests that multi-center or multi-study programs would benefit from creating a dashboard to provide sponsors and network leadership teams with key study monitoring information in a quick, succinct manner. Due to the lack of infrastructure in resource poor settings and despite the strong need for Clinical Research coordination, Information and Communication Technology has not been systematically and sustainably integrated into health research, hospital administration and practice, thus limiting efficiency and outcomes of the efforts to fight poverty related diseases. Cloud-based technology could now overcome infrastructure scarcity problems of resource poor settings and thus greatly support Clinical Research by integrating mobile, online, offline and Voice over IP technology. The Italian Inter University Consortium Cineca, a non-profit Consortium of Italian Universities for high performance computing and ICT, has an over 20 year experience in providing technological support to health research for activities related to the design and development of IT systems and services in the health care and biomedical area, using advanced technologies and methodologies. In particular, since 2008 Cineca has been involved in projects providing and developing technology to collect and analyze clinical and health related data in Africa. The European funded project Medishare, coordinated by Cineca, successfully managed to collect standardized data for over 15000 patients affected by HIV/AIDS, Malaria and TB in Kenya Tanzania and Uganda in one central database. Medishare was recognized supportive of the Millennium Development Goals by the United Nations. The Europe – Africa Research Network for Evaluation of Second-line Therapy (EARNEST) is a trial supported by the European and Developing Countries Clinical Trials Partnership (EDCTP) that is successfully using Cineca integrated technology in Uganda, Malawi, Zimbabwe. Cineca exploits the integration of newer technologies to fill in the infrastructural gaps of Africa and provides a sustainable centralized infrastructure which is fully certified for quality, safety and security procedures. We have developed a LIMS to manage biobank samples for the Scottish Early Rheumatoid Arthritis (SERA) cohort study. Commercial and open-source solutions were evaluated and it was concluded that development of a bespoke LIMS system was needed to provide scalability and flexibility to support future projects with different requirements. The system developed includes a generic workflow which is capable of modelling: • Complex storage configurations with up to 1024 dimensions over multiple geographic sites • Storage rules for sample type, method and time of storage • Multi-event work flow processing from site collection to arrival at labs • Lab technician workflows incorporating working plate practices • Manual and automated validation methods using a combination of 1D and 2D barcode technologies The system is based on a tree structure; all samples are stored in a parentage tree that shows the lineage of the sample allowing for exploration of any number of generations of derived samples. Our tree based approach allows the LIMS to be largely generic. By defining the storage, site configuration and the collection protocols the system can operate with scenarios ranging from a single lab collecting ad-hoc samples, to a multi-lab, multi collection site protocol hosting thousands of samples. We will discuss the functionality provided by the LIMS system, the tree based design and the possibilities it allows, the pros and cons of the development, its integration with two additional large multicentre, multinational, clinical trials and future functionality enhancements. The Glasgow Safe Haven aims to provide the physical location and informatics infrastructure to permit confidential and secure access to routinely collected healthcare data for research purposes. We will discuss the design of this infrastructure illustrating how the availability of routinely collected data can potentially enhance the design and delivery of clinical trials. This will include: • Extraction of prescription and laboratory data during the follow-up phase of a clinical trial • Identification of potentially eligible participants for a multicentre clinical trial in sub-clinical hypothyroidism using centrally held routinely collected laboratory data • Building a clinical data warehouse in specific disease areas to support clinical trial feasibility assessment. We will discuss the proposed governance and access structure. The presentation will be illustrated with examples of trials that will benefit from this infrastructure. The Robertson Centre for Biostatistics, University of Glasgow have been involved in the development of electronic solutions to enhance the conduct of clinical research projects for many years. This includes integrated secure electronic solutions for enrolment and randomisation (web and IVRS), electronic data capture and management via e-CRF (on- and off-line), trial management, information delivery and reporting, eg Independent Data Monitoring Committees. Our systems have been implemented for many clinical trials both commercial and academic. To enhance our ability to produce robust and validated systems efficiently we foresaw a requirement for a standardized library of components. In addition, our systems have to be able to provide specific functionality in accordance with our trial requirements. Included in our system: Standard screens and functionality to allow trial co-ordinators to register their own users and assign the appropriate access rights. Inclusion of previously developed screens eg standard questionnaires Demographics Concomitant medications Medical history Standard measurements eg Height, Weight, Blood Pressure Adverse Events and Serious Adverse Events, and regulatory reporting requirements Metrics required for reporting to funders within the United Kingdom Implementation of CDISC SDTM to standardize our databases Notification and communication to users, either by email or bulletin style within the system Document repository to ensure all sites have correct versions of approved documents We will discuss the implementation of these components in our recent clinical trial systems and how this has improved efficiency in delivering validated systems to our clients. In addition, we will discuss and further explore our implementation of CDISC formats Objective. We present a community developed system focussing on clinical sites, operated as a service by the Clinical Trials Center of the University of Cologne, which employs collaborative peer reviewing in order to support timeliness, efficacy, accuracy and safety in site management. Background. Managing clinical sites in a trial can be very resource-intensive for sponsor and for study site, this, largely due to disconnect in systems and processes, impeding different parties from collaboratively capturing organizational data meaningful to all. Methods. Sponsors, Coordinating Centers, Sites, Study Groups, Societies, Networks and other relevant parties participating in clinical trials each are individually best positioned and incentivized to capture and/or monitor certain organizational trial data. On clinicalsite.org role-based access and specific views enhance individual value and incentives. Reusability of data, meaningful to all, provides the strongest incentive. As a case in point, roles and responsibilities in a trial can be syndicated to be displayed on institutional websites. The data can also be aggregated to the trial track record by the investigator as part of his CV. The sponsor, in turn, can review site and investigator qualifications electronically, cutting turnaround considerably. Furthermore, with roles formally captured and confirmed, the system can act as a Physician Master Index and provide authentication and authorization credentials for external systems. A plugin framework enables parties to embed their own apps while saving the data locally. An API for various services and formats can serve data for aggregation within external systems. Both options provide added value and minimize centrally hosted data, ensuring better data privacy compliance. Results. Currently 1282 trials with 1681 investigators and 997 organizational units involved are managed in the system. Conclusion. Our approach provides critical organizational data of quality with less effort. Exposing the data for reuse and aggregation unlocks new possibilities within and outside the system. Within any organization the workflow structure and departments are clearly defined. Through a case study we will examine the lessons learned and outcomes when staff within specific departments are trained round robin to understand the functions of other departments. By specifically reviewing the cross training of Clinical Research Associates (CRAs) to periodically perform data entry within a CRO, we show meaningful subject matter expertise improvements, positive study impact, and a gain in project efficiency. Pros: Learning Experience: The trainer becomes more familiar with details of his or her position when outlining a training program to teach others. The trainee once completing the instructional information has a broader picture of the entire process and often is able to see the bigger picture and the results of where their work is going. Error Identification: CRA’s have the skills to notice consistency errors that a site may be making emphasizing the need for additional training or reeducation in a particular area. CRAs have the expertise to identify clinical errors including but not limited to the incorrect use of a specific form and inclusion/exclusion criteria errors. Cons: Time and Funds: CRA’s are often paid at a higher rate than data entry staff and their talents may be better utilized in their respective area of expertise. Analysis of Data: CRA’s in most cases have a significant understanding of site operations and clinical information which usually equates to a slow data entry process as they are always scouring data for errors and not just entering what is documented. After considering both pros and cons and the review of a case study where CRA’s perform data entry tasks for quality purposes this often proves to be beneficial not only to the project, but Corporate Structure and capabilities. The National Lung Screening Trial (NLST) was a randomized controlled trial, funded by the National Cancer Institute, to determine whether screening with low-dose helical computed tomography reduces lung cancer mortality relative to screening with conventional chest x-ray in persons at elevated risk of lung cancer. It was comprised of two components: the Lung Screening Study (LSS) and the American College of Radiology Imaging Network (ACRIN). The Coordinating Center (CC) for the LSS (n=34,614) provided coordination and data management support for 10 screening sites. One CC challenge was to ensure completeness and accuracy of data collected throughout all stages of the study (e.g., recruitment, screening, follow up, endpoint verification, close-out). As part of monitoring, the CC conducted visits to each site annually for 9 years. Visits included observation of procedures and random chart audits. Chart audits served to confirm source documents for consents; eligibility and randomization verification; screening results; and diagnostic evaluation and followup. On-site monitoring also provided an opportunity to review and reinforce key study elements with project staff. Annual assessments of the CC audit plan resulted in 6 revisions to the audit form and 4 revisions to the chart selection, reflecting evolving focus from recruitment through close- out. Throughout the trial, 2457 randomly selected charts (7%) were reviewed. Source documents verified ~ 50 data elements during Year 1 and >150 data elements in later years. While chart audits confirmed overall data integrity, selected findings included 17 changes in eligibility status, 18 new protocol violations, and >200 requests for data clarification and modification. The CC found that chart audits were an important part of the total monitoring plan, and that it was important to develop a flexible approach. We will present the challenges of chart audits as part of a monitoring plan throughout the lifecycle of this complex long-term multi- center trial. Clinical trial sponsors are required to set up appropriate measures to monitor the conduct of the trial. The aim of monitoring is to ensure the patients’ well being, compliance with the approved protocol and regulatory requirements, as well as data accuracy and completeness. Classical monitoring approaches that rely on on-site visits are useful for some of these purposes, but extensive source data verification is extremely time consuming and may have only a limited impact on data quality. It is therefore not surprising that the current practice of performing intensive on-site monitoring is coming into question and that interest focuses on more pragmatic, risk- based approaches that improve the cost- effectiveness ratio without compromising the quality and integrity of clinical trials. A recent draft guidance of the Food and Drug Administration (FDA) reflects this trend and states unequivocally: “FDA encourages greater reliance on centralized monitoring practices than has been the case historically, with correspondingly less emphasis on on-site monitoring”. In this presentation, we first review the potential sources of data errors in clinical trials. We then outline the principles of central statistical monitoring and the challenges of its implementation in actual trials. Results from both terminated and on-going trials are presented to illustrate typical findings that can be expected from the monitoring approach. We conclude by a discussion of the potential role and limitations of central statistical monitoring, and we argue that it can both optimize on-site monitoring and improve the quality of clinical trial data. Reference: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM269919.pdf (accessed 10 November 2011). Due to a dramatic increase in the number and complexity of clinical trials and the costs associated with them, the FDA recently introduced a draft guidance “Oversight of Clinical Investigations - A Risk-based Approach to Monitoring”. This new guidance suggests it is acceptable to use alternative approaches to monitoring clinical trials including remote monitoring, centralized monitoring, and risk-based monitoring. The guidance also suggests that source data verification should be focused on critical fields (key efficacy and safety variables) and less than 100% source data verification for less important fields may be acceptable. The guidance gives a clear signal encouraging sponsors to explore cost-effective ways to conduct clinical monitoring instead of relying solely on the on-site monitoring. KAI Research, Inc. has used a variety of alternative approaches to assess data quality and patient safety in its support of NIH clinical trials which often do not have budgets for 100% monitoring. We will describe lessons learned from implementing risk based monitoring approaches that include a central data review to identify potential anomalies. The anomalies help to focus site visits. Focusing on critical data elements has also helped to streamline monitoring visits. We have also implemented centralized monitoring techniques on a sample of data from each of the sites in a study. Through a description of these examples, we will demonstrate alternative methods for ascertaining data quality and subject protection that are both cost effective and efficient. We will also describe how our monitoring plans consider complexity of the study, risk associated with the intervention as well as its safety profile, severity of the illness of the study subjects and experience of the investigators and study sites. We will demonstrate that well thought out alternative approaches to monitoring can meet FDA’s expectations for quality assurance and participant safety. Considerable time and expense are incurred to ensure data quality in a long-term clinical trial. The Protocol Monitor assumes much of the scientific responsibility of closely checking the accuracy and completeness of data collected prior to construction of the final data set. Administrative responsibilities involve assuring clinical sites’ adherence to the protocol, federal guidelines and regulations. Eighty-two clinical sites participate in the Age- Related Eye Disease Study 2 (AREDS2) and are following approximately 4,000 participants throughout five years of follow-up. Data pertaining to effects of supplemental doses of xanthophylls and omega-3 fatty acids on the progression of age-related macular degeneration and vision loss, among other clinical outcomes, are collected at annual visits. Four AREDS2 Protocol Monitors serve as the front line of communication between the study leadership and the clinical sites. The Monitors employ multiple strategies to assist the sites in meeting project deliverables. Routine performance includes standardized protocol review calls on a variety of topics and frequent review of multiple types of data anomaly reports. Monitoring responsibilities constantly change in nature and effort with the revolving door of new Clinic Coordinators and amendments or additions to the protocol through the life of the study. When a new Clinic Coordinator is hired by a site, the Monitor must quickly determine the Coordinator’s knowledge of the protocol, disease topic, and general skills required for the proper conduct of clinical trials. Some clinical sites have educational sessions in place or access to the previous Clinic Coordinator, and others do not. The monitor also carries much historical knowledge of a site’s subject population and may serve as a valuable resource to the new Clinic Coordinator. The many different hats of a Protocol Monitor, as experienced in AREDS2, will be described as both a challenging and rewarding experience. It has been suggested that central statistical monitoring may serve as the foundation for quality assurance and center monitoring in multicenter clinical trials (Knatterud 1998; Buyse 1999; Baiget 2008; Eisenstein 2008). Authors have proposed that central statistical monitoring may identify procedural errors, data errors and data fabrication at centers. Since there are no published evaluations of the different ways of performing central statistical monitoring to detect fraud, we constructed several prognostic models using data from a multi-center trial in which the data from 9 of 196 centers were documented to be fabricated. These data were used to build a series of risk scores to predict fraud at centers. Five different risk scores were identified and each had the ability to discriminate well between centers with and without fabricated data (area under the curve values ranged from 0.90 to 0.95). True and false positive rates are presented for each risk score to arrive at a recommended cut off of 7 or above (high risk score) out of a possible score of 12. We validated these risk scores, using an independent multi-center trial database that contained no data fabrication and found the occurrence of false positive high risk scores to be low and comparable to the model-building data set. With further validation, these risk scores may become part of a series of tools that provide evidence- based central statistical monitoring, which in turn may improve the efficiency of trials, and minimize the need for expensive on-site monitoring. Phase II studies in cancer usually consider short-term outcomes such as response, that are to some degree ‘surrogates’ of phase III outcomes, most typically overall survival (OS). Understanding the relationships between phase II and phase III outcome measures is essential in designing phase II trials and understanding the degree of reliability with which we can move between phases. We consider the association between alternative phase II outcome measures and OS in advanced colorectal cancer (aCRC), to identify measures which may predict treatment effect on OS in phase III trials. Phase II outcomes of response, disease control, continuous tumour measurements and progression-free survival (PFS) were considered. Using surrogacy methodology originally proposed by Buyse et al (2000), estimates of R-squared(R2) were calculated to assess the relationship with OS for each outcome. Individual patient data on 5435 patients from seven trials of aCRC, recruited between 1999 and 2007, were obtained. Data from three 3- arm trials were split to form two treatment comparisons each, resulting in a total of 10 grouping units. As the reference outcome, response was found to have poor predictive ability of the treatment effect on OS, with R2 (trial)=0.14, 95% CI (-0.28,0.56). A much stronger relationship was observed for continuous tumour measurements (R2(trial) =0.65 (0, 0.86)) and PFS (R2(trial)=0.59 (0.18,1.00)). The relationship between OS and disease control was slightly weaker, but still stronger than for response, with R2(trial) =0.44, 95% CI (-0.05,0.93). Sensitivity analyses were performed to investigate alternative grouping units and the impact of covariates. Issues faced in applying surrogacy methodology to the phase II setting, and using these results to identify an optimal outcome measure for phase II aCRC trials, will be discussed. Specific issues include the use of OS as an appropriate phase III outcome, comparability of R2 values for differing outcome measures, and the number of grouping units incorporated. Since 2004, we have developed five phase II trials in Chronic Lymphocytic Leukaemia (CLL), utilising six different statistical methods. Two trials have closed to recruitment and three are currently open. The rationale behind the different designs and methodologies used will be explained. Difficulties and learning experiences with the implementation, wider understanding and interpretation of the trials will be discussed. CLL201 used Gehan’s two-stage approach to assess response, and randomised to a control arm which was not included for formal comparison, but to give validity of the study results. The two stage approach was difficult to implement and of limited use as it only considered efficacy and not safety. Challenges were faced regarding the inclusion and interpretation of the control arm, although randomisation proved to be valuable. CLL207 was a single arm trial designed using Bryant and Day’s two-stage design, incorporating toxicity considerations as well as efficacy. The two-stage aspect worked well, but the implementation of a toxicity stopping guide proved problematic. ARCTIC and ADMIRE are two large, randomised phase IIb trials, both formally powered to compare responses against a common control arm. One of the trials assesses non-inferiority. The choice of the unusual phase II design, and difficulties in justifying it to reviewers will be discussed. COSMIC is a randomised selection design with two experimental arms, combining two phase II trial designs to firstly assess efficacy, and secondly select the treatment to be taken forward. The A’Hern one-stage design is used to determine which treatments are eligible to be further investigated. In the case where both are acceptable, Sargent & Goldberg’s selection criteria will be used. The sample size was inflated to ensure acceptable power for selecting the best treatment. Single-arm Phase II consolidation/maintenance trials in oncology are currently designed to show an improvement in Progression-Free Survival (PFS) when considered against historical controls. PFS includes duration of second line therapy (SLT), treatment free interval (TFI), and time on the investigational therapy (IT). We hypothesize that the duration of SLT, the time on the investigational therapy and patient enrollment plan can affect efficacy measures from maintenance trials and might result in underpowered studies. Efficacy data from four published single-arm consolidation therapies in second remission in ovarian cancer (Sabbatini et al, Gyn Onc 2010, 116(1):66-71) are used for illustration. The studies were designed to show increase in estimated median PFS from 9 to 13.5 months (mos). We partitioned PFS as the sum of the duration of SLT, TFI, and duration of IT. We calculated the statistical power when IT is given concurrently with SLT or following SLT by varying the start of IT. Required sample sizes varied with duration of SLT. If IT starts with initiation of SLT, only 34 patients are needed to provide 80% power to detect a 33% hazard reduction. Alternatively, if one accrues 34 patients at start of SLT with expected 33% hazard reduction and the duration of SLT delays the start of IT to 7.5 months, the power drops to 50%. To maintain 80% power in this scenario, either the protocol therapy would have to reduce the hazard by 50%, or sample size would have to increase to 104. A longer duration of SLT is associated with lower statistical power unless the magnitude of benefit with the new investigational treatment is greater than expected, or the sample size is increased. Designs of consolidation trials should take into account the duration of SLT, by either excluding it from the definition of PFS or restricting it per protocol. Background: RCTs frequently use mortality review committees to assign a cause of death (COD) rather than relying on COD assignments on death certificates (DCs). The National Lung Screening Trial (NLST), an RCT of lung cancer screening with low dose radiation computed tomography versus chest x-ray among heavy and/or long-term smokers, used a committee blinded to arm assignment to determine whether COD was due to lung cancer. Methods: NLST’s committee reviewed a subset of deaths, chosen by a pre-determined algorithm. The algorithm selected deaths with a DC COD that were most likely to represent a death due to lung cancer (either directly or indirectly) and deaths that might have been erroneously assigned a DC lung cancer COD. Also included were deaths within six months of a screen suspicious for lung cancer and within 60 days of certain lung cancer diagnostic evaluation procedures. Using the committee assignment as the gold standard and a lung cancer COD as the outcome of interest, we calculated positive predictive value (PPV) and negative predictive value (NPV) of the death certificate COD assignment (lung cancer vs. non-lung cancer). Results: The committee examined and assigned COD for 1643 deaths (42% of the 3877 NLST deaths). Sensitivity was 90%; specificity, 97%; PPV, 97%; and NPV, 88%. The kappa statistic was 0.87. 40% of the deaths with a DC non-lung cancer COD reclassified to lung cancer had a DC COD of a neoplasm other than lung. Limitations: 2141 deaths with a DC-non lung cancer COD were not reviewed because the DC COD was unlikely to have been in error. Had these deaths been reviewed and assigned a non-lung cancer COD, the specificity would have been 99% and the NPV 97%. Conclusions: When assigning lung cancer COD among heavy/long-term smokers, death certificates provide accurate information nearly all the time. To streamline the implementation of clinical studies, reduce start-up time and accelerate data aggregation across studies, the National Institutes of Health (NIH) together with KAI Research, Inc. (KAI), embarked on common data element (CDE) projects. KAI has developed a systematic approach to the development and implementation of data standards. These standards span the entire study lifecycle, facilitating study start up, data collection and management, and data archiving and sharing. With more and more CDEs developed, establishing standard approaches to manage CDEs becomes critical. This abstract will provide information on the challenges with creating, implementing and managing standards. The challenges as well as our best practices in overcoming these challenges are: – Keeping up with current standards (CDISC, caBIG, etc.) ◦ Harmonization efforts ongoing ◦ Regular communications with these groups key – Reviewing the Literature ◦ Including references for the CDEs adds credibility ◦ Publications may lack detailed data management information – Limiting the universe ◦ Identification of core CDEs (those most essential for data collection) ◦ Build from existing forms to gain buy in from the clinical research community – Creating a team of experts ◦ Identifying who should be part of the team ◦ Advantages and disadvantages to top down vs. bottom up development – Establishing standard procedures for standard development ◦ Development of standard procedures in parallel to standards ◦ Internal standard review committee for quality and consistency – Tools to enhance use ◦ Dictionary, template forms, suggested edits, data management plan (DMP), etc. ◦ Website design/ layout - making it user friendly. The process for creating, implementing and managing standards has been met with many challenges but the benefits outweigh the frustrations. CDEs and standards have demonstrated enhanced data quality, decreases the time and resources needed to develop a study database, and helps customize the DMP. The Practitioners Engaged in Applied Research and Learning (PEARL) Network, a National Institute of Dental and Craniofacial Research- sponsored network of practice-based dentists have successfully created a data service on the cancer Biomedical Informatics Grid’s (caBIG’s) open source platform (caGrid). PEARL has registered the semantically annotated model and implemented the necessary registration interface steps and posted the de-identified clinical data on the caGrid. PEARL has taken the needed steps to create harmonized, semantic metadata in order to leverage the collaborative nature of caBIG with use of caBIG grid architecture and services. PEARL created the metadata based on the International Organization for Standardization Metadata 11179 Model in the Cancer Data Standards Repository (caDSR) which uses the semantic linkages of the hierarchical ontology contained in the National Cancer Institute’s Enterprise Vocabulary Service. New components created by PEARL are: 1321 Data Elements 999 Data Element Concepts 254 Value Domains PEARL used caBIG grid architecture to share Oral Health Inventory Profile and Tooth Sensitivity data from a study of postoperative hypersensitivity (Study PRL0602) following resin-based composite restoration of posterior teeth. The related Unified Modeling Language (UML) model for this data set has been registered in the caDSR where the related CDEs are classified as part of the UML model and visible in the UML Model Browser as a collection. PEARL encountered some challenges during implementation. Firstly, some issues and bugs had to be resolved with the system configuration and web services communication that required additional steps between PEARL and the caGrid team. Additionally, the caGrid data extraction process is not as intuitive as anticipated, and requires detailed instructions to overcome this barrier to data sharing. The projected benefit is that exposing clinically relevant data and the needed descriptive metadata that underpins those data provides informed, meaningful access to researchers, clinicians, and patients. The Clinical Trials Management Systems (CTMS) workspace within the caBIG® program was tasked to deliver integrated software solutions to the clinical research community. These tools are vetted by the community, who require the ability to securely gather, exchange, explore, integrate, and reuse data and information for clinical trials research. As a result the CTMS Workspace developed modular, interoperable and standards-based software tools designed to meet diverse clinical trials data management needs. The tools developed were configurable to work within trial sites with little or no clinical data management systems in place, as well as those sites with extensive existing systems. These tools took into account the diversity of clinical research activities and local practices that existed among trial sites. CTMS software tools enable management of clinical trials tasks such as: patient registration; patient scheduling; integration of laboratory results; adverse events capture/reporting; and the capture, analysis and sharing of clinical data among relevant systems. All tools are also available as a modular enterprise clinical trials management system designed to facilitate clinical workflows and data sharing in single and multi-site settings for use in trial sites. To continue the development/support of these tools with community and industry participation, the caBIG® program is evolving from its role as an active software developer to a facilitator of open source, community-based development. The NCI will provide a range of mechanisms to support open-source development efforts, including: • Processes to report software defects and suggest feature requests, • Forums for information exchange, • Resources for software version control and distribution, • Support of an open source governance framework with an emphasis on standards. We will discuss the implementation plan and the crucial factors for the continued development of these tools through the open-source model, and how this will serve the ultimate mission of furthering cancer research. When a clinical trial subject has a serious adverse event (SAE) or other significant adverse event (AE), such as those leading to the discontinuation of the study, a narrative is written for the clinical study report. These AE narratives summarize the details surrounding the event to enable understanding of the circumstances that may have led to the occurrence and its subsequent management. Such details may include the dose of study drug at the time of the event, the duration of the dose prior to the event, concomitant medications taken at the time of the event and used to treat the event, and other AEs that may have recently occurred. Other details include demography, medical history, laboratory results, the severity of the event and whether the event was related to study medication. Narratives are typically written from the original SAE report faxed from the clinical site in combination with data listings generated as part of the study deliverables. Information contained in the typical narrative requires manual review of these disparate data sources. This is time- consuming and often will require additional review and quality control. Too often, these narratives are written when the full data becomes available, which may become a rate-limiting factor in completing the study report. Since its inception in 1997, the Clinical Data Interchange Standards Consortium (CDISC) has developed standards for data models, study design and supporting clinical trial documents. CDISC standards have made such gains that the Center for Drug Evaluation and Research strongly encourages their use and implementation for the submission of new drug applications. The benefits of CDISC standards for statistics, programming and data management are well established. We will illustrate how these data standards can be used to facilitate composition of adverse event narratives. Background: Unique characteristics of cluster randomized trials (CRTs) complicate the interpretation of standard research ethics guidelines, including from whom, when, and how informed consent ought to be obtained. As part of a larger project to generate ethics guidelines for CRTs, we reviewed a random sample of 300 published CRTs and found that participant consent was reported in only 63% of trials. Whether this reflects an under- reporting of consent, or actual consent practices, is unclear. We therefore conducted a survey of the CRT investigators to (a) gather detailed information about consent practices in the selected CRT, and (b) investigate factors associated with obtaining participant consent in CRTs. Methods: A web-based survey was administered to corresponding authors of the sample of CRTs, in a series of six contacts. Trialists surveyed were based in USA/Canada (47%), UK/Ireland (18%), elsewhere in Europe (21%), Australia/New Zealand (5%), and various low/middle income countries (9%). Results: The survey response rate was 64%. Participant consent had been sought for some aspect of the study in 93% of trials: in 79% of trials consent had been sought from participants at the individual level (56% for the experimental intervention(s), 75% for data collection). Among CRTs with participants at the cluster level, 82% indicated that consent had been sought from cluster level participants (62% for the experimental intervention(s), 75% for data collection). Factors associated with individual level consent for the experimental intervention(s) were a smaller cluster size, and whether experimental and data collection interventions were targeted at the individual level. In addition, setting (healthcare vs. non-healthcare) was associated with seeking individual level consent for data collection. Conclusion: Unique characteristics of the CRT design are associated with participant consent practices, which are under-reported in CRT publications. Further ethical analysis is required to determine whether CRT consent practices meet the highest ethical standards. In some cases it is difficult for study participants to obtain access to a successful study intervention after the conclusion of a clinical trial. A number of international ethical guidelines call for researchers to provide access to products post trial for study participants when they are proven effective and unavailable through local mechanisms. The nature of this ethical obligation is debated, and the moral basis for such an obligation should help determine who bears responsibility for fulfilling it. There is no consensus among bioethicists about whether this is a firm obligation and if so, whether it is based on the idea of reciprocity or gratitude, as a safeguard against exploitation, as compensation for risk or burden, or some other basis. Also, ethical analysis is complicated by the fact that other community members who need the same medical care may have a competing claim for access to the product. Moreover, practical challenges in ensuring post trial access are numerous, and often multiple stakeholders need to be engaged. An interesting consequence of the call for post trial access is that many pharmaceutical companies decline to participate in research studies in countries where they have no plans to register their products. When studies are conducted in developing countries, challenges may arise due to slow or non-existent regulatory structures, lack of funds for product procurement and delivery, and lack of infrastructure. For research sponsors such as NIH, lack of authorization to use funds for non- research purposes may hamper efforts to support post trial access. These ethical and practical challenges will be discussed in detail, with recommendations for future policies to address them. Background: Cluster randomized trials (CRTs) have unique characteristics that complicate the interpretation of standard research ethics guidelines. We conducted an international web-based survey of research ethics chairs in advance of a consensus meeting to generate ethics guidelines for CRTs. Methods: We included all biomedical ethics committees in Canada and the UK, and a random sample from the USA (one per institution). After the initial e-mail invitation, we sent three e-mail reminders and a final postal reminder. The 45-minute questionnaire presented three hypothetical CRTs evaluating: a community-level health promotion intervention, an educational intervention targeted at health professionals, and distribution of insecticide-treated bed nets to villages in a low income country. Using closed- and open-ended items, participants were asked to indicate the type of review required (full board or expedited), whom they would consider research subjects, and objections that might be raised “Use of Lorazepam for the Treatment of Pediatric Status Epilepticus: A Randomized, Double-Blinded Trial of Lorazepam and Diazepam” (Status2), is a protocol conducted under the Best Pharmaceutical for Children Act (BPCA) to evaluate efficacy and safety of lorazepam compared to diazepam in pediatric patients presenting to the emergency department (ED) in status epilepticus (SE). Because of the potentially life threatening presentation, the narrow therapeutic window, and the lack of proven, effective treatments for SE in the pediatric population, this study was approved under the auspices of 21 CFR 50.24, the Exception from Informed Consent (EFIC) for Emergency Research. Navigating the regulatory requirements required significant sponsor, study management and site resources. The design of the protocol included informed consent, data collection and follow up procedures to assure protection of human subjects under EFIC. All sites (14) conducted community consultation and public disclosure activities that were reviewed and approved by the sponsor and local IRBs. The sponsor created training materials on the EFIC regulations to educate study personnel and local IRB members and to assist with community consultation planning. Opt-out procedures were implemented at each site and nationally prior to study initiation. Due to the emergency nature of SE, protocol deviations were common. These included incorrect weight estimation and therefore inaccurate dosing, and failure to adhere to timely study procedures. Study monitoring included assuring that legally authorized representatives of study subjects were approached in a timely manner and that opt-out and study procedures were followed in accordance with the study protocol. In conclusion, significant resources are required to meet the requirements of the Exception from Informed Consent regulations. Emergency conditions may preclude a carefully controlled study environment, so good training and monitoring are required to ensure that all protocol deviations are accurately recorded and reported to local IRBs. Drug Induced Liver Injury (DILI) is the most common safety-related reason for withdrawing a drug from the market. Therefore, it is critical to assess potential hepatotoxicity from treatment using an analysis of liver-related laboratory tests. A starting point is to plot peak values of lab measurements with regard to the upper limit of normal reference ranges to identify patients with elevated levels of bilirubin and alanine aminotransferase as potential Hy’s Law cases. However, this plot alone does not acknowledge the temporal considerations of Hy’s Law and liver injury. Following the FDA Guidance of DILI evaluation, this view can be augmented to flag patients based on the occurrence of elevated liver tests within a clinically- relevant time period. Moreover, by incorporating the temporal dependence of liver test elevations, the frequency and duration of Hy’s Law instances across treatment groups can be interrogated and reported. This presentation will highlight a statistically-driven interactive visualization approach implemented in JMP Clinical software to streamline the detection of potential DILI cases through summary views of lab test elevations and Hy’s Law occurrence that lead to targeted patient time trend plots and cross-domain profiles for further clinical insight. It can be challenging to perform a concise and easily-interpretable analysis of adverse events (AEs). For therapeutic areas where patients have frequent AEs, understanding the safety profile of a new intervention is critical. In these instances, however, understanding the analysis is made more difficult by the sheer number and variety of AEs that occur. Traditionally, the incidence of adverse events is presented in lengthy tables, with coded AE terms grouped by body system or system organ class and presented in order of descending frequency. Many times, statistical testing is not be performed due to the high likelihood of committing multiple Type I errors. Further tables may highlight AEs that occur for specific periods of time (for example, treatment phase or off-treatment follow-up). Needless to say, it is difficult to summarize this information to gain a clear understanding of the risk (or benefit) a new therapy may present. Incidence analyses of adverse events are straightforward to present graphically using Volcano Plots to highlight treatment differences. Color indicates the treatment with higher incidence, with bubble size representing the total number of events that occur during the trial. Adjustments for multiple comparisons can be presented in a manner to clearly indicate which events exhibit statistically significant treatment differences. Further, the study can be broken into distinct time windows, with multiple plots or animation displaying changes in adverse event risk over time. This presentation can emphasize early differences across treatments that may eventually resolve or highlight events that could approach significance given a longer study duration. The use of Volcano Plots is illustrated with an analysis of data on aneurysmal subarachnoid hemorrhage. We describe the challenge of collecting adverse events (AE) data in an unmasked randomized treatment trial. The Ocular Hypertension Treatment Study randomized 1,636 participants to either observation or to treatment with eye drops to lower intraocular pressure (IOP) to prevent the development of glaucoma. Masking would have been impossible to maintain. Participants could become unmasked at free IOP screenings or non-study eye examinations. Clinic personnel would become unmasked by measuring IOP to adjust medication to achieve IOP treatment targets. Efficacy outcomes were masked, but AE data were collected by unmasked clinic personnel and participant self-report. At 24 months, 774 AE’s had been received, 31% in the observation group and 69% in the medication group. The clinic to clinic rate in AE reporting ranged from 7% to 60% signaling a standardization problem. To improve standardization, we reduced branching logic, consolidated forms, included more check boxes and used a scripted interview. The Data Safety Monitoring Committee recommended that AE data received prior to these protocol changes be excluded from future publications. In the 18 months after these protocol changes, 1,688 AE reports were received, more than double the previous rate, 44% of the AE’s were in the observation group and 56% in the medication group. It is striking that the differential between groups decreased from 38% to 12%. This suggests that potential surveillance or detection bias in an unmasked treatment trial can be reduced by a rigorous protocol. Stem cell clinical trials for cardiovascular disease in the last decade have been well tolerated with preliminary evidence suggesting efficacy. Standardization of endpoints is important to facilitate comparisons across studies. We describe our methods of safety and efficacy assessment based on our experience coordinating multiple studies using various cell populations and administration methods. Adverse events (AEs) are collected after enrollment, evaluated by independent medical monitors throughout the trial and coded using the most current version of the Medical Dictionary for Regulatory Activities (MedDRA) and reviewed every 6-months by an independent Data and Safety Monitoring Board. Major Adverse Cardiac Events (MACE) are defined as death, hospitalization for heart failure or non-fatal recurrent myocardial infarction. Patients are evaluated for complications during study product administration and within 24-48 hours. Cardiac enzymes (CK-MB and Troponin) are collected every 12 hours for the first 48 hours. Echocardiograms are performed following the procedure and within 24-48 hours to assess for pericardial effusions. Incidence of arrhythmias is assessed by ambulatory electrocardiograms. Hospitalizations are adjudicated independently and classified as cardiac related or not. Ectopic tissue formation is assessed using CT scan. Functional benefit is evaluated using New York Heart Association Class, Six Minute Walk Test, Peak VO2, and forced expiratory volume in 1 second. Quality of life is measured by the Minnesota Living with Heart Failure Questionnaire (MLHFQ). Cardiac magnetic resonance imaging is used for assessing global and regional function. Patients who are device dependent undergo cardiac CT at baseline and 12 months. Clinical trials of cardiovascular cell therapy are expanding due to the excellent safety profile and early signs of efficacy. Safety is paramount and constantly monitored. Centralized cores such as imaging labs are important to maintain consistency and independence. As therapies evolve, consistency in the endpoint assessment should be considered to ensure comparability across studies. Randomised Controlled Trials (RCTs) require efficient subject recruitment and retention. Often the management of such a trial poses significant challenges. EAGLE (Effectiveness, in Angle closure Glaucoma, of Lens Extraction) is an international multi-centre, pragmatic, publicly funded randomised controlled trial (RCT). EAGLE addresses whether removal of the lens of the eye for newly diagnosed Primary Angle Closure Glaucoma results in better clinical, economic and patient focussed outcomes compared with standard management. The EAGLE management team are UK based; however identification, recruitment and three year participant follow up is executed by staff working in sites across the UK, Malaysia, Hong Kong, Singapore, (Peoples Republic Of) China and Australia. The EAGLE study design requires minimal direct contact between the UK management team and the participant. The maintenance of effective working relationships between the UK management team and the international recruiting teams is, therefore, crucial to project success. To conduct EAGLE across different health care, legal, and cultural settings, a robust but pragmatic protocol was designed. A process for individual site assessment in terms of suitability as a recruiting site was developed involving honest discussion of trial budget and realistic timelines at outset. Maintaining momentum and focus on the research question among collaborators was addressed through site visits, regular scheduled video conferences, newsletter bulletins and an ongoing training regime of bespoke design that altered as appropriate when the trial matured and entered new phases. Global teleconferences were negotiated across time zones. Escalating costs of printing and translating trial paperwork into three different languages required re-profiling of a fixed budget without compromising trial delivery. This paper describes the challenges encountered in EAGLE across the first half of its data collection journey and the strategies employed to overcome them. The EAGLE study experience is widely applicable to any centrally managed international multi-centre RCT. Background: Multicenter clinical trials are designed to generate data that may be generalized across diverse populations with the goal of facilitating unbiased answers to important medical questions. The Food and Drug Administration, the Office of Human Research Protections, and the Department of Health and Human Services (DHHS) support the use of a centralized IRB review process, that is, a single IRB of record for a given protocol, to improve multicenter trial efficiency. Most recently, the DHHS proposed to change the Common Rule to include mandated centralized review for multicenter trials; however, there are concerns that the benefits of centralized review have not been sufficiently demonstrated. Methods: We identified published articles relating to centralized review through PubMed and hand searching. We categorized articles by topic and type (commentary or empirical research). Results: Our review includes 78 sources. Of these, 22 articles describe inefficiencies and inconsistencies with multiple review (12 empirical), 4 articles compare multiple to central review (1 empirical), and 5 articles describe barriers to the adoption of central review (1 empirical). The single empirical study directly comparing multiple and central review showed that affiliation with a central IRB was associated with faster reviews and financial savings. Several limitations of the empirical work were noted: 1) Absent a standardized metric for measuring review quality, all studies focus on efficiency, without informing the debate over the ethical quality of central versus local review, 2) Methodological quality is weak, as the majority of articles are case studies using a sample size of less than 20, and 3) The existing empirical work is not generalizable, as it is unclear how descriptions of prior practices would carry over to a still-undefined system of mandated centralized review. Conclusions: The limitations of the existing empirical work constrain meaningful conclusions about centralized review. The storage and tracking of regulatory documents in a large, long-term clinical trial can quickly become unmanageable without a proper tracking system. Gone are the days of manual files, as clinical trials move to the use of electronic regulatory tracking databases. The Age-Related Eye Disease Study 2 (AREDS2) is a multi-centered clinical trial designed to assess the effects of oral supplementation of high doses of macular xanthophylls and/or omega -3 fatty acids for the treatment of age-related macular degeneration and cataract in just over 4,000 participants. The Administrator and Protocol Monitors at the AREDS2 Coordinating Center have the immense task of tracking regulatory documents for the 82 clinical sites and the 1200 past and present clinical site personnel. A customized electronic regulatory tracking system, Site Management Utility (SMU), was implemented to manage all regulatory documents including those required for study activation at participating sites. SMU enables staff to efficiently track such documents as IRB approvals, medical licenses, Human Subject Protection Training notifications, and protocol specific certifications. The utility also allows the Coordinating Center to maintain a full project contact list and archive of present and former staff. During the study start-up phase, SAS reports linked to regulatory information in SMU were generated to monitor completion of all requirements for study activation. Throughout study follow- up, a SAS program was run to generate automatic email reminders to the clinical sites at 60, 30, and 7 days prior to IRB and/or medical license expirations. SMU and other effective tools utilized at the Coordinating Center will be highlighted during this discussion of efficient methods of regulatory tracking and storage. EHR4CR (Electronic Health Records for Clinical Research) is a research project funded by the European Union Commission and the European Federation of Pharmaceutical Industries and Associations (EFPIA) within the Innovative Medicines Initiative framework. EHR4CR is a collaborative project involving primarily universities, the pharmaceutical industry with input from organisations and groups representing patients and covering aspects of the law, ethics and technologies associated with the use of electronic data for research purposes. The EHR4CR project aims to improve the efficiency of, and reduce the cost of, conducting clinical trials, by better leveraging routinely collected clinical data through development of both an EHR4CR platform (primarily tools and services) and implementing an innovative and sustainable EHR4CR business model. Introducing innovative technology involves cost, new ways of working and organisational change. The multinational scope of the project means that every effort must be made to ensure that the approaches, methodologies and technologies conform to local and international legal and ethical requirements related to data protection, privacy and confidentiality. The project must also conform to the standards required by national and international drug regulators (MHRA, EMA, FDA, etc.) and satisfy the needs of the pharmaceutical industry. The approaches taken also need to be acceptable to the general public, patients, medical professionals (both study investigators and otherwise), and also to politicians and professional bodies sensitive to the concerns of their constituencies. Proactive management of the complex network of European stakeholders external to the EHR4CR project team is an essential task which will continue throughout the full four year lifecycle of the EHR4CR project. This presentation describes a pilot exercise that has been carried out in Scotland in order to develop a best practice approach for local national stakeholder identification, analysis engagement and management during the development phase of the EHR4CR project. A summary of the key outcomes of the Scottish pilot activity and the planned next steps for roll out of the stakeholder management process across Europe will be presented. Randomized Phase 2 oncology trials are mainly aimed at demonstration of improved efficacy (e.g. improved Progression Free Survival, Overall Survival) and acceptable safety profile of the target therapy versus the control therapy. With the development of the new bio-assay technology, biomarkers may provide an important role in demonstrating mechanism of action and pathway intervention of drugs and provide preliminary evidence of treatment effect of the target therapies. In addition, biomarkers may also be used to select patients who are mostly likely to benefit from the treatments. Based on preliminary biological hypothesis of the biomarkers, data generated from Phase 2 studies can be used to identify biomarkers that have predictive potentials on treatment effect. The potential predictive effect of biomarkers can then be further tested in large Phase 3 trials with adequate planning to provide more definitive evidence of the utility of the biomarker for patient selection. In this presentation, we will discuss the exploratory statistical analysis of biomarker data for randomized Phase 2 oncology trials and the impact of the analysis on Phase 3 study designs. Both continuous and categorical biomarker data will be discussed. Applications of statistical methodology including survival analysis, subgroup analysis, logistic regression analysis and time dependent ROC curves will be presented. Potential confounding factors of the biomarkers will be discussed. Experience on practical considerations in exploratory biomarker data analysis such as data collection and transfer, potential un-blinding, ascertainment rate will be shared. The sponsor owns the responsibility for a clinical study to comply with the protocol, including enrollment of patients who meet eligibility requirements. It is the responsibility of the investigator to give the best care available to patients. Struggles sometimes result when investigators, who have enthusiasm and optimism that a particular investigational drug will benefit their patients, have disappointment when trial eligibility criteria exclude patients that they want to enroll in a study. Patients who are enrolled into studies when they are on the edge of eligibility can chip away at the statistical power of a study to demonstrate treatment benefit. From the statistical perspective, having a homogeneous population with a smaller variance can have more statistical power for analysis than having a larger sample size when the population is more heterogeneous and the variability is greater. The enrichment design (referenced in FDA’s Adaptive Design Clinical Trials for Drugs and Biologics, Draft Guidance; CBER, CDER, 10 February 2010) when supported by a conditional sequence of hypothesis tests using methodologies proposed by Bauer (cited in the FDA’s draft guidance on adaptive designs) offer an opportunity for a trial to enroll heterogeneous populations while protecting the ability for the study to be positive. Strategic utilization of the enrichment design can improve the likelihood of success of a study by using Bauer closed procedures. The proposed methodology has two major advantages: (1) using the conditional sequence of hypothesis tests to control multiplicity, alpha can be 0.05 for each hypothesis test while preserving the overall study-wise alpha at 0.05; and (2) it allows patient recruitment to permit “all comers” in the study protocol while protecting the primary analysis to be done in a subpopulation enriched with patients in whom the benefit may be easier to measure, hence, demonstrate Biomarkers are laboratory measures of biological processes. Because they are viewed to be objective and quantifiable by providing valuable information in both assessing exposure and disease status, they are increasingly used in biomedical and public health sciences, drug development programs, testing for targeted therapeutics and personalized medicine. This increased use creates a venue for the development of methods to address important, unexplored analytic issues such as properly handling biomarker measurements below the limit of detection (LOD). Researchers working with biomarker data inevitably have to deal with data containing non-detects, and how to combine non-detects with values above the LOD for data analysis. The common statistical analysis approach has been either to exclude measurements with LOD or perform ad-hoc single imputation techniques. Ad-hoc imputation techniques include replacing each measurement below the LOD with a value such as zero, a half of the detection limit, or LOD value itself, and then conducting the analysis under the assumption that the imputed values are the actual observed values. However, there are other parametric and nonparametric statistical methods, such as model based single imputation methods, Reverse Kaplan-Meier method, and multiple imputation methods that are not widely used in the literature but are better alternatives to ad-hoc single imputation techniques. We will facilitate broader use of all of these methods by describing their properties, illustrating their use with population-based data for secondhand smoke exposure research and showing how they can be calculated using standard software. This work was funded in by FAMRI, NIH, NIOSH. Biomarkers are a critical component of targeted therapies as they can be used to identify patients who are more likely to benefit from a particular treatment. Several prospective clinical trial designs for biomarker directed therapy have been previously proposed including the marker-stratified, sequential testing strategy, adaptive, and enrichment designs. These designs differ primarily in study populations (marker-defined or all-comers) and randomization scheme (fixed or adaptive). Recognizing the need for randomization yet acknowledging the possibility of remarkably promising results at interim, we propose a two-stage Phase II marker-positive design that allows for direct assignment in Stage II. In particular, Stage I of our proposed design randomizes marker- positive patients equally to receive targeted therapy or control, while Stage II has the option to adopt “direct assignment” whereby all Stage II patients receive the targeted therapy. Through simulation, we study the effect of varying the alpha decision cutoffs at interim and timing of interim analysis on power and Type I error rate. We compare our design with a balanced randomized two-stage design. Our results suggest relatively minimal loss in power (<2%) and increase in Type I error rate (<5%). A sensitivity analysis to examine the possible effects of a population shift also suggests relatively minimal inflation of power, compared with no population shift. Our design has a greater appeal to clinicians and patients with its direct assignment option, while maintaining relatively desirable statistical properties. The direct assignment, if adopted in stage II, provides for an “extended confirmation phase” as an alternative to stopping the trial early for efficacy which may help to avoid possibly prematurely launching into a Phase III trial, thereby potentially addressing the high failure rates of Phase III trials. Introduction: Major protocol violations occur more frequently during the early stages of a clinical trial, when investigators are less familiar with study processes. Recruitment errors arise when study eligibility criteria are violated and can account for 50% of all major protocol violations. Study Objectives: To determine whether a formal study run-in phase can improve familiarity with study eligibility criteria and reduce recruitment errors. Methods: Prior to starting a multi-centre clinical trial, participating centres were required to submit de-identified potentially eligible patients to a study web site. The run-in web site did not allocate patients to treatment or control groups however, the information captured allowed true eligibility to be assessed. Appropriateness of enrolment was fed back to the participating centre. Each centre was required to demonstrate proficiency at identifying consecutive truly eligible patients before being allowed to start the trial. Results: Thirty-two centres submitted 199 potentially eligible patients to the run-in site. 32 of 199 (16%) patients did not meet eligibility criteria (recruitment errors). After successful completion of the run-in phase, at the time of this analysis participating centres had enrolled 409 patients into the trial, with four (1%) recruitment errors (16% vs 1%, p<0.001). This is significantly lower than published benchmarks obtained from FDA Phase III licensing trials (PROWESS 159/1690, p<0.001 and INTERSEPT 77/531, p<0.001). Conclusions: A formal run-in phase that provides performance feedback can improve familiarity with trial eligibility criteria and significantly reduces recruitment errors. If a multi-centre trial does not offer a formal run-in phase, we strongly urge local investigators to conduct their own formal run-in phase. It will be explain the results of a large communication project performed by the Italian Medicines Agency (AIFA), together with the Italian Society of General Practitioners and the Italian Federation of Family Doctors, on topics related to clinical trial of drugs. To promote knowledge on clinical trials’ mechanisms, to explain how subjects involved are protected by specific rules, to define potential risks and benefits deriving from the use of unregistered drugs were among the goals of this campaign. A special informative poster had been distributed to a huge number of family doctors; citizens had been asked to fill in a questionnaire, regardless they had read the poster or not, and their answers analyzed. The number of filled-in questionnaires was high (around 25,000) and results show a fair distribution among population. Information campaign showed to be effective: people who read the poster gave answers up to 15 times more correct, especially in the case of technical questions. Instead, to questions regarding the possible involvement in clinical trials, all answered more or less in the same way. We have also performed the same process in different therapeutic area within public hospital and the results will be presented. The encouraging results of this first information campaign show the direction for future years programs: on one hand, conceiving an always more detailed mechanism to inform citizens and, on the other hand, expanding such program to health structures in which clinical trials are carried out, focusing on special populations (i.e. paediatric patients, elderly, pregnant women). Abstract Objective: Achieving required enrollment and diversity in a multi-center international randomized clinical trial is an arduous mission. We will outline the strategies leading to BARI 2D’s success in reaching the final participant target number and doubling the minority participant enrollment recommended in the National Institutes of Health (NIH) guidelines. Methods: Recruitment data were collected from pre-trial, Vanguard and recruitment phase assessments, site visit reports, screening logs and cost analyses before and during the four-year enrollment period to evaluate enrollment rates. Data were analyzed periodically throughout the enrollment phase, identifying barriers and strategies to reach recruitment targets. Results: Pre-trial recruitment assessments of the pool of eligible participants from selected sites did not provide an accurate portrayal of the total number of sites required to reach recruitment goals. Recruitment during Vanguard was half what was expected due to delayed site regulatory approvals and ineffective recruitment procedures. The recruitment period was extended, sites were added including several carefully selected foreign institutions, and a composite end point utilized to adjust total recruitment necessary to power conclusions. Eighty-two percent of sites faced recruitment source barriers, notably reluctance from physicians to refer and difficulty identifying patients before clinical treatment decisions were made. Sixty-nine percent reported inadequate staffing and staff turnover as detriments to recruitment. Extra funding, although useful, did not necessarily equate to higher recruitment numbers. Recruitment and Minority Recruitment working groups were effective in fostering strategies to overcome barriers. Best practices were identified through site visits and a centralized recruitment coordinator disseminated this information to struggling sites. Conclusion: Time and resources must be invested early to complete a realistic pre-assessment of potential sites’ capabilities and initiate an adequate Vanguard phase to pre-test recruitment strategies ensuring success. Proficient study teams with dedicated internal and external support are crucial to meeting recruitment goals in large multi-center trials. Patient information leaflets (PILs) are a core component of the informed consent process in clinical trials. Most evaluation of PILs, if conducted at all, is based on indices of readability, yet several writers argue that readability is not enough. We propose a framework for evaluating PILs that reflects the central role of the patient perspective in communication, based on simple principles from linguistic theory. The framework has three elements: a) Readability (attribute of text) which may be assessed using well- established procedures that assess superficial features of the text; b) Comprehensibility (attribute of reader-and- text) which may be assessed using multiple choice questions based on the lexical and semantic features of the text; and c) Communicative effectiveness (attribute of reader relative to writer’s intent) which focuses on discrepancies between responses of the reader and intentions of the writer. We will discuss the implementation of this framework using the case study of the CLASS trial: Comparison of LAser, Surgery and foam Sclerotherapy, a UK multi-centre randomised controlled trial comparing three treatments for varicose veins. Using the proposed framework, the overall reading grade of the CLASS PIL was observed to be borderline for reading ease acceptability. Readability differed across the descriptions of the three trial interventions with information about laser therapy being most readable while surgery text was least readable. Similarly, communicative effectiveness differed across the descriptions. Foam sclerotherapy was the most communicative, with higher levels of agreement between participants’ and the clinician/author’s views. The observations from this study could be used to revise the PIL to promote equal understanding across the descriptions of the interventions, and thus minimise a potentially hidden bias in trial paperwork. This approach has broad applicability to all clinical trials and has the potential to move the field beyond readability to broader indicators of quality, facilitating enhanced informed consent. Background: There is no shared language in the research community for describing the ‘active ingredients’ of behaviour change interventions. Hence, the body of evidence from randomized studies to assess the effectiveness of interventions for changing health-related behaviour is limited. Without a precise nomenclature, it is impossible to replicate effective interventions, discard ineffective interventions, validly synthesize evidence about behaviour change interventions, or propose causal mechanisms underlying behaviour change. The project reported here aims to develop a reliable method for specifying behaviour change techniques (BCTs), how they work and how to identify when they have been delivered. Methods: In several stages, we generated lists of BCT labels based on a) systematic reviews of behaviour change interventions; b) systematic text-book search; c) expert brainstorming. We also identified definitions from textbooks and dictionaries. The resulting draft list of BCTs and definitions was refined using consensus methods to identify redundancy and improve clarity. The nomenclature was used to code published descriptions of complex behaviour change interventions and inter-coder reliability was assessed. Results: To date, we have agreed specification for 87 BCTs relevant to changing health behaviours. These can now be adopted to accurately specify future trials of these interventions, facilitating appropriate reporting as required by CONSORT. Conclusion: International consensus is required to facilitate the adoption of this nomenclature into standard reporting practice. An international advisory board has been convened. Further research will test the value of the developed system for increasing the reliability of coding of published interventions for evidence synthesis and increasing the replicability of effective interventions. Background: Key components of healthcare interventions such as ‘active ingredients’ should be reported in titles and abstracts of published reports of RCTs. Evidence suggests that reporting of non- pharmacologic interventions (NPIs) is inadequate compared to pharmacologic interventions (PIs). There are particular challenges for specifying the active ingredients of behaviour change interventions. Aim: This review compared the reporting of components of PIs and NPIs in the titles and abstracts of published reports of RCTs. It was hypothesised that active ingredients would be reported more often in PIs than NPIs and less often in behaviour change interventions than in other NPIs. Methods: MEDLINE and EMBASE were searched from 2009 to March 2011 for randomised studies published in the BMJ, JAMA, NEJM, Lancet, and Annals of Behavioral Medicine. All types of interventions, participants, and outcomes were included. Data were extracted from titles and abstracts of 198 randomly selected reports. The papers were coded for the presence or absence of key intervention components. The resulting frequency data were subjected to chi-square analysis. Results: The search strategy returned 1,454 papers. Of 198 papers randomly selected, 98 reported PIs, 96 reported NPIs, and four reported both PIs and NPIs. There were significant associations between intervention type and reporting of active ingredients: 88.2/94.6% of PI papers and 39.6/67.7% of NPI papers reported active ingredients in the title/abstract. Furthermore, 29.3/61.0% of behaviour change papers and 52.0/78.0% of other NPI papers reported active ingredients in the title/abstract. Reporting practice differed for other components such as trial setting, intervention provider and comparator interventions. Conclusions: This review provides further evidence of the need for improved reporting of NPIs. Researchers need a shared language for describing active ingredients of behaviour change interventions. This would ensure that interventions could be faithfully replicated and evidence validly synthesized. Background: Excessive protocol violations (PVs), defined as preventable mistakes in study conduct, may result in patient harm and may dilute statistical power. Purpose: To gain a better understanding of reported PV rates, to describe interventions used to reduce PVs and to investigate relationships between trial characteristics and PVs. Methods: We reviewed consecutive trials published in four major journals identified using a PubMed search. Two authors independently abstracted information on trial characteristics, PV reporting, PV rates and interventions used to reduce PVs. PVs were categorized into one of five distinct types: enrolment, randomization, study intervention, patient compliance and data collection. Associations between PVs and trial characteristics were investigated. Results: Eighty clinical trials (20 from each journal) were identified from 101 consecutive abstracts. Median number of participants was 701 (range: 20-162,367) and median number of participating sites was 15 (range: 1-701). Nineteen percent (15/80) of included trials were single centred. Median study duration was 24 months (range: 5.81-127 months) and 74% (59/80) of trials were primarily academic funded. Thirty two percent (26/80) of trials failed to provide explicit reporting of any type of PV and none (0/80) of provided explicit reporting of all five types of PVs. Larger trials (more patients, more sites, longer duration, more complex management structure) were more likely to have more complete reporting of PV’s. Only 9% (7/80) of trials reported the use of a specific study method to prevent PVs. Use of a run-in phase was the only method reported. Conclusions: PVs are under-reported. Although the CONSORT statement provides guidance on the reporting of PVs, reporting requirements are not explicit for all types of PVs. As a first step towards improved reporting by authors, we recommend the CONSORT statement highlight the importance of PVs by making reporting requirements more explicit. There is growing recognition that insufficient attention is paid to the outcomes measured and reported in clinical trials. Selection of outcomes is crucial to trials designed to compare the effects of different interventions. For the findings to influence policy and practice, the chosen outcomes need to be relevant to patients and the public, healthcare professionals and others making decisions about health care. Trials in a specific condition often report different outcomes, or address the same outcome in different ways. Inconsistency in reported outcomes causes well known problems for those who attempt to synthesise evidence, and many meta-analyses have to exclude key studies because relevant outcomes are not reported. Furthermore, the measured outcomes may not always be important to patients or health service users. Much could be gained if an agreed core outcome set (COS) of a minimum number of appropriate and important outcomes was measured and reported in all clinical trials in a specific condition. Key stakeholders, including patients, should be involved in establishing COS, to ensure consideration of appropriate outcomes. COS may encompass all stages or severities of a condition or may focus on a particular disease category. Likewise, a COS may be for use in trials of all treatment types or only trials of a particular intervention. The scope of a COS should be defined to identify the relevant health condition, population and types of interventions. The COMET Initiative (http://www.comet-initiative.org/) aims to foster and facilitate methodological research in the area of standardising outcomes, to develop much needed standards for methods of COS development and to develop and maintain a publically available internet-based resource to collate the knowledge base for COS development. Work on COS has been identified in over 80 clinical areas. The database will be demonstrated, progress to date presented, and the impact of COS discussed. It is not uncommon in allergy field that a drug can be approved if two out of three pivotal clinical trials are successful. To limit the cost, many sponsors start with two pivotal trials. In situations where one of the two trials is successful and the other one only demonstrates numerical improvement, the decision has to be made as whether a third study should be conducted for the pursuit of approval. In this presentation, we apply the Bayesian approach to assist the decision making. Specifically, a Bayesian hierarchical model was utilized to evaluate the probability of success of the third trial. The posterior probability and predictive probability are derived for assessing the chances for a successful third study. A real example is used to illustrate such approach. Background: Central to the validity of a randomised controlled trial (RCT) is a calculation of the number of participants needed (the sample size). This provides reassurance that the trial will identify a difference of a particular magnitude if such a difference exists. Given its importance, determination of this target difference, as opposed to statistical approaches to calculating the sample size, has been greatly neglected. A variety of approaches have been proposed for formally determining what an important difference is (such as the “minimum clinically important difference” approach). However, in practice the target difference may be driven by convenience or some other informal basis. The awareness and use of formal methods in the trial community is unclear. Aim: To assess awareness and use of formal methods for determining the target difference and thereby the RCT sample size. Methods: Members of the Society for Clinical Trials were sent an email invitation to complete an online survey through the society’s email distribution list. The survey collected information about the individual responding (e.g. position, affiliation and location). Seven formal methods were identified that could potentially be used. Respondents were asked about their awareness and use of, and willingness to recommend, methods. Results: 180 responses were received representing 13 countries and a variety of professions and institutions. Awareness of methods varied from 69 (38%) for health economic methods to 162 (90%) for using pilot data. A majority (96, 53%) had used no more than three of the available methods. Recommendation of methods tended to be lower than use except for health economic and reviewing the evidence base methods. Conclusions: Awareness, use and willingness to recommend varied greatly between methods. Trial specific guidance documentation may increase both awareness and use of formal methods. There is clearly increasing national interest in comparative effectiveness research (CER). Randomized controlled trials must have a prominent place in CER due to their reliable information and well respected standard; however, improved approaches and CER-focused rethinking are needed to ensure their feasibility and overcome tendencies to be slow, expensive, and homogeneous. As in other clinical research, randomized CE trials there may have limited information to guide initial design choices including the patient population, the primary outcome, or the target effect size. In the general RCT setting, adaptive designs have been proposed to address these concerns. There are potential advantages to expanding adaptive designs to within the CE context. However, although there are many similarities between the two, CE trials have some fundamental differences from standard clinical trials. For one, the heterogeneity in the population studied in CE creates higher variability in outcomes. CE studies could be underpowered if they use planning values obtained from tightly controlled clinical trials. Additionally, the concept of a ‘minimum clinically meaningful difference’ is hard to define in the CE context. Even assuming equal cost and safety, a range of meaningful effect sizes could be defined with upper limit the largest effect with reasonable chance of being observed and lower limit the minimal effect deemed sizable enough to change practice in the study context. Following a brief introduction to clinical CER, this talk will focus on additional decisions and methodology needed before the promise of adaptive designs can be achieved in the CE setting. We will describe the evaluative process to determine the usefulness of these designs in CER, assess whether or not we are ready as a scientific community to move on from basic designs to more advanced methods, and discuss items that must be addressed in order to achieve this objective. The DIA adaptive design scientific working group have conducted a survey of the perception and use of adaptive designs (ADs) for clinical development programs in the industry and academia. In this presentation, the results of the survey will be presented and compared to the results of a previous survey carried out by the same group in 2008 under the auspices of PhRMA [1]. The key objectives of the survey were to identify any persistent barriers to implementing such designs and provide recommendations to overcome these challenges. The questionnaire was sent out to a wide selection of organizations in September 2011 to enquire about ADs that were planned, ongoing or completed in their organization from 1st January 2008 to 1st September 2011. In parallel, the medical and statistical literature and clinical trial registries were reviewed by the group to identify published AD case studies and consider to what extent the literature is representative of the information gathered in the questionnaire. The results of this review will also be presented. [1] Quinlan J. et al. Barriers and opportunities for implementation of adaptive designs in pharmaceutical product development. Clin Trials 2010; 7:167-173 Conducting follow-up studies in older adults presents particular challenges, especially if the research is done remotely. The SELenium and vitamin E prostate Cancer prevention Trial (SELECT) opened in 2001 and recruited 35,533 participants age 50 and older at over 400 study sites. In 2009, after study supplementation ended, 17,748 participants consented to continue follow-up by mail conducted centrally by the Coordinating Center. Older adults may retire, travel extensively or buy second homes. Frail adults may need more supportive environments such as assisted living centers. Many face co-morbid conditions that may affect their ability to read, understand or complete the study questionnaires. Although older adults may be reluctant to use newer data collection technology that is widely accepted in younger populations, some expect access to the study through cell phones, email and the web. Others may have limited options for contact, such as mail or land lines. Tracking and maintaining participant contact information requires extensive staff time. In early 2011, we administered a mailed survey to our participants. Despite strong interest in a web data submission option, to date only 550 participants (3% of total) have accessed our website, and only 410 forms have been submitted online. We are now pursuing various strategies to understand barriers to website use and to encourage participants to submit data online. Supported by NIH/NCI/DCP grant CA37429 and in part by the National Center for Complementary and Alternative Medicine (NCCAM). Determining the value of an intervention in a specific population, particularly in neonatal or pediatric care, often requires extended follow- up to monitor patients for effects that may only present themselves years later. The knowledge to be gained through extended follow-up in clinical research can be impeded by several practical difficulties associated with connecting with the original patient population. These challenges include locating past participants, and repatriating them back to clinic. This presentation will outline strategies in extended follow-up to mitigate these issues and maximize the potential for complete data collection. The first step involves contacting the local Research Ethics Board or Institutional Review Board to inquire about acceptable methods for locating patients who may have moved or lost touch with their follow-up team. Once study approval has been granted it can be helpful to get in touch with the coordinator or staff who managed the trial at earlier phases. These individuals may be able to share established practices and identify participants with special circumstances; patients may also prefer this continuity of contact with the research team. Additionally, patients several years out of the clinical setting may be reluctant to return; connecting these potential participants with familiar staff may be helpful when updating them on the aims of the current trial. Subsequent strategies for repatriating patients to the clinic for assessment can include sending a letter outlining the next phase of the trial; following up with a phone call to address questions or concerns; and mailing out a recruitment package to potential participants. This package can include consent or assent forms and a pre-addressed, pre- stamped envelope to make the process of returning signed documents easier for the participant. Finally, a follow-up call prior to the appointment to discuss logistical details and possibly offer transportation assistance can decrease the likelihood of attrition. Complex intervention trials within vulnerable populations are challenging to undertake. They are often difficult to design, implement and evaluate, and suffer from poor recruitment. Feasibility studies are key to testing the viability of trial designs, recruitment potential and other implementation aspects of larger scale trials. We illustrate practical and logistical challenges encountered in the design and set-up of two feasibility trials: OBI (Optimised Behavioural Intervention – in chronic low back pain patients) and WATCH-IT (a community-based obesity treatment intervention for children and adolescents). OBI is currently in recruitment, whereas the evaluation of the feasibility of WATCH-IT is now complete . Both studies were set up to determine: recruitment rates, acceptability of randomisation, data collection, drop-out rate, optimal outcome measures, and aid robust sample size calculation. In addition, OBI also aims to optimise the intervention and to measure its acceptability. Both trials recruited to target and the methods of randomisation and data collection were acceptable to participants. We found that for WATCH-IT self referral (via media promotion) resulted in a higher consent rate than professional referral, but that this did not necessarily translate to greater participant retention. Also, calculation for ultimate sample size was significantly greater than that reported in published trials. Importantly, WATCH-IT highlighted how the trial impacted on service delivery which was an important lesson for the design of the phase III trial. For OBI we found that two recruitment pathways were needed dependent on whether the physiotherapy service was involved in assessment and diagnosis as well as treatment. In addition we found that training of therapists took longer than anticipated and there was a paucity of therapists in general who could provide clinical cover for the OBI therapists. Feasibility studies of complex interventions are often overlooked. However, they are essential to assess the viability of larger definitive trials. There are two types of therapeutic trials in the search of agents that can treat people with Alzheimer’s disease (AD): symptomatic and disease-modifying trials. The former includes these for symptomatic agents with a primary objective of improving cognition, function, and global measures or deferring decline over a short period of time. The latter consists of those for disease-modifying agents which strive to show that the course of AD has been altered and the rate of disease progression has been slowed. Randomized start and randomized withdrawal designs are two popular designs of disease-modifying trials on AD. Crucial design parameters such as sample size allocations and treatment switch time are important to understand in designing such clinical trials. A general linear mixed effects model is proposed to formulate the appropriate hypothesis for the test of disease- modifying efficacy of novel therapeutic agents on AD. This model employs a piecewise linear growth pattern for those in the delayed treatment or early withdrawal arm, and incorporates a potential correlation on the rates of change on efficacy outcome before and after the treatment switch. Optimum sample size allocations and treatment switch time of such trials are determined according to the model. An intersection-union test through an optimally chosen test statistic is used for the test of treatment efficacy. Finally, summary statistics from several reported symptomatic trials on AD in the literature are used to apply the proposed methodology for designing future optimum disease-modifying trials on AD. Background. Often, despite evidence from large and/or high quality randomized clinical trials (RCT) is not available, there are numerous large, high quality cohort studies presenting adjusted risk ratio (RR). For instance, we considered studies evaluating efficacy of intravascular ultrasound (IVUS) guidance in drug eluting stent positioning in percutaneous coronary intervention and compared evidence derived from RCT and observational studies. Methods. We performed a literature search on Pubmed, Embase, Cinhal, Web of Science and the Cochrane Library. Inclusion/exclusion criteria were: full-text articles in peer-reviewed journals (2003-2011), we excluded uncontrolled studies. The search included both RCT and observational cohort studies. The following outcome was metanalyzed: major adverse cardiac events (MACE: death, acute myocardial infarction and/or revascularization). Pooled fixed or random effect RR and 95% confidence intervals (95%CI) were computed. Results. Overall 217 abstract were evaluated and 26 full texts were retrieved; 17 of them were excluded, (abstract, review, metanalysis, MACE missing, no pertinence). Of the 9 articles included in our study 1 was a RCT and 8 were observational studies (5 with adjusted estimates). A total of 17541 patients were enrolled, women were 28%. The median age was 64 years. Median follow-up duration was 18 months (25th-75th percentiles 12-24). The RR for the RCT was 0.92 (0.39-2.19, N=210 pts); for the ‘adjusted’ cohort studies it was 0.79 (0.69-0.91, N=15405 pts) and for the ‘unadjusted cohort” it was 0.89 (0.70-1.13, N=2286 pts). Discussion. The small RCT does not provide sound evidence, as shown by wide 95%CIs, while the large well conducted cohort studies with adjusted estimates indicate a protective effect of IVUS towards MACE, with high level of confidence. Being the IVUS patients more critical, its effect would be diluted in unadjusted studies. Although large RCTs are needed to confirm IVUS role, good quality cohort studies might better reflect real life. Background: Estimates of the fecal occult blood test (FOBT) (Hemoccult II) sensitivity differ widely between screening trials, and will lead to divergent conclusions on the effects of FOBT screening. We used microsimulation modeling to estimate a preclinical colorectal cancer (CRC) duration and sensitivity for unrehydrated FOBT from the data of 3 randomized controlled trials of Minnesota, Nottingham and Funen. In addition to two usual hypotheses on the sensitivity of FOBT, we tested a novel hypothesis where sensitivity is linked to the stage of clinical diagnosis in the situation without screening. Methods: We used the MISCAN-Colon microsimulation model to estimate sensitivity and duration, accounting for differences between the trials in demography, background incidence and trial design. We tested three hypotheses for FOBT sensitivity: sensitivity is the same for all preclinical CRC stages, sensitivity increases with each stage, and sensitivity is higher for the stage in which the cancer would have been diagnosed in the absence of screening than for earlier stages. Goodness of fit was evaluated by comparing expected and observed rates of screen- detected and interval CRC. Results: The hypothesis with a higher sensitivity in the stage of clinical diagnosis gave the best fit. Under this hypothesis, sensitivity of FOBT was 51% in the stage of clinical diagnosis and 19% in earlier stages. The average duration of preclinical CRC was estimated at 6.7 years. Conclusion: Our analysis corroborates a long duration of preclinical CRC, with FOBT most sensitive in the stage of clinical diagnosis. Background: Aprotinin was used to minimize blood loss in cardiac surgery patients before withdrawal from the market in 2008 for safety reasons. The drug has again become available. Methods: We performed a systematic review and network meta-analyses to estimate the relative risks of death, myocardial infarction (MI), stroke and renal failure/dysfunction between aprotinin, tranexamic acid (TXA), epsilon-aminocaproic acid (EACA), and no therapy. A 2011 Cochrane review was used to identify relevant randomized controlled trials (RCTs), and a search of Medline, Embase and the Cochrane Register of Trials was conducted to identify propensity matched/adjusted observational studies. Odds ratios and 95% credible intervals for comparisons between therapies were estimated, as were the average rank and probability that each therapy was most safe. The probabilities of odds ratios excluding a null difference were also estimated. Network meta-analyses based on RCTs were fit first, and then observational evidence was incorporated. Results: 83 RCTs and 11 obervational studies (>41,000 patients) were included (Figure 1). Based on RCTs, TXA was associated with a reduced risk of death versus aprotinin, while pairwise comparisons were inconclusive for MI, stroke, and renal failure/dysfunction; point estimates and coverage probabilities of these intervals suggested aprotinin was often associated with an elevated probability of increased risk (Table 1). When observational data were incorporated, pairwise comparisons showed increased risks of death with aprotinin relative to TXA and EACA, as well an increased risk of renal failure/dysfunction relative to all comparators. There were also probabilities suggestive of increased risks of MI with aprotinin compared to TXA and EACA (Table 1). Conclusions: Data suggests there remains reason for concern regarding aprotinin safety. While meta-analyses of RCTs can lack sufficient sample size to definitively identify harms imbalances, appropriate incorporation of observational evidence and use of network meta-analysis can help reduce uncertainty in analyses of such data. The success of translational research requires sound judgment in the planning and implementation of trials. Such planning involves synthesis of previous literature. We sought to determine the accessibility of preclinical efficacy studies in a cohort of novel investigational agents entering clinical development between 2000 and 2003. A cohort of initial human trials of new agents reported as full journal publications was identified through systematic searches of MEDLINE and EMBASE. The cohort included 100 investigational agents, spanning 10 different indications. Identified agents were linked to preclinical studies by a search of references, EMBASE and PubMed. Preclinical evidence was considered published if at least one full journal article reported disease response in live, non-human animals. Of the 100 agents, 89 had published preclinical work, 80 of which tested the identical intervention and another 9 of which were based on closely related agents. Thirteen percent of agents published preclinical studies only after the initial human trial publication. Fifty-five percent of initial human trial articles referenced published animal work. Of the 89 agents, 57% had five or more animal studies available, 33% had between 2-4, and 10% had one preclinical study available. Of the 13 investigational agents that received eventual FDA approval, 12 had five or more preclinical studies available in the published record. The probability that preclinical evidence was publicly reported in more than five reports was significantly greater for FDA approved drugs than for drugs that did not receive licensure (85% vs. 38%; Yates’ chi-squared test, p=0.002). This work demonstrates that a large proportion of investigational agents had at least some preclinical studies available. This suggests that preclinical knowledge synthesis for many new drugs is feasible—especially for those receiving licensure. However, we are unable to estimate what proportion of preclinical efficacy studies go unpublished. Smoking is directly responsible for 4500 deaths each year in New Zealand (NZ) including 22% of all Maori deaths. Despite the proven efficacy of various cessation approaches, long-term cessation rates are still below 25%. Quitline is a national smoking cessation service that offers telephone delivered behavioral support (3 sessions) and an eight-week supply of nicotine replacement therapy (NRT) at a subsidized rate for smokers in NZ. The Fit2Quit trial aims to determine the effects of a home and community-based exercise intervention on increasing smoking cessation rates at six months when added to usual care. A two-arm parallel randomized controlled clinical trial was conducted in 2010-11. Eligible participants were recruited through Quitline who were interested in quitting, had their first cigarette within 30 minutes of waking, aged 18 years and over, and wanted to be physically active. The intervention group received telephone-based exercise counseling programme delivered by existing Green Prescription (GRx) services over six months, in addition to usual smoking cessation support. The control group received usual care alone. Self-reported smoking abstinence, mood and physical symptoms (MPSS), and physical activity level (IPAQ) were assessed at 8 and 24 weeks. The primary endpoint was self-reported point-prevalence at 24 weeks, confirmed by salivary cotinine reading. Treatment evaluations were performed on the principle of intention-to-treat assuming missing as smoking. Sensitivity analyses were also conducted. A total of 906 smokers were randomized (intervention N=455; control N=451). Participants were aged 37 years (18-78yrs), 31% Maori, 46% males, 83% smoked >10 cigarettes per day, 79% made previous quit attempts, and 55% didn’t use NRT at baseline. Relative risk and adjusted odds ratio were calculated to assess the smoking abstinence. Repeated measures models were used to evaluate change in MPSS and IPAQ total score between groups. Statistical analyses and full trial results will be presented and discussed. Trajectory-based models are increasingly applied in clinical research to understand the etiology and developmental course of different types of disorder [1, 2, 3]. More recently, the range of applications has been extended to capture heterogeneity in treatment responses to clinical and randomized trials [4, 5]. A double-blind, randomized controlled clinical trial in nutrition was carried out on 197 subjects to investigate whether the consumption of an active dairy product can improve different clinical parameters, such as well-being scores. For exploratory purpose Group-based trajectory models (GBTM) [6] were used to map the developmental course of these distinct but related outcomes individually, and then according to a joint approach. The statistical analysis was performed with the Proc Traj [7] with SAS® software release 9.2 (SAS Institute, Inc, Cary, NC). GBTM provided empirical ways of identifying clusters of individuals in response of the product intake, following both typical and atypical courses of development. Overall, results suggested that the active product was associated with trajectories relating clinical improvements. In this application, GBTM also allowed to highlight different responders’ patterns in the active product group by identifying developmentally meaningful subgroups in the population for whom product effects vary. In this communication, authors will discuss the results of applying such methods to a clinical trial in nutrition. Strengths and limits associated to this approach as well as the interpretation will be further detailed. The Lifestyle Interventions and Independence for Elders (LIFE) Study is a Phase 3, multi-center randomized controlled trial (RCT) designed to compare a moderate-intensity physical activity program to a successful aging health education program in 1,600 sedentary older persons, age 70-89 years, across eight field centers. The primary outcome is major mobility disability, defined as the inability to walk 400 meters. Secondary and tertiary outcomes include serious fall injuries, pulmonary events, and cardiovascular events. Given the number of participants and their age range, we expect to collect data on many outcomes throughout the length of the trial. The need for a central tracking and monitoring system initiated the construction of the Outcomes Management Tool (OMT). The OMT allows the Data Management and Quality Control (DMAQC) Center to centrally track status of outcomes, compile cases for assignment to adjudication committee members, and track the adjudications via an online interface system. The web system registers each outcome reported, eliminates duplicate events identified, and enables field centers and the DMAQC to track outcomes as they move through the outcomes process. Making this process accessible on the web is efficient for real-time reporting, gives administrators the ability to manage documents, assign adjudicators, monitor active adjudication status, communicate with adjudicators when needed, and provides a central repository for outcome details. Generating outcome events from participant reports begins with the data entry of a case report form (CRF). Upon submission of the CRF, if specific criteria are met, this triggers the creation of a unique outcome ID for each event reported by the participant and initiates the tracking of outcomes. This utilizes IBM’s WebSphere ILOG-JRules business management system for rules validation and event provocation. This presentation will outline the flow of outcomes in the Outcomes Management Tool from the entry of the CRF to the final adjudication. Patient registries may support a variety of functions, ranging from clinical care to quality measurement and reporting, to clinical and translational research. Compared to clinical trials, quality reporting registries typically involve more data collection (all patients and all visits within a set of practices), often with fewer resources (funding, personnel). Data coordinating centers for registries must be aware of these differences, assist practices in embedding data collection into clinical care, and provide time-efficient resources for monitoring and improving data quality. To support ImproveCareNow, a quality reporting registry for pediatric IBD, we worked with an established multicenter network to standardize data definitions and review their processes of data collection and capture. Additionally, we developed a set of data quality metrics to define areas where data quality is critical to desired outcomes or known areas of deficiencies for the registry. Among the measures of interest are: – proportion of patients enrolled in the registry – proportion of visits captured in the registry – proportion of visits with all critical data elements captured – proportion of visits entered in the registry within 30 days of visit date – proportion of visits where data values are suspicious, based on inter or intra-visit inconsistencies We also developed a pair of data quality reports– a set of run charts that allow sites to track data quality metrics over time and a report that allows sites easily identify potential errors to a level of detail that allows for quick updates to be made to the registry. These reports were designed to have a look and feel similar to tools already in use by these sites to increase the probability of their integration into normal clinical practice. Analysis of the effectiveness of these strategies is ongoing. Preliminary results show an increase from 80% to 89% of visits with all critical variables recorded. One major aspect of multi-center clinical trials coordinating centers includes the work, data flow and monitoring of central units which analyze samples and transmit data to the coordinating center. These include laboratories for blood samples and reading centers for electrocardiographs, retinal photos, CT scans, DEXA scans, etc. Data travels from clinical site to collection site (e.g. scan center) to central reading center and finally to the coordinating center. It is critical to ensure the integrity of the data while streamlining the process when data collection and management can be challenging. The first step in the design process is planning the methods of participant flow from clinical site to collection site, including obtaining informed consent, verifying eligibility and participant scheduling. The second step is developing the method of data flow from collection site to reading center using tools such as web-based portals, transmission software or courier services. The third step is for the reading center to securely transfer the data to the coordinating center. Data transmissions should have preapproved formats and sets of components for consistency and accuracy such as dates of analysis and receipt, and unique identifiers for the individuals performing the analysis and the specific image or sample. The fourth step is to develop methods for data validation to ensure the correct attribution of assessment to the corresponding participant. The coordinating center should have a checklist for the incoming central unit data where the data must check and pass each step before moving to the next step. Finally, a process for the reporting of alert values to sites in a timely manner must be developed. Several options for these processes will be described and various methods in which data is collected and transferred from reading centers to coordinating centers will be compared, and future suggestions will be discussed. With the popularity of electronic data capture (EDC) increasing in clinical trials, there is a growing demand for stronger and more efficient data validation tools. Traditionally, online single field range checks and batched offline logic checks performed by statisticians are used to identify data issues related to data transcription errors and protocol compliance discrepancies. Our home-grown, web-based Clinical Trial Management System (CTMS) is equipped with a database driven data validation process to perform rule-based data checks at the time of data entry. This process uses a centralized database table to manage all data validation checks for all Case Report Forms (CRFs). Initial data checks against invalid data types, missing required data items and impossible data values are performed prior to saving the record to the database. Records that pass this initial process are saved to the database and further data validation is performed by using database queries involving one or more related data items to ensure sound data collection logic is enforced. Depending on the sources of these data items, such data validation queries can be defined within the current data record, across multiple records within the same data table, or across multiple records from different data tables. Violations of these data checks will be saved in a central rule violation table along with the values of the associated data items. Based on the level of the violation, an alert, warning, or rejection flag will be placed on the data form as an indicator for users to take the next appropriate action. Our data management experiences for more than ten large, multi-center phase III clinical trials demonstrated that this comprehensive, real-time data validation function has significantly reduced data errors occurring during the data entry process and shortened the time required for data cleaning and database lock. Background: Source data verification is the process of comparing information on source records to data recorded in a Case Report Form as part of a research study. Researchers are required to perform source data verification in order to comply with National Regulations and Good Clinical Practice Guidelines. There is little guidance for researchers as to the amount and frequency of this process or the effect of source data verification on study outcomes. Recently there has been debate regarding the value of source data verification on data quality. There is little research on the beliefs and attitudes of Investigators and Research Coordinators regarding the amount, frequency, value and effect of source data verification and how it may be optimized. This project is part of a research program to determine the effect of source data verification on study outcomes and the amount of source data verification that should be done. Objectives: The primary objective is to describe the current beliefs and attitudes regarding source data verification of members of the Canadian Critical Care Trials Group. Methods: We developed a self-administered on-line survey using focus groups of Critical Care Investigators and Research Coordinators. The survey has been pilot tested and is undergoing clinical sensibility and reliability testing. The final survey will be sent electronically to the 330 members of the Canadian Critical Care Trials Group between December 2011 and February 2011. Results: We will present descriptive summaries for all items and then separately for Investigator and Research Coordinator groups. We will evaluate associations between variables. Significance: The survey results will provide general guidance to adult and pediatric critical care researchers in determining data quality assurance procedures and will inform the next phase of the research program: assessing methods of source data verification and its effect on study outcomes in Canadian Critical Care research. The concept of reproducible research was introduced in the computational sciences some time ago, but has begun to be considered in biostatistics, bioinformatic, and other areas of medically related data analysis. Reproducible research, in this context, involves performing data analysis and statistical evaluation of data in a manner in which the relevant results (tables, figures, incidental text) can be recovered with minimal effort, analyses can be conveyed to data repositories with minimal effort, and analyses can survive the inevitable changes in personnel at the data management center. This has strong implications for clinical trials, which are usually conducted in a “reproducible research” concept, but which are not always analyzed in this manner. From the perspective of data analysis in SAS (a dominant tool in the clinical trials field), structural approaches to project setup, disciplined and consistent uses of macros, and archival approaches to manuscript construction are the first steps to reproducible research. These will be discussed and illustrated with several examples. The use of technology in clinical trials is increasing where ancillary software is utilized to capture data beyond that typically collected via case report forms in an electronic data capture (EDC) system. In recent studies by the Clinical Trials Network of the National Institute on Drug Abuse, software systems have been used to administer and monitor adherence to trial interventions and perform study assessments. Utilization of ancillary systems requires appropriate advanced planning to address key issues. Ancillary systems vary in complexity and sophistication, ranging from simple and home-grown to complex and well-validated. Upon identifying systems deemed necessary for the trial, a communication plan should be developed outlining expectations for software developer support, as well as the workflow and timeline for identifying and resolving issues. The software and hardware needed for hosting and using the system must be identified and procured. Understanding the structure of the data collected in the system, which is facilitated by obtaining a system diagram and data dictionary, is key. If the system will be used to capture data for analysis, it is important to verify that those data are collected and identifiable in the system. The data dictionary can assist in identifying if protected health information is being collected and if so, to develop appropriate security measures, such as encryption. It is also important to explore whether the system can communicate with the EDC system, or at a minimum that the data captured in the system can be linked to those from the EDC system. Finally, a plan for providing initial and ongoing training in the use of the system must be established. The plan should outline who is responsible for the training, the method of training (e.g., face-to-face, webcast), and the development of a user’s guide. Despite best practice guidelines, patients presenting for primary care are often not screened for medical issues other than the presenting problem due to time and personnel constraints. This presentation details an electronic screening protocol and technology platform that: a) allows patients to complete a brief screening form prior to their appointment, and b) provides immediate feedback on health risks and conditions for clinicians to use in the patient visit. Patients presenting at the primary care clinic self report via a Comprehensive Primary Care Screening (CPCS) instrument using Android-based tablets. The CPCS assesses risks related to falls, colon cancer, sexual health, domestic violence, oral health, alcohol, depression, and basic needs. Responses are entered via a touch-screen interface in English or Spanish, or via audio versions of the instrument. It incorporates dynamic branch logic to capture detailed information where needed from patients. Upon completion, data is saved into a EDC database where it is scored and analyzed for recommendations, including referrals and standing orders. Results are immediately accessible for physician use; scores and raw data may also be extracted for incorporation into de-identified research datasets. We use a CDISC ODM-based form definition created using the OpenClinica open source EDC platform. The CDISC ODM protocol and data capture definitions are automatically transformed into to XForm definitions that are rendered via an open source medical data capture “app” on the tablets for patient entry and physician retrieval. The EDC engine provides data storage, validation, scoring, reporting, and analysis capability. The project objectives are to: - increase referrals for health risks and conditions, - increase access to and use of de-identified screening results by researchers, and - promote adoption, dissemination, and adaptation of the screening tool by using reusable, configurable open source platform. The methods for data capture have advanced as technology has evolved. Often studies utilize an Electronic Data Capture (EDC) system, but many studies may still only be paper based. The costs and regulatory requirements of deploying an EDC system may not outweigh the benefits of conducting a paper based study through the use of fax forms, scanning optical character recognition software, telephone data entry, bubble forms, or digital pen capture. However, these methods are thought of as “old technology,” but with the proper technical approach, a paper based study may actually be more efficient and productive. A model project was conducted in which digital pen data collection technology was examined for benefits and drawbacks. The digital pen was used by study staff members and/or subjects to complete CRFs, the pen is docked in a docking station attached to a computer, and data is uploaded to the data coordinating center. After project evaluation, it was found that the use of digital pen technology to instantly generate electronic copies of paper CRFs benefited the study by: eliminating the need to ship completed CRFs, eliminating the need to scan paper forms for electronic storage and transmission, dramatically reducing data collection training for staff, using a source document as a CRF, and having the ability to track CRF completion virtually. While there are multiple ways to implement digital pen technology, each with their own drawbacks, this project found that data still had to be entered into an electronic database manually, printing CRFs required using print vendors with previous experience with the software, and a method to transfer and track files and store back-up copies was needed. These lessons learned will only benefit future studies using digital pen technology corporately, and help reduce Sponsor costs for paper based studies. The technology of personal computing has undergone great changes the last few years. Naturally, those conducting clinical research would like to take advantage of these advancements. The Clinical Research Database (CRDB) at Memorial Sloan-Kettering Cancer Center (MSKCC) was created as a client server application in 1992. In 2003, the first web based platform was made available to facilitate data entry from remote locations. During the past year, it became necessary to expand our application scope to keep up with the latest needs of the researchers. There are two main areas of our system expansion: 1) the multicenter internet platform and 2) mobile devices. Historically at MSKCC, data management for our multicenter protocols was mainly a paper based system. It became evident that participating sites needed access to CRDB to enter data directly. This brought about security and privacy concerns beyond what was needed for an internal database. The conversion to a new external platform to deal with these concerns brought its own set of challenges. In parallel, a mobile option was needed for survey completion by participants in clinic because desktop computers are not always available. However, CRDB was designed for data entry by clinical research professionals, not participants. We were able to design a tablet survey that addressed this as well as other concerns. In this presentation, we will review the need for alternative platforms for collecting clinical research data. We will also address the various challenges we faced while expanding our system from the traditional data entry model, as well as our suggested solutions. Clinical outcomes are an important component of randomised controlled trials (RCTs) and are often used to complement patient reported outcome measures such as health-related quality of life. Although clinical outcomes were traditionally collected through clinical examination or laboratory results, routine data sources and self-reporting by patients are increasingly used. We examined four RCTs that collected patient reported clinical outcomes through postal questionnaires. In each RCT the patient reported clinical outcome was verified using either medical records, routine data sources or by contacting the patient’s family doctor or hospital physician to ascertain the accuracy. The accuracy of patient reporting of clinical outcomes is dependent on a number of factors including the nature and timing of the clinical outcome and the phrasing of the clinical questions. For example, it may be easier for a patient to report a knee-related hospital re-admission than self-report a urinary tract infection. Nevertheless, approximately 15% of patient reported knee- related hospital re-admissions (collected through annual postal questionnaires) could not be verified through routine data sources and/or medical records. Such inconsistencies were shown to be a combination of misunderstanding by the patient and inaccuracies of the routine data sets. Obtaining clinical information from patients is feasible, especially if the outcome of interest is a symptomatic one. However with the potential inaccuracies associated with patient reporting of clinical outcomes, it may be necessary to consider verifying such outcomes with medical professionals and/or routine data sources. Such a strategy has implications in terms of staff time and cost and therefore has to be considered during the design stage of the RCT. We will discuss some challenges and inconsistencies between self-reporting and medically confirmed clinical outcomes. We will highlight processes involved in verifying patient reported clinical outcomes and how adopting such a verification strategy may impact on the overall trial results. Background: The availability of platforms and devices for electronic data capture has been a major step forward for the assessment of patient-reported outcomes (PROs) in clinical trials. The advantages of electronic PROs include real-time tracking of compliance, automated patient reminders, and date-time data for individual assessments. Multiple modes of ePRO data capture (e.g. PC/tablet, handheld, automated phone survey) allow flexibility in clinical trial design. This has raised the question of whether use of multiple modes in a single trial might affect statistical power. Methods: We review the growing body of research on the measurement reliability across modes, and estimate the effect on statistical power in several common PRO endpoint analyses. We consider this effect compared to other common and PRO-related threats to power. Results: Peer-reviewed studies indicate the measurement reliability across modes of administration is high (approx 0.90) and consistent with the test-retest reliability of individual modes. We have calculated that even if measurement reliability were more moderate (0.70 – 0.80), effects on statistical power would be small. The estimated reduction in statistical power from using multiple modes of data collection is dwarfed by the estimated loss of power associated with missing data (which can be prevented by allowing a second mode as alternate or backup). The threat of multiple modes to power is also far less than mis-estimation of variance in sample size calculation or use of PRO measures with low responsiveness. Conclusions: Very small reductions in measurement reliability from the use of multiple modes of ePRO are outweighed by the advantage of data completeness. To preserve power in clinical trials, researchers would be better off focusing on well-established areas of methodologic concern, such as accurate estimation of variance and choice of responsive outcome measures. Suppose a limited number (N) of patients is available for study and the traditional randomized trial design comparing ‘standard’ to ‘new’ is not feasible. This work was motivated by an interest in comparing two treatments for recurrent Wilm’s tumor, where about 40% of patients are event-free long- term with standard treatment. One option is a study design which seeks to maximize the number of study subjects receiving the better treatment. Consider a design for which n1/2 patients are randomized to each of 2 treatments and then time-to-event is compared using the log-rank test. All remaining (N-n1) patients are then assigned treatment 1 (treatment 2) based on whether the log-rank test is positive (negative). We assumed an available sample size of 100. For exponential failure-time distributions with a baseline ? of 0.50 to 0.75, and for a range of relative risks from 0.65 to 0.80, the design which maximized the number of subjects receiving the better treatment was to randomize about 40 patients (having observed about 20% of the total expected information) and then assign the remaining 60 patients to the regimen with the better observed outcome. We hypothesized that, in other settings, the n1 corresponding to observing about 20% of the expected information would be optimal. We ran simulation studies assuming a cure model outcome: S(t) = 0.4 + (0.6)*exp(- 0.75t). We simulated studies of 100 patients, with n1 values from 20 to 80 and cure rate increases corresponding to relative risks from 0.65 to 0.80. The number of study subjects receiving the better treatment was maximized at an n1 value of 40-45 (corresponding to 17- 20% of the total expected number of events). In non-inferiority (NI) trials, the evaluation of the efficacy of a new experimental treatment allows for some defined level of reduced effect as compared to an active control standard. Serial use of NI trials may lead to erosion in the level of improvement provided by newly approved therapies; this phenomenon is called ‘bio-creep’. Simulations were designed to facilitate understanding of bio-creep risk when approval of a new treatment with efficacy less than some proportion of the effect of the active control treatment would constitute harm, such as sulfonamides or penicillin for treatment of Community-Acquired Bacterial Pneumonia. In this setting, risk of approval of insufficiently effective therapies may be great, even when the standard treatment effect is ‘constant’ across trials. Among the many possible factors contributing to this manifestation of bio-creep, the most influential were the method for selecting the active control, the choice of non-inferiority margin, and bias in the active control effect estimate. Therefore, when non-inferiority testing is performed, margins should be based upon the estimated effect of the active control, should account for the variability and for likely sources of bias in this estimate, and should address the importance of preservation of some portion of the effect of the standard. Purpose: This study demonstrates how errors in estimating the SD of a population can lead to inaccurate sample sizes and underpowered studies, and offers recommendations for maximizing the likelihood of achieving adequate statistical power. Results: Our simulated data show that greater sample size provides a more reliable estimate of the SD of the population than a smaller one. All minimal and 25th percentile sample SDs fell below 44 (the population SD) no matter how big the sampling sample size (from 2 to 100). For samples sizes 10 and 100, the minimum sample SDs underestimate the population SD by 47.1% ([44 – 23.27] / 44) and 18.3% ([44 35.93] / 44), respectively. For all sample sizes below 40, the mean sample SDs fell below the population SD; for sample sizes of 40 or greater, the mean sample SDs were close to the population SD. All maximum and 75th percentile sample SDs exceeded the population SD. Based on a published underpowered trial (n1=13, n2=17 and power < 30%), we found the reported SDs ranged from 1.8 to 9.3 from the literature, with a weighted average SD of 8.1. This study needs to have n=66 for each group to reach a power of 80% according to our calculation. Conclusion: No explicit guidelines have been available for choosing an appropriate SD to use in the calculation of sample size for a clinical trial. To remedy this problem, we have developed an algorithm and recommendations for making a judicious choice intended to minimize the risk of conducting an underpowered study. Many depression trials used the Hamilton Rating Scale for Depression (HRSD) as the primary measure of depression, although its severe measurement issues had been well-known. In a trail (Kellner, Kanpp, et al., 2006) that compares continuation electroconvulsive therapy (C-ECT) vs. Pharmacotherapy (C-Pharm) for relapse prevention in major depression, the total score of the 24-item HRSD (HRSD24) was used to define relapse or remission. That trail found no statistically significant differences between the two arms. However, the results might be misleading, simply because the uni-dimensional assumption made for the HRSD24, i.e., all of the 24 items are measuring a single domain (depression) and therefore the 24 item scores can be summed to a single total score as a measure of depression, might be wrong. In this study, the original data from that trial (201 patients with 98 C-ECT, 103 C-Pharm) were utilized to investigate measurement issues in the HRSD24. Confirmatory factor analyses (CFA) for the uni-dimensional assumption of the HRSD24 were implemented on the data at visits 1, 2, 3, and 4. Results showed that this uni-dimensional assumption failed at each of the 4 visits. This indicated that the single total HRSD24 score should not be used as the defining variable for relapse or remission at any of the visits, thus the analyses using this total score was misleading, and may conceal some true positive findings in that trial. Exploratory factor analysis (EFA) on all of the 24 items failed to yield a consistent factor structure across the 4 visits. Then item-level analyses of the 24 items were implemented, and results showed that statistically significant differences exist on 3 of the 24 items between the two arms. It might be worthy to re-investigate the findings from some of the large depression trials that used the HRSD, especially those with negative findings. The tumor-node-metastasis (TNM) staging system has been the lynchpin of cancer diagnosis, treatment, and prognosis for many years. For meaningful clinical use, an orderly, progressive grouping of the T and N categories into an overall staging system needs to be defined, usually with respect to a time-to-event outcome. This can be reframed as a model selection problem for a censored response grouped with respect to features arranged on a partially ordered two- way grid (the TN table), and a L1 penalized regression method is proposed for selecting the optimal grouping. Instead of penalizing the L1-norm of the coefficients like lasso, in order for the grouping to occur, we place L1 constraints on the differences between neighboring coefficients. The underlying mechanism is the sparsity-enforcing property of the L1 penalty, which is expected to give a reduced number of unique coefficients that represent different groups. A partial ordering constraint is also required as both the T and N categories are ordinal. A series of optimal groupings with different numbers of groups can be obtained by varying the tuning parameter. This gives a tree-like structure for partitioning the TN table, and thus offers a visual aid on how the groupings are made progressively. We hence call the proposed method the lasso tree. We illustrate the utility of our method by applying it to the stage grouping of colorectal cancer. Simulation studies are carried out to examine the finite sample performance of the selection procedure. We demonstrate that the lasso tree is able to give the right grouping with moderate sample size, is robust with regard to changes in the data, and is not affected by random censorship. Non-inferiority trials are appropriate when there is clear clinical rationale that a placebo cannot be used, and inference needs to be made to a putative placebo group to ensure the test compound is effective. However, with the development of drugs that have greater efficacy than the existing gold standard therapy, the use of this standard as a comparator treatment in future trials with new novel drugs may be inappropriate. This is particularly true in the area of antithrombotics for stroke prevention in patients with atrial fibrillation. With the existing gold standard, Warfarin, likely to be replaced by one of possibly three new compounds in coming years, newer drugs for the treatment of SPAF will be faced with the challenge of showing non-inferiority to a new gold standard, that was never tested in a placebo-controlled setting. This leads to the challenge of defining a NI bound based on preserving a certain percentage of the placebo effect, where no placebo-controlled studies exist. We propose a strategy that incorporates placebo controlled data for the existing gold standard (warfarin) and warfarin controlled data for the new gold standard in defining the NI bounds for a time to event endpoint. Interventions to improve patient care are sometimes delivered to whole primary care medical centres. In such instances, it is natural to use a cluster randomized clinical trial design to test the intervention effect. Ideally, outcome data are collected at the individual level, though follow-up rates can be poor, particularly for patients in non- intervention centres who may have little involvement with the study beyond providing data. However, after an initial contact with patients to obtain consent and baseline information not otherwise available, utilization of routinely collected data, in the form of medical centre records, allows for near complete follow-up at an individual level with minimum additional contact. In Scotland, data on all hospitalizations and deaths are also routinely collected and can be linked to individual patients, allowing for long-term passive follow-up. These data include reasons for hospitalization, length of hospital stay, and causes of death. There are several advantages to cluster randomized studies, such as simplified trial organization and reduced contamination between intervention groups. Disadvantages include the increased sample size required for the same statistical power and the need for more complex analyses. Use of routinely collected data is advantageous for several reasons, including reduced trial costs and maximization of follow-up, though limitations include the accuracy and completeness of the data collected. These issues will be discussed in relation to the Heart failure Optimal Outcomes from Pharmacy Study (HOOPS), recently completed in Glasgow, Scotland, which used primary care medical centres to recruit patients with left ventricular systolic dysfunction, and to deliver an intervention of pharmacist-led medication review, to optimize the use of evidence-based medicines. Details of medication use up to two years post- randomization were collected through medical centre records. Routinely collected national hospitalization and death records were also used to follow patients up for a median of 4.7 years. We explore the possibility of obtaining routinely available data for use in clinical trials, the reliability of this data, and the benefits of this approach if found to be successful. The SHIFT Trial (a randomised controlled trial of family therapy vs. treatment as usual for adolescents following self-harm) is used to illustrate the process. The Trial’s primary endpoint (repetition of self-harm leading to hospital attendance) requires timely and regular collection of hospital attendance data to inform the timing of analysis, and is thus a resource intensive task requiring researcher visits to Hospitals to interrogate local medical records. We are exploring the feasibility of collecting this data from a central source, the NHS Information Centre (IC). The IC holds data provided periodically by English Hospitals, their main aim being to provide England-wide statistics to inform frontline decision makers. However we wish to assess whether this is also a complete and reliable means of acquiring data for research. Benefits include: a) Regular, fast England-wide data retrieval rather than collection from an identified, limited hospital ‘pool’; b) avoidance of potentially biased data collection due to, for example, more frequent visits to some Hospitals than others; and c) to free up researcher resources. After comparing data gathered by the Researchers from pre-identified, representative Hospitals with data retrieved from the IC, a change to the method of primary outcome data collection may be instigated after consideration of: a) The percentage of self-harm episodes recorded & coded appropriately; b) The percentage of required data items retrieved for each episode; c) Data quality for Hospitals with diverse catchment areas - to ensure recommendations for appropriate methods of data collection can be made at a study level and a site level. Preliminary findings relating to the reliability of routine data for use in clinical trials will be presented. To describe the broad portfolio of cardiovascular clinical research, the Duke Clinical Research Institute (DCRI) and the Clinical Trials Transformation Initiative (CTTI) derived a dataset for aggregate analysis from ClinicalTrials.gov. We identified 40,970 clinical research studies registered after September 2007 where human subjects received diagnostic, therapeutic or other interventions per protocol. By analyzing 18,491 descriptors from the National Library of Medicine’s Medical Subject Heading (MeSH) thesaurus and 1,220 free-text terms, we included only studies related to the diagnosis, treatment or prevention of diseases of the heart and peripheral arteries which enrolled adults ?18 years (N=2,325 studies). The identified studies were 74% ongoing, 22% completed, and 4% terminated, withdrawn, or suspended. The study intervention was drug in 45%, device or procedural in 39%, behavioral in 8%, and biologic or genetic in 3%. Development phase was Phase 4 in 25.6%, Phase 3 in 19.1%, Phase 2 in 15.9%, Phase 0 or 1 in 4.9%, and inapplicable in 34.5%. Many studies (46.3%) anticipated enrolling fewer than 100 subjects. Only 16.8% of phase 3 studies anticipated more than 1,000 subjects. In phase 3 studies, the median anticipated enrollment was 200 subjects. Most studies had one (26.8%) or two arms (59.9%); 94.7% of studies with two or more arms were randomized. Only 32.0% were double blind, 15.1% were single blind, and 52.9% were open label. Industry was the most frequent sponsor overall (32.0%) and across development phases. Non-US oversight authority was listed for 60% of studies. Of studies with US oversight, 47.9% reported FDA oversight. Cardiovascular medicine is widely regarded as a vanguard for evidence-based drug and technology development. This survey of cardiovascular studies reported in the clinicaltrials.gov registry reveals substantial heterogeneity in study design and sponsorship. The preponderance of small trials represents an opportunity to encourage collaboration and foster research networks. Background: Many clinical trials have endpoints that are changes in a measurement over time. An example is change in kidney function as measured by glomerular filtration rate (GFR). Sample size and power depend on the standard deviation of the change, but this is difficult to estimate from published data. ClinicalTrials.gov now requires the submission of results for completed trials. The objective of this analysis is to determine the number of trials with GFR as an outcome, the number with results, and the usefulness of the results data for planning new trials. Methods: A search was performed on November 10, 2011 using the terms “(NOT epidermal growth factor receptor) [ALL- FIELDS] AND (NOT cancer) [DISEASE] AND (gfr OR creatinine clearance OR egfr) [OUTCOME]”. SAS data sets were created from protocol descriptive data downloaded from ClinicalTrials.gov. These data sets were used to classify the protocols and to identify the outcome variables. Posted results data were reviewed and coded and publications were identified by searching for the NCT number. Results: Of 584 interventional studies that met the above criteria, 122 had completion dates in 2008-2009 and results were posted for 39 (32%). Results were more likely to be posted for studies funded by industry (35/75 versus 4/47), p<0.0001. In total, 61 studies had results and change in GFR was an outcome for 22 of these (6 kidney disease, 9 post- transplant, and 7 other diseases). The duration of follow-up ranged from 24 hours to 5 years and the SD ranged from 3 to 35. Thirteen of the 61 studies had publications that included the NCT number. Conclusions: Only 32% of eligible trials have results at this time. Review of results is time- consuming and baseline data are often incomplete, but the results database should be a useful resource in the future. Sequential meta-analyses, those in which power is determined via group sequential boundaries, have been suggested as a method to evaluate the status of the current evidence as well as tools for the trial planning process. Regarding the latter, some may consider counterproductive to use design parameters from a meta-analysis rather than from conventional sample size calculations. Here we examine conditions under which this could be actually of gain. The minimum sample size for a target effect size of small magnitude (odds ratio up to 1.5) was estimated using conventional formula for superiority and equivalence, considering alpha=0.05 and 80% power. This size was estimated for a minimal proportion varying between 0.1 and 0.9, and adjusted for various values of the inconsistency across studies in the meta-analysis ranging between 0 and 0.5. Meta-analyses subjected to sequential monitoring with 5 and 20 studies were considered. For a hypothesized difference of 20%, the average sample size for the trial ranges between 112 and 224 for superiority and 206 and 412 for equivalence (a 95% reduction) within a meta-analysis of 20 studies. Likewise, for a difference of 50% the average sample size for the trial ranges between 92 and 184 for superiority and 170 to 338 for equivalence (an 80% reduction) within a meta-analysis of 5 studies. The smaller the target effect size, the greater the number of studies required in the meta-analysis. It is important to highlight that the individual trial at this reduced sample size is underpowered but not the meta- analysis that will include it. Here it is shown that incorporating meta- analytic information in trials comparing binary outcomes may be beneficial in cases where the target effect size is small as well as when heterogeneity of estimates across studies is moderate. This is true from both superiority and equivalence points of view. Randomised trials are difficult to design, expensive to conduct, and frequently fail to deliver precise enough answers to relevant healthcare questions. Particularly for complex intervention studies, in which there may be poorer background knowledge going into the trial, being confident that all design elements (PICO - Population, Intervention, Control, and Outcomes) are properly understood and optimised is a considerable challenge. In addition, the majority of trials struggle to recruit on time and to budget, and there are often issues of retention that can undermine a trial. Increasingly public funders (e.g. UK NIHR HTA/MRC) are requiring reassurance on many of these issues via an internal pilot or feasibility study, before releasing funding for the full trial. Although there has been an explosion of interest in adaptive designs, much of this has been from a theoretical statistical perspective, without so far corresponding attention paid to the practical challenges of how you embed such design flexibility into a real trial, and make reliable decisions on the basis of an often small amount of information of uncertain quality, probably tainted by many sources of bias as the trial struggles to get itself established. This talk will discuss the issues around developing ‘stop-go’ criteria to confirm whether a trial should expand to a full trial. These usually include performance metrics for recruitment, but may also include fidelity to the randomised intervention, completeness of follow up (retention), and cost profiles. This will be from a practical perspective and use real examples of such ‘stop-go’ criteria from publicly funded trials. It will contrast ‘information’ (waiting until a pre-specified level of information is accrued) vs. ‘time’ (information accrued in a fixed time) approaches to stop-go algorithms, and will emphasise the need for careful interpretation of the signals from a pilot/feasibility study to inform progression to full trial. Introduction: Conducting research with an aging population in a longitudinal clinical trial can pose unique challenges, particularly when cognitive decline is the primary endpoint. Unanticipated issues can arise due to use of data gathering methods that may be routinely used with younger research participants, but are less effective when used with older adults. Such differences may lead to false impressions of participant status and incidence of endpoints. The analysis of procedural methods that may disproportionately affect older adults is of critical importance at this time when many clinical trials are being undertaken to investigate interventions aimed at the older adult population. Further, research challenges can be magnified when a clinical trial is undertaken as an ancillary study to another larger trial. The directions of the larger study affect the ancillary study in profound ways that may also lead to dramatic changes in research procedures. The NIA-sponsored PREADViSE study is an ancillary study of the prostate cancer prevention trial, SELECT. The goal of PREADViSE is to examine the effectiveness of the antioxidants Vitamin E and Selenium in preventing Alzheimer ’s disease in older men in the United States, Canada, and Puerto Rico. Objectives: The purposes of this poster are to: 1. Discuss challenges facing an ancillary clinical trial when the parent study undergoes major changes. 2, Elucidate some of the challenges encountered with cognitive evaluations of older men via telephone contact. Such as: a) L ogistics of contacting aging men who are still actively working outside the home b)Obtaining meaningful test results from participants who have sensory deficits c)Recognizing the impact of health issues, including recent surgery, chemotherapy, or serious illness on cognitive measurement d)Accelerated lostto-follow-up, due to death, chronic illness, or placement in a nursing home. Background: In Korea, many stroke patients receive traditional medical care because the country has its own system of traditional alternative medicine called Traditional Korean Medicine (TKM). Observation of the tongue, also known as tongue diagnosis, is an important procedure in diagnosis by inspection in TKM. However, the clinical competence of tongue diagnosis was determined by the experience and knowledge of the clinicians who used tongue diagnosis. Much of the experiences in traditional tongue diagnosis have not been verified scientifically or quantitatively. We investigated the reliability of TKM tongue diagnosis in stroke patients by evaluating interobserver reliability regarding tongue indicators as achieved by TKM practitioners. Methodology/Principal Findings: A total of 658 patients with stroke admitted to 9 oriental medical university hospitals participated in this study between February 2010 and December 2010(Figure 1). Each patient was independently seen by two experts from the same department for an examination of the status of the tongue. Interobserver reliability was measured in three ways: simple percentage agreements, Cohen’s kappa coefficient and Gwet’s AC1 statistic. Interobserver agreement for the tongue indicators among all subjects (n=628) was generally high, ranging from “moderate” to “excellent” (AC1=0.43~0.97), while the interobserver agreement about subjects regarding pattern-identification with the same opinion between the raters (n=451) was also generally high, ranging from “moderate” to “excellent” (AC1=0.5~0.98). Interobserver agreement was nearly perfect for certain signs of special tongue appearance (mirror [AC1=0.95~1], spotted [AC1=0.96~0.98], bluish purple [AC1=0.85~0.95]), poor for one of the tongue colors (pale [AC1=0.32~0.66]) and moderate for others(Table1, Table2). Conclusions: Clinicians displayed measurable agreement regarding tongue indicators via both observation and pattern identification consistency. However, interobserver reliability regarding tongue color and fur quality was relatively low. Therefore, it is necessary to improve objectivity and reproducibility of tongue diagnosis through the development of detail- oriented criteria and enhanced training of clinicians. INTRODUCTION AND OBJECTIVES: Saw palmetto (SP) extracts are widely used by men as phytotherapy for lower urinary tract symptoms (LUTS). Factors associated with perceptions about their urinary function in men who use phytotherapy for their LUTS have not been well described. METHODS: The Complementary and Alternative Medicine for Urologic Symptoms (CAMUS) trial was a randomized, placebo- controlled double blind multi-center trial of 369 men ? 45 years of age with AUA (American Urological Association) symptom index score ? 8 and ? 24 at study entry who were treated with SP at single, double, and triple usual dose or placebo (P). Assessments were made at baseline and at 24, 48 and 72 weeks after randomization. These included questions about how the participant felt about his urinary function. We used multivariate ordinal logistic regression to assess the effects of changes in AUA Symptom Index score, quality of life (QOL), and BPH (benign prostatic hyperplasia) Index Score, age, and randomized treatment assignment on the participants’ perceptions of their urinary function. RESULTS: Improvement of AUA Symptom Index Score, QOL, and BPH Index Score were associated with subjects’ perceptions of better urinary function from baseline to weeks 24, 48 and 72 (Table, change at 24 weeks shown). Older age at enrollment was associated with more negative perceptions about the participants’ urinary function. Randomized treatment assignment was not associated with these perceptions. Conclusion: The main determinants of improved perception of urinary function seen in the CAMUS study were improvements in AUA Symptom Index Score, BPH Index, and QOL. Older age was marginally associated with poor perception after adjusting for these other factors. Treatment with saw palmetto was not associated with better perceptions about their urinary function. Purpose To present an objective and relatively straightforward method of assessing compliance in a randomized clinical trial using eye drops. Background Achieving good compliance is of crucial importance to success of clinical trials. Various methods have been used to assess compliance with eye drops in ophthalmic studies. Subjective methods, such as patient questionnaires or diaries, are prone to bias. Objective methods like digital monitoring devices require additional resources and logistics to implement. Design The topical non-steroidal anti-inflammatory drugs for non-central diabetic macular edema study is a 1-year phase II randomized trial evaluating effects of 3-times daily nepafenac ® 0.01% drops compared with placebo. In an attempt to minimize the number of poorly-compliant participants randomized, a run-in phase for 30-60 days prior to randomization was implemented to assess compliance. At the end of the run-in phase only participants who achieved 80% or more compliance were randomized (provided other eligibility criteria were met). Each study bottle was weighed using a calibrated sensitive scale prior to dispensing to participants, and at each follow-up visit after use. Since the length of the run-in phase will vary, a database of expected bottle weight after each required dose was created. A compliance formula that takes the following into consideration was defined: expected initial bottle weight, expected follow-up weight, observed initial bottle weight, observed follow-up weight. Based on previous reports, 80% was chosen as the compliance threshold below which participants would not be eligible for randomization. This approach has many limitations; for example, weight change may not necessarily reflect the proper dosage or installation timing of drops. Furthermore, short-term compliance may not necessarily translate into long-term compliance. Conclusion Compliance with study drops poses challenges in ophthalmic clinical trials. We present an objective and relatively easy method to evaluate compliance that may help exclude potentially non-compliant participants. The Twin Birth Study (TBS) is an international multicentre randomised controlled trial that recruited 2804 patients from 106 centres in 27 countries. TBS seeks to determine, in women expecting twins, whether a policy of planned Caesarean section decreases the likelihood of perinatal or neonatal mortality or serious neonatal morbidity, compared to a policy of planned vaginal birth. Recruitment for the trial ended in April 2011 and the final analysis is scheduled to be completed by January 2012. In order to prepare for the final analysis and answer this important research question, an intensive process of cleaning the data began. In May 2011, specific processes were implemented to facilitate the data cleaning. All the raw data was extracted from the database and multiple reports were compiled to ensure that all missing answers had been queried. In addition, multiple data reviews were scheduled, to analyse the data by the allocated groups according to the analysis tables. Outliers and missing variables were reviewed, and centres were asked to confirm or correct their responses. To keep centres informed of the status of their data, a “Complete and Clean” report listing the percentage of data forms returned and queries resolved were emailed to each centre. An “Overdue Primary Queries” report was sent monthly to the centres listing the queries that were still outstanding. These strategies led to TBS achieving 99.6% clean data for the final analysis, as of November 30, 2011. This presentation will share data cleaning strategies implemented by TBS to ensure data accuracy for the final analysis. The Centre for Mother, Infant, and Child Research (CMICR) is the coordinating centre for several large, international randomised controlled trials (RCTs). To streamline data collection in the clinical trials, a web-based Electronic Data Capture (EDC) system was acquired and implemented. To facilitate the operational side of clinical trials, CMICR has been developing in-house data management systems, including a clinical trial management system (CTMS), an interactive web response randomisation system (IWRS), and a drug supply management system (DSMS). One of the biggest challenges of leveraging the new EDC technology is having it be able to interchange and integrate data with CTMS, IWRS, DSMS and other technologies being used in the studies. Data integration is vital for large RCTs because of the volume of data generated. The integrated data provide a more comprehensive overview of trial progress, allow a closer monitoring of recruitment and data collection, improve trial analysis and allow for better-informed decisions across the variety of tools and sources of information. From a data-handling perspective, integration reduces redundancies and improves data consistency and quality through less human interaction. This presentation will discuss the design and implementation of the system integration and data exchange between the different management systems. It will describe and demonstrate the use of the “interfacing” approach that allows the flexibility and scalability to incorporate new components into data management systems. The Centre for Mother, Infant, and Child Research (CMICR) is the data and clinical coordinating centre for several multi-centre, international randomised controlled trials (RCTs). CMICR uses a web-based Electronic Data Capture (EDC) system to collect study data on electronic Case Report Forms (eCRFs).The EDC system has a summary screen that displays the status of each eCRF. Each eCRF has a scheduled timeframe for when it should be completed, and those completed outside the timeframe are considered overdue or invalid if completed too early. The EDC summary screen does not display the date when the eCRF was completed. To view the eCRF completion date, the user must go through the time-consuming task of opening each eCRF. The date is then manually entered into a standalone Access database to track the completion status each eCRF. This method can be both labor intensive and error-prone. Therefore, a method to accurately and efficiently track each eCRF completion date was needed. A solution to this problem was to develop a process to automate the transfer of data from the EDC system to the Access database. The process involved downloading the data from the remote EDC system to a local copy of the EDC database, on weekly basis. This local database was then directly linked to the Access database, allowing the data to by automatically transferred. This allowed the eCRF completion date to be automatically entered while also eliminating the potential security risk of directly linking the EDC system to the Access database. The introduction of this new process has eliminated the need to manually enter the date into the Access database. It has increased efficiency at the coordinating centre by ensuring that accurate records of the eCRF completion date are being kept, ultimately allowing for more effective data management. The Lifestyle Interventions and Independence for Elders (LIFE) Study is a Phase 3, multi- center randomized controlled trial (RCT) designed to compare a moderate-intensity physical activity program to a successful aging health education program in 1,600 sedentary older persons, age 70-89 years, across eight field centers. The primary outcome is major mobility disability, defined as the inability to walk 400 meters. Secondary and tertiary outcomes include serious fall injuries, pulmonary events, and cardiovascular events. Given the number of participants and their age range, we expect to collect data on many outcomes throughout the length of the trial. These outcomes require central adjudication. The need for a central tracking and monitoring system initiated the construction of the Outcomes Management Tool (OMT) Adjudication System. While the OMT manages and tracks outcomes, it also supports central adjudication of those outcomes. The Adjudication System is an online system which tracks adjudicator assignments, allows medical records to be reviewed online, allows voting by assigned adjudication committee members, and allows tracking of full committee votes. Administrators at the Data Management and Quality Control (DMAQC) Center also have the ability to assign specific cases to adjudicators on a predetermined date and assign the full committee access to the case, if required. The final adjudication of the LIFE outcomes is made by the central adjudication committee based on satisfaction of diagnostic criteria delineated in adjudication report forms. The electronic access allows adjudicators the freedom to review records online, communicate with administrators, and vote without being tied to a stack of papers. This presentation will highlight the flow and functions of the Adjudication System within the Outcome Management Tool. Management of drug inventory in a multi-site, randomized, double-blinded study is of utmost importance. In many traditional settings, a plentiful amount of drug kits are shipped to each participating site, with the expectation that few drug shipments will be performed during the study. When the study drug(s) has short expiration dates, is in short supply, is expensive to manufacture, or accrual rates are difficult to predict, careful monitoring of sites’ inventory is necessary to avoid study drug shortage or wastage. As the Coordinating Center (CC) of two studies with some or all of these features, we integrated a utility for managing drug inventory into the enrollment module of our electronic data capture (EDC) system. In both studies, a master treatment table (MTT) houses the sequence in which treatments are assigned and a kit inventory table (KIT) houses the kit numbers, their corresponding treatment group and availability status. An EDC user interface allows the CC to request the number of drug kits to be shipped to a site. An email requesting shipment of drug is automatically sent to the drug supplier. As shipments occur, the KIT is updated to indicate that requested kits have been allocated to a particular site and are available for randomization. As randomization occurs, the enrollment module accesses the drug utility tables to provide the appropriate blinded treatment assignment to the enrolling site. An additional feature of the utility for the second study is an automated nightly process in which site inventory levels and expiration dates are monitored and when pre-defined thresholds are crossed, a shipment request is emailed to the drug supplier. The benefits and risks of implementing a complex drug utility within an EDC deployment will be discussed, along with the technical challenges, role of the drug supplier and necessary elements to make it successful. The BID Pilot Study is the test phase of a randomized clinical trial comparing effects of two levels of blood pressure control. Sixteen dialysis units at 5 participating sites will participate. This study is funded jointly by NIH/NIDDK via R01 and the non-profit Dialysis Corporation Inc. Because the study’s protocol requires standardized blood pressure measurements and data acquisition by distributed data entry, trial leadership determined that face to face training was required before patient enrollment could begin. Minimal funding was available. Starting the training session at 11 am near an O’Hare airport saved money on air fare, let most attendees find direct flights, and allowed same-day fly in. Since the coordinators attending training needed to learn standardized blood pressure measurement, ABPM, and calibration of the machines used for both types of measures, staff needed to bring multiple pieces of equipment. Planning training for data entry was challenging. BID Coordinators use the internet to access Oracle forms housed in an Oracle 10g database at the DCC. A location with strong and reliable but inexpensive wireless in the meeting room was required. Each coordinator needed a laptop computer with java installed. The system did not support Apple Macintosh computers. Because several coordinators did not have laptops, we brought loaners from the DCC. The training session, held in October 2011, successfully trained staff in blood pressure measurement, equipment calibration and on- line data entry and data discrepancy resolution. Only two glitches occurred: 1) the team could not find a soda can needed for a calibration procedure and 2) one coordinator brought a laptop running Windows Vista and had to use a DCC staffer’s 10-inch netbook. The training session was judged to be cost- effective and highly successful. Patient enrollment began in November 2011. BID is the pilot for an RCT comparing two levels of blood pressure control. Sixteen dialysis units at five sites began enrollment in November 2011, collecting extensive data on safety, feasibility, and optimal use of four blood pressure estimates (dialysis machine, standardized, home, and ambulatory blood pressure measurement). Efficiently integrating data from disparate sources maximizes data quality and minimizes cost. Patients are identified by ID and a random alpha code, and the DCC is blinded to patient names. Site study coordinators enter blood pressure data, oral medications, dialysis details and other clinical information via Oracle forms to the Oracle 10g database at the DCC. Central MRI personnel similarly enter cardiac MRI results. Extensive edit checks are applied at the data entry level. BID is funded by NIH/NIDDK and Not for Profit Dialysis Clinic, Inc. (DCI), facilitating access to laboratory, dialysis dose, and injectable medication data stored by DCI ID in DCI’s Nashville medical information system DARWIN. As patients enroll, participating sites securely notify the Clinical Coordinating Center (CCC) staff, who then securely provides the DARWIN database team a link between DCI ID and the BID ID and alpha code. Once a month, each BID patient’s new data are securely transmitted to Cleveland for batch loading into the DCC database. Data files from ABPM devices are securely transmitted from participating sites to the CCC, where data are cleaned, formatted, and then securely transmitted to Cleveland for batch loading into the DCC database. These batch loading processes save coordinators time and motion and increase data quality since coordinators do not need to enter data that was previously machine-generated and stored. The DCC emails weekly reports to Steering Committee members and participating sites showing the status of missed visits, missing forms, missing data, and discrepant data. Thus, data acquisition is continually monitored. In accordance with the provisions of the national legislation issued in 2007, the National Monitoring Centre for Clinical Trials - OsSC (http://ricerca-clinica.agenziafarmaco.it/en/node/22 ) supervised by the Italian Medicines Agency (AIFA) has become a legally binding tool for applicants who wish to submit a clinical trial application to the Regulatory Authority and the Ethics Committee. Within 2012 a switch over from paper to electronic submission is expected. A Telematics Implementation Group (TIG) chaired by AIFA has been set up in 2008. In the TIG are represented Ethics Committees, National Regulatory Authorities, sponsors (both commercial and non commercial) and CROs. The TIG has the mandate of defining user requirements with respect to e-submission; validate software development through user acceptance testing; actively participate in the pilot phase of the project; provide support to AIFA initiatives to promote harmonization and guidelines in compliance with the EU legislation. The user requirements and technical specifications of the single portal have been agreed and the software development and testing is on-going (Beta-version). Within 2011 a pilot phase with the participation of the TIG delegates has been completed. Either the initial clinical trial application and the substantial amendment module of the single portal are being extensively tested in terms of functionality, reliability and performance. The new portal is expected to be launched in early 2012 with the ultimate aim of reducing bureaucracy, harmonize procedures and shortening delays in the assessment process of clinical trials with investigational medicinal products. This tool will represent the unique access point for all the parties involved managing clinical trials on medicines in Italy. Purpose: The efficient distribution and tracking of Investigational Medicinal Product (IMP) and Non-Investigational Medicinal Product (NIMP) in clinical trials confers certain legal requirements together with meticulous record- keeping. Manual systems that update a paper record each time a product shipment arrives, is supplied to a study participant, or is returned for reconciliation and destruction are time consuming, prone to errors and expensive. Implementing a system whereby records are maintained electronically and updated automatically can reduce costs whilst improving record-keeping. Methods: We examined state-of-the-art in pharmacy systems and concluded that a barcode-based product tracking system was the most practical solution. Each IMP and NIMP are bar-coded and products are tracked from arrival at the dispensary storage facility, supplied to patients, unused drugs returned with tablet counts and destruction records. The system is integrated into the electronic Clinical Report Form (eCRF) and study web portal from which regimen changes can be made. The system also produces bespoke medication labels, Patient Information Leaflets (PILs), and other paperwork, both in English and other languages to be dispatched to patients. Packages are also bar-coded to facilitate identification. Technicians assemble distribution packs in one work-stream and then check packs in another work-stream to reduce errors. Results: We have built an inexpensive, robust and legally-compliant system of managing supply of medication to study participants. The system is currently being used in one large multi-national multi-centre study with a complex IMP and NIMP regimen. Conclusions: We have produced a semi-automated system for drug distribution that we predict will reduce cost, errors, provide a robust audit trail and improve the quality and efficiency of drug supply in the clinical trial setting. Gynecologic Oncology Group Statistical & Data Center, Buffalo, New York The GOG Statistical and Data Center (SDC) developed the SDC Electronic Data Entry System (SEDES) to capture clinical data electronically via a secure website. Over 800 sites have used SEDES to enter and amend over 500,000 electronic Case Report Forms (eCRFs). One of the foundations of this dynamic database driven application is the use of proprietary Adobe Forms Data Format (FDF) files to programmatically manipulate the Adobe Acroform pdf web-based eCRFs created with Cardiff Teleform software. FDF files can be used to import data into a pdf form. The SDC chose to use static FDF files created and stored on the server to import patient database data into eCRFs stored on the server. These FDF files contain the eCRF’s form field names and values to be imported as well as the web link to the pdf form. A patient registered to a GOG Protocol has a set of study specific expectations assigned and accessible as links in an electronic Patient Form Schedule available via SEDES. Clicking on an eCRF link generates a FDF file on the server for that patient and form type which in turn automatically opens the semi-prefilled eCRF allowing the user to complete and submit the data for processing and storage. The FDF file minimally consists of data needed to prefill the form header with patient study identifier, today’s date and web account username. Upon submission, data are immediately available for use through the same process via a FDF file generated from the submitted data prefilling the eCRF for updating. Determining the cause of persistent eosinophilia in immigrants to the United States can be hampered by costs needed to evaluate suspected parasitic infections. Thus, diagnosing eosinophilia- causing helminth infections by stool examination or serology is often beyond the means of community health clinics that commonly serve immigrant populations. To define the causes of persistent eosinophilia among an immigrant population seen at a single community free health clinic, 41 patients (originally from Central and South America, Africa, Asia and the Middle East) – who arrived in the United States 1-27 years (median 7 years) previously–were found to have an absolute eosinophil count (AEC) >500/uL and were referred to the National Institutes of Health for further testing. Of the 41 referred patients, 33 (80.4%) had positive Strongyloides stercoralis-specific serology. Nine of these 33 (27%) also had schistosomiasis (n=6) or hookworm (n=3) infection. Although there was no statistical difference (p=0.15) in the baseline eosinophil levels between those with strongyloidiasis and those without, serum IgE levels differed dramatically between the two groups (geometric mean levels 74 U/ml vs. 566 IU/ml, p<0.01). Interestingly, 16/33 (48.7%) Strongyloides positive patients had an AEC greater than 1000/uL during screening, whereas only 2/8 (25%) without Strongyloides had eosinophil levels that exceeded 1000/uL. All 33 patients with a definitive parasitologic diagnosis received ivermectin and (when appropriate praziquantel and/or albendazole) treatment; 11 have been evaluated one-year post treatment. Not unexpectedly, there was a dramatic and significant (p<0.01) decrease in AEC following treatment, with all returning to normal levels. IgE levels also fell dramatically following treatment. Thus, in community clinics that provide health care to immigrants well after arrival in the United States, an AEC can be used as a surrogate for stool examination and serology and may be a trigger for empiric treatment when testing is limited by cost. The Food and Drug Administration requires most clinical trials to post results to ClinicalTrials.gov within one year after completion of data collection for the study’s primary endpoint measure. The study’s sponsor is responsible for posting results to ClinicalTrials.gov, however the statistical and data coordinating center at Rho has been employed to help with this process. Novice users may find the system difficult to navigate. Often users are unsure of what information is required or allowable. Additionally, many study teams want to review and approve results before they are posted. In an effort to help both the person posting results and the reviewer(s), Rho created a fill-able Word template. This template provides instruction and structure for the person posting results. The template also informs the user of character limits, possible selections, what information is required versus optional, and guidance on typical entries. For the reviewer(s), the template provides a means to see the results without having to be given login-access to ClinicalTrials.gov. Additionally, the reviewer can see the system limitations and what options are available, which can aide in constructive feedback. Part of posting study results to ClinicalTrials.gov is to post adverse events. Hand-entering adverse events for most trials can be a tedious task. ClinicalTrials.gov allows results to be uploaded in a prescribed XML format. Software developers at Rho created a program that will take a SAS® data table and convert it into an XML file that is compatible with the structure provided by ClinicalTrials.gov. Once study results are released to ClinicalTrials.gov, the website staff review study entries and provide feedback and suggestions. During this process we have modified our tools to better meet those expectations. From the lessons learned in the development of results for ClinicalTrials.gov, our new tools can greatly aid and expedite the process of posting study results. The leadership of most clinical trials, including the Antihypertensive and Lipid- Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), facilitate manuscripts and ancillary studies by investigators within the study. To enable data use by outside investigators, studies sponsored by the National Institutes of Health provide a limited-access public use data set, and ALLHAT has done this. Additionally, ALLHAT policies allowed outside investigators to submit data requests directly, but limited the releases to those data that were already publicly available or not being pursued by its investigators. To facilitate and encourage use of ALLHAT’s dataset by outside investigators, the National Heart, Lung, and Blood Institute has funded a new phase entitled ALLHAT Continuation and Outreach. Its purposes are to: (1) develop processes that assure optimal scientific contributions, build on to- date publications and explore new areas, in part through involving appropriate consultants as leaders of relevant scientific areas; (2) encourage productive interdisciplinary collaboration and mentor early-stage investigators from a diverse set of institutions, especially investigators with no recent history of participating in ALLHAT; and (3) promote awareness of the ALLHAT resources among research and clinical communities. This represents a change in direction for ALLHAT. Previous ALLHAT policies protected the interests of study investigators and reserved most Coordinating Center resources for study investigators. ALLHAT’s new phase will widen participation in paper- writing and ancillary studies to investigators within and outside of ALLHAT. It also represents a new approach among clinical trials. While several epidemiologic studies have successfully adopted such an approach, ALLHAT will be the first clinical trial to do this. Additional details of the rationale for ALLHAT’s new phase will be discussed, as well as details of the new study governance structure and the Engagement Plan, which provides guidance for achieving the above aims. Background: Multiple groups conduct national and international clinical trials in obstetrics. The rarity of the most compelling outcomes, such as maternal or neonatal death and childhood disability, at least in developed countries, has led to the use of less compelling outcomes and/or lower power. Meta-analyses are often employed in this field but these are challenging when different definitions of outcomes are used or different data fields collected. In addition, trials have been unwittingly duplicated, leading to a poor use of global resources. For example, seven large overlapping trials of antioxidants to prevent preeclampsia were conducted — all were negative. The process of setting up an international collaboration to address these issues is described. Methods: In 2010, investigators representing different study groups met for the first time, and formed a core group who defined the goals of the collaboration. These include developing a database of ongoing and planned trials, harmonizing obstetrical terms and definitions, defining common endpoints, coordinating study protocols to facilitate meta-analyses, setting up an education program on trial design and performance, establishing priority areas for future trials, and obtaining funding to support all activities including international collaborative trials. All clinical researchers are invited to participate, especially researchers from resource poor countries. Results: GONet has formalized its mission, charter and structure, and held a formal launch meeting. The Board includes representation from Europe, North America, Australia and Asia. A members’ website has been developed (www.globalobstetricnetwork.org), including a database of ongoing/planned trials. A course on clinical trial design has been organized. A meeting with representatives from international funding agencies was held; a funding proposal is underway. Priorities for future trials will be decided at the second annual general meeting. Conclusion: Progress has been made towards global collaboration and more efficient use of resources in obstetric trials. Interpreting results of safety reporting in substance use disorder (SUD) clinical trials is difficult when the baseline rate of serious events like deaths in this population is not well described. Particularly for Data and Safety Monitoring Board reviews, defining a threshold event rate could provide guidance to recognize increased risk in a given clinical trial. Often, the reported number of deaths in any given trial is too small to characterize this issue. National death rates by age and gender are publically available and provide a baseline. Since 1999, the National Institute on Drug Abuse, National Drug Abuse Treatment Clinical Trial Network (CTN) has posted 21 SUD pharmacologic and/or psychosocial intervention clinical trials spanning various SUD populations to the CTN public data share. Protocols and safety data from these completed trials were reviewed. A total of 61 deaths (0.6%) were identified from adverse event reports across 9,396 enrolled participants (5,543 male, 3,844 female, 9 unknown). The overall death rate in the CTN SUD clinical trials was 14.4 deaths per 1,000 person years (PY), and was numerically higher in females (17.1 per 1,000 PY) than males (12.1 per 1,000 PY). By age, the highest death rate was observed in the 25-34 year olds (17.6 per 1,000 PY). These death rates are numerically higher when compared with age- and gender- specific rates to the national death rate information and likely represent underlying social issues surrounding SUD populations, particularly in women, and likely are not attributable to clinical trial participation. Estimating death rates within types of substance use disorders may further contribute to the understanding of underlying risks in SUD populations Progression- free survival (PFS) is a composite endpoint incorporating the time to either death or progression, with varying definitions of progression. Time to death is a continuous variable where death occurs at random times, and the date of death is known specifically. Time to progression is, strictly speaking, an interval measure, as the date of progression is unknown so the time to progression is not precisely known. When the assessment of progression is made, a patient is classified as having progressed or not having progressed, but the time of progression is unknown. In our experience in oncology, the most common statistical methods for PFS include a graphical display of a Kaplan-Meier curve and analysis using a logrank test or a Wilcoxon test with some type of accommodation for censoring (usually). When PFS in the particular dataset is composed mostly of time to death (as a continuous measure) and time to progression (not measured continuously), these analysis methods perform well. When, however, PFS is dominated by time to progression, and when the assessment times are infrequent, the Kaplan-Meier graph resembles a step function, and these analysis methods may be far from optimal. A life-table extension of the Mantel-Haenszel test may perform better, and Cutler-Ederer estimates for interval data may describe the data more accurately. This paper presents a discussion of the appropriateness of the analytical methods for PFS with examples of comparisons from PFS datasets. Objective The development of biological valve prostheses with lifetime native-like performance and optimal host engraftment is a goal of heart valve tissue-engineering. We describe a new concept for autologous graft coating based on a CD133+-stem-cells-plus-fibrin-complex (SC+F) processed from bone marrow and peripheral blood of one and the same patient. Methods CD133+-SC (1x106cells/ml) from human bone marrow and autologous fibrin (20mg/ml) were administered simultaneously via spray administration. During static cultivation, SC+F performance was monitored about 20 days after delivery and compared to controls. For dynamic testing SC+F-composite was sprayed on a decellularized porcine pulmonary valve and transferred to a bioreactor under pulsatile flow conditions for 7 days. Quantitative analysis of the data is given for experimental group containing fibrin and control group without fibrin, respectively. Because measurements were made several times within three independent groups we applied the GLM Repeated Measures procedure for statistical analysis to test the null hypotheses about the effects of both the between-subject factor (group) and the within-subject factor (time). For the endothelial differentiation parameters during endothelial colony forming assay differences between day 0 and day 28 were investigated, using paired t test or Wilcoxon’s rank test, as appropriate. Test selection was based on evaluation of differences for normal distribution using the Shapiro-Wilk test. Results Static cultivation of SC+F-composite induced significant improvements in stem cell proliferation as compared to controls. For dynamic testing, microscopic analyses on a smooth engineered heart valve surface detected homogenous distribution of stem cells. Ultrasonic analysis executed native-like valve performance. Applied CD133+ stem cells differentiated into endothelial-like cells positive for CD31 and VEGF Receptor 2 and engrafted the valve. However, occasional delamination was observed. Conclusion SC+F serves as an excellent autologous matrix for intra-operative tissue-engineering of valve protheses promising optimal in-vivo integration. However, stability remains an issue. References: Kaminski et al. (2011) In clinical trials, paired binary data often arise from crossover study design or pre-test/post- test comparisons. Missing data due to drop- outs and other reasons may lead to incomplete paired binary data for a subgroup of subjects. Interval estimation for the proportion difference can be problematic in these situations. In this article we propose an extension of the method of variance of estimates recovery (MOVER) to construct confidence intervals (CIs) for the correlated proportion difference based on paired and unpaired data. Two sets of CI estimators, one based on paired data, the other based on pooled paired and unpaired data, are utilized in the double-MOVER procedure to construct the asymptotic CI. Extensive simulations show that the double-MOVER estimator performs well under various degrees of missingness and correlations, even with small to moderate sample sizes. Two real examples of clinical studies are used to demonstrate the proposed method. Isotonic regression is a useful tool to investigate the relationship between a continuous covariate and a time-to-event outcome. The resulting non-parametric model is a monotonic step function and the steps can be viewed as change points. However, when there are too many steps, over-fitting can occur and further reduction is desirable. Here we propose adaptive partitioning to allow combination of small steps which do not differ greatly. In this approach, a second step, the reducing step, is integrated into the usual monotonic step building by appropriate statistical testing of the adjacent steps. Adjacent steps not significantly different at a pre-specified alpha level are combined. We achieve this through a modified dynamic programming algorithm. Two popular parametric survival distributions, exponential and its extension Weibull, are implemented first. Then we explored a more robust piece- wise exponential distribution, which is less stringent in model assumptions and the model can potentially be used in interval censored survival data. We apply this methodology to the Diabetes Control and Complication Trial (DCCT) data set to study the relationship between HbA1C and the time to a severe hypoglycemia event. All three approaches are used and the results compared. The models are very similar to each other and all suggest a negative non- linear association between HbA1C and time to severe hypoglycemia. When the alpha level is set at a modest level (for example, 0.05), HbA1C at 6.2, 7.3 and 9.6 are identified as the potential change points in its association with severe hypoglycemia. Sometimes one picture is worth one thousand words, especially in the scientific community. Coding and managing SAS Graphics is difficult even for experienced SAS programmers. It is common to use other software to visually represent data even though the analyses were performed using SAS. Enterprise Guide, a newer facility within SAS, has emerged as a remedy to these problems and creates automated and/or semi-automated graphical displays via a point and click user interface with minimal or no coding required on the user’s part. More importantly, the produced graphics are of high quality and SAS statements that produced the graphs are automatically generated. This added flexibility allows the user to modify the code if needed or to incorporate it into existing SAS programs. This presentation will first investigate traditional SAS coding used to generate graphics. Second, step-by-step instructions will be provided to obtain the same (and even better) graphics by using Enterprise Guide’s built-in graph facility. Finally, use of the Enterprise Guide applications in the University of Iowa Clinical Trials Statistical and Data Management Center to accelerate producing graphics will be discussed. Cost-effectiveness is an important method for gauging the impact of one health care strategy over another. Cost-effectiveness is defined by the incremental cost-effectiveness ratio (ICER), that is, the difference in cost of the two strategies divided by the difference in the effect of interest. Because the ICER is a ratio of two random variables, its sampling distribution cannot be easily derived, and quantification of the uncertainty in ICER estimates can be problematic. Recommended procedures for calculation of a confidence interval on the ICER can only be applied in the case where there is a significant difference in both cost and effect. The bootstrap method of sampling with replacement is a relatively straightforward and popular approach for estimating uncertainty in the ICER, but it has been criticized for giving overly narrow confidence regions and involving many unstated assumptions. The purpose of this presentation is to explore methods for calculating and displaying the variability of the ICER using data from a recently completed clinical trial conducted by the Pediatric Eye Disease Investigator Group (PEDIG) to illustrate the methods. Background: The primary efficacy study for a vaccine to prevent herpes zoster (HZ) and its complications followed subjects for a median of 3.1 years post-vaccination and analyses of vaccine durability showed that the vaccine was effective for four-years post-vaccination. Of interest is the assessment of how long the zoster vaccine remains effective against HZ. Extended follow-up of study subjects was carried out in one substudy (3.3 to 7.8 years post-vaccination), but once placebo subjects were vaccinated, there was no longer a concurrent control group for direct estimation of vaccine efficacy. Objectives: A method for determining historical control rates for long-term follow-up study was needed in order to provide estimates of long-term efficacy for the three study outcomes: burden of illness due to HZ associated pain or discomfort; incidence of HZ; and incidence of postherpetic neuralgia (PHN). Methods: The extended follow-up protocol initially planned to use historical controls adjusted for increasing age of the study population. However, the final analysis of the primary efficacy study showed that the incidence of HZ in placebo recipients increased over the study duration even after the adjustment for increasing age. Poisson regression methodology was used to develop and select models to estimate the age and calendar time effects. Age and calendar time adjusted controls were then calculated by multiplying the model-based estimates times the number of person-years of follow-up observed for each age year in the long-term follow-up of vaccinated subjects (4.7 to 11.6 years post-vaccination). Results: Models selected for calculating the historical controls for the incidence of HZ included effects for both age and study (calendar) time, while models for outcomes related to severity of HZ only included age. Methods for assessing age and calendar time effects, for model selection, for the calculation of controls and for calculating vaccine effectiveness will be presented. The Environmental Polymorphisms Registry (EPR) of the National Institute of Environmental Health Sciences (NIEHS) was established as a “recruit-by-genotype” resource which houses a databank of DNA linked to participants who may be recontacted for phenotype information. The EPR is designed to screen for functionally significant polymorphisms by identifying individuals with shared genotypes at candidate genes and then recontacting them for phenotype information that can be used for genetic association testing. We show that sufficient power is obtained by equally sampling few numbers of individuals per genotype group. For example, assuming a 25% change in TNF-alpha release between each genotype group (and a standard deviation of 200 units), we calculated 89% power by including a total of 12 subjects (4 per genotype group). In contrast, if we randomly sampled subjects and measured both their continuous phenotypes and genotypes using the same distributional assumptions and mode-of-inheritance, a rare susceptibility allele of 2% would require 166 subjects to achieve similar power, a substantial increase in the number of subjects as compared to the EPR approach. This suggests that contrasting equal numbers of homozygotes dramatically improves power for less common alleles as opposed to sampling genotypes at random. We note that there is no need to include heterozygotes if investigators seek only to test the presence of an effect, rather than including this group to distinguish among inheritance models. When inclusion/exclusion criteria are considered during protocol development for a clinical trial, previous study experience is routinely considered exclusionary only if a previous drug or intervention is likely to have a direct effect on the intervention under study. However, patients who have participated previously in a research study may respond differently to an intervention compared to new participants, even when the interventions are not similar. We examined data from the National Cooperative Inner-City Asthma Study (NCICAS) and the Inner-City Anti-IgE Therapy for Asthma (ICATA) Study to assess the impact of previous research experience on the response to the intervention. The NCICAS Phase II study enrolled 1033 participants, 457 (44%) of whom were recruited from NCICAS Phase I, a 12-month observational study designed to identify risk factors for asthma morbidity. The ICATA Study enrolled 419 participants, 39 (9%) of whom were previously enrolled in the Asthma Control Evaluation (ACE) Study, a trial that used measurement of exhaled nitric oxide as an adjunct to clinical care in the intervention group. In both studies, the intervention was more effective in the study-naïve population compared to the study-experienced population. In NCICAS II, there was a strong intervention effect for the study-naïve population and the experience-by-intervention interaction was statistically significant. The results of ICATA were consistent with NCICAS and showed a significant decrease in asthma symptom days for the study-naïve participants in contrast to study-experienced participants, but the interaction term was non-significant. In both NCICAS and ICATA, study-naïve participants responded more favorably to the intervention than study-experienced participants. These study-experienced participants began the second study with a lower level of symptoms, which may reflect better asthma management learned during the previous study. Researchers should consider the effect of prior study experience even when it is not directly related to the current intervention. Background: Imaging in clinical trials has become more important than ever to help predict treatment outcomes, monitor disease progress and document treatment effectiveness. With the ever-present need for shorter, more cost effective trials quality assurance in imaging is an essential component of ensuring trial success. It is generally recognized that the implementation of digital imaging systems and cloud computing may be improve remote imaging methods, but data is limited. Purpose: To report on remote monitoring techniques use to ensure protocol adherence and high image quality in a study where quantification of small peripheral retinal changes was the key endpoint. Methods: This trial is an ancillary imaging study to a large multi-center randomized interventional trial for Age-related Macular Degeneration. Images were obtained using a widefield imaging system in which peripheral changes were to be observed and quantified. The primary aim of monitoring was to ensure each site was obtaining full field images taken in the appropriate sequence for evaluation. Images were submitted remotely to the reading center for evaluation and mirrored to the sponsor for monitoring purposes. Images were reviewed by the sponsor on a weekly basis and feedback to the site occurred immediately. If issues arose with image quality a monitor was dispatched to the site for a traditional site visit. Results: Trial is ongoing however the average amount of retakes requested has been reduced by 30% with a reduction in site visits by 50%. Prior trials relied on on-site monitoring visits in 8 week intervals. Monitors would review study image databases and request reimaging as well as re-train imagers if necessary. This often led to delays in assessing quality issues and required more recall visits for reimaging. Monitoring in this trial allowed for more swift feedback to be given to sites thus reducing reimaging requests and technical failures. Introduction: Excessive protocol violations (PVs), defined as preventable mistakes in study conduct, may result in patient harm and may dilute statistical power. PVs are more likely to occur early during trial conduct, while research staff are still ‘learning’ the trial protocol. Incorporation of interactive workshops into start-up meetings addresses the needs of research staff as adult learners and may lead to improved study conduct. Purpose: To evaluate the attendees responses towards interactive workshops as part of clinical trial start-up meetings. Methods: In 2010 we commenced two novel multi- centre clinical trials (Study A and Study B) at 29 sites across Australia and New Zealand. We held five sequential start-up meetings over seven months, each attended by research staff from three to five study sites. Didactic lectures were followed by interactive group workshops to reinforce new skills and study processes. Both Study A and Study B introduced new study processes however only Study B taught a ‘new skill’, which involved a specific technique for measuring QT intervals. At the final start-up meeting for each study, attendees evaluated the interactive workshops. Results: Study A and Study B: New Study Processes When asked whether the interactive workshops were useful for reinforcing the learning of new study processes, 16/16 (100%) participants responded YES. Study B: New Skills When asked whether the interactive workshops improved confidence with new skills, 8/8 (100%) participants responded YES. When asked whether the interactive workshops should be held in future meetings if new skills were being taught, 10/10 (100%) participants responded YES. Conclusions: Interactive workshops provide a protected learning environment for research staff learn new trial protocols prior to enrolling their first patient. Attendees reported improved confidence in the mastery of new skills, which may translate to improved study conduct. Additional research is required in this field. After clinical trials end, continued follow-up of the assembled cohort is often pursued for additional research. Factors influencing participants’ decisions to consent to additional follow-up and how these shape post-trial cohorts have not been broadly studied. The WHI Hormone Therapy (WHI HT) clinical trials were designed to assess the impact of two regimens of postmenopausal hormone therapies compared to placebo. Postmenopausal women, 50-79 years of age at initial screening, were eligible for participation. We examine how two re- enrollment campaigns, occurring in 2004-2005 and 2009-2010, and the passage of time altered broad features of the post-trial cohorts compared with the original WHI cohort, which was recruited in 1993-1998. Associations that markers of socio- demography, health, lifestyle and on-trial experiences had with re-enrollment were examined and the characteristics of successive post-trial cohorts were contrasted with those of the original enrollees. The post-trial enrollment campaigns re- enrolled 81.1% and 82.5% of available women, respectively. Women who re- enrolled tended to have better health characteristics than those not re-enrolled. Compared to women of comparable age in the original cohort, women retained for the second post-trial follow-up were less likely to have a history of cardiovascular disease [odds ratio=0.36: 95% confidence interval: 0.32,0.41], hypertension [0.57: 0.54,0.61], diabetes [0.59: 0.54,0.61], or measured cognitive deficit [0.40: 0.26,0.64]. These women were more likely to have graduated from high school [1.72: 1.54,1.92] and to have participated in other WHI trials [1.76: 1.66,1.87]. Although we have examined a single study and cannot clearly generalize how our findings might apply to other cohorts and protocols, these methods may be used to estimate re- enrollment in other studies. Post-trial enrollment can be successful, however the characteristics of the resulting cohort may differ substantially from the originally assembled cohort and it may be important to collect predictors of differential re-enrollment during the original trial to facilitate re-enrollment. The Department of Veterans Affairs Cooperative Studies Program (CSP) Clinical Research Pharmacy Coordinating Center (Center) is responsible for clinical supplies used in multicenter clinical trials conducted by the CSP. Clinical supplies lay at the core of drug and device trials. They must meet stringent quality standards to ensure patient safety. The integrity of the clinical supplies and the controls on their processing must provide a guarantee that the results of the trial, whether positive or negative, cannot be questioned. The Center integrates quality regulations, standards and guidelines into all their processes to ensure the integrity of not only the clinical supplies, but all their processes. These regulations, standards and guidelines include: • current Good Manufacturing Practice Regulation 21 CFR 210 and 211 for drugs, • Quality System Regulation, 21CFR Part 820 for medical devices, • Quality Management System Standard, ISO 9001:2008, • Primary Packaging Materials for Medicinal Products Standard, ISO 15378:2007, • Organizational Quality processes, Malcolm Baldrige Criteria, and • ICH guidelines for Stability testing and other processes. This poster demonstrates how quality standards and guidelines apply to all processes associated with the clinical supply chain including the typical cGMP regulated product development, manufacturing, laboratory testing, packaging, labeling, distribution, corrective and preventative actions, training and document controls. But equally important, in addition to using regulations, standards, and guidelines to control the supply chain processes, they also support administrative processes including management responsibilities, planning and design, pharmaceutical project management, budget and finance, treatment assignment, continuous improvement, document control, and study closeout. The organizational model for the Center’s quality management system melds industry standards, quality models and federal regulations into an effective infrastructure, resulting in a high-performing organization that designs and controls all processes to assure the successful conduct of large and small scale clinical trials from design to closeout. The Centre for Mother, Infant, and Child Research (CMICR) designs and conducts investigator-initiated academic multi-centre randomized controlled trials and disseminates the findings in order to improve clinical practice and the health outcomes of women and their children. Currently, CMICR is managing six international and national multi- centre trials. To ensure the success of these studies, the acquisition of a sufficient number of collaborative centres is critical. CMICR relies on these collaborative centres to recruit patients and collect data concerning these patients, which will be analyzed and reported at the conclusion of the studies. To join a study, potential centres send several documents to CMICR which demonstrate their ability to participate. Among these documents is a contract called the Clinical Study Agreement (CSA/CTA). Acting on behalf of each Sponsor-Investigator, CMICR prepares a CSA in the form of a standard template. This template is sent to each potential centre, after which the approval process begins with a legal negotiation and ends with the execution of the agreement. In some negotiations, no revisions are requested and the template is accepted, however, most negotiations require a series of revisions prior to execution. In these cases, efficiency becomes vital because delays can be costly and impact recruitment. The handling of one agreement for one study is straightforward, but when multiple studies and agreements have to be managed, the process becomes complex. To facilitate the process, CMICR 1) prepared a flow chart for guidance, 2) created a filing system for record keeping, and 3) kept a detailed spreadsheet to log and document the steps. This presentation will illustrate an efficient way of handling and monitoring several Clinical Study Agreements for the respective CMICR trials, as centres move from Potential status to becoming Active collaborative centres in a process that is sometimes convoluted. Sunnybrook Health Sciences Centre is the sponsoring institution and lead site for several multi-centred clinical trials. As such, investigator-initiated trials are conducted by clinical research teams generally comprised of the Sponsor-Investigator, Site Principal Investigator (PI), and Research Coordinator. As a policy at Sunnybrook, the initial approach for participation in a trial must be made through a member of a patient’s circle of care. Because of their access to the patients, their knowledge of the question and commitment to the study, Sponsor- Investigators that are clinicians are most suitable for the approach phase of the recruitment process. However, due to ethical standards, their exclusion from the consent process during recruitment is imperative. The requirements for autonomy and informed consent in research are competence, information, understanding and voluntariness. Due to factors of trust and dependency in a physician-patient relationship, the imposition of undue influence is commonly an area of ethical concern. To avoid this, the inclusion of a Site PI and Research Coordinator to the research team has become vital. As a task delegated by the Sponsor- Investigator, the Site PI assumes responsibility for the oversight of the consent process, which establishes a chain of command that results in the Coordinator reporting consent to the Site PI instead of the Sponsor-Investigator. This order alters the physician-patient relationship and eliminates any obligation the Coordinator may have to the Sponsor-Investigator, allowing an unbiased approach to potential participants. The implementation of these precautionary measures assists in building trust between the Coordinator and each participant, which leads to increased study compliance and decreased rates of attrition. Furthermore, they allow the Sponsor-Investigator to continuously drive the study with enthusiasm and commitment without ethical compromise during the consent process. The National Lung Screening Trial (NLST) is a randomized controlled trial, funded by the National Cancer Institute, to determine whether screening with low-dose helical computed tomography reduces lung cancer mortality relative to screening with conventional chest x- ray in persons at elevated risk of lung cancer. It was comprised of the Lung Screening Study (LSS) and the American College of Radiology Imaging Network (ACRIN). Lung cancer diagnosis, a critical study outcome, relied on data abstracted from medical records. The Coordinating Center (CC) for the LSS was responsible for training and monitoring abstraction at 10 sites. A medical record abstractor (MRA) training program was developed and implemented by the CC to include centralized initial training, refresher training at annual meetings, quarterly conference calls and individual training as needed. Training sessions focused on comprehension of specifications and proper completion of MRA forms, and emphasized the responsibility of the abstractor in collecting quality data. Effectiveness of training was assessed during the on-going QA process: central re-abstraction of all primary lung cancer cases and a random sample of positive screens without a lung cancer diagnosis. Over 2000 sets of medical records were reviewed. MRA QA reports monitoring 13 key data elements were compiled quarterly, depicting the number and type of errors. Findings from the QA reports changed during the study. Overall adherence to abstraction specifications and reporting of diagnostic procedures improved over time. In studies requiring medical record abstraction, it is vital to develop thorough standardized training approaches and to engage in continual abstraction training and monitoring. The use of quarterly QA reports enabled us to focus training efforts on problem areas and to identify individual MRAs and/or sites requiring additional training. We will demonstrate how our on-going training program was an integral part of the MRA process. The CHIPS Trial (Control of Hypertension in Pregnancy Study) is an international multi-centre randomised controlled trial recruiting 1028 women from 100 centres internationally. CHIPS aims to determine whether ‘less tight’ or ‘tight’ control of non-proteinuric hypertension will decrease the likelihood of pregnancy loss or ‘high level’ neonatal care, without increasing serious complications for the mother. A committee of clinicians (representing obstetrics, medicine, and neonatology), masked to treatment, convenes quarterly to review perinatal and maternal outcomes and adjudicate whether or not the outcomes have occurred. As the Trial expects approximately 30% of babies and 2% of mothers to have a serious complications, an efficient adjudication preparation process is required. To accurately adjudicate the outcomes, it is necessary to review all reported cases with fetal, neonatal, and/or maternal complications to ensure that case report forms (CRF) have been accurately completed, and the hospital documents received. These cases are assessed in a preliminary review to confirm they meet the relevant trial outcome definition. The following are resolved prior to adjudication: CRF inaccuracies, missing hospital documents, discrepancies between the CRFs and hospital records, and translation of international hospital documents. At the adjudication meeting, all data forms for each case are reviewed along with the corresponding hospital records. Due to the high volume of outcomes and various steps required for preparation, a Microsoft Access tracking system was developed. Cases requiring adjudication are added to the tracking system, and issues requiring clarifications are recorded. Lists are generated to identify outstanding issues, which are resolved by corresponding with the centre, and subsequently tracked to monitor case progess. Once all issues have been resolved, case identifiers are blocked, an identifying code is assigned, and cases are mailed to the adjudication committee members. The tracking system allows for systematic collection of information resulting in an efficient adjudication preparation process. Treatment Options for type 2 Diabetes in Adolescents and Youth (TODAY) was a multi- site randomized clinical trial comparing three treatment arms on time to treatment failure, funded by the National Institute of Diabetes and Digestive and Kidney Disease (NIDDK) of the National Institutes of Health. Participants were randomized between the ages of 10-17 and diagnosed with type 2 diabetes (T2D) less than 2 years at baseline. As the first large national study of this vulnerable population of youth diagnosed with T2D, it was considered important to continue to follow this unique, well-characterized, ethnically diverse cohort in order to understand the persistence of effects of the treatment regimens used in TODAY and the development of vascular complications. The youth are now participating in a post- intervention, prospective follow-up study called TODAY2. While other studies have undergone the transition from a clinical trial of blinded experimental intervention regimens to a post- intervention follow-up study, TODAY provides insights into procedures and strategies specific to this cohort. We will describe our experience and impart ‘lessons learned’, which have resulted in several important study-wide changes intended to increase compliance during the follow-up period. We encountered challenges due to major life changes occurring in participants shifting from adolescence to young adulthood. Further, we were providing standard practice patient care and management in TODAY2, which meant a switch from blinded pill packets to open-label bottled pills and from an over-encapsulated pill size (described as a ‘horse pill’) to a smaller tablet. These changes had ramifications for encouraging and tracking study drug adherence at the start of TODAY2. We make recommendations and suggestions for other study groups facing a similar situation. The Blood and Marrow Transplant Clinical Trials Network (BMT CTN), sponsored by the NHLBI and NCI, was established in 2001 to conduct multi-institutional clinical trials in hematopoietic stem cell transplantation (HSCT). The Data and Coordinating Center (DCC) is responsible for data collection, review, monitoring, and reporting of adverse events (AE). Significant regimen-related adverse events are anticipated in this patient population that warrants a systematic approach to reporting adverse events. The BMT CTN utilizes the AdvantageEDCSM electronic data management system to capture unexpected grades 3-5 AEs in six case report forms (CRFs): AE1-Initial Report, AE2-Summary, AE3-Therapy/Concomitant Medications, AE4-Laboratory/Diagnostics, AE5- PI Review, and AE6-Medical Monitor Review. AEs are graded according to the National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE). Grades 3-5 AEs not listed in the protocol, informed consent, or drug label are determined to be “Unexpected” and are reported expeditiously regardless of attribution. Expected adverse events including protocol- specific toxicities are captured on calendar and event driven CRFs. Examples of expected data collected on these CRFs include infectious complications, graft-versus-host disease, relapse, readmission, graft failure and other protocol-specific events. AdvantageEDCSM generates an automated e- mail notification when an unexpected grade 3- 5 AE has been entered. The AE coordinator notifies the DCC Medical Monitor who makes a determination of the expectedness and grade. The AE coordinator is responsible for managing the process including queries to the transplant center and filing reports to the NHLBI Project Officer and the FDA in compliance with reporting timeframes. The BMT CTN AE process was developed to manage unexpected grades 3-5 AEs without burdening the system with expected events that are common among HSCT clinical trials. The AE process is coordinated by the DCC and requires constant vigilance to ensure events are reported, processed, and reviewed expeditiously with the objective of protecting the safety of study participants. The need to independently develop infrastructure for each trial is one of the impediments to efficiently implementing phase 2 trials. To improve efficiencies, the National Institute of Neurological Disorders and Stroke (NINDS) recently initiated the Network for Excellence in Neuroscience Clinical Trials (NeuroNEXT). The goals of NeuroNEXT are to lower barriers for individuals to bring good ideas forward in adult and pediatric neurological disorders, and to provide a flexible and accessible infrastructure to facilitate the development and deployment of phase 2 clinical trials and biomarker studies. To accomplish this goal, a Data Coordinating Center (DCC) is funded at the University of Iowa and a Clinical Coordinating Center (CCC) at Massachusetts General Hospital (MGH) and 25 clinical sites are funded throughout the United States. The first year of the project is aimed at establishing Standard Operating Procedures between the DCC and the CCC, Master Trial Agreements with each site to facilitate contractual requirements, and Reliance Agreements between the Central Institutional Review Board (IRB) at MGH and the IRBs at each site to facilitate IRB review. In addition, there are Protocol Working Groups (PWGs) that provide CCC and DCC assistance to academic, industry and advocacy partners in order to design and finalize robust protocols and grant applications. Data entry systems are being developed using the NINDS Common Data Elements to harmonize data sets at the end of each trial. The NeuroNEXT will develop and implement at least 7 clinical trials over the next 7 years, with at least one trial expected to start within the first year. Metrics on site performance, protocol development, and success of implementation are being collected to examine efficiencies and identify barriers. This innovative approach to improving efficiencies in phase 2 clinical trials may lead to more rapid development of therapies for adults and children with neurological disorders. The concept of serious unexpected events (SUEs) is rooted in drug research in which unintended effects of medication (e.g., skin rash) must be identified. Although routine reporting of SUEs is an essential part of conducting a clinical trial, how to define a SUE in a pragmatic trial of high risk women is problematic. The CHIPS Trial (Control of Hypertension In Pregnancy Study) is a multicentre randomised controlled trial recruiting 1028 women in 100 centres internationally (2008-14). CHIPS aims to determine whether ‘less tight’ control [target diastolic blood pressure (dBP) 100mmHg] or ‘tight’ control [target dBP 85mmHg] of non-proteinuric hypertension in pregnancy is better for the baby (primary outcome: pregnancy loss or high level neonatal care for > 48hr) without increasing risk to the mother. Women eligible for CHIPS are high risk from perinatal and maternal perspectives, such as serious neonatal complications (16%), admission to high level neonatal care (30%), pre-eclampsia (35%), or serious maternal complications (2-3%). We needed to further develop our method of SUE review to prevent anticipated outcomes from being sent to our Data Safety Monitoring Board (DSMB) for review as SUEs. In CHIPS, reported SUEs are reviewed by the Working Group (WG) to determine both the urgency of further review and what supporting documents are required. The Outcomes Adjudication Committee (OAC) reviews each SUE to determine whether the event is a pre- specified trial outcome or a ‘true’ SUE. This decision is reviewed by the Steering Committee (SC). A summary report is submitted to the DSMB at the interim and final analyses or immediately, as decided by the WG, OAC, and/or SC. To date in CHIPS, three SUEs have been reported, all of which were determined to be pre-specified outcomes and will be reviewed by the DSMB at the time of the first interim analysis. The T1DGC was an international effort to identify genes that determine risk of type 1 diabetes. The Consortium created a resource of 2,836 affected sibling pair families as well as 493 trios, 830 cases and 968 controls worldwide. Data and sample collection was conducted in four networks (Asia-Pacific, European, North American and United Kingdom) from January 2004 - January 2010, with support from a Coordinating Center for various activities. While implementing standardized data collection for this international study presented many challenges, study closure proved to be as formidable. Disassembling this complex infrastructure (composed of 214 clinics in 34 countries, four Network Centers and 17 laboratories) while preparing all data and samples for transfer to the NIDDK Central Repository proved as difficult and required as much effort as initiating the study. At the time that we started study closure activities, there were no established guidelines or experience to direct our actions or identify potential difficulties. Study close out was staggered, beginning in October 2006 and concluding in December 2011. We developed a clinic close-out data entry form, delineating key elements required for closure. Additionally, comprehensive checklists of closure tasks for laboratories and Network Centers were developed and utilized. Finally, the Coordinating Center held a meeting with the NIDDK Central Repository to outline the data, samples and study documentation to be deposited. By clearly defining close-out tasks, providing timelines, and performing close-out site visits and/or conference calls, staff at each facility can understand their responsibilities and ensure that all items are completed in this lengthy and complex process. Implementing these steps permits an orderly and efficient study closure and should be undertaken as soon as data collection has been initiated. We will present our procedures and solutions to problems encountered to assist other studies in the close-out process. Clinical Trials of Investigational Medicinal Products (CTIMPs) are regulated in the UK via the Medicines for Human Use (Clinical Trials) Regulations. This is implemented via a) Research Ethics Committee (REC) review, b) Research Management & Governance review by the NIHR Clinical Research Networks followed by c) local Trust R&D department review, and d) Competent Authority (CA) review. Non-CTIMPs are approved via the same routes (excluding CA review). However, these studies are not governed by the Clinical Trial Regulations; instead they are conducted in line with the UK Department of Health’s Research Governance Framework, and in accordance with the Medical Research Council’s Good Clinical Practice guidelines. Robust processes to ensure ethical and governance compliance are implemented but we have observed that low-risk non-CTIMP trials are subject to high levels of scrutiny by RECs and R&D departments, which is not necessarily commensurate with the risk to participants. This can detract from addressing key logistical issues associated with trial implementation (e.g. clinician training, optimising complex data collection processes from many sources), and can considerably extend the set-up phase of such trials, delay the start of recruitment, and even discourage site and patient participation. It is apparent that the measurement of risk in non-CTIMP trials needs to include the risk to the participant per se, but also the complexity of trial conduct (which impacts on participant risk if processes for implementing the trial are not thoroughly planned). We will discuss the risk to the participant and the complexity of trial conduct identified in 3 complex intervention trials co-ordinated by the CTRU, and present the impact this had on the set-up and implementation of each trial. We will present a risk-based approach for review and implementation of complex intervention trials, based on the type of intervention, participant, patient pathways and outcome measures to be used. Response rates of patient reported outcome measures administered postally influence robustness of trial results. We aim to present ways in which postal response rates can be maximised in pragmatic trials using experience gained in the implementation of postal follow-up process for two large multi- centre randomised controlled trials within stroke - Training Caregivers after Stroke (TRACS) and Longer Term Stroke Care (LoTS care) co-ordinated by the University of Leeds’ Clinical Trials Research Unit. Both trials obtain their primary outcome measures via patient self-completed measures administered postally. Follow-up procedures include a mixture of postal and telephone reminders, typically undertaken at two weekly intervals. Response rates are carefully monitored and, where necessary, amendments to the process are made and accounted for in later trials. For example in TRACS, following initial monitoring of data to enhance completeness, the team instigated the collection of the primary endpoint via a telephone interview for patients who did not return questionnaires despite earlier standard postal and telephone reminders. This process was adopted much earlier in the subsequent LoTS care trial. In addition, after the publication of a systematic review on methods to increase response rates, trial pens were enclosed with questionnaire packs. Lessons from LoTS care and TRACS led to implementation of strategies to enhance compliance in subsequent trials (e.g. sending questionnaires to participants in advance of a face-to-face interview in participants’ homes) Response rates during specific postal and telephone reminder stages will be reported as well as the characteristics (such as age, gender, level of anxiety and depression) of participants responding at the different stages. Many aspects need consideration when developing follow-up response strategies. Above all, strategies need to be easily adaptable, closely monitored from the outset and any lessons learnt disseminated. OBJECTIVE: Patient recruitment in trials often takes longer than expected. This is costly and if the sample size remains insufficient it might lead to indecisive conclusions for practice. The field of obstetrics is unique in the way that there are two patients: the mother and her baby. We aimed to identify reasons why women participated or declined participation in an RCT during or shortly after pregnancy. DESIGN: We performed a qualitative study with semi-structured interviews in women who were asked to participate in a RCT in obstetrics. We randomly selected both women who consented and those who did not for one of a set of 7 different RCT’s. Transcripts of the interviews were analyzed anonymously using a constant-comparative approach. Two researchers independently identified barriers and facilitators for participation. Interviews were continued until saturation was reached. RESULTS: Of 22 women approached, 20 (90%) consented to be interviewed. Respondents varied by educational level, ethnicity, geographical area, age and parity. The main motivation for trial participation was either to contribute to research, as the women were convinced of it’s importance (50%) and/or preference for a treatment arm not available outside the trial (60%). Key barriers for participation in non-consenters were a negative association with the intervention (100%) and/or with randomization (45%). Fear and uncertainty, probably due to unfamiliarity with research and the feeling of being inappropriate because of unique personal characteristics, also played a role. Stress and doubt about the decision were present in both groups. CONCLUSION: Trial participation is a tough decision for most women. The final choice about participation is made intuitively, but will be influenced positively if women are well informed about the importance of research in general and the specific trial in particular. A personal, complete, well-timed dialogue may facilitate to make a balanced decision and improve trial participation. Good clinical practice (GCP) compliance is required for state of the art quality in clinical trials (CT). For HASTA trial (Hand suture vs stapling for closure of loop ileostomy) most participating centers are non-university institutions with little experience in CT. Therefore, we established a quality assurance system with on site monitoring visits and audits in all 27 centers. Based on monitoring reports, GCP violations are analyzed and their impact on patient safety and data quality is assessed. Preparation for HASTA included a 2-day investigator meeting with surgical training and individual on site initiation visits. During recruitment 48 regular on site monitoring visits were conducted. Close out visits and independent audits are currently carried out and will be finished in February 2012. Most GCP violations concerned patient informed consent (IC), although no IC was missing. 16 findings were rated major, 42 minor. Safety: 27 events were identified requiring serious adverse event reporting according to protocol. Primary endpoint needed to be corrected in 6 cases following monitoring findings. Violation of eligibility criteria were reported 4 times only. Further violations were regarding performance of trial intervention (wrong treatment group, procedure not performed according to protocol) and time window for follow-up visits. Quantitative and qualitative results of monitoring findings will be presented and compared to findings discovered by on site audits. As on site monitoring is not routinely established for investigator initiated surgical trials, relevance of monitoring and auditing for conduct of HASTA trial and validity of trial data will be discussed. Like most clinical research studies, minority enrollment rates are low in the EPR, particularly for African American males. In addition, African Americans have higher attrition rates compared with Caucasians. Therefore, in the EPR, we have employed special methods for recruiting and retaining minority subjects with varying levels of success. In this report, we will describe these methods, the lessons learned, and strategies used to improve minority recruitment and retention. These include: relationship-building with key stakeholders; capitalizing on personal and professional contacts; utilizing community “champions” and leaders; and targeting health-related and faith-based organizations/events. Success rates for the various strategies are presented by sex, age, race, and ethnicity. Objectives: During participation in a Randomized Controlled Trial (RCT) adherence to the intervention and to other protocol requirements (e.g. visit attendance and completion of questionnaires) is essential for the production scientifically robust and clinically meaningful outcomes. Yet little is known about the issues that affect adherence during trial participation. This study aimed to explore the factors that influenced adherence to research procedures by parents who were invited to enrol their infants in longitudinal RCTs aimed at allergic disease prevention. Methods: Two RCTs of an intervention to prevent food allergy in infants were used as case studies for this research. Data were collected using ethnographic methods: participant observation (130 hours) was carried out on the clinical trials unit; staff (n=26) who worked on the RCT and parents (n=55) who considered participation or had participated in the RCT took part in semi-structured interviews; and documents were gathered. Data were analysed thematically. Results: Parents appeared to commit to a contract in their participation in the studies. The potential to gain direct benefit from participation improved adherence to the protocol. However even when no personal benefit was obtained, parents put considerable effort into adhering. This reflected their beliefs in the worth of the RCT and the importance they placed on fulfilling the perceived contract. Staff took proactive and reactive approaches to improving adherence. This involved ‘negotiating’ adherence with colleagues and parents to balance manageable participation with achieving scientifically robust outcomes. Conclusion: Many factors are important in achieving optimal adherence to the intervention and other protocol requirements. Parental and staff willingness to invest considerable time and resources is essential to achieving optimal adherence. The results suggest that staff and parents considered that research participation represented a ‘contract.’ Maintaining an open dialogue regarding the nature of this contract helps to promote adherence in longitudinal RCTs. Objectives: Successful recruitment to clinical trials is essential for scientifically and clinically robust findings. Additionally problems with accrual results in increased trial costs and the inefficient use of resources. Yet little is known about effective recruitment to pediatric studies. This study aimed to explore the factors that influenced recruitment to non-therapeutic, pediatric longitudinal randomized controlled trials (RCTs). Methods: Two RCTs of an intervention to prevent food allergy in infants were used as case studies for this research. Data were collected using ethnographic methods: participant observation (130 hours) was carried out on the clinical trials unit; staff (n=26) who worked on the RCTs and parents (n=55) who considered participation or had participated in the RCT took part in semi-structured interviews; and documents were gathered. Data were analyzed thematically. Results: The potential value of participation had a substantial influence on recruitment. Parents considered possible benefits for their child and the scientific worth and integrity of the RCT before agreeing to participate. The possibility that participation would be mutually beneficial was relevant to parents and staff. The ability to manage the demands of participation was also important; parents delayed making a decision until they considered they could cope. The potential for participation to cause harm had a substantial influence; fathers were often more concerned about harm and placed less importance on the scientific worth of the study than mothers. Conclusion: Recruitment to non-therapeutic longitudinal RCTs in early childhood is influenced by complex inter-related factors. Holding detailed discussions with parents regarding the scientific value of the studies and the support that can be offered to improve the manageability of participation may help to promote recruitment. Mothers and fathers often appear to have different opinions about the potential for harm and benefit; holding discussions with both parents (where appropriate) may also increase recruitment. The Environmental Polymorphisms Registry (EPR) is a unique resource of subjects and their DNA samples that was created to facilitate genotype-driven translational research of complex disease. EPR goals are to recruit 20,000 individuals from North Carolina, collect their blood for DNA isolation, and make these DNA samples available to scientists to screen for genetic variants in environmental response genes. Once individuals with the “genotypes-of-interest” are identified, they are invited to participate in various types of follow-up studies ranging from basic laboratory ex vivo cell phenotyping investigations to observational studies and clinical trials. Since 2005, over 15,000 subjects of diverse sex, age, race, and ethnicity have been enrolled in the EPR from clinics at local hospitals, community-based events (health fairs, health conferences, community meetings, churches, etc.) and the internet. To minimize attrition and ensure high response rates for follow-up studies, these subjects are contacted annually (through multiple mailings and phone calls) and asked to update their contact information and other basic data. Nonresponders are traced using LexisNexis® and other public databases. Using these methods, our overall attrition rate is <20%. A major benefit of long-term clinical trials is the ability to collect large amounts of prospective data on individual participants. With such large amounts of data, it is critical to ensure the integrity of the data and the methods with which it is collected. Long-term clinical trials may span much of a participant’s lifetime. To accommodate for this aging transition, minor adjustments may need to be considered during the course of the trial in order to collect valid data while ensuring systematic collection. With an aging cohort, it is important to consider your methods of data collection to ensure the best quality of the data. Such methods include specialized staff trainings to deal with age-related sensory, functional, and physiologic decline and cognitive impairment. To increase functional capability, modified equipment such as sitting scales, wider examination tables and assistive devices may be necessary. Altering phlebotomy techniques such as utilizing smaller needles and tubes may help alleviate poor venous access caused by physiologic decline. Cognitive impairment can affect the accuracy of the data collected and a proxy may need to be consulted for verification. Prolonged illness and decreased mobility of aging participants create barriers to clinic access for data collection. In such cases, performing collection at convenient locations for participants including their home or nursing home may combat possible retention issues. Specific equipment such as a portable centrifuge or scale may be needed for non-clinic study visits, so long as consistent methods of data collection and types of equipment between the home and clinic setting are maintained. Prioritizing data collection for non-clinic visits will help staff capture the essential outcomes while reducing burden on participants. This presentation will discuss data collection and retention issues and the important considerations necessary to retain data quality for an aging cohort in a long-term clinical trial. Benign prostatic hyperplasia (BPH), a common condition among older men, is characterized by lower urinary tract symptoms (LUTS). Symptom severity is measured by the American Urological Association Symptom Index (AUA-SI) with higher scores reflecting greater symptom burden. The Complementary and Alternative Medicines for Urological Symptoms (CAMUS) trial, a randomized, double-blinded trial, found that saw palmetto, a phytotherapy, did not differ from placebo with respect to symptom relief over a 72-week period. Saw palmetto participants received a single daily 320 mg gelcap for the first 24 weeks, two gelcaps from weeks 25-48, and three gelcaps from weeks 49-72. Placebo participants received the same number of matching gelcaps for each period. At the end of each 24-week period, participants were asked to guess their treatment assignment and the AUA-SI was administered. Across assessments, no significant difference was found between treatment groups with respect to the participants’ guesses of their treatment assignments (P=0.638) (Table 1). Regardless of actual treatment assignment, participants who thought they were on saw palmetto had significantly greater decreases in AUA-SI than those who thought they were on placebo (Table 2). Using a general estimating equations analysis that adjusted for intrapatient variation, the effects of actual treatment assignment, time and AUA- SI change on participant guess were evaluated. Only change in AUA-SI was significantly associated with participant guess of treatment assignment (P<0.001). Men with greater improvement in LUTS were more likely to believe they were assigned to saw palmetto than placebo. It is unclear whether the perception of treatment assignment had an effect on AUA- SI or AUA-SI change influenced perception of treatment assignment. BACKGROUND: CREST received five-years’ funding ($21,268,366) from NIH/NINDS to compare stenting to surgery for stroke prevention in 2500 randomized participants at 40 sites. Reimbursement denial by Medicare and device malfunction delayed start-up. Strategies to conserve funding when the trial was halted and recruitment lagged are described and may benefit other trials. METHODS: Immediately, site-payment was changed from line-items for physicians’, coordinators’ and technicians’ salaries, mailings, and indirect costs, to a flat fee-for- service reimbursement for trial visit data. Reimbursement was highest for enrollment visits and increased ($1500-$2500 per participant) when recruitment lagged. Core- centers were combined and reduced (eight to three); payments were converted to fee-for- service for consultants and readings of angiogram and ultrasound studies. Nonessential visits were eliminated. Credentialing study follow-up was decreased from four-years to one-year. With additional funding from Abbott Vascular, CREST assumed the Investigational Device sponsorship, centralizing regulatory and site management by adding previously- unbudgeted research associates, a regulatory expert, paralegal, site monitoring organization and services of Abbott field personnel. RESULTS: NINDS funding was extended by no-cost extension from five to eight years. During these three years, credentialing-study sites increased from 52 to 98, and 773 (50%) additional patients were enrolled. Randomizing sites tripled (34 to 109), and 9,528 follow-up visits occurred. Of the 2500 sample size, 138 (5.5%) were randomized during the first five years and 1387 (55.5%) during the no-cost extensions, contributing to completion of enrollment (2008) and publication of results (2010). CONCLUSIONS: Performance-based budgets for data submitted are cost-effective, conserve funding during slow recruitment, promote visit and data compliance, and allow for additional sites at little additional cost. Partnering with industry for unbudgeted resources enhances productivity. Costs of large-scale clinical trials can thus be reduced through effective management without compromising scientific integrity. The central adjudication of primary safety and efficacy outcomes has been frequently used in multicenter clinical trials in order to ensure the accuracy and consistency of outcome assessments.[1] A recent review indicated roughly one third of randomized controlled trials used central adjudication. [2] In addition, a task-dedicated, web- based outcome adjudication system has been used in clinical trial practice.[3] The NIH NINDS-funded Platelet-Oriented Inhibition in New TIA and Minor Ischemic Stroke (POINT) Trial plans to enroll 4150 subjects from up to 150 clinical sites. The trial requires a complex adjudication procedure for primary safety outcomes, involving up to seven review steps. This adjudication procedure has been implemented into a web-based Clinical Trial Management System (CTMS), using its generic form review function, which allows a form record to be reviewed by a designated user based on the contents of the record and the results of previous review steps. To accomplish this in the CTMS, the adjudication procedure was broken down into seven steps with clear definitions of triggering conditions and review questions. A transition logic matrix was used to avoid dead loops and broken chains. Scheduled and action driven email notifications and main menu alerts were designed to notify users of action items. The preliminary experience of using this web-based adjudication function demonstrates that the average time from a site’s first submission of the safety event case report form to final adjudication was trimmed about 50 percent compared to manual coordination, in addition to benefits in automatic documentation of the review process. While the advantages of web-based trial operation management are easy to show, the complexity level of the adjudication procedure must be justified based upon cost- benefit considerations. Also, the close monitoring of the adjudication process by a designated project manager is needed in order to address unforeseeable situations. Some medical conditions may resolve with a period of observation and/or conservative treatment but may require a more aggressive treatment if symptoms persist. In such conditions it is often of interest to design studies to compare clinical and economic outcomes of immediate treatment versus deferred treatment strategies. Evaluating cost-effectiveness can be particularly important when the timing of treatment affects how much the treatment costs. One example is treatment of nasolacrimal duct obstruction (blocked tear ducts) in infants. This common condition often resolves without surgery, either spontaneously or with non-surgical treatment, and is sometimes treated with a surgical probing of the nasolacrimal duct. In infants younger than one year, the surgical probing can be performed with the infant awake in the office setting. In infants older than one year, the surgical probing must usually be performed under general anesthesia in a more costly surgical facility setting. We designed a randomized trial to address the question ‘is it more cost- effective to perform the less expensive surgery on all patients or to perform the more expensive procedure on the subset of patients whose condition doesn’t resolve with observation and non-surgical treatment?’ Using our randomized trial as an example, we explore the impact of the rate of resolution in the observation/deferred treatment group on cost-effectiveness and illustrate study design issues related to defining outcomes, choosing an outcome timepoint, deciding on a subject-level or eye-level analysis, estimating sample size, and the impact of loss to follow up. We also discuss the advantages and disadvantages to the approach we have taken to handling these issues. Genetic Repositories are increasingly common components of clinical trials since identifying genetic risk factors for disease is an area of great interest to the research community. In the process of setting up a Genetic Repository for the Age-Related Eye Disease Study 2 (AREDS2), a large multi-center study sponsored by the National Eye Institute, we found a number of areas needed to be considered both in the design of the sample collection and the selection process of the laboratory to serve as the Repository. Maximizing research goals given budgetary considerations will determine what type of samples will be collected (blood and/or saliva) and which materials (DNA, RNA, serum, and/or plasma) will be isolated and stored in the Repository. Participants may be offered a choice in the consent document whether their samples are to be used for research for any disease or a specific disease. However, if this choice is provided, this means that data from the different consent groups will need to be provided in separate data tables when sent to the Database of Genotype and Phenotype (dbGaP). While assigning a different number to the material than the one on the blood/saliva sample helps protect participant confidentiality, this should be weighed against the potential for labeling error at this step. Quality control procedures should be put into place to identify labeling errors. One possibility is DNA fingerprinting the materials obtained from multiple samples from the same participant. In evaluating the laboratory that will serve as the Repository, it is important to assess how sample history is documented and to evaluate the procedures for adjudication and documentation of mishandled/mislabeled samples. In addition, the experience of the laboratory in handling a similar volume of samples and whether staff will be dedicated to processing samples for the study should be considered. This trial was designed to determine whether intravitreal injections of ranibizumab would facilitate clearing of vitreous hemorrhage and avoidance of vitrectomy (surgical removal of the vitreous) in eyes with proliferative diabetic retinopathy and vitreous hemorrhage. Vitrectomy by 16 weeks was the primary outcome. Due to uncertainty in the outcome proportions used for the sample size calculation, an interim sample size re- estimation was proposed, in addition to monitoring for efficacy and futility. This presentation will describe the interim monitoring and sample size re-estimation plan, focusing on the concepts and implementation rather than statistical details, and discuss points to be considered and pitfalls to be avoided. The plan used the method proposed by Li et al [Biostatistics, 2002], and consisted of a single interim analysis when outcome data was available for approximately ½ of the initial sample size. If this analysis showed a strong benefit of treatment meeting a pre- defined criterion, the plan recommended early stopping for treatment benefit. Otherwise, the sample size needed to give 90% conditional power to detect the treatment effect of interest (2:1 ratio of events control:treated) was calculated based on the observed outcome proportion, pooling both treatment groups, at the time of interim analysis. If the new sample size was the same or less than the original sample size, the plan recommended continuing with the original sample size. If the new sample size was greater than the original sample size but less than a pre- specified maximum, the plan recommended continuing using the new sample size. If the new sample size was greater than the pre-specified maximum sample size, the conditional power at the pre-specified maximum was computed. If this conditional power was less than a pre-specified minimum acceptable power, the plan recommended stopping for futility. Otherwise, the plan recommended continuing to the maximum sample size. Single arm studies with intervention delivered to clusters of participants have been used to assess strategies for the prevention and control of chronic disease, but sample size approach for such a design have not been available. In this paper, we propose sample size approaches for pre- post comparison with intervention delivered to clusters. Both equal and unequal cluster sizes are considered. With assumptions that the pre-post difference is normally distributed and that the paired t-test is used to test the intervention effect, the t statistic has a central and non-central t distribution under the null and alternative hypotheses, respectively, while taking account clustering. For unequal cluster sizes, the relative efficiency (RE) of unequal versus equal cluster size, under the alternative hypothesis, is defined as the ratio of the noncentrality parameter of unequal versus equal cluster sizes. The RE is straight forward to calculate and easy to interpret. Furthermore, the proposed RE connects directly the required mean cluster size to the required sample size with equal cluster sizes for assessing study power. Consequently we can calculate the required mean cluster size to achieve the same power as an equal cluster sizes trial. There exists a minimum required number of clusters for a pre-specified power, which provides a quick evaluation of feasibility as to achieving a statistical power. All results are obtained analytically. Randomized trials designed to determine effective treatments for alcoholism have produced, at best, minimally positive, short- term findings. To obtain more clinically relevant results, studies need to address: the heterogeneity and diverse treatment goals of the population with alcohol problems, narrow eligibility criteria that limit generalizability; subjects’ reluctance to enter and remain in studies; and short-term interventions that, even if successful, have little impact on long- term courses and recoveries. We propose an adaptive study design which would be potentially feasible within the large cohort of veterans seen at Department of Veterans Affairs (VA) medical facilities. The VA electronic data capture (EDC) system collects nearly all data from clinical visits, pharmacy, procedures, radiology, and follow-up visits. Utilization of this system would allow for the potential to passively follow a cohort of perhaps 50,000 veterans for a decade. Some randomization is needed to detect and minimize confounding by indication; this process would entail attention to eligibility and informed consent. Physicians must comply with rules about regular follow-up evaluations. With these provisos, the range of problems, goals, and treatments could be evaluated as in a phase II design with statistical adjustment for missing data and dropout. Adaptively, new patients can receive more of the promising treatments and weaker treatments can be dropped as if seeking an optimal dose, with the ongoing cohort providing long-term results for the promising treatments. If a particularly promising treatment emerges and some past randomization has occurred, then a formal phase III randomized trial can begin making use of the past accrued data and following the analysis plan in Berry’s seamless designs. A decision of the Federal Joint Committee Germany states that negative pressure wound therapy is not accepted as a standard therapy with full reimbursement by the health insurance companies in Germany. This decision is based on the rapid report and the final report of the Institute for Quality and Efficiency in Health Care, which demonstrated through systematic reviews and meta-analysis of previous studies projects that an insufficient state of evidence regarding the use of negative pressure wound therapy (NPWT) for treatment of acute and chronic wounds exists. The Institute for Research in Operative Medicine (IFOM) as part of the University of Witten / Herdecke gGmbH is an independent scientific institute that is responsible for the planning, implementation, analysis and publication of trial projects regarding the efficacy and effectiveness of negative pressure wound therapy for acute and chronic wounds in both medical sectors (in- and outpatient care) in Germany. The study projects are designed and conducted with the aim to provide solid evidence regarding the efficacy of NPWT. The trials evaluate the treatment outcome of the application of a technical medical device which is based on the principle of negative pressure wound therapy (Intervention Group) in comparison to standard wound therapy (Control group) in the treatment of chronic foot wounds and acute subcutaneous abdominal wounds after surgery. All used treatment systems bear the CE mark and will be used within normal conditions of clinical routine and according to manufacturer’s instructions. The aim of the trial projects is to compare the clinical, safety and economic results of both treatment arms. Study results will be provided until the end of 2014 to contribute to the final decision of the Federal Joint Committee Germany regarding the general admission of negative pressure wound therapy as a standard of performance within both medical sectors. The management of multicentre trials is a tremendous undertaking from an administrative perspective. Too often, multicentre trials operate as silos and there may be a lack of standardization in how administrative aspects of the trials are organized. Presently, it is extremely inefficient for the development, implementation and management of a multicentre trial using a paper-based data capture system. The bigger the multicentre trial, the bigger the challenges for the management of a paper-based, non-standardized system. The OHRI Methods Centre Electronic Data Capture System Platform is an integrated, electronic multicentre trial management software package (thin-client /web-based) that is comprehensive in scope, yet general enough that it can be adapted for use in multicentre trials of various designs and in different clinical areas. It has many advanced features and fully compatible with all the popular internet browsers. The Human Intervention Studies Database (HISD), developed by KAI Research, Inc. for the National Institute on Aging (NIA), National Institutes of Health, helps NIA program officials to manage their scientific portfolio. The database tracks all NIA funded clinical trials and provides detailed information and customized reports. We have used the NIA HISD data to analyze the frequency of health conditions, diseases and interventions the NIA targets in its mission to improve the health and quality of life of older adults. We compare these data to aging statistics and key indicators of well- being available from the National Center for Health Statistics and to the strategic initiatives of the NIA. Indicators of chronic health conditions, mortality, age-related diseases and disabilities and various health care indicators are used for the comparison. The HISD frequency data are well aligned with the strategic goals of the NIA and the national health statistics. We will provide case studies of uses of the data and illustrate how an administrative tracking tool can yield valuable information for describing current initiatives and planning future directions. Since the database tracks only interventional studies, these data show an important part but not all of the clinical research and strategic initiatives sponsored by the NIA in support of an aging society. Relationships between deterioration in health status and exposure to stressful life events have been well-documented among patients with various chronic illnesses; however, little is known about the impact of life stressors on the course and management of type 2 diabetes (T2D) in youth. The TODAY clinical trial provided an opportunity to examine associations between stressful life events and physiologic markers, specifically, elevations in HbA1c and BMI among youth with T2D. TODAY (Treatment Options for type 2 Diabetes in Adolescents and Youth) was a multi-center NIH-funded clinical trial to test treatment regimens for youth-onset T2D. Participants were age 10-17 inclusive and diagnosed with T2D < 2 years at baseline. They were treated and followed for 2-6 years. During the trial, the study teams had to address multiple challenges to retention and participation in the study due to the psychosocial environment of the participant and the youth’s family. In the final year of the trial, participants completed a self-report life stressor form adapted from the Yeaworth Adolescent Life Change Event Scale(1980) and inclusive of selected items from the Holmes and Rahe (1967) Social Readjustment Rating Scale. The 33-item measure assessed frequency and self-rated level of distress associated with stressful life events over the past year. Data are available from 517 overweight youths from predominantly minority backgrounds (41.1% Hispanic; 31.5% Non-Hispanic Black; 19.6% Non- Hispanic White). After presenting descriptive statistics and correlations to describe the associations between stressful life events and select physiological markers, we report the methods and results of factor analysis to identify any relevant subgroups of stressor types. Findings will increase our understanding of the psychosocial challenges faced by youth with T2D which will help shape and optimize clinical practice, as well as inform the development of intervention programs to maximize self-care and functional status among this vulnerable population. The Multiple Courses of Antenatal Corticosteroids for Preterm Birth Study: A 5-year Follow-up (MACS-5), is an international multi-centre trial assessing 1200 children from the original MACS study at 5 years of age. The MACS-5 study aims to determine the long-term effects of antenatal corticosteroids with emphasis on cognitive, behavioral and motor development. It was important to implement specific data retrieval strategies in the early phases of the study, since challenges with long term follow-up were anticipated. Throughout the study, regular communication via email was maintained with all sites. Monthly reminder reports listing all overdue data were mailed to collaborators, and reminders to send in all data were also included in a bi-monthly newsletter. Efforts to ensure data accuracy were maintained by generating quarterly query reports for each site and conducting annual frequency reviews for all MACS-5 data. In the final year of follow-up, efforts to retrieve data and resolve queries were intensified and specific strategies were implemented to collect the remaining data. A site status report was created to regularly monitor outstanding data and queries for each site. Focus on retrieving query replies was increased by sending monthly query reports to sites, followed by telephone calls if no reply was received within a three week period. For sites who still had not returned their data, the Principal Investigator contacted the sites by email and telephone. Continued delays of receiving data led to the implementation of a pre-arranged courier to have data picked up from the sites. Site visits were undertaken as a last resort where communication remained unsuccessful. The approaches implemented in the MACS-5 study were successful in surpassing the original target follow-up goal of 1200 children. As of December 1, 2011, 1539 children had completed MACS-5 follow-up assessments. In addition, 98.6% of these follow-up cases had no outstanding queries. In randomised controlled trials (RCTs) involving study drug, supply management is a challenge that is often faced by the site and trial coordinating centre. The drug supplier and drug distributor may work independently, and as a result, a combined effort is essential to maintain a consistent flow of study drug for the duration of the trial. The Centre for Mother, Infant, and Child Research (CMICR), is the data and clinical coordinating centre for several multi-centre RCTs. Several of the RCTs evaluate study drug intervention. The approach to study drug management was to design and develop a software program to facilitate the process of ordering treatment kits containing study drug, tracking shipment, and maintaining precise details of all assigned treatment kits. In the study drug management program, treatment kits are uniquely numbered and a corresponding document is generated as a reference for the drug distributor. As new sites join the trial, treatment kits are ordered and shipped to prepare for patient recruitment. Additional kits are ordered for participating sites on an as-needed basis. The process is completed using automated electronic communication with instant update of the site inventory that is available for patient randomisation. For the trial coordinator, the program gives them the control over the amount of treatment kits to store at a particular site and better manage the overall supply parameters. The program is also able to facilitate drug reconciliation, drug recall, and set triggers for low inventory and drug expiry. The study drug management program greatly reduces the administrative effort needed by the trial coordinator while improving the efficiency of the drug supply management process. Fidelity and adherence with respect to an intervention can be tracked using readily available software and devices without burdening study participants. The Coping Peer Intervention for Adherence Trial (ADEPT) is a behavioral trial aiming to improve medication adherence in underserved minority youth with asthma. The study’s primary intervention involves listening to medication adherence messages on an iPod © that is provided to each participant and pre-loaded with songs. Participants in the intervention arm meet as a group weekly to record messages to each other that encourages his/her peers to take their medication as prescribed. These messages are overlaid on a song of the participant’s choice. While mirroring in content, the control arm’s messages are recorded by a physician and don’t incorporate music. Participant’s iPods© are updated weekly with five new songs of his/her choice as well as new adherence messages. During these weekly visits the iPod© statistics, including song name, artist, length of song, number of times played, and number of times skipped, is extracted from the iPods©. Usage statistics are reset between individuals to ensure clean data. These data are stored in iTunes© and can be exported and stored in Microsoft Excel sheets by participant and time period, and can be exported to any database of choice. This data allows investigators to monitor intervention fidelity by comparing listening habits, particularly to medication adherence messages. In ADEPT, the iPod© serves as not only the intervention delivery method, but also the means to remotely monitor intervention fidelity and dosage. American Indian and Alaska Native (AI/AN) populations within the United States are underserved in their access to health care as well as clinical trials. Our goal was to establish a clinical site in a rural Tribal community for an observational trial and train staff at the clinical site in remote data entry. In order to conduct a clinical trial with members of a tribe, approval must first be obtained from the Tribal Research Review Board and the Tribal Institutional Review Board of record. To facilitate interactions with the Tribe, the Principal Investigator contracted with a researcher who has an established professional history of working in and among AI/AN populations to serve as a liaison. A clinical site coordinator was hired who did additional hirings of staff for recruitment, clinical assessments and data entry. A data entry system was developed with up to 15 electronic case report forms collected at each of 8 visits. Because of the likelihood of missing data or missing visits, none of the forms or fields were developed as required or mandatory data entry. When the data entry system was completed and validated, we traveled to the clinical site on the reservation and conducted data entry training with the staff, all of whom were Tribal members. The clinical site staff quickly learned the requirements of data entry and have entered data on almost 250 subjects into the data entry system. The tribal liaison has traveled to the site on several occasions to review data entry and perform quality control. Follow-up is expected to be completed by May 2014 and the data will be ready to be analyzed by statisticians at the University of Iowa. Objectives: HIV/AIDS-related illnesses are key endpoints in multi-site HIV prevention trials. Accurate reporting of these endpoints requires systematic procedures for collecting, reviewing, and cleaning the clinical data. The U.S. Center of Disease Control (CDC) and World Health Organization (WHO) provide guidelines that define HIV/AIDS-related illnesses; yet clinical data management and operational challenges are rarely discussed. Methods: The Statistical Center for HIV/AIDS Prevention and Research (SCHARP) was responsible for collection and analysis of clinical data from the National Institute of Allergy and Infectious Diseases (NIAID)-sponsored HIV Prevention Trials Network protocol 052 (HPTN 052). In HPTN 052 primary clinical endpoints were reviewed by blinded protocol team physicians for acceptance. Clinical data reconciliation challenges included reporting of multiple symptoms instead of one known diagnosis, inaccurate onset and outcome date, inconsistent reporting on worsening of illness, and over reporting of events. Discrepancies in reporting any of the above key data elements ultimately affect the medical review and data analysis. Results: SCHARP clinicians and statisticians routinely partnered with protocol team physicians to assess and reconcile discrepancies in HIV/AIDS-related illnesses using multiple data sources. The SCHARP clinicians then worked with study sites to resolve identified issues. SCHARP designed a web-based tool to facilitate and track the clinical medical review. A consistent work flow was developed and utilized for the clinical data management and operational process for review of HIV/AIDS-related illnesses. Conclusion: Close collaboration between the SCHARP statistical and clinical groups and the protocol team physicians ensured accuracy and completeness in HIV/AIDS-related illnesses for endpoint analysis. The web-based tool, though in its nascence, proved useful to streamline the review process and reduce potential data errors. The utilization of this web-based tool, providing blinded clinical cases for protocol team member review, can serve as a useful model for other clinical trials. Open-ended responses are common in many questionnaires, and coding of responses is challenging. La Familia Sana Promotora Program was a demonstration project conducted to evaluate the translation of an intervention designed to improve farmworker family pesticide-related knowledge and practices. The Promotora Program used open- ended responses to measure changes in behavior and knowledge from pre- to post- test. This paper identifies complexities and solutions in the process of ensuring reliability in coding of responses in order to create discrete categories for evaluation of results. The evaluation included 20 questions used to address 18 learning objectives. Seven questions relied completely on participants’ narrative responses and were coded as “correct”, “incorrect”, or “partially correct”. Thirteen questions had “other” as a response field, and the text was then coded as “correct” or “incorrect”. Primary and secondary coders were trained to code the open-ended responses. The primary coder coded all participant responses; the secondary coder coded a 10% random sample. Kappa statistics were used to gauge the inter-coder reliability. The seven questions based solely on narrative responses were highly discordant (range: 21% to 60%). The “partially correct” and “correct” codes were combined into a single “correct” code, and the discordance decreased (range: 8% to 27%). Despite this improvement, the investigators decided that an expert should code these problematic questions. For the thirteen partial narrative responses, the kappa statistics showed favorable reliability between coders and did not require expert coding (range: 0% to 2% discordant). This study illustrates the need for content experts to code responses to completely open-ended questions; training staff is not sufficient to ensure quality data is coded. Data structure can be straightforward for a single protocol in a project. However, when sub-studies need customization, there can be many obstacles when variation is required for each protocol. The Autism Treatment Network (ATN) began a data registry to collect retrospective and prospective data on children with autism spectrum disorders (ASD) between the ages of 3-18 with a primary focus on nutrition and sleep issues. With a grant from the Health Resources and Services Administration (HRSA), the ATN conducted two sub- studies: The Diet and Nutrition in Children with ASD (Nutrition) and Sleep Education Program for Parents of Children with Autism (Sleep). Both studies had assessments that overlapped with the Registry as well as assessments that were exclusive to each protocol. Data from the Registry was used for the sub-study if collected within a specific window. For example, the Child Behavior Checklist was completed for the Registry and required for both the Nutrition and Sleep protocols but did not need to be repeated as long as the data were collected within 60 and 90 days, respectively. Initially, only assessments exclusive to the Nutrition and Sleep protocols were required in the database for each participant at the time of enrollment. Assessments that overlapped with the Registry were already developed and as the assessment was only repeated for a portion of the participants, any repeated assessments were entered under the Registry protocol. This allowed the data to be pulled from one database for all three protocols using the form date to determine most recent data. However, when the Registry protocol was amended and the database was restructured, the Nutrition and Sleep databases also required restructuring. This presentation will also highlight other obstacles related to the management of sub-studies. Accurate, high quality data are essential to any research study, making it critical to proactively identify and eliminate errors before they enter the database. A major challenge to using optical character recognition (OCR) software for data entry is identifying errors produced by stray marks, respondent corrections, or improperly completed fields. While many errors can be caught by carefully defining fields through built-in field property settings, we have found that the error reducing capacity of OCR software is greatly expanded by use of customizable scripts. Common checks scripted into data collection forms typically verify skip pattern logic, logical comparisons between fields, and restrict numeric entries to discrete values. If data read by the OCR fails these checks, they are flagged for review. Customizable scripting allows developers to add additional checks and flag data during processing. In instances where noise or an unintended mark was interpreted as data, or valid data was not detected by the OCR, use of customizable scripts enables the data manager now to prevent such errors from migrating to the database. Furthermore, if the data reported were captured accurately but fails a check, the data manager can query the reporting site and rectify the issue. We selected an OCR software system (TeleForm®) as our primary tool for data collection and management in a multi-center study investigating the safety and efficacy of bariatric surgery in adolescents. We conducted an in-depth evaluation of the database, encompassing more than 250,000 data fields, and found very few errors (0.065%). Among critical data elements, the error rate was 0.048%. By optimizing TeleForm® through robust error checking scripts, we produced accurate, high quality data in a timely fashion for a multi-site clinical study. The Centre for Mother, Infant, and Child Research (CMICR) is a central coordinating centre for several large national and international multi-centre randomized controlled trials to improve the health of women and their children. The unit has conducted research trials since 1988, and has evolved in clinical research practices alongside technological advancements. The data collection process has transitioned with each new trial, from paper-based data collection to online PDF data forms to electronic data capture. With every transition, it is important to maintain the integrity of data quality and ensure that the process is in compliance with International Conference of Harmonisation (ICH)/ Good Clinical Practice (GCP) standards. The challenges and considerations in transitioning the data collection process include programming to set up the database and training for both the CMICR data coordinator and site coordinators. The transition requires changes in programming for how the data are captured in the database and updating various reporting systems such as data form tracking, queries, funding and compliance reports. Furthermore, it is necessary to consider training each site coordinator in the new process of reporting data, and how to interpret the associated reports. This presentation will illustrate the approaches used in the technological transition of the data collection process. The Women’s Health Initiative Memory Study (WHIMS) is an ancillary study to the Women’s Health Initiative (WHI), designed to determine the short and long-term effects of hormone therapy on the development and progression of dementia symptoms in postmenopausal women. The Women’s Health Initiative Memory Study - Epidemiology of Cognitive Health Outcomes (WHIMS-ECHO) provides annual cognitive assessments of women who were aged 65 or older at the time of randomization to WHI, and The Women’s Health Initiative Memory Study of Younger Women (WHIMS-Y) assesses the long-term impact of random assignment to postmenopausal hormone therapy among women who were aged 50-54 at the time of randomization into the WHI hormone trials. To increase efficiency, lower participant burden, and reduce costs, cognitive data on consenting women from WHIMS-ECHO and WHIMS-Y are obtained annually through centralized cognitive telephone interviews conducted by trained and certified staff at the WHIMS Coordinating Center. To assist in this process, a web-based telephone call tracking system has been implemented, identifying administrative tasks and the order in which participants should be called based on study rules and priorities that have been established. Additionally, it is flexible enough to allow interviewers to schedule calls in accordance with participant needs. When a participant is selected, the interviewer is provided the contact information, pertinent study data and call history. After a call has been made the interviewer inputs the date, start time, end time, and outcome of the call allowing the system to track each attempt. Real-time reports monitoring study calls are used to detect any issues that may have a significant impact on data collection, personnel needs, or costs. The poster will describe the Telephone Call Tracking System in detail discussing the benefits and challenges. Background: The use of web sites to conduct patient allocation is increasing in popularity. Embedding encrypted information into a data element that is fed- back to the user after a patient is allocated is an inexpensive and effective way to provide an off-site back-up to protect machine rooms from catastrophic failures. Methods: Assume a multi-centre trial generates an allocation sequence using variable sized permuted blocks, stratified by site and one patient factor. Each site requires a unique allocation sequence matrix (Table 1). Table 1. Allocation sequence, Site 1 Table 1. Allocation sequence, Site 1 View larger version To allocate patients, the web server must track two counters at each site: the number of patients previously assigned to StrataA (AX) and StrataB (BY). Using a hash function based on a block cipher, or a trapdoor function, the study server can generate a variable representing these counters (Site1_AXBY) and pass the variable back to the participating site as an encrypted ‘transaction code’ (enc- Site1_AXBY). This transaction code is recorded on paper and stored locally at the participating site. Results: To randomise a patient during a catastrophic server failure, the site research coordinator telephones the study coordinating centre (SCC) and provides the ‘transaction code’ (enc-Site1_AXBY) for the patient randomised immediately prior to the server failure. The transaction code is decrypted and matched against the site- specific allocation sequence matrix which enables the SCC to make the appropriate treatment allocation for the next patient to be enrolled. Conclusion: This approach to backing-up trial status is inexpensive, robust and enables a study to recover from catastrophic server failures with zero down time with no reliance on archived electronic information. It can be extended to work with any type of allocation sequence (Ex. blocked or adaptive, such as minimization). Optimal healthcare is often accomplished by enlisting a team of interventionists. However, an intervention team does not have to meet in person to achieve effective communication. The intervention arm of the BRIGHTEN Heart behavioral trial employs an interdisciplinary virtual team strategy to enable experts to jointly develop and implement treatment plans for older adult trial participants with depression. The virtual team consists of BRIGHTEN Heart interventionists and the participant’s primary care physician (PCP). The PCP’s access is limited to their patients using built in security. In order to facilitate secured remote communications between team members, the Data Core of the Rush Center for Urban Health Equity has developed a Microsoft SharePoint web server to host an intervention communication platform that is HIPAA compliant. All data and written communications are held in a series of connected “lists” that appear on the webpage in a set of interactive, easily readable forms. Data from the initial research assistant interview is combined with a report from an evaluation by a social worker to create a patient description. Initial communications about the participant are handled in a separate linked list. The platform allows each virtual team member to have an individual entry line to record their recommendations for each participant. This communication platform is used to keep the intervention team informed of each participant’s progress in the study by the Data Core and treatment personnel, offers the means to make new recommendations, and allows effective communication among team members. The lists also enable the Data Core to track specific intervention fidelity measures (e.g. level of engagement and quality of care for each virtual team member). By utilizing standard components of SharePoint we have created a tailored communication platform that is easy to use, efficient, cost effective, tracks intervention fidelity and protects participant information. Introduction: Evidence demonstrates the effectiveness of clinical trial interventions may not become apparent until major protocol violations are minimised. Certain change management strategies are proven to promote adherence to clinical protocols and guidelines and can successfully narrow gaps between knowledge and practice. These strategies may also be useful in minimising major protocol violations in a clinical trial. Study Objectives: To design a clinical trial implementation and education strategy using change management strategies and evaluate its impact on major protocol violations. Methods: The change management and clinical trials literature was reviewed. The resultant implementation and education strategy included: formal training for local recruitment coordinators; educational outreach visits with one-on-one education (academic detailing); a formal run-in phase and; staggered site start-up meetings. This strategy was used to initiate a multi-centre clinical trial. Major protocol violation rates are reported and compared to published figures. Results: At the time of analysis, 409 patients had been enrolled from 32 sites, with 9 (2.2%) major protocol violations: four recruitment errors; three study process errors and; two technical website errors (2/9). This is significantly lower than published Phase III FDA licensing trial figures (PROWESS 351/1690 vs 9/409, p <0.001). Conclusions: An implementation and education strategy based on change management strategies can reduce major protocol violations significantly lower than published rates. We believe that formally trained local trial recruitment coordinators, who retain primary responsibility for patient identification, recruitment and protocol implementation, are the most effective component of our strategy. Background: Recruitment fatigue is known to occur in numerous trials. Many strategies have been utilised in the past in order to improve and/or maintain steady recruitment, such as use of patient screening logs. Screening logs have been shown to be effective in increasing trial recruitment but require dedicated persons to screen, which may not always be feasible. Methods: In preparation for a 26-hospital randomised controlled trial, potentially eligible sites interested in participating in the trial were required to complete a detailed site selection survey. Interested sites recorded trial eligibility criteria information on 15 consecutive patients in a de-identified manner. Realistic minimum and maximum expected recruitment rates were then calculated for each site. Results: After study onset, certain sites were found to be recruiting sub-optimally. Although many sites kept accurate screening logs, others were incomplete, as research coordinators worked part-time. Where no complete screening logs were available, site selection surveys were used to facilitate open communication between the site and the lead investigators. We found that site investigators, who had not previously seen the completed site selection surveys, were very responsive to the results of the survey and were motivated to work actively to improve recruitment. Minimum recruitment rates estimated by the site selection survey were then set as their future recruitment rate targets. Conclusion: When accurate screening logs are not available, detailed site selection surveys can be used to improve sub-optimal recruitment and combat trial fatigue. Data Safety Monitoring Boards (DSMBs) perform an important role in overseeing participant safety during randomized clinical trials. DSMBs generally review unblinded data and report back to the sponsor or Steering Committee. Interim reports to the DMSB typically include data by study arm that encompass enrollment, treatment, patient disposition, data collection, laboratory values, adverse events, and/or efficacy. While DSMB reports generally include tables and listings, figures can be particularly effective in communicating overall trends. A CONSORT like (Consolidated Standards of Reporting Trials) flowchart can provide an effective way for DSMB members to easily review the time course of participant status with respect to randomization, time on- study, treatment, and other events. Like the CONSORT diagram used in publications of final study data, a CONSORT- like figure in a DSMB report would typically include the number of participants enrolled, randomized, treated, and discontinued by study arm relative to their time on-study. For a DSMB review, this type of figure provides an interim “snapshot” of relevant study data illustrating the study flow and can incorporate important reference events such as achieving study goals, completion of milestone visits, deaths, or withdrawal of consent. The report could also contain separate figures with the same layout to display events of interest particular to the study or the DSMB, such as data lag, secondary outcomes, elements of a composite outcome, or adverse events on a timescale. We will present several examples of CONSORT-like flowcharts that can be used to enhance interim reports to DSMBs to help them monitor the ongoing safety, conduct, and efficacy of a trial in a succinct and clear manner. We will also provide an outline highlighting select SAS version 9.2 statistical graphing procedures that can be used to automate the generation of the flowcharts, saving time and reducing the possibility of transcription errors. Specimens collected in clinical trials are an important resource for biomedical investigation. Frequently secondary analyses are conducted on specimens, or they may be banked in repositories for future access by researchers. Researchers and oversight bodies often struggle with the question of consent for secondary uses of specimens. Frequent questions include 1) whether it is acceptable to anonymize specimens which were collected for a specific purpose and use them for research outside the original scope of the project; 2) whether general consent for future unspecified use is ethically acceptable; 3) whether genetics research requires a separate specific consent; 4) when re-consent may be needed when research aims change. As recent developments in the Advance Notice of Proposed Rule-Making (ANPRM) on 45 CFR 46 have demonstrated, this area of research oversight is hotly debated. This presentation will describe three models for addressing consent for research with specimens: 1) the risk based model, where oversight is tied solely to identifiability of the materials; 2) the contribution model, where oversight is based on donors’ interests in controlled future use of materials, regardless of identifiability and 3) the public health model, where consent is presumed when research serves public health goals and when risks of information breaches are appropriated reduced through secure data management and in some cases, anonymization. The ethical and practical implications of these three models will be discussed. Introduction: The Ralph Lauren Center for Cancer Care and Prevention in New York has led current research on health-care delivery models to reduce time to treatment in oncology patients through patient navigation. Patient navigation resolves barriers, e.g. financial, socio-economic, personal obstacles that preclude adherence or completion to screening, cancer care and trial participation. This activity, begun by Harold Freeman, MD led to the “Improving Access to Clinical Trials Act” (IACTA) (S.1674 [111th]), signed into United States public law on October 5, 2010, amending title XVI (Supplemental Security Income) of the Social Security Act and title XIX (Medicaid) through lessening financial barriers to participation in clinical trials and increasing enrollment in clinical trials for individuals with rare diseases or conditions. Objective: Here we explore specific barriers to care faced by cancer care patients, review the number of patients receiving navigation, its impact on treatment adherence, demonstrate the navigation process, its operational integration for decreasing barriers from their cancer care and clinical trial participation. Methods: The following subsets of data were analyzed: number of: clinical trial participants; patients receiving navigation; patients with barriers to care; and time duration for resolving these barriers. We compared the number who received navigation to those who did not and their compliance to cancer care. Results: Of 293 participants studied at the Ralph Lauren Center for Cancer Care and Prevention, barriers to care were identified as English-language proficiency issues (insurance/financial issues), cancelled appointments, scheduling issues (lack of proper documentation as United States residents), treatment complications and pain. Conclusions: In this poster presentation we will review the outcomes for participation in clinical trials and cancer care adherence with navigation. We suggest increasing the resources for on-site foreign language translation, reducing administrative barriers for legal and non-legal residents, and continuing support of patient navigation services and hospital compliance evaluation. CIOMS International Ethical Guidelines for Biomedical Research Involving Human Subjects (2002) states that investigators should ensure that research subjects who suffer injury as a result of their participation are entitled to free medical treatment for such injury and to such financial or other assistance as would compensate them equitably for any resultant impairment, disability or handicap. While Helsinki Declaration of World Medical Association (2008) describes that the protocol should include information regarding provisions for treating and/or compensating subjects who are harmed as a consequence of participation in the research study. This (no- fault) compensation is different notion from legal liability/indemnity/reparation due to malpractice or negligence in clinical trials. Un-notified clinical trials to the authorities are still allowed in Japan, other than IND/IDE trials. The Ethical Guidelines for Clinical Studies, the only regulation for those, were fundamentally revised and enacted in April 2009, which obligate researchers to take measures on compensation such as insurance in clinical trials to assess pharmaceuticals or medical devices. Since casualty insurance companies have not accumulated know-how to estimate the risk of un-notified trials besides IND/IDE trials, compensation insurance remains inadequate in quality; 1) Medical expense or medical allowance cannot be paid. 2) There are considerable exceptions for such clinical trials as employ anticancer agents, immunosuppressants, and implantable devices in this insurance. In order to overcome the situation, we have set a working group of ‘Risk-based protection of trial participants in un-notified clinical trials’ among academic clinical institutes and casualty company. The working group has investigated legal restriction for insurance, medical expense reduction system in the academic hospital, and an academic guideline for compensation in researcher-initiated un-notified clinical trials. As of September 2011, there were 79 active MSKCC-coordinated multicenter protocols (MCPs) spanning six clinical departments and involving 104 unique active sites in 29 states and 5 countries. Until recently, regulatory oversight of MCPs was investigator dependent and staff efforts were duplicated around the Center. In December 2009, the MSKCC Office of Clinical Research established the Multicenter Protocols Group (MCPG) to provide institutional support for MSKCC- coordinated MCPs. The goals of the MCPG are to institute policies and procedures, as well as provide tools and training to ensure participant safety, protocol and regulatory compliance, and data integrity for all MSKCC- led MCPs. The MCPG participates in protocol development, tracks regulatory documentation for all participating sites in MSKCC’s Protocol Information Management Database, trains MSKCC research staff, manages an intranet page housing information and tools for MCP management, conducts MSKCC-based MCP regulatory audits, and advises on and reviews all participating sites audits. Since the establishment of the MCPG, there has been an increase in the number of new participating sites opened to accrual as well as the number of inactive participating sites closed; a decrease in the average time for sites to obtain amendment approvals, from 82 to 67 days. Thirty sites were suspended for regulatory non-compliance; 93% were re- opened following corrective action. The MCPG conducted the first two regulatory audits of MSKCC-coordinated MCPs; and the study teams conducted fifteen audits of participating sites on eight protocols across four clinical departments. Staff report that MCPG oversight has improved overall regulatory management and streamlined MCP development and oversight. The MCPG will continue tool development, quality assurance, and work closely with Clinical Research Informatics towards web- based data entry by participating sites. This poster will compare trends in regulatory compliance on MSKCC led MCPs since the inception of the MCPG and discuss MCP best practices and challenges. Although Barnard’s exact test is not as commonly known or used as Fisher’s exact test, it typically offers more statistical power for analyzing 2x2 tables. The difference between the two tests is how they handle the nuisance parameter of the common success probability under the null hypothesis. Fisher’s test avoids estimating this parameter by conditioning on the margins thereby considering fewer tables and restricting the number of distinct values of the test statistic. Barnard’s test considers all possible values of the nuisance parameter and chooses the one that maximizes the p-value. This typically provides greater power because the discreteness is less pronounced. A previous barrier to the widespread use of Barnard’s test was likely the computational burden of considering all possible values of the nuisance parameter. However, with the continuing improvements in computing speed and the implementation of Barnard’s test now accessible in StatXact, Matlab, and R, the computational burden is no longer a barrier. The advantages of using Barnard’s test were shown when comparing the two tests by running 10,000 simulations with different sample sizes and proportions. Both tests give type 1 error rates at or below the nominal alpha level, but Fisher’s test was often more conservative. This resulted in less power under the alternative hypothesis. For example, when Fisher’s test required a sample size of 50 to achieve 90% power, Barnard’s test required a sample size of 44 to attain the same power (Figure). The limitation with using Barnard’s test is that it only applies to 2x2 tables and often requires more computation time. However, implementing Barnard’s test using R with an Intel® vPro™ processor took around 0.1, 1, and 20 seconds for sample sizes of 30, 100, and 500 respectively. Using Barnard’s test can require fewer observations to achieve the same level of statistical power. Background: Frequentist group-sequential and Bayesian interim analysis techniques are increasingly promoted for improved efficiency and ethical conduct of clinical trials. However, use of these methods is seldom reported outside of data and safety monitoring boards. In addition, there are few published reviews of the different techniques. This project is a comparative effectiveness study which compares and contrasts Bayesian analyses with frequentist group-sequential analyses using examples of data from two published clinical trials. We offer interpretation and evaluation of these two types of sequential analyses. Methods: For each trial, we use a commonly utilized frequentist method (O’Brien-Fleming) and implement Bayesian sequential analyses with non-informative prior. Because current practice dictates the preservation of an experiment-wise false positive rate (Type I error), we approximate these error rates for our Bayesian analyses with the posterior probability of detecting an effect in a simulated null sample. In order to consider stopping for futility, we use a frequentist group sequential inner wedge and conditional power, a concept that translates easily to the Bayesian realm. Thus for these trials, we are able to compare the relative performance of these techniques and evaluate both the Bayesian and frequentist methods. Results: In the two trials we present, Bayesian and frequentist methods lead to the same decisions about treatment efficacy or futility, however the interpretations differ. The frequentist method, more commonly reported, uses observed data to refute the hypothesis of no effect, while the Bayesian method uses the data to build a probability distribution from which to determine likelihood of effectiveness of the treatment. Conclusions: Bayesian methods provide an easily interpretable outcome distribution, however there is little emphasis on decision making. Frequentist methods provide an easy decision-making tool, but tend to be more difficult to explain. Code is provided for all analyses. Background: Interim analyses provided to data and safety monitoring boards in clinical trials address ethical, cost and efficacy issues. Both frequentist group-sequential and Bayesian interim monitoring approaches address these issues; however their implementation and verification involve very different steps. Methods: SAS 9.2 has recently developed procedures to facilitate both frequentist and Bayesian interim monitoring. For frequentists, PROC SEQDESIGN and PROC SEQTEST will develop and identify interim monitoring bounds and account for sequential testing of interim results. The effects of various parameters, and the interplay between the two procedures is explained and demonstrated on a real data set. Tips and troubleshooting for these procedures are discussed. For Bayesians, a BAYES option has been introduced into PROC GENMOD, PROC PHREG and PROC LIFEREG. This option executes a Monte Carlo Markov Chain (MCMC) technique to obtain draws from the posterior distribution. Tools needed to monitor convergence and techniques for improving convergence are presented and discussed. Results: Bayesian and frequentist methods of interim monitoring will often, though not always, agree. Frequentist methods have the advantage of familiarity among trialists. Bayesian methods have the advantage of a more natural interpretation and the ability to take interim looks at any time during the analysis. They have different diagnostics and options that must be understood before implementing either method. Conclusions: Beginning with the dissemination of SAS9.2, group sequential and Bayesian methods are accessible without high-level programming expertise. SAS9.2 provides a relatively easy-to-use set of procedures that allows flexible analysis plans for clinical trials analysis. Understanding how to implement both sequential method approaches should help trialists adopt a flexible approach to choosing among available options. Data and safety monitoring committees (DMC) become more common to evaluate the safety of new therapies, especially in clinical trials with long duration, or life-threatening diseases requiring more or less dangerous therapies. The task of the DMC is to monitor independently the course of the study and recommend whether an intervention is too dangerous for patients to continue the trial or it is not ethical to continue due to efficacy of the new therapy. While efficacy analysis is mainly focused on one primary endpoint looking for differences, analysis of safety has to consider several adverse events which may occur during the course of a trial and therefore the focus lies more on rates and confidence intervals. We developed SAS-macros to calculate measures for safety analyses such as crude rates, exposure adjusted incidence rates and poisson rate ratios. Graphical methods to illustrate the course of adverse events over time include Kaplan-Meier time to event plot and mean cumulative function (1). The application of the SAS macros will be given by an example comparing two different surgical interventions within a randomized clinical trial (2). The proposed SAS-macros provide measures to evaluate the safety of an intervention with standardized tables and plots, which can be applied several times over the course of a trial. We hope this is a contribution to the wish to a DMC “helping them to do their job well” (3). 1. Siddiqui O: Statistical Methods to Analyse Adverse Events Data of Randomized Clinical Trials. J Biopharm Stat; 19: 889-899, 2009. 2. Diener MK, et al: Efficacy of stapler versus hand-sewn closure after distal pancreatectomy (DISPACT): a randomised, controlled multicentre trial. Lancet; 377:1514-22, 2011. 3. Damocles study group: A proposed charter for clinical trial data monitoring committees: helping them to do their job well. Lancet: 365: 711-22, 2005. Adverse effect assessment often involves many clinical symptoms, and each individual may experience multiple adverse events (AE) in a clinical trial. The common practice for AE analysis compares the proportions between the treatment arms by forming confidence intervals of the proportional differences or ratios for each individual AE type. However, this analysis is often not adequate, and sometimes, it may not be methodologically appropriate. We propose to analyze multiple AE data by comparing the multiple AE incidence densities and constructing simultaneous confidence intervals of both the relative risks (RR) and the hazard ratios (HR) derived from Cox proportional hazard models. This approach usually involves multiple testing procedures that control for inflation of type I error, such as using family-wise error rate (FWER) across all AE types. Commonly used procedures include the Bonferroni-based corrections. These methods are straightforward and easy to implement, but they are conservative and reduce sensitivity. A less conservative method that considers dependencies among the various tests is employed to control for type I error by using false discovery rate (FDR). Multiple AE data from a VA/NIDA double-blinded randomized multi-center clinical trial was used to test the proposed method’s clinical relevance. From the analysis, among thirty AEs, three of them were significant between two treatment arms using an unadjusted confidence interval level or significance level; when adjusted for multiple hypotheses testing, no AE was significant by controlling either FDR or FWER. Among the procedures used, the Benjamini- Hochberg-Yekutieli procedure (BHY), which controls for FDR, was more powerful when assuming a positive dependency among the multiple tests than both the Bonferroni-Holm step-down method and the simple Bonferrni procedure by controlling for FWER. However, the BHY procedure was more conservative when assuming a negative dependency among the multiple tests. Randomized trials typically enroll a convenience sample of eligible patients without regard to formal probability sampling. However, trial results often substantively change clinical practice for the population at large, without systematic evaluation of the generalizability of results. The Surfactant Positive Airway Pressure and Pulse Oximetry Trial (SUPPORT) used a 2X2 factorial design to test different ventilation and oxygenation strategies for respiratory management of extremely premature babies. This influential and largest of its kind trial demonstrated that a less invasive ventilator strategy may be safe to use and indicated increased mortality at lower oxygen saturation. Because intervention started upon delivery, antenatal consent was required, restricting the ability to enroll some eligible infants, including those who were born precipitously following the mother’s admission. Enrolled babies had significantly higher socioeconomic status and greater exposure to antenatal steroids compared to the non-enrolled, which raised questions about the generalizability of the trial results. We conducted a sensitivity analysis by incorporating enrollment propensity weights so the analysis would better reflect the eligible population. We used Classification and Regression Trees to model enrollment based on maternal and infant characteristics at delivery. Using the groups created by the trees, we constructed enrollment propensity weights. Then we analyzed the weighted data using models that reduced the variance based on a finite population correction. The results were largely similar to the original unweighted analysis. Although weighting to reflect the characteristics of the larger population is common in survey statistics, to our knowledge the approach has not been used to explore the generalizability of results from randomized trials conducted on convenience samples of eligible patients. When adequate data on the eligible population are available and enrolled individuals are known to differ from those not enrolled, these methods provide a means to assess the sensitivity of trial results to such differences. The aim of RCTs is to obtain unbiased estimates of treatment effects to answer the question of interest. In a cluster randomised trial (CRT), maximum statistical efficiency is obtained with equally sized clusters. In practice, the ability to achieve such balance is an exception rather than the norm. Two CRTs run by the Leeds CTRU; TRACS (Training Caregivers After Stroke) and LoTS Care (Longer Term Stroke Care) demonstrate potential solutions to this issue. Unequal cluster sizes decreases the statistical power in CRTs and leads to underestimated sample size. In the TRACS trial, all centres were randomised simultaneously and unequal cluster size was not anticipated. However, the recruitment attrition rates differed by centre and the attrition rate was higher than expected. Including more clusters, each recruiting the same number of patients, would be an optimal solution. Due to time, logistics and budget constraints, the number of centres was fixed, so overall more participants were recruited and the maximum cluster size was capped. In the LoTS Care trial, imbalances were expected; centres were randomised in two phases but the recruitment period was fixed. In both trials, we re-assessed the sample size calculations and studied the effect of conservative, typical and extreme cluster size scenarios on the statistical power. Various values of the drop-out rate, design effect and coefficient of variation were considered and the impact on statistical power was calculated. Using the most conservative estimates, the overall power dropped by 2-3% when compared to the power based on equal cluster size. For both trials, statistical power based on equal cluster size was estimated to be 90% - despite unequal cluster size, power above 80% was maintained. Using TRACS and LoTS Care trials as examples, we have demonstrated the importance of incorporating unequal cluster sizes into calculations of robust sample size for CRTs. Background: Futility analysis is common when seeking to demonstrate non-inferiority. When the expected event is rare careful monitoring is required. 111 is a single group trial in a very good prognosis group of patients, evaluating one cycle of adjuvant chemotherapy (BEPx1) in high risk, stage 1 non-seminomatous germ cell tumours of the testis. BEPx2 (standard treatment) is associated with a 2% 2-year recurrence rate (Cullen et al. Journal Clinical Oncology 1996). Methods: 111 was designed to exclude a 2- year recurrence rate of >=5% and requires 236 patients (based on exact binomial probabilities). If >=230/236 patients (>=97.5%) are recurrence-free at 2 years then the trial will conclude that the event rate is <5% (power=80%, alpha=5%). A trivial futility boundary is therefore defined: if >=7 recurrences were observed the trial would be terminated, but more flexibility and guidance for interim monitoring was sought. In calculating the recurrence rate, total follow up for each patient is taken into account. As recurrences are more likely to occur in the early months following treatment, follow up is weighted according to recurrence rates observed in previous studies. To derive appropriate beta values for each analysis we simulated several design parameters. A repeated confidence interval approach was adopted. The beta spending function was selected so that: 1) overall alpha=0.05, beta=0.20 and 2) probability of stopping after 1 recurrence is very low. Results: The chosen rule was that at each recurrence the estimated rate and its exact confidence interval (CI) would be calculated. If the lower limit of this CI exceeded 2% then the trial would stop recruiting. Conclusions: A sequential interim monitoring plan has been developed. This methodology could be adopted in other studies to ensure that a rare event rate does not exceed a pre- specified threshold. Background: Reliable intermediate outcome measures are needed to assess clinical interventions in the current era of HIV treatment. We evaluated the performance of the Veterans Aging Cohort Study (VACS) Index that uses routine clinical biomarkers to predict mortality as a surrogate outcome for randomized clinical trials. Methods: VACS Index scores were determined from data collected in the Options In Management with Antiretrovirals (OPTIMA) multi-national study of treatment strategies in patients with advanced HIV (Holodniy, 2011). OPTIMA data were used to validate the use of the VACS Index in a cohort of late-stage HIV patients. Proportional hazards regression and survival analysis using baseline VACS Index scores and changes in score during treatment were used to assess the utility of the index as an outcome for future clinical trials. Results: VACS Index scores at baseline correlated highly with mortality and with time to death in the advanced HIV patients enrolled in OPTIMA (Kirkwood, 2011; c = 0.749). Baseline score and change in score over 48 weeks of treatment were equally associated with all-cause mortality (c = 0.783). Conclusions: The VACS Index accurately predicted mortality and responded to changes in treatment. Measuring changes in VACS Index scores over relatively short time periods may offer an efficient alternative endpoint for the design of randomized clinical interventions among patients with HIV. References: Holodniy M, Brown ST, Cameron DW, et al. Results of antiretroviral treatment interruption and intensification in advanced multi-drug resistant HIV infection from the OPTIMA trial. PLoS One 2011; 6(3):e14764. Katherine A. Kirkwood, Tassos Kyriakides, Sheldon T. Brown, Amy C. Justice, Mark Holodniy, Janet Tate, Joseph Goulet. The VACS Risk Index Responds to Treatment Interventions and is Highly Correlated with and Predictive of Mortality Events in the OPTIMA study. 2011 Joint Statistical Meetings, Miami, FL, July 30-August 4, 2011. Session #423. Recruiting to target sample size is key to the success of all research yet continues to pose a significant problem. Potential barriers are often unknown at the time of securing funding and trials must have processes in place to identify and respond to these barriers. We present our experience from three UK publically funded multicentre RCTs of stroke patients and discuss ways in which recruitment strategies were adapted and refined to help achieve target recruitment. The main approaches we used focused on the site feasibility and the collection of screening data. Site feasibility and selection is used to understand the implementation and delivery of trials within existing services and provides recruitment estimates. This process revealed potential capacity and resource constraints which were subsequently used to facilitate discussion with service providers and allow appropriate allocation of funds. The recruitment estimates are used to engage in discussion with site during the recruitment phase to identify further barriers to target. Systematic collection of screening data from time of recruitment is reviewed monthly and is used to identify trends in the reasons for non-recruitment. This process led to changes in the recruitment and consent procedures to simplify the patient information and to allow obtaining consent on the same day as the provision of information. Eligibility criteria are reviewed and amended where clinically and scientifically valid. Amendments included increasing the time window of recruitment and allowing co-enrolment into other trials. Barriers to trials achieving target recruitment are multifaceted and trial teams need to be responsive to adapt trial processes accordingly. We have shown that the systematic identification of barriers from the time of site selection and during recruitment is essential and allows the trial to adapt to meet target sample size. Data on the impact of the changes on recruitment will be presented. Mucopurulent cervicitis (MPC), a condition characterized by cervical discharge and inflammation, can be caused by sexually transmitted infections (STIs). For the 50- 80% of cases where the cause of MPC is unknown, it is unclear whether empiric antibiotic treatment is effective. Concerns that unnecessary antibiotic use may contribute to antibiotic resistance motivated a noninferiority clinical trial to compare the efficacy of empiric treatment with two antibiotics, azithromycin and cefixime, with placebo for MPC not attributed to STIs. The plan was to screen 3,357 women to enroll 772 participants with MPC, based on an estimated MPC prevalence of 23%. In one year, 577 women were screened at one family planning (FP) and 3 sexually transmitted diseases (STD) clinics, and 131 (23%) met the definition of clinical MPC: presence of cervical discharge and/or easily induced cervical bleeding on pelvic exam. Eighty-seven (66%) women enrolled on the trial, 31 (24%) were excluded due to the presence of an STI, 10 (8%) met another exclusion criterion, and 3 (2%) chose not to participate. The number of women screened per enrolled participant was 4.6 and 7.0 from FP and STD clinics, respectively. Some entry laboratory test results were not available until after enrollment. They identified 32 participants who were ineligible for the study: 24 had positive results for STIs and 8 did not have > 30 WBCs per high power field on the cervical gram stain, a measure of inflammation. Fifty-five women met all eligibility criteria. The study was terminated due to low accrual. Among women with MPC, 42% had one or more STIs, mainly among those enrolled at STD clinics. Future studies of MPC with unknown etiology should consider recruiting from FP clinics or factor in the high rate of exclusions due to STIs among women recruited through STD clinics. In Europe requirements for conducting clinical trials with medical devices were changed fundamentally by the council directive 2007/47/EC (2007). Meanwhile the international standard ISO 14155:2011 and the German Medical Device Act have been amended accordingly. All these regulations demand appropriately qualified investigational site teams. A certified training curriculum was established in order to develop the required operational competence. It was not only harmonized within the German Clinical Trials Center (CTC)-network, but also reconciled with ethics committees as they examine the qualification of investigators. The curriculum was designed as a two-stage process. It was based on an already well established 2-day training course for investigators focusing on ICH-GCP E6 and German regulations for clinical trials with medicinal products. This course was supplemented by an advanced half-day course for medical devices. This half-day course concentrates on the characteristic requirements of the federal authority and ethics committees, the newly set up electronic application procedure, and the electronic SAE- reporting obligation of sponsor and investigators. In the next step we established a standalone training course qualifying investigators for trials with medical devices without previous knowledge of conducting clinical trials. The curriculum covers all regulatory specifications, good clinical practice, uses incorporated didactic quizzes to upgrade learning effects, and considers the characteristics of Investigator Initiated Trials (IITs). With their know-how and academy CTCs are important components in the field of qualifying for conducting clinical trials with medical devices even beyond the scope of IITs. Our report will show the challenges of developing these certified training courses for investigational site teams. And it presents the experiences of the CCT Cologne conducting these courses and the evaluation by attending investigators. The Network of Hubs for Trials Methodology Research (HTMR Network) was established by the Medical Research Council to create a UK- wide regionally distributed research resource to improve the design, conduct, analysis, interpretation, and reporting of clinical trials. Consisting of eight Hubs, the HTMR Network possesses methodological expertise and fosters links with clinical trials units and other methodological groups in universities, the National Health Service, industry and relevant professional bodies. The HTMR Network aims to (1) promote and fund high quality collaborative methodological research, both between Hubs and with other groups, UK-wide and internationally; (2) provide methodological advice to the clinical trials community; (3) encourage the implementation of the most effective and appropriate methodological practice, for example by providing education and training; and (4) work with stakeholders, in particular to agree on shared priorities for research and guidance and to advocate for improvements in the conduct of clinical trials. The HTMR Network has funded several successful projects involving the Hubs and other groups, including COMET (Core Outcome Measures in Effectiveness Trials), DIRUM (Database of Instruments for Resource Use Measurement) and workshops on topics such as randomised trials in surgery and on recruiting children to trials. We are working on methodological issues raised by stakeholders through consultation and provide the Methodology Advisory Service for Trials (MAST), a “second-line” advisory service for clinical trials units and Research Design Services in the UK. In February 2010, the Office of Clinical Research (OCR), at Memorial Sloan-Kettering Cancer Center (MSKCC), established an institutional clinical research monitoring program for MSKCC-sponsored studies. Although MSKCC has an extensive Data and Safety Monitoring plan which includes institutional and clinical department auditing, the focus on protocol compliance auditing only comprises a small percentage of the institutional portfolio in a given year (<15%). The new clinical research monitoring program focuses on in-house, investigator-initiated, MSKCC-sponsored studies, for which no outside organization provides monitoring or oversight. The goal of the monitoring program is to provide “real time” oversight of approximately 40% of MSKCC investigator initiated studies. In order to achieve this goal we developed the following: a set of criteria to evaluate protocols in our institutional portfolio for monitoring priority; documents to conduct protocol compliance, data verification, and regulatory documentation review; and guidelines for selection of participants to be monitored per visit. In addition, we identified department monitors and established written procedures for them to follow in the program. To determine the outcome of the monitoring visits, we created a rating system. Consistently high quality monitoring visits for a particular protocol would allow transition of that protocol off the monitored list and allow for us to initiate review of another protocol. Guidelines for serious and unresolved issues were also established which included suspension of new accruals for continuous unacceptable ratings. To keep a record of the monitoring visits, an access database to collect the summary letters of each visit and enter relevant data was created. In this poster presentation we will review our evaluation of the program and determine whether we are achieving our goal of “real time” oversight of 40% of our investigator initiated studies. We will also outline the challenges we encountered and methods we are developing to improve the program. Metformin in Women with Type 2 Diabetes in Pregnancy Trial (MiTy) is a multi-center, double masked, randomized placebo-controlled trial for pregnant women with type 2 diabetes. With 500 recruits expected from 25 participating centers across Canada, MiTy seeks to determine the effect of the addition of Metformin to a standard regimen of insulin on perinatal morbidity and mortality. Study sites are required to record specific outcomes for each study participant in prenatal visit forms every four weeks, based on gestational age at randomization. In order to accurately capture the complex nature of this large amount of data, it was decided that electronic data capture (EDC) would be more effective than paper data collection. Since EDC is a recent system, a strategy for disseminating user training was required. Collaborators participated in a three hour training session in which they were guided through a simulation that involved entering patient data into an EDC program. Due to differences in technical skills, abilities and learning styles, it was evident that the majority of the participants did not gain a comprehensive knowledge of how to use the EDC program. Following this, three separate strategies were implemented to provide more effective training. First, an EDC test website was created in which collaborators could enter simulated data to become proficient with the system. This allowed collaborators to familiarize themselves with the program according to their own skill level and pace. Next, a section in the Study Manual was created with step-by-step instructions and troubleshooting for further clarification. Lastly, email and phone support was offered to remedy problems that could not be resolved by the Study Manual. These strategies proved both cost-effective and successful in providing collaborators with comprehensive EDC training. INTRODUCTION: Patient recruitment in clinical trials often takes longer than expected. Trials with slow recruitment are more costly and an insufficient sample size leads to indecisive conclusions. We identified factors influencing recruitment in clinical trials. DESIGN We sent a questionnaire to principle investigators of all 1130 trials prospectively registered in the Netherlands Trial Register with an expected date of completing recruitment between 2005 and 2010. We used logistic regression analysis to assess whether characteristics of the trial or the principle investigator were associated with successful recruitment, i.e. 80% of the targeted number of patients was recruited within the planned time. RESULTS Of 392 trials questionnaires were completed. For these trials 232,707 persons were to be recruited. In half of the trials recruitment was unsuccessful. Although 42% of the trials were extended for ?6 months, when closing recruitment 46% still had recruited fewer patients than originally intended. Of the investigators 67% stated recruitment was more difficult than expected. Factors univariably associated with unsuccessful recruitment were: clear responsibilities for recruitment, investigator not PhD, multicenter trial, presentation at start for recruiters, newsletter for recruiters, pocket cards, email at start of the trial and a low expected number of randomizations per month. In multivariable analysis, adjusted for potential confounders, associated with recruitment were having a PhD (OR 0.37; 95% CI 0,17-0,83) and giving a presentation at start of the trial for recruiters (OR 0.24; 95% CI 0.09-0.64) - the latter might be indicative for complexity of a trial. CONCLUSION Investigators frequently overestimate recruitment success when starting a trial. Although we identified factors associated with recruitment, we are unable to make general recommendations for improving recruitment. A possible limitation of this study is the risk for selective responses and unmeasured confounders. Investigators should be aware of potential recruitment difficulties when starting a new trial. The role of glucocorticoid receptor single nucleotide polymorphisms (SNPs) in receptor function and metabolic disease is being studied at the National Institute of Environmental Health Sciences (NIEHS). This in vivo and in vitro observational gene association study investigates functional relevance of SNPs in the NR3C1 gene. Individuals with and without functionally relevant SNPs are identified and recruited from the Environmental Polymorphism Registry (EPR). The EPR is a DNA biorepository of over 15,000 participants developed at NIEHS to help researchers better understand relationships between environmental exposures, genetic susceptibility and disease. The primary objective is to investigate in vivo the role of hGR SNPs (hGR9b A3669G, hGR N363S) in steroid responsiveness using dexamethasone and comparing serum cortisol levels by genotype. Initial comparisons indicate some hGR9b A3669G carriers are resistant to dexamethasone. The secondary objective is to investigate the role of hGR SNPs (hGR9b A3669G, hGR N363S) in human steroid responsiveness by comparing gene expression profiles of macrophages exposed ex vivo to corticosteroids. A pilot microarray study on macrophages without hGR SNPs revealed that although there is great variability in gene regulation between individuals, treatment with dexamethasone significantly altered gene regulation. These initial study results suggest functional relevance of hGR SNPs in the NR3C1 gene. The study is also the first EPR study conducted at the NIEHS Clinical Research Unit (CRU) and will help define and enhance operations of future EPR studies, including recruitment strategies. Metformin in Women with Type 2 Diabetes in Pregnancy Trial (MiTy) is a multi-centre, double-masked, randomized placebo-controlled trial for pregnant women with type 2 diabetes. With 500 recruits expected from 25 participating centres across Canada, MiTy seeks to determine whether the addition of Metformin to their usual insulin regimen, will decrease the incidence of adverse perinatal outcomes. Study sites are required to schedule prenatal visits for study participants every four weeks, based on gestational age at randomization. At MiTy prenatal visits, collaborators are required to supply study drug, distribute glucose strips, take specific blood tests, measure serum creatinine, download glucometer readings, and count pills to verify patient compliance, in addition to routine standard care. Prenatal visit information is then entered on the prenatal visit form following each patient appointment. Specific strategies were needed to help collaborators keep track of prenatal visit schedules; identify which tasks were required to be completed at each visit; and to indicate when to complete the prenatal forms. Therefore, several tools were developed to ensure successful adherence to the study requirements. A study manual was created to explain the specifics on how to manage the trial. This included a timeline with descriptions of tasks to be completed at particular stages during the patient’s pregnancy, as well as checklists of tasks to be completed at each appointment. Prenatal visit appointment schedules were programmed upon randomization, notifying study coordinators of the exact week to schedule prenatal visits according to gestational age. Appointment reminders were emailed one week prior to the appointment, along with consistent email and telephone communication. Furthermore, centres receive tips and reminders through the monthly MiTy newsletter, group teleconferences, and regional meetings. This presentation will share successful approaches used to ensure adherence to study requirements. Assessment of the change in tumor size in response to treatment is critical in the evaluation of anti-cancer agents. The purpose of this presentation is to review the status of standardized solid tumor response criteria published since the early 1980s including WHO criteria, RECIST 1.0, RECIST 1.1 and PERCIST 1.0. The existing criteria (WHO, RECIST 1.0, RECIST 1.1) will be compared in terms of number of target lesion requirement at baseline, lymph node usage in the calculation of tumor size, the requirement of confirmatory responses as well as other specific criteria used in the existing guidelines. Similarities and differences between the tumor assessment criteria will be illustrated using specific examples. Consideration of unidimensional and bidimensional measurements as well as the possibility of three-dimensional analyses will be discussed. New imaging techniques such as FDG-PET and MRI in tumor assessment as well as treatment classification, specifically cytostatic, will be examined. Non-measurable disease at baseline will also be considered with respect to each criterion. Other discussions will include “modified RECIST” and why it should be avoided. The limitations of standardization and possible future development will also be discussed. Background: Historically, there have been significant barriers to conducting research in tribal communities, including a lack of understanding of American Indian (AI) culture and research. Methods: Spring of 2002, University of Oklahoma Health Science Center (OUHSC) and four AI Tribe/Nations with Institutional Review Boards (IRB) (Absentee Shawnee Tribe, Cherokee Nation, the Chickasaw Nation, Choctaw Nation) and OKC Area Indian Health Service (IHS) IRB (collectively AI Partner(s)) proposed to participate in one of the largest pediatric diabetes research trials supported by the National Institutes of Health (NIH). This venture required an agreement between AI Nations and academic university addressing essential elements: trust, collaboration, and successful outcome. Results: The four AI Tribe/Nations executed a Memorandum of Agreement (MOA). The MOA stipulated: 1) AI Coordinator hired to foster relationships, assist with recruitment, and facilitate pre- review by AI Partners of publications containing AI descriptive data. Sixty-one (61) AI patients were screened and forty (40) of those patients randomized into the TODAY Study. 2) Establish a committee comprised of AI Partner representatives to review publications. The committee meets for monthly publication updates with regular communication between meetings. The committee has met approximately 20 times and has reviewed approximately 50 items. Presentation will describe the efforts in retaining tribal partnerships including barriers that arose and resolutions. Conclusions: After 9 years, 4 AI IRBs and OKC IHS IRB have created and maintained a successful collaboration with an academic university in a large clinical trial. This multi-tribal research partnership is an excellent model of how community and multi-ethnic collaborative research can be effective. Purpose: To ascertain the experiences and perceptions of participation in cancer clinical trials by Hispanic and Black cancer survivors. Methods: As part of a larger study to develop a decision aid for participation in cancer clinical trials, we conducted semi-structured interviews with English-speaking Hispanic (15), Spanish- speaking Hispanic (15), and Black (15) cancer survivors. We employed quantitative content analysis to code responses to 16 specific questions based on transcripts from the interviews. Results: The average age of participants was 56.0 (SD 10.6) years; 93.3% were female, with most having had breast cancer (66.7%). Years since diagnosis ranged from 1-16, with an average of 3.3 (SD 3.0). Only 4 had been asked to participate in a clinical trial, and of these, 3 had joined a trial. Although only 4 had talked with a provider about participating, 23 (54.8%) wished that such a conversation had happened. In addition, 32 (71.1%) said that they would be willing to participate in a trial, and another 8 (17.8%) said that they may be willing. Almost all, 97.8%, said that it would be helpful to hear other cancer patients’ experiences with clinical trials. However, most (81.8%) stated that it would not make any difference if the people relaying the experiences were of the participant’s own race/ethnicity. Rather, information from someone with their own type of cancer was viewed as very important by 36%. Conclusions: Although few participants had even talked about clinical trials with their health care providers, most cancer survivors expressed willingness to participate. These findings highlight the need for discussions about clinical trials between physicians and patients. Increasing cancer patients’ knowledge and self-efficacy to discuss clinical trials with healthcare providers, for example through a decision aid, may lead to a much needed increase in participation rates. Purpose: Participation in cancer clinical trials is very low, particularly for minority cancer patients. The goal of this research was to develop a web-based decision aid to improve minority cancer patients’ decision making about whether to participate in a clinical trial. Methods: Steps in developing the decision aid were: a) collecting information on barriers and promoters of participation; b) developing a theoretical framework to conceptualize the decision aid components; c) using cognitive interviewing to pilot test the decision aid; and d) testing the decision aid to determine its effects on attitudes and knowledge about cancer clinical trials. Information was collected using 3 complementary approaches: a) data from a telephone survey of 1100 cancer survivors; b) a literature review; and most importantly, c) semi-structured interviews with cancer survivors. All participants were either Hispanic or Black. Results: For the semi-structured interviews, 45 participants were recruited. The information from the 3 sources was used to develop the KEV (Knowledge, Empowerment, and Values) framework for the decision aid. The website was developed, and successfully pilot tested with 8 participants. Currently, we are assessing the effectiveness of the decision aid to improve patients’ knowledge about clinical trials, empower them to talk with their health care provider about trials, and clarify their values and preferences related to trials participation in 100 participants. Assessment of the effect of the decision aid is done by comparing knowledge, attitudes, and decisional conflict about clinical trials before and after viewing the decision aid. Conclusions: While quantitative data on the effectiveness and acceptability of the decision aid is not yet available, feedback from the participants has been extremely positive. Participants knowledge increases, and many state that they wish they had had such an instrument when they were first diagnosed. Our ultimate goal is to make this decision aid widely available. The Centre for Mother, Infant, and Child Research (CMICR) was established to improve the health of women and their children through clinical research and education. As the data coordinating centre for several international randomized controlled trials (RCTs), it is imperative that all trial operations are completed in a consistent manner to ensure maintenance of a quality management system. Thus, in adherence to Good Clinical Practice (GCP) guidelines, CMICR uses written standard operating procedures (SOPs) in its daily operations. In a team setting, it is especially important to have a standard set of detailed SOPs, along with clear, well-established rules to follow. This allows for the management of various aspects of the organization to be structured and easily maintained. In previous years, each trial team created and maintained their individual SOPs. This resulted in varying procedures between each study and a lack of detail, which was addressed in the new format. Various methods were employed to identify the most effective way to achieve uniformity of all SOPs. A staff member was designated the specific role of standardizing all trial SOPs to ensure consistency. The existing SOP system was reviewed over a three month period to assist in developing an improved structure. This new structure includes a revised coding system, as well as improved methods for archiving and updating task-specific SOPs. A template was also created to ensure a simple, easy-to-read layout of all newly established SOPs. In addition, to assist in the implementation of the new SOP system, specific guidelines were established on how to create an SOP document. All SOPs are submitted to the designated staff member for review, and are made accessible to all staff. These revisions to the SOP system have addressed inconsistencies, and created cohesion throughout the organization. The Centre for Mother, Infant, and Child Research, Sunnybrook Research Institute Toronto, Ontario, Canada The Centre for Mother, Infant, and Child Research (CMICR) is the data coordinating centre and administrative site for several large, multi-centre, international randomized controlled trials that aim to improve clinical practice and the health outcomes of women and their children. CMICR has established an international collaborative group of research sites that may simultaneously participate in several CMICR-coordinated trials. It has become evident that successful recruitment and protocol adherence of a site in one CMICR-coordinated trial does not guarantee its success in another, and in trials with limited budgets and timelines, identifying suitable sites is vital. To assist in the site evaluation and selection process, CMICR developed an improved trial-specific Site Feasibility Questionnaire (SFQ). The revised SFQ developed by CMICR includes four main sections. The first, “Study Background”, includes the research summary, inclusion/exclusion criteria, and outlines the study requirements. In the second section, “Site Information”, sites are asked to provide information about their experience and performance in other research trials, resource information (i.e. research staff, facilities, equipment), and the total number of patients that meet the study inclusion/exclusion criteria annually. The third section, “Study Protocol Requirements”, is important in determining whether the site will be able to adhere to the study protocol. Sites are asked about their local standard of care and whether they will be able to follow the schedule of events and carry out all the required tasks. The final section covers recruitment topics and sites are asked to explain any anticipated difficulties with the trial. Once the centre has reviewed the protocol and considered all survey responses, the final requirement is to provide their expected annual recruitment goal. The presentation will share how to write an effective Site Feasibility Questionnaire that will help in site evaluation and selection process. Ensuring timely recruitment of appropriate subjects into clinical trials is not only a measure of study discipline, it provides economic efficiencies and also ensures that study results will be relevant since they are reflective of current practice. We have developed processes to ensure the timely recruitment of appropriate subjects and will provide suggestions and examples from several large international trials involving over 500 hospitals in more than 40 countries. Not only is timely recruitment essential to success of the trial, adherence to the trial intervention and complete follow-up on all patients is critical. We have developed tools to encourage compliance and prevent “potential lost to follow-up” . We will also present examples of specific impediments to follow-up. Among these examples is recent privacy legislation enacted in some countries which has fostered an environment of uncertainty and caution at many investigating sites. There is a perception that individual rights can take precedence over the completeness and accuracy of the clinical trial results. Site personnel have refused to follow any patient who has “withdrawn his/her consent”. We will demonstrate how we have been able to overcome this obstacle in most countries by clearing up misconceptions and by re- educating the investigating site personnel. Successful blood sample collection in multi-center trials could be achieved through the implementation of a trial-specific Quality Assurance Monitoring System (QAMS) that ensures the high-quality sample retrieval needed for optimal biomarker measurement. In the SOX Trial, a Canadian Institutes of Health Research and Heart & Stroke Foundation funded multi-center RCT of active vs. placebo stockings to prevent PTS in 803 patients with acute proximal deep vein thrombosis (DVT), blood specimen samples were collected within the first 6-mths of diagnosis to evaluate whether biomarkers reflective of inflammation, coagulation activation and genetic thrombophilia influence PTS development. At baseline, 1 and 6-mth visits, blood samples drawn from consenting patients were processed, aliquoted into microcentrifuge tubes, frozen at -80°C and batched from 24 study sites to the Trial Coordinating Centre (TCC) for central storage according to study Standard Operating Procedure (SOP) instructions. Specimens were then assessed for quality based on these standards: sample recovery (SR), correct sample type (CST), adequate sample volume (ASV) and correct tube labeling (CTL). For buffy-coat collection, all 4 criteria were met for 79% of samples: 99.2%, 79.9%, 99.8% and 100% for SR, CST, ASV and CTL standards, respectively. For platelet-poor plasma collection, all 4 criteria were met for 94.3% of samples: 97.7%, 98.2%, 99% and 100% for SR, CST, ASV and CTL, respectively. In general, we noted that sites with lower overall quality were those with a lower patient retention rate, frequent coordinator turnovers, and sample processing left to study personnel. Development of site-directed SOPs followed by routine TCC quality checks for sample integrity are two economical ways to implement a successful QAMS in budget-limited government-funded studies; these mechanisms help safeguard sample quality, reduce missing data and enhance credibility of study results. Additional site-specific monitoring may also be warranted for study sites with poor performance and limited resources. There are unique challenges faced in the conduct of clinical trials in the relatively new research field of cardiac surgery. Several strategies have been implemented in the start-up phase of a large, international, publically-funded clinical trial in cardiac surgery (the Steroids in Cardiac Surgery (SIRS) trial) to address these challenges. Currently, the largest published clinical trial in cardiac surgery is the ROOBY trial which recruited 2203 patients from 18 centers. SIRS will recruit 7500 patients from approximately 85 centers in 18 countries, making it the largest clinical trial conducted in cardiac surgery to date (recruitment as of Nov. 30, 2011 was 1919). From the experience of starting-up SIRS, many challenges were identified and strategies have been used which can be translated to other research fields, particularly to trials being conducted in new disease research areas. The most common challenges faced during the start-up phase of a large, international clinical trial are: adhering to project timelines; dealing with various country-specific regulatory issues; identifying appropriate clinical sites; and ensuring proper study conduct for the duration of the trial. In SIRS, there were the added unique challenges of being a study in a fairly new clinical research field with many novice investigators and without an established global cardiac surgery research network, as well as having limited funding. These challenges and the implemented strategies used to address these obstacles have been outlined in Table 1. Conducting global clinical research is complex and involves many hurdles, especially in a relatively new area such as cardiac surgery. SIRS will hopefully pave the way for future large clinical trials in cardiac surgery by building a global network of investigators, and by developing tools to meet the challenges (particularly in the start-up phase) which are crucial to the successful and efficient coordination of a large clinical trial. Under the Best Pharmaceutical for Children Act (BPCA), the protocol “Use of Lorazepam for the Treatment of Pediatric Status Epilepticus: A Randomized, Double- Blinded Trial of Lorazepam and Diazepam,” changed from one Clinical Research Organization (CRO) to another CRO mid- study. The transition included moving from paper to electronic case report forms (CRFs). The transition involved four steps: 1) transferring the datasets from the previous CRO’s data format to the new data format, 2) creating a user-friendly electronic data capture system that included all required fields and any new fields desired by the sponsor, 3) accurately mapping and uploading the data from the previous CRO to the new system and 4) transitioning the sites to the new system. To streamline these processes, collaboration with the previous CRO, the study sponsor, and the coordinators performing the data entry was essential. Building the electronic data system required identifying all required fields, determining the format of data fields, changing free text fields to drop-down fields, adding new fields requested by the sponsor, obtaining any data quality checks implemented by the previous CRO, and assuring sponsor approval for all changes made to the data collection procedures. All uploaded data required extensive quality control to assure the accuracy of the mapped data. Site transition involved creating a plan for the collection of ongoing data while the electronic system was in process, including reporting of serious adverse events and protocol deviations, and training the sites on the new system. Soliciting input from the site coordinators on the data collection process was valuable in improving efficiency of and satisfaction with the new data collection system. Background: The National Institute of Allergy and Infectious Diseases (NIAID) invests heavily in the conduct of clinical trials benefiting public health. For the Division of AIDS (DAIDS), trial quality is managed in part through a risk-based monitoring program performed by contract monitors. Prior to 2010, these monitoring functions were not subject to independent review. In 2010, DAIDS began sponsoring an independent auditing program to assess the quality and effectiveness of the contract monitoring. We report on the development and progress of these quality assurance (QA) audits during the first 2 years of the auditing contract, where ~90 QA audits were conducted, and on their preliminary impact on the DAIDS monitoring program. Methods: Using an independent contractor for the NIAID Auditing Services Program (NASP), QA auditors sampled monitoring reports and performed on-site audits at selected clinical research sites (CRSs). About 12 CRSs were chosen quarterly based on several qualifying criteria, including monitoring report content and risk level of protocols. Routine NASP audits are characterized primarily as performance audits rather than Good Clinical Practice (GCP) audits, since DAIDS procedures for the monitoring contractor are very focused due to the risk-based monitoring program design. Teams of NASP auditors based in the United States, South Africa, and Brazil were trained on DAIDS-specific procedures and the same instructions provided to monitors. Auditors reviewed a sample of the same materials that were previously monitored, and reported directly to the DAIDS sponsor on any findings. Results: From over 740 CRSs in nearly 50 countries, over 1170 on-site monitoring visits were performed at about 233 CRSs in 21 countries over the past 2 years. Sampling those visits, ~90 NASP audits were completed in 14 countries, with no critical and few major findings observed. Several findings led directly to significant improvements in DAIDS monitoring policies and procedures. Over the past decade, the administrative burden of clinical trials has become an issue of national concern. One significant component of the bureaucracy is ensuring compliant billing practices. Where a coverage analysis at a single site may take as many as eight hours to complete, the inefficiency of this being repeated by every site in a multi-institution clinical trial highlights an unnecessary burden. One cooperative group, SWOG, piloted providing a coverage analysis to its sites and found the utility of doing so most benefited lower-accruing sites, who reported the most time spent completing analyses for trials they participated in and with the least resources to do so. The pilot also revealed that some sites had not yet adopted a practice of completing a coverage analyses, pointing to the increased risk of clinical billing fraud. With the future of cancer research diverging from standard of care and into innovative new directions such as biomarker discovery, the current federal funding system for cancer research requires processes to ensure compliant billing. In multi-institutional clinical trials, having the sponsor providing a coverage analysis using national guidelines and standards adds efficiency to an administratively over- burdened clinical trial management process. Further, the provision of a sponsor-provided coverage analysis particularly benefits the lower-accruing sites in reducing the risk of fraudulent billing. Cancer research sponsors can easily support success and continued participation by community practices by providing coverage analyses. Project supported by NIH funded grants CA032102 and CA037429 Many dose finding designs including the continual reassessment method (CRM) have been shown as statistically advantageous over traditional up down methods such as the ‘3+3’. However, these designs have been slow to gain popularity as they can be computationally intensive, often need specialized software, and require additional input during protocol development. We present a practical application that utilizes the likelihood-based CRM to monitor an ongoing investigator initiated ‘3+3’ phase I trial for lymphoma patients at the University of Kentucky Markey Cancer Center. During this trial, dose-limiting toxicities were redefined and there were concerns about the flexibility of the ‘3+3’ to accommodate these changes mid-cohort. Since the CRM accommodates various cohort sizes and makes use of all accrued patient toxicity information to estimate the dose closest to the maximum tolerated dose at any point in the trial, we monitored the ongoing trial to confirm dose escalation rules according to the ‘3+3’ were appropriate. Additionally, we compared operating characteristics of various CRM designs that accommodate ordinal or binary toxicity grades. We describe what information was needed to construct the initial design as well as how current accrued toxicity data was included in the model to estimate the next dose needed, using the R package ordcrm. Interestingly, this monitoring exercise gave clinical investigators, biostatisticians, and other research team members the opportunity to learn more about the likelihood-based CRM and gain information regarding the logistical issues and advantages of applying this design in practice. Not only did this verify that the CRM is feasible to implement and incorporates more toxicity information throughout the trial, but it has educated clinical research team members on this alternative dose finding design and has sparked interest in using the likelihood-based CRM as the design in future trials. The utilization of randomization in phase II trials is increasingly being recommended in the evaluation of anticancer therapies prior to pursuing larger phase III trials. Randomized comparisons offer protection against selection bias and are needed when historical control data are not available for designing a new trial. Several candidate randomized designs can be considered depending on trial design and objectives. A phase II randomized trial in breast cancer of two candidate targeted therapies is planned in patients with locally advanced or metastatic breast cancer. The study endpoints are tumor response rate and progression-free survival (PFS). We consider different study designs to determine patient number requirements and assess feasibility and performance of the trial design under different scenarios. Specifically, we utilize i) selection design (design A); ii) two-stage selection design (design B); iii) screening design (design C); and iv) two-stage randomized phase II trial design (design D). Given the null and alternative hypotheses assumptions for response and PFS endpoints of this particular breast cancer trial, patient number requirements are reasonable across all potential trial designs. Designs C and D offer more conclusive results given the hypothesis-testing strategy of these designs. For the particular regimens and clinical problem being addressed in this phase II breast cancer trial, we favor the use of designs C or D which, although requiring slightly bigger patient numbers, affords more definitive conclusions on whether to proceed to a larger phase III trial. It is important to carefully assess a particular clinical trial scenario to decide not only whether to proceed with a single arm, historical control design but also the most appropriate randomized design if the latter is of interest. The Chinese Center for Disease Control and Prevention has successfully completed phase Ia and Ib clinical trials of one of their leading HIV vaccine candidates in 2010. While China is committed to launch its first HIV vaccine efficacy trial, it is important to appropriately choose the most promising vaccine regimen to advance from Phase IIa safety and immunogenicity testing to later phase efficacy testing. The candidate vaccine regimen is comprised of 3 DNA prime immunizations followed by one replicating Tiantan Vaccinia (rTV) virus vector boost. Since the timing of the vector boost will have great impact on the potency and longevity of the vaccine-induced immune responses, the Phase IIa trial is designed to evaluate 3 different boosting schedules and to select the best to advance to efficacy testing. For Phase IIa testing, since multiple immunogenicity and safety endpoints with an ordering of preference are of interest for assessment with respect to clinical and biological relevance and importance, we propose a ranking and selection design that accounts for multiple endpoints of different types and allows for equal ranking when vaccine arms are indistinguishable within certain pre-specified margin from the best arm. Bootstrap-based evaluation of the precision of ranking is performed for each endpoint. In addition, to prevent advancing any vaccine arms from trials with uniformly week candidates, we require that a vaccine regimen to be selected must exceed an absolute minimum response probability and/or magnitude threshold. Sample size calculations are performed to evaluate how many vaccinees per arm are needed to achieve high probability of correctly selecting the arm with the best combination of 4 endpoints. Preliminary simulation results show that such selection design performs well. As the distinction between the best arm(s) and inferior arm(s) becomes larger, the selection accuracy increases. Overall the probabilities of incorrect selection are low. Chinese Center for Disease Control and Prevention plans to launch the country’s first Phase IIb HIV vaccine efficacy trial in 2014. The candidate vaccine regimen is comprised of 3 DNA prime immunizations followed by one replicating Tiantan Vaccinia (rTV) virus vector boost. Supported by a supplementary fund to the HIV Vaccine Trials Network (HVTN) by NIAID, the HVTN statistical team is currently working closely with Chinese researchers on the design and planning of this trial. The proposed trial design is considered to be reviewed by the Chinese State FDA in early 2012. We intend to apply a contingency two-stage evaluation of vaccine efficacy (VE), where durability of VE over a range of 0-3 years will be evaluated in Stage 2 if and only if positive VE proximal (e.g. over the first 18-24 months) to the immunization series is detected in Stage 1. Such a strategy is under consideration for the design of future HVTN trials in South Africa which may launch after the start of this China trial. To foster the most synergetic collaboration, we expect to work with a collaborative trial design team including representatives from trial sponsors, regulators, vaccine developers, manufactures, the Institutional Review Board of China CDC and key HIV vaccine researchers from both US and China. We acknowledge that the design phase will be highly iterative in order to finalize various context-driven trial design parameters, including study population, trial endpoint, type I and type II error rate level, trial outcome monitoring type and monitoring plan. Because the primary HIV vaccine trial sponsors are, for the first time, outside of US and the design expertise is primarily from the US, experience gained and lessons learned during such a process will be a treasure to share with the international clinical trial community. While many Phase III clinical trials have efficacy and safety endpoints, sample size calculations are generally based on efficacy, with safety addressed separately as a secondary endpoint. However, the overall clinical effect often involves efficacy and adverse safety or side effect trade-offs. While joint efficacy and safety endpoints have often been used for Phase II dose finding studies, such approaches have found limited application in Phase III trials. Bronchopulmonary dysplasia (BPD) is a leading morbidity in extremely preterm infants, and prolonged mechanical ventilation increases BPD risk. Dexamethasone can facilitate extubation and decrease BPD incidence, but risk of neurodevelopment impairment (NDI) has led to decreased use. Because preliminary evidence suggests a better long-term safety profile for hydrocortisone than for dexamethasone, we designed a Phase III trial to sequentially evaluate the composite hypothesis that hydrocortisone (a) reduces the risk of death or BDP at 36 weeks; and (b) has an acceptable safety profile for death or NDI at 18 to 22 months. For the two- stage evaluation using robust Poisson models, we will first estimate relative risk and test the hypothesis that hydrocortisone lowers risk of death or BPD. If this test shows benefit, we will evaluate safety descriptively with safety considered achieved if either (1) the point estimate of risk of death/NDI is lower for hydrocortisone, or (2) death/NDI risk for hydrocortisone is increased, but the lower limit of a one-sided 95% confidence interval for the ratio of death/BPD benefit to death/NDI risk is greater than 4. We describe procedures used for power and sample size calculations for this two- stage testing approach, present results from those calculations, and compare operating characteristics of this approach to one previously reported that involves joint testing of the bivariate death/BPD and death/NDI outcomes using extensions of methods described by Tournoux, et al (Contemporary Clinical Trials; 28:514). The objective was to undertake a phase II trial (TO-PARP) to identify the best CRPC patient group(s) to be studied for sensitivity to olaparib in a phase III trial. A multistage phase II design has been adopted which has a non-randomised component with response as the primary endpoint followed by a randomised component with overall survival as the primary endpoint. Non-randomised component : This allows early progression to a randomised comparison if there is evidence of a high response rate (50% or more) in unselected patients. Biomarker defined groups are investigated for treatment sensitivity if the response rate is weaker than this. The first stage involves entry of 30 patients. If 50% or more respond then no more patients will be entered and the randomised component will be undertaken in unselected patients. If 2 (7%) or less respond then olaparib will be rejected. If between 3-14 (10%-47%) patients respond then a further 15 patients will be entered. Should 23 (51%) or more respond overall then the randomised component will be undertaken in unselected patients but if 5 (11%) or fewer respond then olaparib will be rejected. Otherwise with 6-22 responders (13%-49%), biomarker analysis of tissue collected from all 45 patients will be undertaken with the aim of identifying a sensitive subgroup, with a response rate which is compatible with a 50% response rate. If such a subgroup is found, a confirmatory single stage 44 patient trial will be undertaken in this group; this will also yield experience of prospective biomarker testing in a multi-centre clinical trial setting. Randomised component : a phase II assessment of the results generated in the non-randomised part, offering more secure evidence before proceeding to phase III. 180 patients will be randomised 2:1 to olaparib or an appropriate standard of care (? 1-sided 10%, power 80%). Food can have very different effects in terms of health, because of nutrient composition and dosage. In nutritional field, multiple outcomes or multiple hypotheses related to the inter-relationships among those outcomes are common. This induces several statistical issues related to the inflation of the statistical error. Such aspects are not commonly taken into account in nutritional research, whereas are very commonly accomplished for in the drug-related investigations. In the context of multi-functional food, there can be both multiple effects on the health target of a given nutrient, or multiple nutrients acting toward a single target or perhaps even on multiple targets. In nutritional fields it unrealistic to take hypothesis and outcomes as independent, mostly because nutrient effects are strongly related to each other. This induces a different correlation structure among outcomes and targets. Basically, if no health claims intended for components, then the composite endpoint is tested at a predefined alpha-level and components are not statistically tested: then adjustment for multiplicity is needed, and the basic choice would be on how to consider the concurrent outcomes. If claims are intended for components, then a sequential testing scheme might be applied: the composite endpoint is statistically significant ant the alpha-level, and then tests for components are made in sequence for the same significance level alpha. No claims can be made if sequence breaks. The usage of more complex models allows for instance to approach the problem as a fallback testing strategy: allow to continue testing even if sequence breaks, testing the hypothesis after the failed using a weighted Bonferroni method. This paper discuss the issues arising in multifunctional research by multiplicity in testing, both with regard to multiple nutrients and to multiple effects of a single or more nutrients on the people’s health. BACKGROUND. In the past few decades, it has become increasingly common for scientists based in richer countries to initiate clinical trials to be conducted in higher risk groups, in developing countries. This may creates opportunities for exploitation, as some activists and authors have argued, following early closure of several trials aiming at testing antiretrovirals for HIV prevention (PrEP), in 2004-2005. We aimed to compare the scientific methods of trials involving recruitment in developing countries to trials involving recruitment only in industrialized countries. METHODS. We conducted a systematic review, with a focus on efficacy/effectiveness trials. Based on guidance documents for the ethical conduct of HIV biomedical prevention trials (by UNAIDS/ WHO and by the Institute of Science), we developed a list of 33 key items. We searched Medline, Embase, Central, trial registries and the World Wide Web to identify eligible trials; and we communicated with investigators to obtain protocols and reports. Each protocol was fully reviewed by two independent assessors. RESULTS. We identified nine eligible trials, among which four had closed early at least one study site. We obtained full- text protocols for all efficacy trials. All but one trial had a Jadad score of 3/5 or more. On average, 20/33 validity items were reported in PrEP efficacy trials (61%). This average was 19/33 (58%) for trials closed early versus 2/33 (67%) for other trials. DISCUSSION. Early closures in HIV PrEP efficacy trials might be related to methodological issues. It is unclear how methods might have affected the ethical design and/or conduct of those trials.
TI - Abstracts from the Society for Clinical Trials Annual Meeting, Miami, May 21–23, 2012: 
JF - Clinical Trials
DO - 10.1177/1740774512453224
DA - 2012-08-24
UR - https://www.deepdyve.com/lp/sage/abstracts-from-the-society-for-clinical-trials-annual-meeting-miami-HHcVeWLa0J
SP - 450
EP - 554
VL - 9
IS - 4
DP - DeepDyve
ER -