British Journal of Surgery

doi: 10.1002/bjs.9656pmid: N/A

Article PDF first page preview Close This content is only available as a PDF. © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Alderson, D; Cromwell, D

doi: 10.1002/bjs.9641pmid: 25196475

During 2013, a political directive to publish operative mortality rates for individual surgeons across a variety of procedures in England was met with divided opinion among the surgical community, patients, the media and organizations responsible for regulation and delivery of surgical services. Justification for the initiative from National Health Service England was based on the notion that publishing these figures would identify poor surgical performance, improve the quality of care for surgical patients, help the public to make more informed choices about where to have an operation, and inform healthcare planners about the distribution of surgical resources. The usefulness of this exercise remains open to debate1. Nine surgical specialties published information during the summer of 2013, starting with figures on cardiac and vascular procedures. Some national media focused on a handful of surgeons with apparently high mortality rates after abdominal aortic aneurysm surgery, although their figures were within the expected range for the volume of activity provided by the Vascular Society of Great Britain and Ireland for 2013. The other surgical audits attracted no media interest, and no specific initiatives seem to have resulted from the information gathered. Surgeons in other healthcare systems might wonder whether they will also be required to provide this information in future, what consequences might ensue and how the profession should respond. To date, the issue has not received much attention in other countries but, where it has, there seems little support. A survey2 in the USA indicated that, although surgeons there supported public reporting of quality metrics at an institutional level, only a minority were in favour of individual reporting and only about one-quarter felt that this would improve outcomes. That some surgeons obtained better results than others was first pointed out in 19793, when both higher surgeon and provider (hospital) volume were shown to be associated with lower mortality across a range of complex operations. In the ensuing 35 years, evidence has accumulated in almost every branch of surgery that institutional and/or surgeon experience improves patient care. Reports4,5 suggesting that low-volume hospitals can do complex surgery equally well in terms of clinical outcomes, patient satisfaction and cost have been swamped by literature to the contrary. In some countries, this has resulted in the organization of surgical services that restricts the performance of high-risk operations to hospitals based on catchment populations or operative caseload6,7. No country has yet rationalized services solely on operative mortality rates achieved by individual surgeons. Separating the impact of individual performance on patient outcomes remains difficult. Evaluations of new surgical interventions indicate that a myriad of institutional factors influence outcome8. Can the publication of an individual surgeon's results truly indicate performance, or is this merely a surrogate reflecting available resources and overall performance within an institution? For example, in a study9 of coronary artery bypass grafting, aortic valve repair or abdominal aortic aneurysm repair, a substantially higher risk of death was seen among patients suffering serious complications after surgery in lower-volume hospitals compared with higher-volume hospitals. This suggests that the quality of postoperative care makes a vital contribution to outcome as well as operative skill. In an era of openness and public accountability, however, there is no reason why surgeons' outcomes should not be freely available. Anything else implies that surgeons have something to hide. The notion of publishing a rating of a surgeon's performance may seem unpleasant, but politicians and the media are likely to insist that patients have a right to know that surgeons are performing to a safe standard. The challenge for the surgical community is to ensure that the information is accurate and meaningful, and to work out how best to get this into the public domain. Reporting outcomes of specific operations fairly, in a way that can be accepted by the profession and understood by the public, is challenging. Few surgeons carry out large numbers of a specific operation in a short interval. A small denominator of procedures means that reliably identifying ‘outlying’ performance is difficult10. If a long observation time is needed to acquire this information, other changes in patient management will influence results. There is the issue of selecting an appropriate outcome measure. Last year, the surgical specialties were instructed to publish postoperative mortality. Although this might be relevant for a small number of high-risk operations, it is not meaningful for many other procedures. For groin hernia surgery, recurrence might be the appropriate outcome. Patient selection, referral practices and risk stratification for individual patients are all difficult issues, particularly when surgeons typically undertake operations for conditions affecting older patients with co-morbidities. Finally, the process may introduce some perverse incentives. If the publication of results is not mandatory, poorly performing surgeons may not report, thereby penalizing surgeons who are open and transparent. Conservatism as a means of reducing short-term complications and mortality might deprive some patients of cure. There is also the danger that data quality will be compromised, particularly for complications whose definition is open to interpretation. In all of this, there is risk that training might be affected adversely. The surgical community should meet this challenge. It may seem unfair that surgeons are being evaluated in a way that does not apply to most other hospital doctors, but the surgeon–patient relationship is quite different from that between patient and anaesthetist. Regulatory bodies and specialist associations should work together to deal with the above points. If it is accepted that there is a need for the profession to place consultant-level results in the public domain, then clinical audit must be compulsory at unit and personal level. This demands that data collection, attribution of work and recording are accurate, adjustments are made for differences in case mix, and results that are made public are relevant to each operation and easily interpretable. The last point is worth stressing. The experience of the first round of publications highlighted how the figures can be misrepresented. Education of the public, politicians and surgeons in the interpretation of these data is vital. A small survey11 in the USA suggested that even among surgeons only about half had reasonable comprehension of data validity, accuracy or complexity. The undeserved criticism of individual surgeons by an unscientific popular press that alarms the public can be allayed only by accurate data collection. The surgical community should recognize this as its responsibility, but must be provided with the necessary resources to do this. The notion that publication of surgeons' results helps the profession, as well as enabling patients to make more informed choices, is as yet unproven. Experiments have yet to be performed, and the design of studies to answer these questions will not be easy. If driven by external forces, the scientific rigour essential to do this properly may be compromised. Although it is important for the surgical community to rise to the challenge of knowing the value of each surgeon's results, this should not be confused with improving outcomes for patients. This is always more likely to be achieved by compulsory national clinical audits that report on an institutional basis. Addressing the issues raised above does assist this task, but the delivery of other services essential to good surgical outcomes means that political pressure must be brought to bear so that all aspects of a service are considered, rather than the surgeon in isolation. Insisting on institutional publication of results is far more likely than anything else to help patients in the long run. Disclosure The authors declare no conflict of interest Open in new tabDownload slide References 1 NHS choices . Consultant Treatment Outcomes . http://www.nhs.uk/choiceintheNHS/Yourchoices/consultant-choice/Pages/consultant-data.aspx [accessed 6 February 2014]. 2 Sherman KL , Gordon EJ, Mahvi DM, Chung J, Bentrem DJ, Holl JL et al. Surgeons' perceptions of public reporting of hospital and individual surgeon quality . Med Care 2013 ; 51 : 1069 – 1075 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Luft HS , Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation between surgical volume and mortality . N Engl J Med 1979 ; 301 : 1364 – 1369 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Killeen SD , O'Sullivan MJ, Coffey JC, Kirwan WO, Redmond HP. Provider volume and outcomes for oncological procedures . Br J Surg 2005 ; 92 : 389 – 402 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Yoshioka R , Yasunaga H, Hasegawa K, Horiguchi H, Fushimi K, Aoki T et al. Impact of hospital volume on hospital mortality, length of stay and total costs after pancreaticoduodenectomy . Br J Surg 2014 ; 101 : 523 – 529 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Finks JF , Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery . N Engl J Med 2011 ; 364 : 2128 – 2137 . Google Scholar Crossref Search ADS PubMed WorldCat 7 De Wilde RF , Besselink MGH, van der Tweel I, de Hingh IHJT, van Eijck CHJ, Dejong CHC et al. Impact of nationwide centralization of pancreaticoduodenectomy on hospital mortality . Br J Surg 2012 ; 99 : 404 – 410 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 8 Ergina PL , Cook JA, Blazeby JM, Boutron I, Clavien PA, Reeves BC et al. Challenges in evaluating surgical innovation . Lancet 2009 ; 374 : 1097 – 1104 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Gonzalez AA , Dimick JB, Birkmeyer JD, Ghaferi AA. Understanding the volume–outcome effect in cardiovascular surgery: the role of failure to rescue . JAMA Surg 2014 ; 149 : 119 – 123 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Walker K , Neuburger J, Groene O, Cromwell DA, van der Meulen J. Public reporting of surgeon outcomes: low numbers of procedures lead to false complacency . Lancet 2013 ; 382 : 1674 – 1677 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Yi SG , Wray NP, Jones SL, Bass BL, Nishioki J, Brann S et al. Surgeon-specific performance reports in general surgery: an observational study of initial implementation and adoption . J Am Coll Surg 2013 ; 217 : 636 – 647 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Steele, R J C

doi: 10.1002/bjs.9599pmid: 25088598

Population screening for colorectal cancer has now been rolled out across all the constituent countries of the UK. Although the algorithms differ slightly from one country to another, the first-line screening test is a guaiac-based faecal occult blood test (gFOBT) with the same analytical characteristics as those used in the randomized clinical trials (RCTs) that demonstrated convincing reductions in disease-specific mortality1. A recent matched cohort study has indicated that the current programme is producing a mortality reduction of the expected magnitude2, even though the guaiac-based test has significant weaknesses. The overall uptake of screening remains stubbornly below 60 per cent1. Interval cancers (those arising within 2 years of a negative test result) are being increasingly recognized, with recent data suggesting that interval disease accounts for approximately 50 per cent of the cancers diagnosed in the screened population3. A final disadvantage relates to the lack of efficacy of the current programme in the prevention of colorectal cancer; gFOBT at the analytical sensitivity used in the UK does not identify sufficient numbers of adenomas to have a measurable preventive effect4. Although these UK screening programmes are undoubtedly preventing colorectal cancer deaths, future developments must address these deficiencies. One approach is to move to quantitative faecal immunochemical testing (QFIT) for blood. This has the advantages of being specific for haemoglobin and providing a value for the concentration of haemoglobin in the stool. The best evidence supporting the use of QFIT comes from a randomized study5 in the Netherlands comparing QFIT at a cut-off of 100 ng haemoglobin per ml faeces in buffer with a guaiac-based test that has an analytical detection limit of around 400 ng haemoglobin per ml faeces. This study demonstrated better uptake with QFIT, and a higher sensitivity for both cancer and adenoma detection with a similar number needed to colonoscope. However, this came at the expense of a higher positivity rate and therefore the need for a larger overall number of colonoscopies. If QFIT is used at the same analytical sensitivity as the standard gFOBT, both tests perform in much the same way in terms of disease detection. For QFIT to reduce the interval cancer rate and increase the chances of adenoma detection it seems that it would be necessary to reduce the analytical threshold such that the positive predictive value would decrease and the number of colonoscopies required would increase. This is currently unpalatable given the pressures on endoscopy services in the UK, but there is increasing evidence that using QFIT to triage symptomatic patients could dramatically reduce the number of colonoscopies required in the symptomatic population and free up resources for screening6. Furthermore, it should be possible to use the quantitative power of QFIT in a more intelligent way. For example, it is known that interval cancers are more common in women than in men, and that the concentration of blood in stool is significantly lower in women than in men7. It should therefore be possible to address the problem of interval cancers in women by using a differential cut-off to trigger colonoscopy. It is also feasible to decrease the analytical threshold and increase the interval between screens so that sensitivity of the test may be improved in a cost-neutral manner in terms of colonoscopy resource. With a little more imagination it may also be possible to vary the interval between screens based on the index faecal haemoglobin concentration, as it is known that this is directly proportional to the risk of having significant neoplastic disease8. Thus an individual with a faecal haemoglobin concentration close to a predetermined cut-off may have a repeat test within 6 months, whereas someone with undetectable haemoglobin may have their next test deferred for a few years. Little is known about variations in faecal haemoglobin levels with time and this requires attention if we are to optimize screening based on faecal haemoglobin detection. Of course, there is good evidence from RCTs that using flexible sigmoidoscopy as the primary screening test can reduce both the mortality from, and incidence of, colorectal cancer9. This test is being rolled out at around the age of 55 years across England, but not in Scotland where FOBT screening starts at age 50 years as opposed to 60 years in England. A pilot study of flexible sigmoidoscopy offered at the age of 60 years superimposed on the current FOBT screening programme in Scotland has been commissioned, and is about to commence. One of the problems with flexible sigmoidoscopy screening is that it does not appear to have a preventive effect on right-sided colonic cancer, even when colonoscopy is performed in those with high-risk index lesions found on initial examination9. It follows that colonoscopy might be a better screening test, but observational studies suggest that, although colonoscopy can prevent left-sided cancer, it is not as effective at preventing right-sided cancer9, presumably because the precursor lesions on the right side of the colon are more subtle than those on the left, and because bowel preparation on the right side is often suboptimal. Currently there are four RCTs of colonoscopy as the primary screening test in progress worldwide, but none has yet reported. FOBT screening, as employed in the UK, is effective in reducing colorectal cancer mortality, but can be enhanced by the use of more sophisticated and sensitive testing for blood and by supplementing FOBT with primary endoscopic screening. What, however, might the future hold? For many years there has been interest in testing stool for DNA mutations associated with colorectal cancer. Recent research suggests that, by using an appropriate panel of DNA abnormalities, this can outperform faecal immunochemical testing at a cut-off level of 100 ng haemoglobin per ml, at least in terms of sensitivity10. This technology is not yet sufficiently developed, however, to be used in population screening, where ease of testing and cost are paramount. There has also been interest in using biomarkers in blood as screening tests, but to date no biomarker has been found with a sensitivity and specificity any better than those achieved in testing for blood in stool. For now, the most important development in the UK should be the replacement of gFOBT with QFIT, along with high-quality prospective studies to establish the role of flexible sigmoidoscopy within national programmes. The UK is not the only country to engage in population screening. In Europe, Finland was early on the scene with gFOBT, and Denmark, the Netherlands and the Republic of Ireland have recently started screening with QFIT. France has a gFOBT-based programme, but is exploring the use of QFIT, Germany offers both gFOBT and colonoscopy, and Italy has regional programmes that use QFIT and flexible sigmoidoscopy. Other countries, such as Sweden and Spain, are exploring different options. Outside Europe, there are screening programmes based on faecal testing in Australia, New Zealand, Japan, Taiwan and Canada. In the USA, colorectal screening is delivered largely by opportunistic colonoscopy. The major issue that faces the international colorectal screening community is the plethora of different approaches, not all of them evidence-based, and not all operating as true population screening programmes. Firmly evidence-based guidelines are available11, and for screening to have a truly global impact it is essential that this evidence is translated into public health action. Disclosure The author declares no conflict of interest. Snapshot quiz 14/14 References 1 Logan RFA , Patnick J, Nickeson C, Coleman L, Rutter MD, von Wagner C; English Bowel Cancer Screening Evaluation Committee. Outcomes of the Bowel Cancer Screening Programme (BCSP) in England after the first 1 million tests . Gut 2012 ; 61 : 1439 – 1446 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Libby G , Brewster DH, McClements PL, Carey FA, Black RJ, Birrell J et al. The impact of population-based faecal occult blood test screening on colorectal cancer mortality: a matched cohort study . Br J Cancer 2012 ; 107 : 255 – 259 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Steele RJC , McClements P, Watling C, Libby G, Weller D, Brewster DH et al. Interval cancers in a FOBT-based colorectal cancer population screening programme: implications for stage, gender and tumour site . Gut 2012 ; 61 : 576 – 581 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Scholefield JH , Moss SM, Mangham CM, Whynes DK, Hardcastle JD. Nottingham trial of faecal occult blood testing for colorectal cancer: a 20-year follow-up . Gut 2012 ; 61 : 1036 – 1040 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Van Rossum LG , van Rijn AF, Laheij RJ, van Oijen MG, Fockens P, van Krieken HH et al. Random comparison of guaiac and immunochemical faecal occult blood test for colorectal cancer in a screening population . Gastroenterology 2008 ; 135 : 82 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat 6 McDonald PJ , Digby J, Innes C, Strachan JA, Carey FA, Steele RJC et al. Low faecal haemoglobin concentration potentially rules out significant colorectal disease . Colorectal Dis 2013 ; 15 : e151 – e159 . Google Scholar Crossref Search ADS PubMed WorldCat 7 McDonald PJ , Strachan JA, Digby J, Steele RJC, Fraser CG. Faecal haemoglobin concentrations by gender and age: implications for population-based screening for colorectal cancer . Clin Chem Lab Med 2011 ; 50 : 935 – 940 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 8 Digby J , Fraser CG, Carey FA, McDonald PJ, Strachan JA, Diament R et al. Faecal haemoglobin concentration is related to severity of colorectal neoplasia . J Clin Pathol 2013 ; 66 : 415 – 419 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Brenner H , Stock C, Hoffmeister M. Effect of screening sigmoidoscopy and screening colonoscopy on incidence and mortality: systematic review and meta-analysis of randomised controlled trials and observational studies . BMJ 2014 ; 348 : g2467 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Imperiale TF , Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP et al. Multitarget stool DNA testing for colorectal-cancer screening . N Engl J Med 2014 ; 370 : 1287 – 1297 . Google Scholar Crossref Search ADS PubMed WorldCat 11 European Commission . European Guidelines for Quality Assurance in Colorectal Cancer Screening and Diagnosis (1st edn). Publications Office of the European Union : Luxembourg , 2011 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Bloo, G J A; Hesselink, G J; Oron, A; Emond, E J J M; Damen, J; Dekkers, W J M; Westert, G; Wolff, A P; Calsbeek, H; Wollersheim, H C

doi: 10.1002/bjs.9609pmid: 25093587

Abstract Background Insight into the effects of ethnic disparities on patients' perioperative safety is necessary for the development of tailored improvement strategies. The aim of this study was to review the literature on safety differences between patients from minority ethnic groups and those from the ethnic majority undergoing surgery. Methods PubMed, CINAHL, the Cochrane Library and Embase were searched using predefined inclusion criteria for available studies from January 1990 to January 2013. After quality assessment, the study data were organized on the basis of outcome, statistical significance and the direction of the observed effects. Relative risks for mortality were calculated. Results After screening 3105 studies, 26 studies were identified. Nine of these 26 studies showed statistically significant higher mortality rates for patients from minority ethnic groups. Meta-analysis demonstrated a greater risk of mortality for these patients compared with patients from the Caucasian majority in studies performed both in North America (risk ratio 1·22, 95 per cent confidence interval 1·05 to 1·42) and outside (risk ratio 2·25, 1·40 to 3·62). For patients from minority groups, the length of hospital or intensive care unit stay was significantly longer in five studies, and complication rates were significantly higher in ten. Methods used to identify patient ethnicity were not described in 14 studies. Conclusion Patients from minority ethnic groups, in North America and elsewhere, have an increased risk of perioperative death and complications. More insight is needed into the causes of ethnic disparities to pursue safer perioperative care for patients of minority ethnicity. Introduction Patient safety has been set as a priority by researchers and policy-makers since the 1999 landmark report ‘To Err Is Human’ from the Institute of Medicine1. Patient safety can be defined as ‘the prevention of unintended harm to patients, not due to an inherent medical risk but as a result of failure by healthcare professionals or the healthcare system’1–3. A primary concern in patient safety is the perioperative care process, which often involves handover of care4,5, and complicated surgical treatments in vulnerable patients6–8. An unsafe perioperative care process can result in considerable suffering, unnecessary treatment and extra costs. Some 51–77 per cent of adverse events in hospitals are attributable to the perioperative care process; these can result in postoperative infections, temporary or permanent disability or death9–11. Approximately half of these adverse events are considered preventable9,10. Many factors influence patient safety in the perioperative care process, for example standardization of work processes5,12, use of information technology13, care provider behaviour14 and knowledge15, and organizational structures16. There is growing awareness of the relationship between ethnic background and perioperative patient safety17,18. The literature suggests that a patient's ethnicity can influence the quality of communication between care providers and patients19–21, patient adherence to treatment22,23 and care provider adherence to evidence-based guidelines24, thereby affecting the likelihood of errors, complications and death. Ethnicity can be defined as ‘a collection of common factors such as origin, appearance, religion, language, culture or history shared by a group of people and transmitted from generation to generation’25. Ethnic identity indicates an individual's feelings or emotional attachment toward a specific ethnic group26 and the objective criteria that can be allocated by others, such as common ancestry27. Despite the potential impact of patient ethnicity on perioperative patient safety, insight into the effects of ethnic disparities on health outcomes and the underlying mechanisms that lead to these disparities is lacking. Previous reviews examining variation in surgical care outcomes have several limitations, supporting the need for the present review. Previous reviews did not focus explicitly on ethnic disparity2,28 or evaluate disparities in a single country only2. The aim of this study was to review systematically international literature on differences in patient safety between patients from minority ethnic groups and those of majority ethnicity undergoing surgery, and to identify possible explanations for any effects found. Methods This systematic review was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist29. Data sources English-language studies published between December 1990 and January 2013 were sought using the following databases: PubMed (including MEDLINE), Cumulative Index to Nursing and Allied Health Literature (CINAHL), Embase and the Cochrane Library. Appendix S1 (supporting information) provides a detailed listing of the search terms. The references of the selected studies were checked manually to identify additional relevant studies that were missed in the database search. Study selection Two reviewers assessed the inclusion eligibility of the retrieved studies independently, using the search strategy. The initial selection for inclusion was based on the title and a scan of the abstract of the study. When the title and abstract provided insufficient information, a full-text copy of the article was reviewed. For the final selection, a full-text copy of the study was examined; disagreement about inclusion was resolved by discussion. For inclusion, each study had to meet four criteria: published in English from January 1990 to January 2013 as a full-text article or dissertation with an English-language title and abstract; conducted in patients who had surgical treatment; contained a comparison between a majority ethnic group (a particular ethnic group that comprised the majority of a particular population) and minority ethnic groups; and studied at least one of the following outcomes – postoperative mortality, morbidity or complications, reoperation within 30 days of the first operation, unplanned intensive care unit (ICU) admission, length of stay, physical functioning or pain. Studies were excluded if they involved transplant or dental surgery, mental health or diagnostic coronary angiography, or included patients younger than 18 years or who were cognitively impaired, or where ethnicity was related to organ transplant studies. Quality assessment of methods Two reviewers assessed the methodological quality of the full-text studies independently and discussed the results to obtain consensus. The Agency for Healthcare Research and Quality guidelines for rating the quality of comparative studies were used, and modified to ensure standard scoring. Methodological quality was assessed on the basis of nine criteria: studies had to specify study design, eligibility criteria, selection of participants, minimal baseline differences between groups, sample size calculation, subjects excluded, statistical methods, outcomes controlled for confounding and effect modifiers, and potential sources of bias. Data extraction Each selected article was abstracted independently by two reviewers, using a standard form modified from a checklist developed by Grimshaw and colleagues30. The data extracted from the studies comprised a description of objectives, design, setting, participants and outcome measurements. Any disagreement was resolved by discussion. Data synthesis and analysis Study outcomes have been organized in tabular form and qualitative assessment was based on the methodological quality of the study, the design, setting, sample size, identification of patient ethnicity, outcomes and statistical significance of outcomes observed. The results of the individual studies were combined into a meta-analysis. For this purpose Review Manager 5.0 software was used31. Risk ratios (RRs) were calculated for mortality per minority ethnic group for all of the studies combined to determine overall differences in mortality between patients with minority ethnicity and the majority Caucasian group. The data were organized by ethnicity, in-hospital mortality rate, 30-day mortality rate, origin of the study (countries, region) and the type of surgery before calculation of the RRs. Heterogeneity was tested by the Mantel–Haenszel method, using fixed-effect and random-effects models. Results Search results The initial search identified 3105 citations, of which 998 were in PubMed, 227 in CINAHL, 638 in the Cochrane Library and 1242 in Embase (Fig. 1). The title and abstract scan resulted in 49 papers that appeared to meet the inclusion criteria. Twenty-three papers were excluded after full-text review. No additional papers were identified by manual review of the reference lists of the original papers. The final set consisted of 26 published studies32–57 that underwent full-text evaluation. Fig. 1 Open in new tabDownload slide Flow diagram of the search process Methodological quality The overall methodological quality of the studies was moderate (Table S1, supporting information). The scores ranged from 4 to 9 (mean(s.d.) fulfilled criteria 5·9(1·3)). Seven of the 26 studies scored high (7–9 points) and 19 studies scored moderate (4–6 points) on quality. Twenty-five studies specified their eligibility criteria. The selection procedure of participants and potential sources of bias were described in 19 studies. Minimal baseline differences between groups were established in nine studies. Sample size was calculated in three studies. Eleven studies described the subjects excluded from the analysis. Nineteen studies controlled for confounders or effect modifiers. Characteristics of included studies Table S2 (supporting information) summarizes the characteristics of the included studies. Nineteen studies were conducted in North America. Ten studies concerned cardiac and 16 non-cardiac surgery. The study populations consisted of patients undergoing cardiac, arterial, gastrointestinal, urological, cervical spine or joint replacement surgery. Patients were admitted to various types of hospital (general, veteran, tertiary, teaching and university-affiliated). The sample size ranged from four to 166 858 participants for the majority ethnic and from nine to 14 006 participants for minority ethnic groups. Twenty-two of the 26 studies were retrospective with secondary analyses of patient medical charts from large hospitals or governmental databases. The studies reported various types of outcome (Table 1; Table S2, supporting information). Nineteen studies32–50 reported one or more outcomes related to postoperative mortality, 18 studies32–33,36–39,41–43,45–47,50–55 reported postoperative complications (for example any complications, graft failure, stroke, wound infection), and 11 studies32,38–39,42,44,47,50–51,54,56–57 reported hospital or ICU length of stay. One other reported outcome was functional status56. Table 1 Types of outcome and statistically significant differences between ethnic majority and minority patient cohorts in included studies Reference . Year . Mortality . Length of hospital stay . Length of ICU stay . Complications* . Cardiac surgery Murphy et al.33 2010 ± DiGiorgi et al.46 2008 Yeo et al.40 2007 + Rumsfeld et al.43 2002 − + Nallamothu et al.44 2001 + Verderber et al.53 1999 ± Brister et al.39† 2007 + − + Goldsmith et al.47† 1999 Elahi et al.38† 2008 ± ± ± Hadjinikolaou et al.50† 2010 + ± − Non-cardiac surgery Norwood et al.45† 2009 Tan et al.55† 2008 + Collins et al.54† 2001 + Barocas et al.32 2012 + + − Alosh et al.34 2009 + Bergés et al.56 2008 Cheung et al.35 2009 + Halm et al.36 2009 + Robinson et al.37 2009 + Ibrahim et al.57 2008 ± Neighbors et al.49 2007 Yaghoubian et al.51 2007 Alvord et al.41 2005 + Horner et al.42 2002 + + Toursarkissian et al.52 2002 + Lavery et al.48 1997 Reference . Year . Mortality . Length of hospital stay . Length of ICU stay . Complications* . Cardiac surgery Murphy et al.33 2010 ± DiGiorgi et al.46 2008 Yeo et al.40 2007 + Rumsfeld et al.43 2002 − + Nallamothu et al.44 2001 + Verderber et al.53 1999 ± Brister et al.39† 2007 + − + Goldsmith et al.47† 1999 Elahi et al.38† 2008 ± ± ± Hadjinikolaou et al.50† 2010 + ± − Non-cardiac surgery Norwood et al.45† 2009 Tan et al.55† 2008 + Collins et al.54† 2001 + Barocas et al.32 2012 + + − Alosh et al.34 2009 + Bergés et al.56 2008 Cheung et al.35 2009 + Halm et al.36 2009 + Robinson et al.37 2009 + Ibrahim et al.57 2008 ± Neighbors et al.49 2007 Yaghoubian et al.51 2007 Alvord et al.41 2005 + Horner et al.42 2002 + + Toursarkissian et al.52 2002 + Lavery et al.48 1997 * Unplanned coronary artery bypass graft, postprocedure emergency percutaneous coronary intervention, postprocedure arrhythmia, postprocedure stroke, renal failure, vascular bleeding, congestive heart failure, postprocedure shock, low output syndrome, sternal wound infection, stroke, pulmonary complications, intensive care unit (ICU) readmission, pain, reoperation, postoperative wound infection, graft failure, treated ventricular extrasystoles. † Non-North American study. +, Significantly higher scores for one or more minority ethnic groups; −, significantly lower scores for one or more minority groups; ±, significantly higher and lower scores for one or more minority groups. Open in new tab Table 1 Types of outcome and statistically significant differences between ethnic majority and minority patient cohorts in included studies Reference . Year . Mortality . Length of hospital stay . Length of ICU stay . Complications* . Cardiac surgery Murphy et al.33 2010 ± DiGiorgi et al.46 2008 Yeo et al.40 2007 + Rumsfeld et al.43 2002 − + Nallamothu et al.44 2001 + Verderber et al.53 1999 ± Brister et al.39† 2007 + − + Goldsmith et al.47† 1999 Elahi et al.38† 2008 ± ± ± Hadjinikolaou et al.50† 2010 + ± − Non-cardiac surgery Norwood et al.45† 2009 Tan et al.55† 2008 + Collins et al.54† 2001 + Barocas et al.32 2012 + + − Alosh et al.34 2009 + Bergés et al.56 2008 Cheung et al.35 2009 + Halm et al.36 2009 + Robinson et al.37 2009 + Ibrahim et al.57 2008 ± Neighbors et al.49 2007 Yaghoubian et al.51 2007 Alvord et al.41 2005 + Horner et al.42 2002 + + Toursarkissian et al.52 2002 + Lavery et al.48 1997 Reference . Year . Mortality . Length of hospital stay . Length of ICU stay . Complications* . Cardiac surgery Murphy et al.33 2010 ± DiGiorgi et al.46 2008 Yeo et al.40 2007 + Rumsfeld et al.43 2002 − + Nallamothu et al.44 2001 + Verderber et al.53 1999 ± Brister et al.39† 2007 + − + Goldsmith et al.47† 1999 Elahi et al.38† 2008 ± ± ± Hadjinikolaou et al.50† 2010 + ± − Non-cardiac surgery Norwood et al.45† 2009 Tan et al.55† 2008 + Collins et al.54† 2001 + Barocas et al.32 2012 + + − Alosh et al.34 2009 + Bergés et al.56 2008 Cheung et al.35 2009 + Halm et al.36 2009 + Robinson et al.37 2009 + Ibrahim et al.57 2008 ± Neighbors et al.49 2007 Yaghoubian et al.51 2007 Alvord et al.41 2005 + Horner et al.42 2002 + + Toursarkissian et al.52 2002 + Lavery et al.48 1997 * Unplanned coronary artery bypass graft, postprocedure emergency percutaneous coronary intervention, postprocedure arrhythmia, postprocedure stroke, renal failure, vascular bleeding, congestive heart failure, postprocedure shock, low output syndrome, sternal wound infection, stroke, pulmonary complications, intensive care unit (ICU) readmission, pain, reoperation, postoperative wound infection, graft failure, treated ventricular extrasystoles. † Non-North American study. +, Significantly higher scores for one or more minority ethnic groups; −, significantly lower scores for one or more minority groups; ±, significantly higher and lower scores for one or more minority groups. Open in new tab Identification of patient ethnicity Table S3 (supporting information) provides an overview of the methods used to identify patient ethnicity and designate patients to a majority or minority ethnic group. Twelve studies reported the ethnic identification method used. Ethnicity was self-reported by patients in six studies37,40,45,50,53,55, and in three studies36,41,54 ethnicity was determined by care providers. In one study57, ethnicity was determined by both the patient and a care provider; in one study46, ethnicity was determined by the patient and family members; and in another study39, ethnicity was determined by assessors. Fourteen studies did not report the methods used to identify patient ethnicity. Only one study39 described a comprehensive, multimethod identification process, using independent assessors to check the patients' first and last names, patient confirmation of ethnicity assessment via a randomly distributed questionnaire and strict inclusion criteria (patients were included only if both parents were of the same ethnicity). Outcome differences between majority and minority ethnic groups Outcome differences in the selected studies between majority and minority ethnic patient groups are presented in Table 1. In nine studies32,34–36,39–41,44,50, perioperative mortality rates were significantly higher for one or more minority patient groups, compared with the majority group. In one study43, mortality rates were significantly higher for the majority group relative to the Hispanic minority group. The overall mortality risk was 30 per cent higher for minority groups than for white Caucasian patients (RR 1·30, 95 per cent confidence interval 1·11 to 1·52) (Fig. 2). Compared with the majority Caucasian group, Asians (RR 1·52, 1·25 to 1·84), Native Americans (RR 1·66, 1·30 to 2·12) and African Americans (RR 1·15, 1·02 to 1·29) had a higher mortality risk (Fig. S1a–c, supporting information). For Hispanic patients, the mortality rate did not differ significantly from that for white Caucasian patients (Fig. S1d, supporting information). Further analyses showed a significantly higher mortality rate for minority ethnic groups in both North American (RR 1·22, 1·05 to 1·42) (Fig. S1e, supporting information) and non-North American (RR 2·25, 1·40 to 3·62) studies (Fig. S1f, supporting information). Fig. 2 Open in new tabDownload slide Forest plot comparing perioperative mortality in patients of minority ethnicity versus white Caucasian patients. A Mantel–Haenszel random-effects model was used for meta-analysis. Risk ratios are shown with 95 per cent confidence intervals In five cardiac surgery studies33,38–39,43,53 and five non-cardiac surgery studies37,42,52,54–55, patients from minority ethnic groups had statistically significant higher complication rates compared with the ethnic majority. Examples included sternal wound infection (minority 2·7 per cent versus majority 1·4 per cent; P < 0·005)39, myocardial infarction (4·6 versus 2·6 per cent respectively; P < 0·05)39 and graft failure within 30 days (11 versus 5 per cent respectively; P = 0·012)37. Ethnic minorities scored significantly higher on perceived pain55 and reoperation within 30 days (13·3 and 17·5 per cent for two minority groups versus majority 11·9 per cent; P < 0·05)42. In five studies32,38,42,50,57 patients from minority groups had a significantly longer postoperative hospital stay than the majority group, and in four studies38–39,50,57 the majority stayed significantly longer in hospital. One study38 showed a significantly longer ICU stay for two ethnic minority groups and a shorter ICU stay for two other minority groups, both compared with the ethnic majority group. Explanations for outcome differences provided by the studies Fifteen studies provided possible explanations for the postoperative outcome differences found between the majority and minority ethnic patient cohorts (Table 2). Ten studies related their findings to differences in presurgical risk (for example prevalence of diabetes or obesity)38,48,50 and socioeconomic status (for instance insurance, accessibility to healthcare)33,46. Respectively, nine and eight of the studies described cultural factors (such as preferences, attitudes, habits, lifestyle) and biological or genetic factors (for example cholesterol level, quality of coronary arteries) as possible reasons32,37. Table 2 Explanations for perioperative outcome differences between ethnic majority and minority patients suggested in included studies Presurgical risks38,48,50 Biological or genetic factors32,37 Socioeconomic status33,46 Language barriers56 Cultural factors (for example preferences, attitudes, habits, lifestyle)32,37 Accessibility to healthcare (insurance)33,46 Biased healthcare providers (actual or perceived discrimination, stereotyping or prejudice towards minorities)55 Hospital volume34,36,49 Healthcare provider's experience and skills36,49,57 Geographical factors41,48 Presurgical risks38,48,50 Biological or genetic factors32,37 Socioeconomic status33,46 Language barriers56 Cultural factors (for example preferences, attitudes, habits, lifestyle)32,37 Accessibility to healthcare (insurance)33,46 Biased healthcare providers (actual or perceived discrimination, stereotyping or prejudice towards minorities)55 Hospital volume34,36,49 Healthcare provider's experience and skills36,49,57 Geographical factors41,48 Open in new tab Table 2 Explanations for perioperative outcome differences between ethnic majority and minority patients suggested in included studies Presurgical risks38,48,50 Biological or genetic factors32,37 Socioeconomic status33,46 Language barriers56 Cultural factors (for example preferences, attitudes, habits, lifestyle)32,37 Accessibility to healthcare (insurance)33,46 Biased healthcare providers (actual or perceived discrimination, stereotyping or prejudice towards minorities)55 Hospital volume34,36,49 Healthcare provider's experience and skills36,49,57 Geographical factors41,48 Presurgical risks38,48,50 Biological or genetic factors32,37 Socioeconomic status33,46 Language barriers56 Cultural factors (for example preferences, attitudes, habits, lifestyle)32,37 Accessibility to healthcare (insurance)33,46 Biased healthcare providers (actual or perceived discrimination, stereotyping or prejudice towards minorities)55 Hospital volume34,36,49 Healthcare provider's experience and skills36,49,57 Geographical factors41,48 Open in new tab Less frequently mentioned explanations include language barrier56, geographical factors (such as access to preferred hospital)41,48, biased healthcare provider (actual or perceived discrimination, stereotyping or prejudice toward minorities)55, hospital volume34,36,49, and the healthcare providers' experience and skills36,49,57. However, as explanations were provided post hoc, they remain speculative and warrant further investigation. Discussion This systematic review suggests that there is an association between patient ethnicity and less safe perioperative care. Most studies (19 of 26) showed statistically significant differences in a variety of outcomes (postoperative mortality, length of ICU or hospital stay, ICU readmission, complications, pain and reoperation) between the majority ethnic group and one or more minority ethnic groups. In all of these 19 studies, patients from minority groups scored less favourably than those from the majority group for one or more outcomes. The findings show that, both in and outside North America, patients from minority ethnic groups tend to experience a higher risk of dying in the perioperative period compared with the Caucasian majority. Although this was known for minority groups in North America3,17,56, this study provides a novel insight into the position of patients from ethnic minorities in healthcare systems elsewhere. Based on the present data, the risk of dying in the perioperative care process is even higher for those with minority ethnicity outside North America. The observation of differences in perioperative safety for minority ethnic groups warrants additional research on the underlying contributing factors at patient, provider and systemic levels2,58. Many of the included studies in the present review provided a variety of explanations for the outcome differences, ranging from biological characteristics (such as genetic factors) and presurgical risks to sociocultural features and biased care providers. The explanations, however, remain speculative, although they are in line with the US Agency for Healthcare Research and Quality 2012 National Healthcare Disparities Report59, which reported that physician decision-making, hospital characteristics and interpersonal processes (the way in which patients and clinicians interact) affect the healthcare received by patients in North America and the outcomes of that care. Furthermore, Haider and colleagues2 mentioned patient willingness to undergo surgery, late presentation of disease, disparity in disease burden and surgeon quality as factors that may contribute to patient disparities in surgical outcomes. Interventions should therefore target patients from minority ethnic groups who, based on this review, face higher perioperative safety risks, with a focus on improving care provider and patient communication, for instance by involving relatives in the decision-making process, and changing attitudes, for example avoiding care provider prejudice and increasing patient empowerment. These findings must be interpreted in the light of the limitations of this study. First, the included studies show a great variability in types of ethnic group, outcomes and outcome definitions, as well as a high degree of clinical diversity. These factors made it difficult to synthesize results and to draw general conclusions. Second, when measuring patient safety, it is suggested that measuring organizational or professional failures is more effective than measuring clinical outcomes60. The retrospective, secondary analysis of clinical data, performed in many of the included studies, increased the likelihood of biased reporting (selective reporting and underdetection) of outcomes, such as postoperative complications60–62. Interestingly, none of the included studies used adverse events, near misses or incidents as outcome measures, whereas such measures are considered to be primary safety measures9,63. Third, the review might have been influenced by selection bias because the study participants were selected without randomization, and differences between the baseline characteristics of the ethnic groups existed for most of the included studies. The data sets used were predominantly from patients from the ethnic majority, in combination with a small sample size of minority ethnic cohorts. Further, information bias might have occurred in the identification of patient ethnicity in the included studies. The representation of an ethnic cohort was questionable when different ethnic groups were classified into one category (for example Japanese, Chinese and Philippine into one ‘Asian’ category, different Latino-American populations into one ‘Hispanic’ category or different Native American tribes into one ‘Native American’ category). As ethnicities, the terms Asian, Hispanic and Native American encompass a variety of racial, geographic, genetic and cultural backgrounds, making it difficult to draw conclusions to all patients from these designated categories. The validity of patient ethnicity was also challenged by the poor identification methods used in the studies, thereby increasing the risk of misclassification64,65. For example, most of the studies lacked strict ethnicity inclusion criteria or ethnicity assessment techniques, and instead relied on the available data. Fourth, the review might have been influenced by publication bias; studies might not have been published, because no differences between ethnic groups were found. As the funnel plot was roughly symmetrical (Fig. S2, supporting information), publication bias was not expected to be present. Fifth, although 19 of the 26 included studies controlled outcomes for confounding variables, only eight controlled for the following important factors that might contribute to additional safety risks for patients with minority ethnicity: socioeconomic status, hospital volume and quality, accessibility of services and patient lifestyles. Therefore, safety differences found in this review may relate to factors other than ethnicity alone. The importance of including all potential confounders in ethnic disparity studies was addressed in a recent study by Rangrass and co-workers66, who showed that 53 per cent of the observed disparities could be explained by socioeconomic status and hospital quality. Further studies are needed to draw general conclusions regarding the effects of ethnic disparity on perioperative safety. Attention should be paid to defining key concepts such as ‘ethnicity’ and ‘ethnic minority’, using a comprehensive and rigorous ethnicity identif ication method, and uniform and valid outcomes to ensure accurate evaluation of effects among studies. A prospective, mixed-methods design including all potential confounders, combining quantitative methods to compare the incidence of perioperative adverse events among ethnic groups and qualitative methods to examine the underlying mechanisms, would be a useful strategy for investigating safety differences among ethnic groups58 and for identifying effective solutions for the safety differences found. Acknowledgements This work was funded by ZonMw, The Netherlands Organization for Health Research and Development. Disclosure: The authors declare no conflict of interest. References 1 Kohn L , Corrigan J, Donaldson M (eds). To Err is Human: Building a Safer Health System . Institute of Medicine of the National Academies : Washington, DC , 2000 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 2 Haider A , Scott V, Rehman K, Velopulos C, Bentley J, Cornwell E III et al. Racial disparities in surgical care and outcomes in the United States: a comprehensive review of patient, provider, and systemic factors . J Am Coll Surg 2013 ; 216 : 482 – 492 . Google Scholar Crossref Search ADS PubMed WorldCat 3 van der Sande R , Mintjes J, van Harten W. Patient Safety from an International Perspective. Patient Safety Toolbox . Bohn Stafleu van Loghem : Houten , 2007 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 4 Kalkman C Handover in the perioperative care process . Curr Opin Anaesthesiol 2010 ; 23 : 749 – 753 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Nagpal K , Vats A, Ahmed K, Vincent C, Moorthy K. An evaluation of information transfer through the continuum of surgical care: a feasibility study . Ann Surg 2010 ; 252 : 402 – 407 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Kanonidou Z , Karystianou G. Anesthesia for the elderly . Hippokratia 2007 ; 11 : 175 – 177 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 7 Arora V , McGory M, Fung C. Quality indicators for hospitalization and surgery in vulnerable elders . J Am Geriatr Soc 2007 ; 55 ( Suppl 2 ): 347 – 358 . Google Scholar Crossref Search ADS WorldCat 8 O'Regan N , Fitzgerald J, Timmons S, O'Connell H, Meagher D. Delirium: a key challenge for perioperative care . Int J Surg 2013 ; 11 : 136 – 144 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Zegers M , de Bruijne M, de Keizer B, Merten H, Groenewegen P, van der Wal G et al. The incidence, root-causes, and outcomes of adverse events in surgical units: implication for potential prevention strategies . Patient Saf Surg 2011 ; 5 : 1 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Vincent C , Neale G, Woloshynowych M. Adverse events in British hospitals: preliminary retrospective record review . BMJ 2001 ; 322 : 517 – 519 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Bellomo R , Goldsmith D, Russell S, Uchino S. Postoperative serious adverse events in a teaching hospital: a prospective study . Med J Aust 2002 ; 176 : 216 – 218 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Catchpole K , Giddings A, Wilkinson M, Hirst G, Dale T, de Leval M. Improving patient safety by identifying latent failures in successful operations . Surgery 2007 ; 142 : 102 – 110 . Google Scholar Crossref Search ADS PubMed WorldCat 13 St Jacques PJ , Minear MN. Improving perioperative patient safety through the use of information technology. In Advances in Patient Safety: New Directions and Alternative Approaches (Vol 4: Technology and Medication Safety) , Henriksen K, Battles JB, Keyes MA, Grady ML (eds). Agency for Healthcare Research and Quality : Rockville , 2008 . http://www.ncbi.nlm.nih.gov/books/NBK43757/ [accessed 17 June 2014]. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 14 Smith A , Mishra K. Interaction between anaesthetists, their patients, and the anaesthesia team . Br J Anaesth 2010 ; 105 : 60 – 68 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Rejeh N , Ahmadi F, Mohammadi E, Kazemnejad A, Anoosheh M. Nurses' experiences and perceptions of influencing barriers to postoperative pain management . Scand J Caring Sci 2009 ; 23 : 274 – 281 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Smith A , Pope C, Goodwin D, Mort M. Interprofessional handover and patient safety in anaesthesia: observational study of handovers in the recovery room . Br J Anaesth 2008 ; 101 : 332 – 337 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Russo CA , Andrews RM, Barrett M. Racial and ethnic disparities in hospital patient safety events, 2005: statistical brief #53. In Healthcare Cost and Utilization Project (HCUP) Statistical Briefs . Agendy for Healthcare Policy and Research : Rockville , 2006 . http://www.ncbi.nlm.nih.gov/books/NBK56034/ [accessed 17 June 2014]. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 18 Suurmond J , Uiters E, de Bruijne M, Stronks K, Essink-Bot M. Explaining ethnic disparities in patient safety: a qualitative analysis . Am J Public Health 2010 ; 100 : 113 – 117 . Google Scholar Crossref Search ADS WorldCat 19 Schinkel S , Schouten B, van Weert J. Are GP patients' needs being met? Unfulfilled information needs among native-Dutch and Turkish-Dutch patients . Patient Educ Couns 2013 ; 90 : 261 – 267 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Fransen M , Van Schaik T, Twickler T, Essink-Bot M. Applicability of internationally available health literacy measures in the Netherlands . J Health Commun 2011 ; 16 : 134 – 149 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Johnson R , Roter D, Powe N, Cooper L. Patient race/ethnicity and quality of patient–physician communication during medical visits . Am J Public Health 2004 ; 94 : 2084 – 2090 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Schioler T , Lipczak H, Pedersen B, Mogensen T, Bech K, Stockmarr A et al. [ Incidence of adverse events in hospitals. A retrospective study of medical records .] Ugeskr Laeger 2001 ; 163 : 5370 – 5378 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 23 King D , Mainous A III, Carnemolla M, Everett C. Adherence to healthy lifestyle habits in US Adults, 1988–2006 . Am J Med 2009 ; 122 : 528 – 534 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Manna D Standardisation in a Multi Ethnic World: a Paradox? The Validity of Guidelines in the Treatment of Patients with Different Ethnic Backgrounds (Dissertation). Erasmus University : Rotterdam , 2006 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 25 Oppenheimer G Paradigm lost: race, ethnicity, and the search for a new population taxonomy . Am J Public Health 2001 ; 91 : 1049 – 1055 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 26 Weber M An Outline of Interpretive Sociology . Bedminster Press : New York , 1968 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 27 Aspinall P Operationalising the collection of ethnicity data in studies of the sociology of health and illness . Sociol Health Illn 2001 ; 23 : 829 – 862 . Google Scholar Crossref Search ADS WorldCat 28 Morris AM , Rhoads KF, Stain SC, Birkmeyer JD. Understanding racial disparities in cancer treatment and outcomes . J Am Coll Surg 2010 ; 211 : 105 – 113 . Google Scholar Crossref Search ADS PubMed WorldCat 29 von Elm E , Altman D, Egger M, Pocock S, Gøtzsche P, Vandenbroucke J. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies . Bull World Health Organ 2007 ; 85 : 867 – 872 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Grimshaw J , McAuley L, Bero L, Grilli R, Oxman A, Ramsay C et al. Systematic reviews of the effectiveness of quality improvement strategies and programmes . Qual Saf Health Care 2003 ; 12 : 298 – 303 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Dutch Cochrane Centre . Revman 5.2 . http://ims.cochrane.org/revman [accessed 1 September 2013]. 32 Barocas DA , Gray DT, Fowke JH, Mercaldo ND, Blume JD, Chang SS et al. Racial variation in the quality of surgical care for prostate cancer . J Urol 2012 ; 188 : 1279 – 1285 . Google Scholar Crossref Search ADS PubMed WorldCat 33 Murphy E , Davis C, Modrall J, Clagett G, Arko F. Effects of ethnicity and insurance status on outcomes after thoracic endoluminal aortic aneurysm repair (TEVAR) . J Vasc Surg 2010 ; 51 ( Suppl ): 14 – 20 . Google Scholar Crossref Search ADS WorldCat 34 Alosh H , Riley LH III, Skolasky R. Insurance status, geography, race, and ethnicity as predictors of anterior cervical spine surgery rates and in-hospital mortality: an examination of United States trends from 1992 to 2005 . Spine (Phil Pa 1976) 2009 ; 34 : 1956 – 1962 . Google Scholar Crossref Search ADS WorldCat 35 Cheung MC , Zhuge Y, Yang R, Koniaris LG. Disappearance of racial disparities in gastrointestinal stromal tumor outcomes . J Am Coll Surg 2009 ; 209 : 7 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat 36 Halm EA , Tuhrim S, Wang J, Rojas M, Rockman C, Riles T et al. Racial and ethnic disparities in outcomes and appropriateness of carotid endarterectomy: impact of patient and provider factors . Stroke 2009 ; 40 : 2493 – 2501 . Google Scholar Crossref Search ADS PubMed WorldCat 37 Robinson WP III, Owens CD, Nguyen LL, Chong TT, Conte MS, Belkin M. Inferior outcomes of autogenous infrainguinal bypass in Hispanics: an analysis of ethnicity, graft function, and limb salvage . J Vasc Surg 2009 ; 49 : 1416 – 1425 . Google Scholar Crossref Search ADS PubMed WorldCat 38 Elahi M , Matata B, Yii M. Ethnicity and adverse operative outcomes among Australian patients undergoing first-time isolated coronary artery bypass graft surgery . Int Surg 2008 ; 93 : 358 – 365 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 39 Brister S , Hamdulay Z, Verma S, Maganti M, Buchanan M. Ethnic diversity: South Asian ethnicity is associated with increased coronary artery bypass grafting mortality . J Thorac Cardiovasc Surg 2007 ; 133 : 150 – 154 . Google Scholar Crossref Search ADS PubMed WorldCat 40 Yeo KK Li Z, Amsterdam E. Clinical characteristics and 30-day mortality among Caucasians, Hispanics, Asians, and African–Americans in the 2003 California coronary artery bypass graft surgery outcomes reporting program . Am J Cardiol 2007 ; 100 : 59 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat 41 Alvord L , Rhoades D, Henderson W, Goldberg J, Hur K, Khuri S et al. Surgical morbidity and mortality among American Indian and Alaska Native veterans: a comparative analysis . J Am Coll Surg 2005 ; 200 : 837 – 844 . Google Scholar Crossref Search ADS PubMed WorldCat 42 Horner RD , Oddone EZ, Stechuchak KM, Grambow SC, Gray J, Khuri SF et al. Racial variations in postoperative outcomes of carotid endarterectomy: evidence from the Veterans Affairs National Surgical Quality Improvement Program . Med Care 2002 ; 40 ( Suppl ): 35 – 43 . Google Scholar Crossref Search ADS WorldCat 43 Rumsfeld J , Plomondon M, Peterson E, Shlipak M, Maynard C, Grunwald G et al. The impact of ethnicity on outcomes following coronary artery bypass graft surgery in the Veterans Health Administration . J Am Coll Cardiol 2002 ; 40 : 1786 – 1793 . Google Scholar Crossref Search ADS PubMed WorldCat 44 Nallamothu B , Saint S, Saha S, Fendrick A, Kelley K, Ramsey S. Coronary artery bypass grafting in Native Americans: a higher risk of death compared to other ethnic groups? J Gen Intern Med 2001 ; 16 : 554 – 559 . Google Scholar Crossref Search ADS PubMed WorldCat 45 Norwood MG , Mann CD, West K, Miller AS, Hemingway D. Restorative proctocolectomy. Does ethnicity affect outcome? Colorectal Dis 2009 ; 11 : 972 – 975 . Google Scholar Crossref Search ADS PubMed WorldCat 46 DiGiorgi P , Baumann F, O'Leary A, Schwartz C, Grossi E, Ribakove G et al. Original reports: Cardiovascular disease and risk factors differences in mitral valve disease presentation and surgical treatment outcome between Hispanic and non-Hispanic patients . Ethn Dis 2008 ; 18 : 306 – 310 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 47 Goldsmith I , Lip GY, Tsang G, Patel RL. Comparison of primary coronary artery bypass surgery in a British Indo-Asian and white Caucasian population . Eur Heart J 1999 ; 20 : 1094 – 1100 . Google Scholar Crossref Search ADS PubMed WorldCat 48 Lavery LA , van Houtum WH, Armstrong DG, Harkless LB, Ashry HR, Walker SC. Mortality following lower extremity amputation in minorities with diabetes mellitus . Diabetes Res Clin Pract 1997 ; 37 : 41 – 47 . Google Scholar Crossref Search ADS PubMed WorldCat 49 Neighbors CJ , Rogers ML, Shenassa ED, Sciamanna CN, Clark MA, Novak SP. Ethnic/racial disparities in hospital procedure volume for lung resection for lung cancer . Med Care 2007 ; 45 : 655 – 663 . Google Scholar Crossref Search ADS PubMed WorldCat 50 Hadjinikolaou L , Klimatsidas M, Lacona GM, Spyt T, Samani NJ. Short- and medium-term survival following coronary artery bypass surgery in British Indo-Asian and white Caucasian individuals: impact of diabetes mellitus . Interact Cardiovasc Thorac Surg 2010 ; 10 : 389 – 393 . Google Scholar Crossref Search ADS PubMed WorldCat 51 Yaghoubian A , De Virgilio C, El-Masry M, Lewis RJ, Stabile BE. Gallstone pancreatitis: a benign disease in Hispanics . Am Surg 2007 ; 73 : 1071 – 1074 . Google Scholar Crossref Search ADS PubMed WorldCat 52 Toursarkissian B , Jones WT, D'Ayala MD, Shireman PK, Harrison A, Schoolfield J et al. Does the efficacy of dorsalis pedis artery bypasses vary among diabetic patients of different ethnic backgrounds? Vasc Endovasc Surg 2002 ; 36 : 207 – 212 . Google Scholar Crossref Search ADS WorldCat 53 Verderber A , Castelfranco AM, Nishioka D, Johnson KG. Cardiovascular risk factors and cardiac surgery outcomes in a multiethnic sample of men and women . Am J Crit Care 1999 ; 8 : 140 – 148 . Google Scholar Crossref Search ADS PubMed WorldCat 54 Collins L , Padda J, Vaghadia H. Mini-audits facilitate quality assurance in outpatient units . Can J Anaesth 2001 ; 48 : 737 – 741 . Google Scholar Crossref Search ADS PubMed WorldCat 55 Tan EC , Lim Y, Teo YY, Goh R, Law HY, Sia AT. Ethnic differences in pain perception and patient-controlled analgesia usage for postoperative pain . J Pain 2008 ; 9 : 849 – 855 . Google Scholar Crossref Search ADS PubMed WorldCat 56 Bergés IM , Kuo YF, Ostir GV, Granger CV, Graham JE, Ottenbacher KJ. Gender and ethnic differences in rehabilitation outcomes after hip-replacement surgery . Am J Phys Med Rehabil 2008 ; 87 : 567 – 572 . Google Scholar Crossref Search ADS PubMed WorldCat 57 Ibrahim S , Stone R, Cohen P, Henderson W, Khuri S, Kwoh C. Race, ethnicity and length of hospital stay after knee or hip arthroplasty . Curr Orthop Pract 2008 ; 19 : 556 – 563 . Google Scholar Crossref Search ADS WorldCat 58 van Rosse F , de Bruijne MC, Wagner C, Stronks K, Essink-Bot ML. Design of a prospective cohort study to assess ethnic inequalities in patient safety in hospital care using mixed methods . BMC Health Serv Res 2012 ; 12 : 450 . Google Scholar Crossref Search ADS PubMed WorldCat 59 Agency for Healthcare Research and Quality . 2012 National Health Disparities Report . http://www.ahrq.gov/research/findings/nhqrdr/nhdr12/2012nhdr.pdf [accessed 1 August 2013]. 60 Agency for Healthcare Research and Quality . Are We Getting Better at Measuring Patient Safety? http://webmm.ahrq.gov/perspective.aspx?perspectiveID=94 [accessed 1 August 2013]. 61 Aspden P , Corrigan JM, Wolcott J, Erickson SM (eds). Patient Safety: Achieving a New Standard for Care . Institute of Medicine of the National Academies : Washington, DC , 2004 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 62 McGauran N , Wieseler B, Kreis J, Schüler YB, Kölsch H, Kaiser T. Reporting bias in medical research – a narrative review . Trials 2010 ; 11 : 37 . Google Scholar Crossref Search ADS PubMed WorldCat 63 Wollersheim H , Bakker PJM, Bijnen AB, Gouma DJ, Wagner C, van der Weijden T. Kwaliteit en veiligheid in patiëntenzorg . Bohn Stafleu van Loghum : Houten , 2011 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC 64 Bradby H Describing ethnicity in health research . Ethn Health 2003 ; 8 : 5 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat 65 Stronks K , Kulu-Glasgow I, Agyemang C. The utility of ‘country of birth’ for the classification of ethnic groups in health research: the Dutch experience . Ethn Health 2009 ; 14 : 255 – 269 . Google Scholar Crossref Search ADS PubMed WorldCat 66 Rangrass G , Ghaferi AA, Dimick JB. Explaining racial disparities in outcomes after cardiac surgery: the role of hospital quality . JAMA Surg 2014 ; 149 : 223 – 227 . Google Scholar Crossref Search ADS PubMed WorldCat 67 Egger M , Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test . BMJ Open 1997 ; 315 : 629 – 634 . Google Scholar OpenURL Placeholder Text WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Khosla, S; Morris, D R; Moxon, J V; Walker, P J; Gasser, T C; Golledge, J

doi: 10.1002/bjs.9578pmid: 25131598

Abstract Background Abdominal aortic aneurysm (AAA) is an important cause of sudden death; however, there are currently incomplete means to predict the risk of AAA rupture. AAA peak wall stress (PWS) can be estimated using finite element analysis (FEA) methods from computed tomography (CT) scans. The question is whether AAA PWS can predict AAA rupture. The aim of this systematic review was to compare PWS in patients with ruptured and intact AAA. Methods The MEDLINE database was searched on 25 May 2013. Case–control studies assessing PWS in asymptomatic intact, and acutely symptomatic or ruptured AAA from CT scans using FEA were included. Data were extracted independently. A random-effects model was used to calculate standard mean differences (SMDs) for PWS measurements. Results Nine studies assessing 348 individuals were identified and used in the meta-analysis. Results from 204 asymptomatic intact and 144 symptomatic or ruptured AAAs showed that PWS was significantly greater in the symptomatic/ ruptured AAAs compared with the asymptomatic intact AAAs (SMD 0·95, 95 per cent confidence interval 0·71 to 1·18; P < 0·001). The findings remained significant after adjustment for mean systolic blood pressure, standardized at 120 mmHg (SMD 0·68, 0·39 to 0·96; P < 0·001). Minimal heterogeneity between studies was noted (I2 = 0 per cent). Conclusion This study suggests that PWS is greater in symptomatic or ruptured AAA than in asymptomatic intact AAA. Introduction Abdominal aortic aneurysm (AAA) is a progressive focal dilatation and weakening of the abdominal aorta, and is associated with a risk of fatal rupture. Important risk factors for AAA are advanced age, male sex, smoking and positive family history1–3. The latest results of the National Health Service Abdominal Aortic Aneurysm Screening Programme in England4 suggest that the prevalence of AAA is 1·8 per cent in men aged 65–74 years. AAAs are usually asymptomatic, but AAA rupture has a mortality rate of approximately 65–85 per cent5. Surgical repair and stenting are the only available treatments for AAA. The management is determined largely by maximum AAA diameter, which is monitored routinely by medical imaging assessments6. Elective repair is usually considered when the maximum AAA diameter is greater than 54 mm, as below this diameter intervention has been shown not to reduce AAA-related mortality3. However, rupture of AAAs measuring less than 55 mm has been reported, suggesting that the risk of rupture is not determined by aortic diameter alone1,3. In the UK Small Aneurysm Trial (UKSAT)3, the annual rate of AAA rupture was 2·2 per cent after 3 years of follow-up. Additional means of selecting patients for prophylactic AAA repair could prevent more ruptures. The RESCAN collaborators reported recently7 that the rupture rate of small AAAs was fourfold higher in women, double in smokers and increased with higher mean arterial blood pressure, suggesting that these additional measures could be considered when selecting patients for AAA repair. The precise mechanisms leading to AAA rupture remain unclear; however, a biomechanical wall stress that exceeds the mechanical strength of the weakened arterial wall is thought to be the final common pathway8. Consequently, the highest wall stress within an AAA – peak wall stress (PWS) – has been suggested as an indicator of potential rupture. Factors that influence PWS include blood pressure, aneurysm geometry, vessel wall stiffness and thickness, and the shape and characteristics of the intraluminal thrombus (ILT); these are all patient (AAA) specific. ILT is present in about 75 per cent of all AAAs, and the volume of ILT has the potential to alter PWS by multiple means; it is also associated with AAA growth rates9–12. PWS can be estimated non-invasively by computed tomography (CT) using finite element analysis (FEA) (Fig. 1). This approach has been used to explore the possible association of PWS with aortic rupture, although interpretation of the information is complicated by small sample sizes and heterogeneity between studies13–15. A systematic review was performed of publicly available literature to examine the current evidence supporting PWS for prediction of AAA rupture. The aim of the review was to compare PWS in patients with symptomatic or ruptured versus asymptomatic intact AAAs. A meta-analysis was performed to combine the results from studies that measured PWS in patients with symptomatic or ruptured, and asymptomatic intact AAAs. Fig. 1 Open in new tabDownload slide Peak wall stress (PWS) measurement using finite element analysis. The areas are coloured according to the PWS exerted on the abdominal aortic aneurysm wall, with red representing the point of maximum PWS, followed by yellow, green and blue respectively as shown on the colour scale. Interestingly, the PWS is greater at a site that is not the point of maximum diameter. L, left; PWRR, peak wall rupture risk; max, maximum; Lmn, lumen; ext, external; v. Mises Stress, von Mises Stress Methods Literature search A search strategy was devised according to the 2009 Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement16. The MEDLINE (January 1966–May 2013) database was searched on 25 May 2013. To identify studies assessing the association between PWS and AAA rupture, the following search terms were applied: (‘abdominal aortic aneurysm’ OR ‘AAA’)[Title/Abstract] AND (‘rupture’ OR ‘rupture risk’)[Title/Abstract] AND (‘peak wall stress’ OR ‘stress’ OR ‘shear stress’ OR ‘biomechanic*’)[Title/Abstract]. There were no language restrictions. In addition, reference lists of primary articles and reviews were searched to increase the yield of relevant publications. Titles and abstracts were screened to identify potentially relevant studies. If the suitability of an article was uncertain, the full text was assessed. To be eligible, studies were required to have compared PWS in patients with asymptomatic intact and ruptured AAAs. Studies that recruited patients with symptomatic AAAs requiring urgent repair were also included. Included studies used FEA to measure PWS from CT scans. Studies were excluded if: AAAs were not assessed by abdominal CT; there was no clear division of patients into ruptured/symptomatic and intact groups; or ex vivo methods were used to analyse biomechanical wall properties. Data extraction and quality assessment Data were extracted from included studies by one author using set criteria and recorded in tables. Extracted data were reviewed independently by three other authors. Any inconsistencies in data were recorded and resolved by discussion. The following data were recorded: definitions used for symptomatic, ruptured and asymptomatic intact AAAs; the timing of CT relative to AAA rupture; population characteristics; details of the FEA methodology; and the value of PWS at both population and standard systolic blood pressure (SBP). Methodological quality was assessed using a modified version of the QUADAS-2 quality assessment tool17. Quality measures included a description of patient characteristics, the timing of CT relative to rupture, the measurement of PWS at a standard SBP of 120 mmHg, the incorporation of intraluminal thrombus in calculating PWS, and the sample size used. The quality of studies was categorized as good, fair or poor. Good-quality studies were those with a case-controlled design that had a minimum of 20 individuals in each group; reported at least three major risk factors for AAA rupture including maximum AAA diameter, sex, age or smoking history; reported PWS at both population and standard SBP of 120 mmHg; and included CTs done before rupture in the ruptured AAA group. Fair-quality studies required all of the above except a smaller sample size of ten to 20 patients in both groups and did not necessarily include prerupture CTs for the ruptured group. Studies of poor quality had fewer than ten patients in each group and failed to report a minimum of three major risk factors for AAA rupture including maximum AAA diameter, sex, age or smoking history, even if they included prerupture scans. Statistical analysis A meta-analysis was done comparing PWS in patients with asymptomatic intact AAAs and ruptured (or symptomatic) AAAs. Some studies combined patients with a symptomatic AAA with those with a ruptured AAA. Four studies reported PWS as mean(s.d.) and the remaining five studies reported mean(s.e.m.). Data were imported into the Review Manager (RevMan) version 5·2 software package (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark). Standard deviations were calculated automatically by the RevMan software for studies that cited standard error only. PWS was compared between groups of patients with intact and ruptured AAA for each included study via a two-sample t test (RevMan). Standard mean differences (SMDs) and 95 per cent confidence intervals (c.i.) were calculated for each included study. Study-specific estimates were combined using inverse-variance weighted average of logarithmic SMDs in a random-effects model. The random-effects model was used to reduce the effect of heterogeneity in FEA methods on the summary statistics. A further analysis looked at the PWS in the two groups at standard SBP. The interstudy heterogeneity was assessed by means of the I2 index. Sensitivity analyses were performed to assess the contribution of each study to the pooled estimate by excluding individual studies one at a time and recalculating the pooled SMD estimates for the remaining studies. Publication bias was assessed using a funnel plot of the logarithm of effect size versus the standard error for each study, but could not be appraised accurately because of the limited number of studies. Results Study selection Initial database searches yielded 67 potentially eligible studies (Fig. 2) including three additional studies that were identified by hand-searching the reference lists. Forty-five articles were excluded based on review of titles and abstracts. The main reason for exclusion was the lack of clear division of patients into asymptomatic intact, and symptomatic or ruptured AAA. Of the 22 full-text articles evaluated for inclusion in the meta-analysis, 13 were excluded as they used ex vivo methods of measuring biomechanical wall properties. The remaining nine studies18–26 were included in the review. Fig. 2 Open in new tabDownload slide PRISMA diagram of the review. AAA, abdominal aortic aneurysm Study characteristics The studies assessed populations mainly from the USA and Europe. All studies used CT as the imaging modality for AAA visualization. Study characteristics and methodological quality of included studies are shown in Table S1 (supporting information). One24 of the nine studies did not present a clear description of the inclusion criteria for the two groups of patients. In patients with a symptomatic/ruptured AAA, PWS was assessed from CT scans performed before the onset of acute symptoms or rupture in five studies18,21–22,25–26, and afterwards in three studies19–20,23. For one study24 it was unclear whether the CT scans had been obtained before or after the onset of acute symptoms or rupture. Five18,20,22–24 of the nine studies estimated PWS in the intact and ruptured groups at a SBP of 120 mmHg. Four studies19,21,25–26 considered ILT volume in their calculation of PWS using FEA. The reporting of risk factors was not uniform across studies. Maximum aortic diameter was reported by eight studies; overall the diameter of intact AAAs ranged from 51 to 70 mm whereas ruptured AAAs had diameters between 53 and 81 mm. Two studies21,25 did not report the age of included patients. Data from the remaining studies suggested that the average age of the studied patients ranged from roughly 69 to 77 years. Seven studies provided details of patient sex, five18–19,22–23,26 of which noted a higher proportion of women in the ruptured AAA group. The proportion of smokers appeared to be higher in the ruptured groups in four18,20,22,24 of the five studies that detailed patient smoking history. The population maximum blood pressure was higher in the ruptured groups in four18,20,22–23 of the five studies that reported this risk factor. In three18,22–23 of these studies, the prevalence of atherosclerotic cardiac disease was higher in people with intact aneurysms than in those with a rupture. Sample sizes of the included studies ranged from five to 40 patients in the intact groups and from eight to 30 in the ruptured AAA groups (Table S1, supporting information). Symptomatic patients were defined as those presenting with acute abdominal pain requiring emergency surgery and were included in the ruptured group for comparison with asymptomatic intact AAAs. Four18–20,22 of the nine studies reported PWS for symptomatic patients, but only three19–20,22 of these included symptomatic patients in the ruptured AAA group when reporting the PWS and were included in the meta-analysis. The fourth study18 reported PWS separately for intact, symptomatic (n = 8) and ruptured (n = 10) AAA. For the purposes of the present meta-analysis, PWS was used from individuals in the ruptured group in the latter study18 only. The meta-analysis included a combined population of 348 individuals: 204 asymptomatic intact and 144 symptomatic/ruptured AAAs (Table 1). Table 1 Comparison of peak wall stress in asymptomatic intact and symptomatic or ruptured abdominal aortic aneurysms Reference . Year . Country . Total no. . Asymptomatic intact AAA . Symptomatic or ruptured AAA . P¶ . n . PWS (N/cm2)* . n . PWS (N/cm2)* . Measured at population systolic blood pressure Fillinger et al.18† 2002 Lebanon 40 30 36·9(8·8) 10  47·7(20·6)   0·024 Fillinger et al.22† 2003 Lebanon 61 39  42·0(12·5) 22  58·0(18·8) < 0·001 Venkatasubramaniam et al.24‡ 2004 UK 27 15  62·0(28·0) 12 102·0(38·0)   0·004 Vande Geest et al.21† 2006 USA 13 5 46·0(9·6) 8  49·9(11·3)   0·537 Truijers et al.23† 2007 USA 20 10  39·7(10·4) 10  51·7(7·6)   0·009 Heng et al.20‡ 2008 UK 70 40  67·0(30·0) 30 111·0(51·0) < 0·001 Vande Geest et al.25† 2008 USA 14 5 46·0(9·5) 9  49·9(12·1)   0·546 Maier et al.19§ 2010 Germany 53 30  34·3(10·5) 23  47·7(12·5) < 0·001 Gasser et al.26§ 2010 Sweden 50 30  27·6(11·7) 20  35·2(12·6)   0·034 Measured at standard systolic blood pressure (120 mmHg) Fillinger et al.18† 2002 Lebanon 40 30 32·2(7·7) 10 38·0(9·5)   0·058 Fillinger et al.22† 2003 Lebanon 61 39  37·0(12·5) 22  46·0(14·1)   0·012 Venkatasubramaniam et al.24‡ 2004 UK 27 15  55·0(24·0) 12  77·0(29·0)   0·041 Truijers et al.23† 2007 USA 20 10 31·7(7·3) 10  36·7(12·6)   0·293 Heng et al.20‡* 2008 UK 70 40  65·0(25·0) 30  84·0(31·0)   0·006 Reference . Year . Country . Total no. . Asymptomatic intact AAA . Symptomatic or ruptured AAA . P¶ . n . PWS (N/cm2)* . n . PWS (N/cm2)* . Measured at population systolic blood pressure Fillinger et al.18† 2002 Lebanon 40 30 36·9(8·8) 10  47·7(20·6)   0·024 Fillinger et al.22† 2003 Lebanon 61 39  42·0(12·5) 22  58·0(18·8) < 0·001 Venkatasubramaniam et al.24‡ 2004 UK 27 15  62·0(28·0) 12 102·0(38·0)   0·004 Vande Geest et al.21† 2006 USA 13 5 46·0(9·6) 8  49·9(11·3)   0·537 Truijers et al.23† 2007 USA 20 10  39·7(10·4) 10  51·7(7·6)   0·009 Heng et al.20‡ 2008 UK 70 40  67·0(30·0) 30 111·0(51·0) < 0·001 Vande Geest et al.25† 2008 USA 14 5 46·0(9·5) 9  49·9(12·1)   0·546 Maier et al.19§ 2010 Germany 53 30  34·3(10·5) 23  47·7(12·5) < 0·001 Gasser et al.26§ 2010 Sweden 50 30  27·6(11·7) 20  35·2(12·6)   0·034 Measured at standard systolic blood pressure (120 mmHg) Fillinger et al.18† 2002 Lebanon 40 30 32·2(7·7) 10 38·0(9·5)   0·058 Fillinger et al.22† 2003 Lebanon 61 39  37·0(12·5) 22  46·0(14·1)   0·012 Venkatasubramaniam et al.24‡ 2004 UK 27 15  55·0(24·0) 12  77·0(29·0)   0·041 Truijers et al.23† 2007 USA 20 10 31·7(7·3) 10  36·7(12·6)   0·293 Heng et al.20‡* 2008 UK 70 40  65·0(25·0) 30  84·0(31·0)   0·006 * Values are mean(s.d.); peak wall stress (PWS) values converted from † s.e.m. to s.d., ‡ MPa to N/cm2 and § kPa to N/cm2. ¶ Two-sample t test. Open in new tab Table 1 Comparison of peak wall stress in asymptomatic intact and symptomatic or ruptured abdominal aortic aneurysms Reference . Year . Country . Total no. . Asymptomatic intact AAA . Symptomatic or ruptured AAA . P¶ . n . PWS (N/cm2)* . n . PWS (N/cm2)* . Measured at population systolic blood pressure Fillinger et al.18† 2002 Lebanon 40 30 36·9(8·8) 10  47·7(20·6)   0·024 Fillinger et al.22† 2003 Lebanon 61 39  42·0(12·5) 22  58·0(18·8) < 0·001 Venkatasubramaniam et al.24‡ 2004 UK 27 15  62·0(28·0) 12 102·0(38·0)   0·004 Vande Geest et al.21† 2006 USA 13 5 46·0(9·6) 8  49·9(11·3)   0·537 Truijers et al.23† 2007 USA 20 10  39·7(10·4) 10  51·7(7·6)   0·009 Heng et al.20‡ 2008 UK 70 40  67·0(30·0) 30 111·0(51·0) < 0·001 Vande Geest et al.25† 2008 USA 14 5 46·0(9·5) 9  49·9(12·1)   0·546 Maier et al.19§ 2010 Germany 53 30  34·3(10·5) 23  47·7(12·5) < 0·001 Gasser et al.26§ 2010 Sweden 50 30  27·6(11·7) 20  35·2(12·6)   0·034 Measured at standard systolic blood pressure (120 mmHg) Fillinger et al.18† 2002 Lebanon 40 30 32·2(7·7) 10 38·0(9·5)   0·058 Fillinger et al.22† 2003 Lebanon 61 39  37·0(12·5) 22  46·0(14·1)   0·012 Venkatasubramaniam et al.24‡ 2004 UK 27 15  55·0(24·0) 12  77·0(29·0)   0·041 Truijers et al.23† 2007 USA 20 10 31·7(7·3) 10  36·7(12·6)   0·293 Heng et al.20‡* 2008 UK 70 40  65·0(25·0) 30  84·0(31·0)   0·006 Reference . Year . Country . Total no. . Asymptomatic intact AAA . Symptomatic or ruptured AAA . P¶ . n . PWS (N/cm2)* . n . PWS (N/cm2)* . Measured at population systolic blood pressure Fillinger et al.18† 2002 Lebanon 40 30 36·9(8·8) 10  47·7(20·6)   0·024 Fillinger et al.22† 2003 Lebanon 61 39  42·0(12·5) 22  58·0(18·8) < 0·001 Venkatasubramaniam et al.24‡ 2004 UK 27 15  62·0(28·0) 12 102·0(38·0)   0·004 Vande Geest et al.21† 2006 USA 13 5 46·0(9·6) 8  49·9(11·3)   0·537 Truijers et al.23† 2007 USA 20 10  39·7(10·4) 10  51·7(7·6)   0·009 Heng et al.20‡ 2008 UK 70 40  67·0(30·0) 30 111·0(51·0) < 0·001 Vande Geest et al.25† 2008 USA 14 5 46·0(9·5) 9  49·9(12·1)   0·546 Maier et al.19§ 2010 Germany 53 30  34·3(10·5) 23  47·7(12·5) < 0·001 Gasser et al.26§ 2010 Sweden 50 30  27·6(11·7) 20  35·2(12·6)   0·034 Measured at standard systolic blood pressure (120 mmHg) Fillinger et al.18† 2002 Lebanon 40 30 32·2(7·7) 10 38·0(9·5)   0·058 Fillinger et al.22† 2003 Lebanon 61 39  37·0(12·5) 22  46·0(14·1)   0·012 Venkatasubramaniam et al.24‡ 2004 UK 27 15  55·0(24·0) 12  77·0(29·0)   0·041 Truijers et al.23† 2007 USA 20 10 31·7(7·3) 10  36·7(12·6)   0·293 Heng et al.20‡* 2008 UK 70 40  65·0(25·0) 30  84·0(31·0)   0·006 * Values are mean(s.d.); peak wall stress (PWS) values converted from † s.e.m. to s.d., ‡ MPa to N/cm2 and § kPa to N/cm2. ¶ Two-sample t test. Open in new tab As some studies reported P values from comparisons of PWS in more than two groups of patients, extracted data were reanalysed via two-sample t tests in order to compare directly groups of patients with intact and ruptured AAA and to standardize statistical comparisons between studies. In all studies mean PWS was higher in patients with symptomatic or ruptured AAA than in patients with intact AAA (Table 1)18–26, but was not statistically significant in two studies21,25. Both of these studies21,25 had small sample sizes of fewer than ten patients in each group and were found to be of poor quality using the quality assessment tool (Table S1, supporting information). Five of the nine studies compared PWS between groups of patients with intact and ruptured AAAs using standard SBP of 120 mmHg; PWS remained higher in the patients with ruptured AAA compared with those with intact AAA, although statistical significance was demonstrated by only four studies18,20,22,24 after this adjustment (Table 1). Data synthesis Three19–20,22 of the nine studies combined symptomatic and ruptured AAA groups to compare PWS with that in the intact group. One study20 defined the symptomatic group of patients as those with an acute or leaking AAA, and these patients were considered to have had a ruptured AAA owing to the high probability of rupture. Two studies20,24 reported the results in megapascals (MPa), two19,26 in kilopascals (kPa), and the remaining studies used newtons per square centimetre (N/cm2). All measurements in MPa and kPa were converted to N/cm2 according to the following equation: 1 MPa = 1000 kPa = 100 N/cm2 (Table 1). A meta-analysis of the nine studies consisting of 204 asymptomatic intact and 144 symptomatic/ruptured AAAs demonstrated significantly higher PWS in patients with a ruptured AAA than in those with an intact AAA using a random-effects model (SMD 0·95, 95 per cent c.i. 0·71 to 1·18; P < 0·001) (Fig. 3a). Analysis suggested that there was little heterogeneity between the included studies (I2 = 0 per cent). Sensitivity analyses demonstrated that multiple studies contributed to the observed difference in PWS between the two groups, and the exclusion of any single study from the analysis did not substantively alter the overall result of the analysis (Table S2, supporting information). A further analysis of PWS measured at SBP of 120 mmHg in 218 AAAs (134 intact and 84 symptomatic/ruptured AAAs) demonstrated that PWS remained significantly higher in the symptomatic/ruptured group than in the intact AAA group (SMD 0·68, 95 per cent c.i. 0·39 to 0·96; P < 0·001) (Fig. 3b). No significant heterogeneity was observed between the studies included in this subanalysis (I2 = 0 per cent). Fig. 3 Open in new tabDownload slide Forest plots of mean(s.d.) peak wall stress (PWS) in symptomatic or ruptured abdominal aortic aneurysms (AAAs) compared with intact AAAs. a PWS measurements were made at normal blood pressure, except in the study by Gasser and colleagues26 which used mean arterial pressure; b measurements made at standard systolic pressure of 120 mmHg. Standard mean differences (SMDs) are shown with 95 per cent confidence intervals. Study-specific estimates were combined using the inverse-variance (IV) weighted average of logarithmic SMDs in a random-effects model Discussion PWS has been suggested as a measure for estimating AAA rupture risk. PWS has been documented to be higher in ruptured than in intact AAAs in a number of independent studies, although this difference was significant in only seven18–20,22–24,26 of nine studies. The results from these small studies have been combined in the present analysis, which suggests that PWS is higher in patients with symptomatic or ruptured AAAs than in those with an intact AAA, at both normal and a standard SBP of 120 mmHg. Confidence in these findings remains restricted owing to the overall small size of both groups. This analysis must be viewed in the context of its limitations. First, there was qualitative heterogeneity in participant selection among the included studies, and also in the FEA software used to analyse CT scans. Three studies19–20,22 combined symptomatic and ruptured AAA, and compared these with the asymptomatic intact AAA. Second, there was heterogeneity in the timing of CT relative to the time of rupture. Five18,21–22,25–26 of the nine studies assessed CT done before rupture to calculate PWS in the ruptured group, whereas three studies19–20,23 used CT done at the time of the rupture, when the patients were in a stable condition. Which of these is representative of PWS as a risk factor is unclear. Differences were also observed in the FEA calculations applied to calculate PWS in the included studies and the units that PWS was reported in. In the present review, all PWS values were converted into N/cm2 to allow easier comparison. A random-effects model was used to calculate a SMD in PWS between the ruptured/symptomatic and intact AAAs. This model was used to minimize the influence of heterogeneity between studies in the meta-analysis, as suggested by Cochrane Collaboration guidelines27. The results from the meta-analysis indicate that PWS is likely to be higher in symptomatic/ruptured than in intact AAAs. This finding is supported by observational studies that assessed whether PWS increases with increasing AAA size. For example, Fillinger and colleagues22 followed 103 patients with a small AAA and found that the initial peak wall stress determined through FEA was 38 N/cm2 for aneurysms that remained stable during observation of 14 months, compared with 42 N/cm2 for expanding aneurysms and 58 N/cm2 for aneurysms that ultimately ruptured. It is well established that smaller AAAs can rupture3,7,28. FEA has gained significant credibility as a reliable measure for calculating PWS; however, this technique relies on some standard assumptions of AAA wall thickness and stiffness, based on previous autopsy studies13,29. Most software packages use a standard wall thickness of 2 mm for all AAAs, which is not specific to each patient. The aneurysmal aorta is assumed to be homogeneous with linear elastic properties. Based on experimental findings suggesting that the mechanical properties of circumferentially and longitudinally oriented aortic tissues do not differ29, isotropic material properties are used to estimate PWS. However, to estimate PWS accurately the strength of the wall should be tested mechanically on excised tissues from various regions of the same aorta in order to account for localized differences in vessel biology. Another concern with the use of commercially available semiautomated FEA programmes is the variability in results between different software packages. A recent investigation30 studied the interobserver and intraobserver variability of a semiautomatic diagnostic software (A4research™; VASCOPS, Graz, Austria) to measure PWS in AAA. The interobserver reproducibility for the three observers showed an interclass coefficient (ICC) of 0·98 (range 0·97–0·99) for PWS. Not all models have been tested for interobserver variability, limiting the confidence in PWS estimated with some models. Finally, although simple geometrical properties can be used to calculate PWS, the use of FEA provides more accurate results as it also incorporates ILT, AAA geometry, wall stiffness and blood pressure15,30–31. The presence of calcification and ILT in an AAA has been shown to increase PWS, suggesting that both should be considered in the evaluation of wall stress for assessment of rupture risk9,31. However, the use of calcification in FEA also requires calculation of the resultant reduction in wall strength at the site of calcification, which is problematic26. It has been suggested that PWS measurements can add to rupture prediction from AAA diameter alone13,19,22,26. Fillinger and co-workers22 showed that PWS was superior to diameter in predicting catastrophic events in patients with an AAA under observation. Receiver operating characteristic (ROC) curves for predicting rupture showed PWS to have higher sensitivity, specificity and accuracy (94, 81 and 85 per cent respectively at a threshold of 44 N/cm2) than diameter (81, 70 and 73 per cent at a 55-mm threshold)22. PWS is considered to be a direct function of critical factors such as patient age, sex, blood pressure, AAA size and shape, as it incorporates AAA geometry and blood pressure in the calculation13,29. PWS analysis is most likely to benefit the management of small aneurysms through the identification of AAAs that may be unsuitable for surveillance owing to a higher risk of rupture. Fillinger et al.22 showed that the smallest aneurysm to rupture in their study had a maximum diameter of 44 mm but had stress equivalent to a typical AAA twice the size. AAA rupture does not necessarily occur at the site of maximum diameter; PWS is often higher at points proximal or distal to this (Fig. 1). Previous studies suggest that inflammatory activity assessed by positron emission tomography or CT is high at sites of high PWS32. Currently there are several biomechanical and postprocessing FEA methods available for calculating PWS. The lack of a standard technique limits the translation of this potentially beneficial technique into clinical practice. Until recently, the calculation of PWS using FEA was an experimental method that was both time-consuming and labour-intensive. There are now a number of commercially available programs that streamline this process and enable analysis of standard CT images. Further investigation of the geometrical and material properties that influence PWS may improve the ability to predict AAA rupture and assist in the development of a patient-specific biomechanical model to help surgeons manage small AAAs. However, before this is possible, a standard technique to measure PWS is required. Acknowledgements J.G. holds a Practitioner Fellowship from the National Health and Medical Research Council and a Senior Clinical Research Fellowship from the Office of Health and Medical Research, Queensland Government. T.C.G. is scientific adviser to VASCOPS. Disclosure: The authors declare no other conflict of interest. Editor's comments References 1 Moll FL , Powell JT, Fraedrich G, Verzini F, Haulon S, Waltham M et al. Management of abdominal aortic aneurysms clinical practice guidelines of the European Society for Vascular Surgery . Eur J Vasc Endovasc Surg 2011 ; 41 : S1 – S58 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Golledge J , Norman PE. Current status of medical management for abdominal aortic aneurysm . Atherosclerosis 2011 ; 217 : 57 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Small UK Aneurysm Trial Participants. Mortality results for randomized controlled trial of early elective surgery or ultrasonographic surveillance for small abdominal aortic aneurysms . Lancet 1998 ; 352 : 649 – 655 . Google Scholar OpenURL Placeholder Text WorldCat 4 NHS Abdominal Aortic Aneurysm Screening Programme Annual Report . http://aaa.screening.nhs.uk/annualreport [accessed 15 February 2014 ]. 5 Moxon JV , Parr A, Emeto TI, Walker P, Norman PE, Golledge J. Diagnosis and monitoring of abdominal aortic aneurysm: current status and future prospects . Curr Probl Cardiol 2010 ; 35 : 512 – 548 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Golledge J , Muller J, Daugherty A, Norman PE. Abdominal aortic aneurysm: pathogenesis and implications for management . Arterioscler Thromb Vasc Biol 2006 ; 26 : 2605 – 2613 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Sweeting MJ , Thompson SG, Brown LC, Powell JT; RESCAN collaborators. Meta-analysis of individual patient data to examine factors affecting growth and rupture of small abdominal aortic aneurysms . Br J Surg 2012 ; 99 : 655 – 665 . Google Scholar Crossref Search ADS PubMed WorldCat 8 McGloughlin TM , Doyle BJ. New approaches to abdominal aortic aneurysm rupture risk assessment: engineering insights with clinical gain . Arterioscler Thromb Vasc Biol 2010 ; 30 : 1687 – 1694 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Li ZY , U-King-Im J, Tang YT, Soh E, See TC, Gillard JH. Impact of calcification and intraluminal thrombus on the computed wall stress of abdominal aortic aneurysm . J Vasc Surg 2008 ; 47 : 928 – 935 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Parr A , Cann MM, Bradshaw B, Shahzad A, Buttner P, Golledge J. Thrombus volume is associated with cardiovascular events and aneurysm growth in patients who have abdominal aortic aneurysms . J Vasc Surg 2011 ; 53 : 28 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Speelman L , Schurink GW, Bosboom EM, Buth J, Breeuwer M, van de Vosee FM et al. The mechanical role of thrombus on the growth rate of an abdominal aortic aneurysm . J Vasc Surg 2010 ; 51 : 19 – 26 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Golledge J , Wolanski P, Parr A, Buttner P. Measurement and determinants of infrarenal aortic thrombus volume . Eur Radiol 2008 ; 18 : 1987 – 1994 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Raghavan ML , Vorp DA, Federle M, Makaroun MS, Webster MW. Wall stress distribution on three-dimensionally reconstructed models of human abdominal aortic aneurysm . J Vasc Surg 2000 ; 31 : 760 – 769 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Vorp DA , Vande Geest JP. Biomechanical determinants of abdominal aortic aneurysm rupture . Arterioscler Thromb Vasc Biol 2005 ; 25 : 1558 – 1566 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Gasser C , Auer M, Labruto F, Roy J, Swedenburg J. Using finite element analysis to assess rupture risk in abdominal aortic aneurysms including the effect of the intraluminal thrombus: PP52 . J Vasc Surg 2009 ; 49 ( Suppl 5 ): S29 . Google Scholar Crossref Search ADS WorldCat 16 Moher D , Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement . Open Med 2009 ; 3 : 123 – 130 . Google Scholar OpenURL Placeholder Text WorldCat 17 Whiting PF , Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB et al.; QUADAS-2 Group. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies . Ann Intern Med 2011 ; 155 : 529 – 536 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Fillinger MF , Raghavan ML, Marra SP, Cornenwett JL, Kennedy FE. In vivo analysis of mechanical wall stress and abdominal aortic aneurysm rupture risk . J Vasc Surg 2002 ; 36 : 589 – 597 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Maier A , Gee MW, Reeps C, Pongratz J, Eckstein HH, Wall WA. A comparison of diameter, wall stress, and rupture potential index for abdominal aortic aneurysm rupture risk prediction . Ann Biomed Eng 2010 ; 38 : 3124 – 3134 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Heng MS , Fagan MJ, Collier JW, Desai G, McPollum PT, Chetter IC. Peak wall stress measurement in elective and acute abdominal aortic aneurysms . J Vasc Surg 2008 ; 47 : 17 – 22 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Vande Geest JP , Martino ES, Bohra A, Mackaroun MS, Vorp DA. A biomechanics-based rupture potential index for abdominal aortic aneurysm risk assessment: demonstrative application . Ann N Y Acad Sci 2006 ; 1085 : 11 – 21 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Fillinger MF , Marra SP, Raghavan ML, Kennedy FE. Prediction of rupture risk in abdominal aortic aneurysm during observation: wall stress versus diameter . J Vasc Surg 2003 ; 37 : 724 – 732 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Truijers M , Pol JA, SchultzeKool LJ, van Strekenburg SM, Fillinger MF, Blankensteijn JD. Wall stress analysis in small asymptomatic, symptomatic and ruptured abdominal aortic aneurysms . Eur J Vasc Endovasc Surg 2007 ; 33 : 401 – 407 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Venkatasubramaniam AK , Fagan MJ, Mehta T, Mylankal KJ, Ray B, Kuhan G et al. A comparative study of aortic wall stress using finite element analysis for ruptured and non-ruptured abdominal aortic aneurysms . Eur J Vasc Endovasc Surg 2004 ; 28 : 168 – 176 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 25 Vande Geest JP , Schmidt DE, Sacks MS, Vorp DA. The effects of anisotropy on the stress analyses of patient-specific abdominal aortic aneurysms . Ann Biomed Eng 2008 ; 36 : 921 – 932 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Gasser TC , Auer M, Labruto F, Swedenborg J, Roy J. Biomechanical rupture risk assessment of abdominal aortic aneurysms: model complexity versus predictability of finite element simulations . Eur J Vasc Endovasc Surg 2010 ; 40 : 176 – 185 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Higgins JPT , Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions . Version 5.1.0, March 2011. http://handbook.cochrane.org/chapter_9/9_4_3_1_random_effects_dersimonian_and_laird_method_for.htm [accessed 7 November 2013 ]. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 28 Nordon IM , Hinchliffe RJ, Loftus IM, Thompson MM. Pathophysiology and epidemiology of abdominal aortic aneurysms . Nat Rev Cardiol 2011 ; 8 : 92 – 102 . Google Scholar Crossref Search ADS PubMed WorldCat 29 Raghavan ML , Webster MW, Vorp DA. Ex vivo biomechanical behavior of abdominal aortic aneurysm: assessment using a new mathematical model . Ann Biomed Eng 1996 ; 24 : 573 – 584 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Hyhlik-Dürr A , Krieger T, Geisbüsch P, Kotelis D, Able T, Böckler D. Reproducibility of deriving parameters of AAA rupture risk from patient-specific 3D finite element models . J Endovasc Ther 2011 ; 18 : 289 – 298 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Hua J , Mower WR. Simple geometric characteristics fail to reliably predict abdominal aortic aneurysm wall stresses . J Vasc Surg 2001 ; 34 : 308 – 315 . Google Scholar Crossref Search ADS PubMed WorldCat 32 Nchimi A , Cheramy-Bien JP, Gasser TC, Namur G, Gomez P, Seidel L et al. Multifactorial relationship between 18 F-fluoro-deoxy-glucose positron emission tomography signaling and biomechanical properties in unruptured aortic aneurysms . Circ Cardiovasc Imaging 2014 ; 7 : 82 – 91 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Stather, P W; Sidloff, D A; Dattani, N; Gokani, V J; Choke, E; Sayers, R D; Bown, M J

doi: 10.1002/bjs.9593pmid: 25131707

Abstract Background Many studies have investigated the systemic and local expression of biomarkers in patients with abdominal aortic aneurysm (AAA). The natural history of AAA varies between patients, and predictors of the presence and diameter of AAA have not been determined consistently. The aim of this study was to perform a systematic review, meta-analysis and meta-regression of studies comparing biomarkers in patients with and without AAA, with the aim of summarizing the association of identified markers with both AAA presence and size. Methods and results Literature review identified 106 studies suitable for inclusion. Meta-analysis demonstrated a significant difference between matrix metalloproteinase (MMP) 9, tissue inhibitor of matrix metalloproteinase 1, interleukin (IL) 6, C-reactive protein (CRP), α1-antitrypsin, triglycerides, lipoprotein(a), apolipoprotein A and high-density lipoprotein in patients with and without AAA. Although meta-analysis was not possible for MMP-2 in aortic tissue, tumour necrosis factor α, osteoprotegerin, osteopontin, interferon γ, intercellular cell adhesion molecule 1 and vascular cell adhesion molecule 1, systematic review suggested an increase in these biomarkers in patients with AAA. Meta-regression analysis identified a significant positive linear correlation between aortic diameter and CRP level. Conclusion A wide variety of biomarkers are dysregulated in patients with AAA, but their clinical value is yet to be established. Future research should focus on the most relevant biomarkers of AAA, and how they could be used clinically. Introduction Abdominal aortic aneurysm (AAA) is a significant burden on healthcare globally1; there is no established medical treatment for small AAA. The natural history of the disease is progressive, with growth and rupture. Screening programmes have been developed internationally2, with the aim of detecting AAA before rupture and undertaking timely elective AAA repair. Understanding the pathophysiology of AAA formation and progression is imperative for creating targeted interventions to slow or halt disease progression. Many studies have investigated the systemic and local levels of biomarkers in patients with AAA, often with conflicting results. A recent meta-analysis and meta-regression3 of coagulation markers in patients with AAA identified a significant increase in circulating levels of fibrinogen, D-dimer and thrombin–antithrombin III complex in people with AAA, and a significant correlation between aortic diameter and D-dimer level. However, although haematological markers associated with coagulation are plausible biomarkers for AAA, there are many other potential biomarkers related to other processes involved in aneurysm formation. The aim of the present study was therefore to undertake a systematic review, and where possible meta-analysis and meta-regression, of studies comparing the clinical value of alternative biomarkers for AAA such as matrix metalloproteinases (MMPs), inflammatory markers, acute-phase reactants and lipids. Methods Search strategy A systematic review comparing biomarkers in patients with and without AAA was conducted. Study titles and abstracts were searched according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines4 using MEDLINE and Embase databases through OVID Online (version: OvidSP_UI03.04.02.112) in February 2013 using the mesh terms ‘aneurysm’, ‘aorta’, ‘abdominal’, ‘miRNA’, ‘microRNA’, ‘mRNA’, ‘RNA’, ‘protein’, ‘biomarker’, ‘human’, NOT ‘thoracic’. Further literature searches for each individual biomarker were undertaken using PubMed in August 2013, along with manual searching of reference lists, for additional articles. Study selection and data extraction Studies included in this analysis were either case–control studies, or studies comparing biomarker levels with aortic diameter. Data extracted included mean or median and s.d., i.q.r., range, s.e.m., 95 per cent confidence interval (c.i.), and/or a P value. In addition, the following data were extracted for each article: year of publication, author, journal and mean aneurysm diameter. Statistical analysis Meta-analysis of summary statistics from the individual biomarker studies was conducted. For each study, data relating to biomarker levels in both the AAA and control groups were used to generate standardized mean differences (SMDs) and 95 per cent c.i. using Review Manager version 5.25. Studies were combined using the inverse-variance method in random-effects models owing to the heterogeneity between included studies; P < 0·050 was used to determine significance. Individual circulating biomarkers (measured in blood, serum or plasma) and aortic tissue studies were analysed separately. Biomarker concentrations were each converted to the same SI unit (depending on the predominant unit for each biomarker), and converted to mean and s.d. in accordance with the Cochrane Handbook for Systematic Reviews6. Standard conversion charts were used to convert mg/dl to mmol/l. For articles reporting the median and i.q.r., the median was taken to be representative of the mean, and the i.q.r. was divided by 1·35 to generate the s.d. For studies reporting the s.e.m., this was multiplied by the square root of the number of subjects within that group to calculate the s.d. For studies reporting the 95 per cent c.i., the s.d. was calculated by multiplying the difference between the upper and lower intervals by the square root of the number of subjects within the group, then dividing this by 3·92. Where a P value had been given but no s.d., i.q.r. or s.e.m., the s.d. was calculated from the P value using the standard formula from the Cochrane Handbook for Systematic Reviews Version 5.1.0 updated March 2011. Where studies had reported more than one subgroup of patients with AAA, a combined mean and s.d. was obtained from standard formulas. Between-study heterogeneity was analysed using the I2 statistic, which describes the percentage of total variation across studies that is due to heterogeneity rather than chance or random error7. Publication bias was assessed by visual inspection of funnel plots8. To determine whether biomarker concentrations were associated with AAA diameter, meta-regression was done using n-weighted linear regression (SPSS® version 20; IBM, Armonk, New York, USA). Biomarker concentrations were compared with the reported mean or median aneurysm size from each study. Where reported biomarker concentrations were stratified by AAA diameter, each size subgroup was entered into the regression as a separate outcome. Where reported diameter was greater than or less than a specified diameter, the specified diameter was used. Where reported diameter was given as a range, the central point of the range was used. Sensitivity analyses for all outcomes were used to assess the influence of each study on the SMD by exclusion of individual studies one at a time. Results Systematic review yielded a total of 1319 relevant studies. Additional searches identified a further 65 studies to include in the analysis. Following removal of duplicates (155), case reports (129), cardiac studies (54) and animal studies (146), a total of 900 studies remained. Some 130 studies were identified that reported biomarkers in subjects with AAA compared with either controls, subjects with aortic occlusive disease (AOD), or compared aortic diameter. Twenty-four studies were excluded because they reported of coagulation markers only (fibrinogen, D-dimer, thrombin antithrombin complex, plasminogen activator inhibitor). Review articles and those pertaining to thoracic aneurysms were also excluded. A total of 106 studies evaluating the levels of proteases, inflammatory markers, acute-phase reactants and lipids formed the basis for this study (Fig. 1; Table S1, supporting information)9–114. Fig. 1 Open in new tabDownload slide PRISMA diagram showing the literature search Proteases Matrix metalloproteinases Six studies9–14 examined MMP-2 in the circulation; all except one9 found no difference in MMP-2 levels. Meta-analysis of these studies9–14 showed no difference in MMP-2 levels between 776 patients with AAA and 502 without: SMD −0·26 (95 per cent c.i. –1·19 to 0·67) ng/ml (P = 0·590; I2 = 98 per cent) (Fig. S1, supporting information). This is in contrast to aortic tissue MMP-2 levels, which were significantly increased in six studies15–20, with no difference in only one21. Circulating MMP-2 activity was measured by zymography in two studies13,18, and demonstrated no difference. Five studies10–12,14,22 of circulating MMP-2 were suitable for meta-regression, which showed no correlation between MMP-2 levels and aneurysm diameter (P = 0·920), consistent with the findings of one previous study23 and in contrast to the results of another24. Circulating MMP-9 levels were significantly increased in patients with AAA in eight studies9–10,25–30 and significantly decreased in another11, with no difference found in four studies12,14,31–32. Meta-analysis revealed a significant increase in circulating MMP-9 levels in 914 patients with AAA compared with 485 controls: SMD 0·67 (0·24 to 1·10) ng/ml (P = 0·002; I2 = 91 per cent) (Fig. 2). This finding was supported by five studies15–16,21,33–34 that analysed aortic tissue MMP-9 levels, although two17,35 did not find a significant increase. Meta-regression of circulating MMP-9 studies10–11,13–14,22,27,31–32 demonstrated no correlation with aneurysm diameter (P = 0·690), consistent with three additional studies23,25,36; however, one study37 did report this correlation. Circulating MMP-9 activity13,18,38 (measured by zymography) showed no clear relationship with AAA presence; however, MMP-9 activity within aortic tissue was increased significantly in six studies18–21,34,39. Fig. 2 Open in new tabDownload slide Forest plot comparing protease expression in the circulation in patients with abdominal aortic aneurysm (AAA) versus controls: a matrix metalloproteinase (MMP) 9, b tissue inhibitor of matrix metalloproteinase (TIMP) 1 and c α1-antitrypsin. An inverse-variance random-effects model was used. Standardized mean differences (SMDs) are shown with 95 per cent confidence intervals Two studies26,27 that compared MMP-3 levels in patients with AAA with those in healthy controls found a significant increase; however, one study40 comparing AAA with AOD found no significant difference in MMP-3 levels, suggesting that these changes may reflect atherosclerotic disease. In aortic tissue, MMP-3 levels were not significantly different in patients with AAA17. A limited number of studies have reported on other MMPs in aortic tissue, but the evidence is inconsistent. No significant difference in the levels of MMP-1 were found between patients with and without AAA17,35. MMP-8 levels35,41 and activity41 were significantly increased in AAA, as were levels of MMP-1242, MMP-1343 and MMP-1444 in single studies comparing AAA with controls; however, no difference was found in additional studies of MMP-13 and MMP-14 levels35. Tissue inhibitors of matrix metalloproteinases Circulating levels of tissue inhibitor of matrix metalloproteinase (TIMP) 1 were significantly raised in patients with AAA in three studies10,13,45, whereas no difference was found in five9,11,14,32,40. Meta-analysis revealed a significant increase in TIMP-1 in 740 patients with AAA compared with 394 controls: SMD 0·52 (0·25 to 0·79) ng/ml (P < 0·001; I2 = 65 per cent) (Fig. 2). However, meta-regression revealed no correlation with aortic diameter9,11,14,32,40 (P = 0·930), consistent with previous findings23. TIMP-1 levels in aortic tissue were significantly downregulated in only a single study41, with no significant difference found in three studies17,24,35. Circulating levels of TIMP-2 were reported to be significantly decreased in patients with AAA14, but not to correlate with AAA growth rates or diameter23. Similarly, TIMP-2 levels in tissue were significantly downregulated in a single study41, whereas no significant difference was found in three others17,24,35. Circulating levels of both TIMP-3 and TIMP-4 were not significantly different between patients with AAA or AOD38, and there were also no significant differences between TIMP-3 levels in aortic tissue35. Cystatins and cathepsins Three studies33,46–47 analysed cathepsin (Cat) L levels in either blood or aortic tissue, but only one46 demonstrated a significantly higher concentration in patients with AAA versus those without. Three studies33,46,48 analysed CatS, with two46,48 finding a significantly higher concentration in blood in those with AAA, whereas the study33 of aortic tissue found no difference. One study46 that analysed CatV levels in plasma demonstrated a significant increase in patients with AAA. Two studies33,47 analysed CatB levels and found no difference between patients with AAA and healthy controls. Cystatin C levels were reported to be lower in patients with AAA in two studies10,33; however, two studies49,50 demonstrated an inverse correlation with aneurysm size, whereas one10 did not. One study33 showed no difference in levels of cystatins A and B in aortic tissue, and another46 reported a significant increase in plasma cystatin S levels among those with AAA. α1-Antitrypsin Although there is no correlation between α1-antitrypsin (α1AT) deficiency and AAA51, several studies have looked at α1AT levels in patients with AAA. Three studies52–54 reported upregulation of α1AT in AAA, and three10,55–56 reported no difference. Meta-analysis of four of these studies10,53–55 revealed a significant upregulation of α1AT in 379 patients with AAA compared with 6539 controls: SMD 1·60 (0·72 to 2·47) g/l (P < 0·001; I2 = 95 per cent) (Fig. 2). Three studies10,23,57 found no correlation between α1AT and aneurysm size. Acute-phase reactants C-reactive protein Seventeen studies compared plasma C-reactive protein (CRP) levels between patients with AAA and controls; ten10,48,55,58–64 found a significant increase, and seven11,40,65–69 reported no significant difference. Meta-analysis10–11,40,48,55,58–67,69 revealed a significant increase in CRP level in 1692 patients with AAA compared with 2960 controls: SMD 0·58 (0·37 to 0·79) mg/l (P < 0·001; I2 = 87 per cent) (Fig. 3). All studies48,63,66,70–72 except one49 found a positive association between CRP and aortic diameter. Meta-regression10–11,57–58,60–62,65–66,73 showed a significant positive linear correlation between CRP level and aortic diameter (adjusted R2 = 0·161, P = 0·001) (Fig. 4). No association was demonstrated between CRP and aneurysm growth rates36,74–76. Golledge and colleagues77 reported an area under the receiver operating characteristic (ROC) curve (AUC) of 0·61 for CRP; therefore, despite the results of this study, CRP may not be a useful clinical marker. Fig. 3 Open in new tabDownload slide Forest plot comparing plasma expression of acute-phase reactants in patients with abdominal aortic aneurysm (AAA) versus controls: a C-reactive protein and b interleukin 6. An inverse-variance random-effects model was used. Standardized mean differences (SMDs) are shown with 95 per cent confidence intervals Fig. 4 Open in new tabDownload slide Scatter plot demonstrating the association between C-reactive protein (CRP) level and abdominal aortic aneursym (AAA) diameter. Dotted lines represent 95 per cent confidence intervals. Adjusted R2 = 0·161, P = 0·006 Markers of inflammation Interleukins Circulating levels of interleukin (IL) 6 were assessed in six studies40,55,58,78–80, with four55,58,78,80 finding a significant increase and two40,79 reporting no significant difference. Meta-analysis of these studies40,55,58,78–80 found a significant increase in circulating IL-6 levels in 698 patients with AAA compared with 389 controls: SMD 0·90 (0·08 to 1·71) pg/ml (P = 0·030; I2 = 97 per cent) (Fig. 3). This finding is supported by one study81 of aortic tissue IL-6 levels that reported a significant increase in patients with AAA. Two studies investigated differences in IL-10 levels in plasma; one58 found a significant decrease in those with AAA whereas the other79 showed no difference. IL-1β, like IL-6, acts as a driver of inflammation82. Two studies78,83 compared concentrations of circulating IL-1β and one compared aortic tissue IL-1β84; all showed IL-1β levels to be raised significantly in those with AAA. These results demonstrate an increase in IL-1β and IL-6 concentration in patients with AAA, and are supported by tissue analyses of IL-1α, IL-1β, IL-4, IL-5 and IL-6 levels81,85–86, most of which showed an upregulation within aneurysm tissue, although one87 reported downregulation of IL-1β and IL-6. Tumour necrosis factor α Eight studies20,78,80,83–85,88–89 showed a significant upregulation of tumour necrosis factor (TNF) in aortic tissue20,83–85,89, plasma80 and serum78,88 from patients with AAA compared with both controls and subjects with AOD. Only one study87 reported significant downregulation of TNF in patients with AAA compared with AOD. Hamano and co-workers89 found TNF levels to be highest in patients with a small AAA. TNF mRNA levels were reported in three articles83,90–91, with two90,91 documenting significant upregulation in patients with AAA. Osteoprotegerin and osteopontin Moran and co-workers92 identified an 8–12-fold increase in osteoprotegerin (OPG) levels in aortic tissue from patients with AAA; patients with AAA also had higher levels than those with AOD. However, these results were not replicated in aortic tissue by Liu et al.93. A more recent study94 revealed that aortic tissue OPG levels increased with aortic diameter. In the circulation, both blood95 and plasma58 samples demonstrated an increase in OPG levels in patients with AAA, and OPG levels correlated with aneurysm growth rates92; however, no correlation was found with aortic diameter22. Levels of osteopontin (OPN) were increased significantly in both aortic tissue15 and the circulation9,58,77. Golledge and colleagues77 reported an AUC of 0·65 for OPN; therefore, despite the findings of this meta-analysis, it may not useful in the clinical setting. Interferon γ Raised interferon (IFN) γ levels were reported in the circulation of women with AAA (levels in men were not significantly different)78 and in aortic tissue81,83,85. Cell adhesion molecules Increased levels of intercellular cell adhesion molecule (ICAM) 1 were reported in both the circulation88,96 and aortic tissue97,98 of people with AAA, whereas levels of vascular cell adhesion molecule (VCAM) 1 were increased significantly in the circulation96,99 but not in aortic tissue97. Caeruloplasmin One study54 quantified serum caeruloplasmin levels in subjects with AAA or AOD, and found no significant difference. However, a later study53 reported a small but significant upregulation of caeruloplasmin in subjects with AAA compared with controls. There, was however, no correlation between aneurysm diameter and caeruloplasmin levels57. Lipids Five studies53,69,100–102 found a significant increase in cholesterol levels in patients with AAA, whereas four62,80,103–104 reported a significant decrease, and 12 no difference10,48,58–59,68,96,105–110. Meta-analysis revealed no significant difference in cholesterol levels between 2551 patients with AAA and 12 549 without: SMD 0·09 (−0·12 to 0·30) mmol/l (P = 0·390; I2 = 93 per cent) (Fig. S1, supporting information). Meta-regression of 22 studies10,48,53,57–59,62,68–69,80,96,100–110 demonstrated no correlation with AAA diameter (P = 0·930). Triglyceride levels were significantly increased in six studies53,62,69,100,106,109, with no difference found in nine48,58–59,96,102–103,105,108–109; however, meta-analysis revealed a significant increase in triglyceride levels in 1493 patients with AAA compared with 11 532 without: 0·28 (0·18 to 0·37) mmol/l (P < 0·001; I2 = 43 per cent) (Fig. 5). Meta-regression of four studies58,62,105,108 demonstrated no correlation with aortic diameter (P = 0·600). Fig. 5 Open in new tabDownload slide Forest plot comparing plasma lipid expression in patients with abdominal aortic aneurysm (AAA) versus controls: a triglycerides, b high-density lipoprotein (HDL), c apolipoprotein A and d lipoprotein(a). An inverse-variance random-effects model was used. Standardized mean differences (SMDs) are shown with 95 per cent confidence intervals High-density lipoprotein (HDL) levels were significantly increased in two studies102,105, decreased in eight58,69,80,96,100,103,106,109, and no difference was found in seven48,59,62,101,103,108,110. Meta-analysis of the data revealed a significant decrease in HDL levels in 2245 patients with AAA compared with 17 243 controls: SMD −0·29 (−0·50 to −0·08) mmol/l (P = 0·006; I2 = 92 per cent) (Fig. 5). However, meta-regression revealed no correlation with aortic diameter58,62,80,105,108 (P = 0·630). Low-density lipoprotein (LDL) levels were increased significantly in three studies58,69,108, decreased significantly in two62,104, and were no different in nine studies48,59,96,102–103,105–106,109–110. Meta-analysis revealed no significant difference in LDL concentration between 1453 people with AAA and 5786 controls: SMD 0·08 (−0·13 to 0·29) mmol/l (P = 0·470; I2 = 86 per cent) (Fig. S1, supporting information). There was no correlation with aneurysm diameter58,62,105,108 (P = 0·750). Apolipoprotein A-I (ApoA-I) was downregulated significantly in patients with AAA in three studies80,103,106. Meta-analysis of these studies found a significant decrease in ApoA-I levels among 454 people with AAA compared with 1751 controls: SMD −0·75 (−1·20 to −0·30) g/l (P = 0·001; I2 = 89 per cent) (Fig. 5). ApoB levels were assessed in five studies80,103–104,106–107, with meta-analysis revealing no significant difference between 542 patients with AAA and 2247 controls: SMD −0·16 (−0·54 to 0·22) g/l (P = 0·410; I2 = 90 per cent) (Fig. S1, supporting information). A single study80 found a significant decrease in ApoM in patients with AAA, and another103 reported no difference in levels of ApoC-III or ApoE. Three studies110–112 reported a significant increase in lipoprotein(a) in patients with AAA, and two showed no significant difference106,107. Meta-analysis revealed a significant increase in lipoprotein(a) among 1016 patients with AAA compared with 1298 controls: SMD 0·41 (0·03 to 0·79) mg/dl (P = 0·030; I2 = 92 per cent) (Fig. 5). Insulin-like growth factors Circulating levels of insulin-like growth factor (IGF) 1 were raised in patients with AAA113. Studies85,114 examining IGF-1 levels within the aortic wall itself, however, yielded conflicting results. In contrast IGF binding protein (BP) 1 levels were raised in both the plasma and aortic wall of patients with AAA65,114, and raised levels of IGFBP-3 were found in patients with AAA114. Sensitivity analysis and publication bias Sensitivity analysis by sequential exclusion of individual studies did not significantly alter the results for any biomarker (Table S2, supporting information). Funnel plots did not identify any publication bias or biases associated with small study size (Fig. S2, supporting information). Discussion An abundance of markers of AAA have been studied; this review found that AAA was associated with increased levels of several proteases, inflammatory markers, acute-phase reactants and lipids, with levels of two lipids (ApoA and HDL) decreased. Meta-analysis demonstrated a significant difference in MMP-9, TIMP-1, IL-6, CRP, α1AT, triglycerides, lipoprotein(a), ApoA and HDL in patients with AAA, compared with those without. Although meta-analysis was not possible for MMP-2 in aortic tissue, TNF-α, OPG, OPN, IFN-γ, ICAM-1 and VCAM-1, systematic review suggested that levels of these biomarkers were also increased in patients with AAA. Meta-regression identified a significant positive linear correlation between aortic diameter and levels of CRP. Previous systematic reviews115–118 have highlighted these biomarkers as potentially being associated with AAA; however, the present study attempted to meta-analyse these data, where possible. This study, in addition to a recent similar study3 that demonstrated a significant association between the presence of AAA and plasma levels of fibrinogen, D-dimer and thrombin–antithrombin III, and a strong correlation between AAA diameter and D-dimer (r2 = 0·94), should act as a guide to focus future research. MMPs degrade extracellular matrix proteins including collagen and elastin, and their expression has been studied widely in relation to AAA. This study found no difference in circulating MMP-2 levels in patients with AAA compared with controls (P = 0·590), whereas several studies15–21 reported that aortic tissue MMP-2 was increased in patients with AAA. Circulating MMP-9 levels were significantly increased in patients with AAA (P = 0·002), a finding supported by many aortic tissue MMP-9 studies. This suggested an overall increase in MMP-9 levels in the circulation and aortic tissue of subjects with AAA. The association between MMP-1, MMP-3, MMP-8, MMP-12, MMP-13 and MMP-14 with AAA was less clear. TIMPs have been studied extensively with regard to AAA pathophysiology. Circulating levels of TIMP-1 were increased significantly in patients with AAA (P < 0·001); however, these results were not replicated in aortic tissue. The association between TIMP-2/TIMP-3 and AAA was less clear. Cathepsins are proteases that act as potent elastases and collagenases. They are highly expressed in human arterial lesions47 and have been implicated in human arterial wall remodelling48. Cystatins are potent intracellular inhibitors of cathepsin activity, and cystatin C is the most abundant and potent. In addition to inhibiting cathepsins, cystatins may interact directly with MMP-9, potentiating its activity49. Although a deficiency of cystatin C may be associated with an increase in aneurysm size and expansion rate52, this review of the literature did not demonstrate a clear association between cystatins or cathepsins and AAA presence or size. α1AT is a protease inhibitor that acts on a variety of serine proteases, including elastase, and therefore may play a role in maintaining the integrity of the aortic wall. Although no correlation appears to exist between α1AT deficiency and AAA51, this study revealed a significant upregulation of α1AT in AAA (P < 0·001). Acute-phase reactants are proteins whose level rises in response to acute or chronic inflammation. Inflammation is thought to be a key driver of elastin destruction and AAA development119, so a rise in baseline CRP would be anticipated in patients with AAA. Both enzyme-linked immunosorbent and turbidimetric assays have been used to quantify CRP levels (they are deemed equivalent120). Meta-analysis revealed a significant increase in CRP levels in patients with AAA (P < 0·001) as well as a significant correlation between CRP and aortic diameter. No association was demonstrated between CRP and aneurysm growth rates36,74–76. One of the key features of AAA presence and progression is inflammation10. The proinflammatory cytokine IL-6 is a key component in driving the systemic inflammatory response. IL-1β, like IL-6, acts as a driver of inflammation82. IL-10 is a potent anti-inflammatory cytokine, and so an imbalance between these mediators could help explain differences in inflammation seen between those with, and without AAA121. In meta-analysis, circulating levels of IL-6 increased significantly in patients with AAA (P = 0·030), a result supported by findings in aortic tissue IL-6 levels81. Similarly, this systematic review suggested an increase in IL-1β in patients with AAA, but no difference in IL-10 levels. TNF is an inflammatory cytokine that causes smooth muscle cell death and release of proteases, leading to degradation of structural proteins in the aortic wall122. This review suggested a significant upregulation of TNF in aortic tissue, plasma and serum in patients with AAA. TNF levels were highest in patients with a small AAA; Satoh and colleagues90 identified the source of TNF to be the media and adventitia of the aortic wall, primarily at the transition point between aneurysmal and non-aneurysmal tissue. TNF may therefore be involved in the early pathogenesis of AAA formation. OPG is a member of the TNF-related family expressed by vascular smooth muscle cells, and is associated with decreased vascular smooth muscle cell proliferation, arterial calcification, and an increase in MMP secretion from monocytes and vascular smooth muscle cells92,123. This review suggested that OPG is a plausible marker of AAA; it was shown to be upregulated in blood, plasma and aortic tissue compared with levels in controls, and potentially correlated with AAA growth rates. OPN increases the expression of MMPs, and is involved in collagen synthesis, angiogenesis, neovascularization and calcification. It was significantly increased in both aortic tissue15 and the circulation9,58,77 of people with AAA. IFN-γ is an inflammatory cytokine that interacts with macrophages. Its production is controlled by interleukins124. ICAM-1 and VCAM-1 are members of the immunoglobulin superfamily mediating the adhesion of leucocytes to endothelial cells125. Caeruloplasmin is an acute-phase reactant linked to inflammatory disease and is an independent marker of cardiovascular risk126. This review demonstrated an increase in IFN-γ, VCAM-1 and ICAM-1 levels in subjects with AAA, with no such findings regarding caeruloplasmin57. Meta-analyses suggested an association with the presence of AAA for triglyceride (P < 0·001), HDL (P = 0·006), ApoA-I (P = 0·001) and lipoprotein(a) (P = 0·030) levels, but not for cholesterol, LDL and ApoB levels. These results are in keeping with previous meta-analyses127,128 that identified significant downregulation of HDL and lipoprotein(a) in subjects with AAA. Takagi and colleagues127 also reported significant upregulation of LDL; however, this effect appeared to be lost with the addition of six studies in the present analysis. Although a dysregulation of lipid levels is seen in subjects with AAA, association may not imply causation, and whether alteration of lipid levels can be used to affect AAA outcomes is unknown. IGF is involved in cell proliferation, collagen synthesis and matrix degradation. Circulating and aortic tissue levels of IGF-1, IGFBP-1 and IGFBP-3 may be raised in patients with AAA. The significance of these findings is unclear, but it is possible that changes in IGF-1 levels may have a role in mediating the protective effect of diabetes on AAA because diabetes is known to dysregulate circulating IGF-1 levels129. One common limitation of these studies is that individual biomarker sensitivity and specificity remain low, as each of these markers has been associated with a wide variety of clinical pathologies, particularly acute inflammatory and atherosclerotic processes. Significant heterogeneity exists between studies; for example, many did not control for smoking or other causes of raised inflammatory markers. Some studies used patients with peripheral arterial disease as controls. Although the findings were significant, many of the studies were small, and so the results need to be taken in context. Although these biomarkers are likely to be related to the underlying pathophysiology of the disease, this approach cannot differentiate between biomarkers that cause AAA, or those with raised levels owing to the presence of AAA. Another limitation of these studies is that levels of inflammatory markers are known to depend on age and smoking habits78,121, which were not controlled for in many of these studies79. Although zymography studies looking at biomarker activity in plasma were included, their inhibitors will also have been present, thus affecting the results. TNF and IL-1 protein levels in aortic tissue are low, and as such are less reliable; however, this study aimed to include all possible data to amalgamate current knowledge concerning AAA biomarkers. Although some of the P values were marginally significant, such as those for IL-6, ApoA-I and lipoprotein(a), meta-analysis used a large number of patients for each of these biomarkers, and systematic review demonstrated consist differences between the aneurysm and control groups, suggesting that these differences are real. An ideal method of reporting biomarker studies would be to use ROC curves. ROC curve analysis has been sparsely reported in studies on AAA biomarkers. However, Golledge and colleagues77 reported an AUC of 0·65 for OPN and 0·61 for CRP, with a combination of OPN and CRP leading to an AUC of 0·66. An additional study by Golledge and co-workers61 reported an AUC of 0·83 for D-dimer. Although these results show promise, they remain insufficient for use as a screening tool for AAA. Using an array of biomarkers, such as some of those identified in this article, may potentially provide a sufficiently sensitive and specific panel for accurate screening of patients with AAA. This would, however, require a large-scale study to determine the value of this approach. Acknowledgements P.W.S. and D.A.S are joint first authors of this article. The study received funding from the National Institute for Health Research Leicester Cardiovascular Biomedical Research Unit. P.W.S. is funded by a Royal College of Surgeons/Dunhill Medical Trust Research Fellowship; and M.J.B. by a Higher Education Funding Council for England Clinical Senior Lecturer Fellowship and the Circulation Foundation. Disclosure: The authors declare no conflict of interest. References 1 Lozano R , Naghavi M, Foreman K, Lim S, Shibuya K, Aboyans V et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010 . Lancet 2012 ; 380 : 2095 – 2128 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Stather PW , Dattani N, Bown MJ, Earnshaw JJ, Lees TA. International variations in AAA screening . Eur J Vasc Endovasc Surg 2013 ; 45 : 231 – 234 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Sidloff D , Stather PW, Choke E, Bown MJ, Sayers RD. A systematic review and meta-analysis of the association between markers of haemostasis and abdominal aortic aneurysm presence and size . J Vasc Surg 2014; 59 : 528 – 535 .e4. Crossref Search ADS PubMed WorldCat 4 Moher D , Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement . Int J Surg 2010 ; 8 : 336 – 341 . Google Scholar Crossref Search ADS PubMed WorldCat 5 The Cochrane Collaboration . Review Manager (RevMan) [computer program]. Version 5.0. The Nordic Cochrane Centre : Copenhagen , 2008 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 6 Higgins JPT , Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011 . 7 Higgins JP , Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses . BMJ 2003 ; 327 : 557 – 560 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Egger M , Smith GD. Bias in location and selection of studies . BMJ 1998 ; 316 : 61 – 66 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Huusko T , Salonurmi T, Taskinen P, Liinamaa J, Juvonen T, Pääkkö P et al. Elevated messenger RNA expression and plasma protein levels of osteopontin and matrix metalloproteinase types 2 and 9 in patients with ascending aortic aneurysms . J Thorac Cardiovasc Surg 2013 ; 145 : 1117 – 1123 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Hellenthal FA , Pulinx B, Welten RJ, Teijink JA, van Dieijen-Visser MP, Wodzig WK et al. Circulating biomarkers and abdominal aortic aneurysm size . J Surg Res 2012 ; 176 : 672 – 678 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Villard C , Wågsäter D, Swedenborg J, Eriksson P, Hultgren R. Biomarkers for abdominal aortic aneurysms from a sex perspective . Gend Med 2012 ; 9 : 259 – 266 .e2. Google Scholar Crossref Search ADS PubMed WorldCat 12 Eugster T , Huber A, Obeid T, Schwegler I, Gürke L, Stierli P. Aminoterminal propeptide of type III procollagen and matrix metalloproteinases-2 and -9 failed to serve as serum markers for abdominal aortic aneurysm . Eur J Vasc Endovasc Surg 2005 ; 29 : 378 – 382 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Flondell-Sité D , Lindblad B, Kölbel T, Gottsäter A. Markers of proteolysis, fibrinolysis, and coagulation in relation to size and growth rate of abdominal aortic aneurysms . Vasc Endovascular Surg 2010 ; 44 : 262 – 268 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Antoniou GA , Tentes IK, Antoniou SA, Georgiadis GS, Giannoukas AD, Simopoulos C et al. Circulating matrix metalloproteinases and their inhibitors in inguinal hernia and abdominal aortic aneurysm . Int Angiol 2011 ; 30 : 123 – 129 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 15 Mi T , Nie B, Zhang C, Zhou H. The elevated expression of osteopontin and NF-kB in human aortic aneurysms and its implication . J Huazhong Univ Sci Technolog Med Sci 2011 ; 31 : 602 – 607 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Davis V , Persidskaia R, Baca-Regen L, Itoh Y, Nagase H, Persidsky Y et al. Matrix metalloproteinase-2 production and its binding to the matrix are increased in abdominal aortic aneurysms . Arterioscler Thromb Vasc Biol 1998 ; 18 : 1625 – 1633 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Crowther M , Goodall S, Jones JL, Bell PR, Thompson MM. Increased matrix metalloproteinase 2 expression in vascular smooth muscle cells cultured from abdominal aortic aneurysms . J Vasc Surg 2000 ; 32 : 575 – 583 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Sakalihasan N , Delvenne P, Nusgens BV, Limet R, Lapière CM. Activated forms of MMP2 and MMP9 in abdominal aortic aneurysms . J Vasc Surg 1996 ; 24 : 127 – 133 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Yamashita A , Noma T, Nakazawa A, Saito S, Fujioka K, Zempo N et al. Enhanced expression of matrix metalloproteinase-9 in abdominal aortic aneurysms . World J Surg 2001 ; 25 : 259 – 265 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Kaneko H , Anzai T, Horiuchi K, Kohno T, Nagai T, Anzai A et al. Tumor necrosis factor-α converting enzyme is a key mediator of abdominal aortic aneurysm development . Atherosclerosis 2011 ; 218 : 470 – 478 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Saito S , Zempo N, Yamashita A, Takenaka H, Fujioka K, Esato K. Matrix metalloproteinase expressions in arteriosclerotic aneurysmal disease . Vasc Endovascular Surg 2002 ; 36 : 1 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Muehling B , Oberhuber A, Schelzig H, Bischoff G, Marx N, Sunder-Plassmann L et al. Effect of statin therapy on serum activity of proteinases and cytokines in patients with abdominal aortic aneurysm . Vasc Health Risk Manag 2008 ; 4 : 1433 – 1437 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Lindholt JS , Vammen S, Fasting H, Henneberg EW, Heickendorff L. The plasma level of matrix metalloproteinase 9 may predict the natural history of small abdominal aortic aneurysms. A preliminary study . Eur J Vasc Endovasc Surg 2000 ; 20 : 281 – 285 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Petersen E , Wågberg F, Angquist KA. Proteolysis of the abdominal aortic aneurysm wall and the association with rupture . Eur J Vasc Endovasc Surg 2002 ; 23 : 153 – 157 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Watanabe T , Sato A, Sawai T, Uzuki M, Goto H, Yamashita H et al. The elevated level of circulating matrix metalloproteinase-9 in patients with abdominal aortic aneurysms decreased to levels equal to those of healthy controls after an aortic repair . Ann Vasc Surg 2006 ; 20 : 317 – 321 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Sangiorgi G , D'Averio R, Mauriello A, Bondio M, Pontillo M, Castelvecchio S et al. Plasma levels of metalloproteinases-3 and -9 as markers of successful abdominal aortic aneurysm exclusion after endovascular graft treatment . Circulation 2001 ; 104 ( Suppl 1 ): I288 – 295 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Tsarouhas K , Tsitsimpikou C, Apostolakis S, Haliassos A, Tzardi M, Panagiotou M et al. Homocysteine and metalloprotease-3 and -9 in patients with ascending aorta aneurysms . Thromb Res 2011 ; 128 : e95 – e99 . Google Scholar Crossref Search ADS PubMed WorldCat 28 Hovsepian DM , Ziporin SJ, Sakurai MK, Lee JK, Curci JA, Thompson RW. Elevated plasma levels of matrix metalloproteinase-9 in patients with abdominal aortic aneurysms: a circulating marker of degenerative aneurysm disease . J Vasc Interv Radiol 2000 ; 11 : 1345 – 1352 . Google Scholar Crossref Search ADS PubMed WorldCat 29 McMillan WD , Pearce WH. Increased plasma levels of metalloproteinase-9 are associated with abdominal aortic aneurysms . J Vasc Surg 1999 ; 29 : 122 – 127 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Taurino M , Visco V, Raffa S, Ricci B, Ruggiero M, Torrisi MR et al. Matrix metalloproteinase 9 activity in patients before and after endovascular or surgical repair of abdominal aortic aneurysms . Vascular 2004 ; 12 : 312 – 317 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Smallwood L , Allcock R, van Bockxmeer F, Warrington N, Palmer LJ, Iacopetta B et al. Polymorphisms of the matrix metalloproteinase 9 gene and abdominal aortic aneurysm . Br J Surg 2008 ; 95 : 1239 – 1244 . Google Scholar Crossref Search ADS PubMed WorldCat 32 van Laake LW , Vainas T, Dammers R, Kitslaar PJ, Hoeks AP, Schurink GW. Systemic dilation diathesis in patients with abdominal aortic aneurysms: a role for matrix metalloproteinase-9? Eur J Vasc Endovasc Surg 2005 ; 29 : 371 – 377 . Google Scholar Crossref Search ADS PubMed WorldCat 33 Abisi S , Burnand KG, Waltham M, Humphries J, Taylor PR, Smith A. Cysteine protease activity in the wall of abdominal aortic aneurysms . J Vasc Surg 2007 ; 46 : 1260 – 1266 . Google Scholar Crossref Search ADS PubMed WorldCat 34 Diehm N , Di Santo S, Schaffner T, Schmidli J, Völzmann J, Jüni P et al. Severe structural damage of the seemingly non-diseased infrarenal aortic aneurysm neck . J Vasc Surg 2008 ; 48 : 425 – 434 . Google Scholar Crossref Search ADS PubMed WorldCat 35 Abdul-Hussien H , Soekhoe RG, Weber E, von der Thüsen JH, Kleemann R, Mulder A et al. Collagen degradation in the abdominal aneurysm: a conspiracy of matrix metalloproteinase and cysteine collagenases . Am J Pathol 2007 ; 170 : 809 – 817 . Google Scholar Crossref Search ADS PubMed WorldCat 36 Karlsson L , Bergqvist D, Lindbäck J, Parsson H. Expansion of small-diameter abdominal aortic aneurysms is not reflected by the release of inflammatory mediators IL-6, MMP-9 and CRP in plasma . Eur J Vasc Endovasc Surg 2009 ; 37 : 420 – 424 . Google Scholar Crossref Search ADS PubMed WorldCat 37 Wilson WR , Anderton M, Choke EC, Dawson J, Loftus IM, Thompson MM. Plasma matrix metalloproteinase levels do not predict tissue levels in abdominal aortic aneurysms suitable for elective repair . Vascular 2008 ; 16 : 248 – 252 . Google Scholar Crossref Search ADS PubMed WorldCat 38 Lamblin N , Ratajczak P, Hot D, Dubois E, Chwastyniak M, Beseme O et al. Profile of macrophages in human abdominal aortic aneurysms: a transcriptomic, proteomic, and antibody protein array study . J Proteome Res 2010 ; 9 : 3720 – 3729 . Google Scholar Crossref Search ADS PubMed WorldCat 39 Kimura T , Yoshimura K, Aoki H, Imanaka-Yoshida K, Yoshida T, Ikeda Y et al. Tenascin-C is expressed in abdominal aortic aneurysm tissue with an active degradation process . Pathol Int 2011 ; 61 : 559 – 564 . Google Scholar Crossref Search ADS PubMed WorldCat 40 Pradhan-Palikhe P , Vikatmaa P, Lajunen T, Palikhe A, Lepantalo M, Tervahartiala T et al. Elevated MMP-8 and decreased myeloperoxidase concentrations associate significantly with the risk for peripheral atherosclerosis disease and abdominal aortic aneurysm . Scand J Immunol 2010 ; 72 : 150 – 157 . Google Scholar Crossref Search ADS PubMed WorldCat 41 Wilson WR , Schwalbe EC, Jones JL, Bell PR, Thompson MM. Matrix metalloproteinase 8 (neutrophil collagenase) in the pathogenesis of abdominal aortic aneurysm . Br J Surg 2005 ; 92 : 828 – 833 . Google Scholar Crossref Search ADS PubMed WorldCat 42 Curci JA , Liao S, Huffman MD, Shapiro SD, Thompson RW. Expression and localization of macrophage elastase (matrix metalloproteinase-12) in abdominal aortic aneurysms . J Clin Invest 1998 ; 102 : 1900 – 1910 . Google Scholar Crossref Search ADS PubMed WorldCat 43 Mao D , Lee JK, VanVickle SJ, Thompson RW. Expression of collagenase-3 (MMP-13) in human abdominal aortic aneurysms and vascular smooth muscle cells in culture . Biochem Biophys Res Commun 1999 ; 261 : 904 – 910 . Google Scholar Crossref Search ADS PubMed WorldCat 44 Nollendorfs A , Greiner TC, Nagase H, Baxter BT. The expression and localization of membrane type-1 matrix metalloproteinase in human abdominal aortic aneurysms . J Vasc Surg 2001 ; 34 : 316 – 322 . Google Scholar Crossref Search ADS PubMed WorldCat 45 Nakamura M , Tachieda R, Niinuma H, Ohira A, Endoh S, Hiramori K et al. Circulating biochemical marker levels of collagen metabolism are abnormal in patients with abdominal aortic aneurysm . Angiology 2000 ; 51 : 385 – 392 . Google Scholar Crossref Search ADS PubMed WorldCat 46 Lv BJ , Lindholt JS, Wang J, Cheng X, Shi GP. Plasma levels of cathepsins L, K, and V and risks of abdominal aortic aneurysms: a randomized population-based study . Atherosclerosis 2013 ; 230 : 100 – 105 . Google Scholar Crossref Search ADS PubMed WorldCat 47 Gacko M , Ostapowicz R, Chrostek L, Worowska A, Kordecki K. Activity of enzymes with different subcellular localization in the blood plasma of patients with aortic aneurysm . Med Sci Monit 2005 ; 11 : CR211 – CR213 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 48 Qin Y , Yang Y, Liu R, Cao X, Liu O, Liu J et al. Combined cathepsin S and hs-CRP predicting inflammation of abdominal aortic aneurysm . Clin Biochem 2013 ; 46 : 1026 – 1029 . Google Scholar Crossref Search ADS PubMed WorldCat 49 Lindholt JS , Erlandsen EJ, Henneberg EW. Cystatin C deficiency is associated with the progression of small abdominal aortic aneurysms . Br J Surg 2001 ; 88 : 1472 – 1475 . Google Scholar Crossref Search ADS PubMed WorldCat 50 Shi GP , Sukhova GK, Grubb A, Ducharme A, Rhode LH, Lee RT et al. Cystatin C deficiency in human atherosclerosis and aortic aneurysms . J Clin Invest 1999 ; 104 : 1191 – 1197 . Google Scholar Crossref Search ADS PubMed WorldCat 51 Elzouki AN , Rydén Ahlgren A, Länne T, Sonesson B, Eriksson S. Is there a relationship between abdominal aortic aneurysms and alpha1-antitrypsin deficiency (PiZ)? Eur J Vasc Endovasc Surg 1999 ; 17 : 149 – 154 . Google Scholar Crossref Search ADS PubMed WorldCat 52 Gamberi T , Puglia M, Guidi F, Magherini F, Bini L, Marzocchini R et al. A proteomic approach to identify plasma proteins in patients with abdominal aortic aneurysm . Mol Biosyst 2011 ; 7 : 2855 – 2862 . Google Scholar Crossref Search ADS PubMed WorldCat 53 Lindblad B , Börner G, Gottsäter A. Factors associated with development of large abdominal aortic aneurysm in middle-aged men . Eur J Vasc Endovasc Surg 2005 ; 30 : 346 – 352 . Google Scholar Crossref Search ADS PubMed WorldCat 54 Powell JT , Muller BR, Greenhalgh RM. Acute phase proteins in patients with abdominal aortic aneurysms . J Cardiovasc Surg (Torino) 1987 ; 28 : 528 – 530 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 55 Lindqvist M , Wallinder J, Henriksson AE. Soluble urokinase plasminogen activator receptor in patients with abdominal aortic aneurysm . Thromb Res 2012 ; 130 : 511 – 513 . Google Scholar Crossref Search ADS PubMed WorldCat 56 Ramsbottom D , Fitzgerald P, Grace PA, McAnena O, Burke P, Collins P et al. Biochemical and molecular genetic studies of abdominal aortic aneurysm in an Irish population . Eur J Vasc Surg 1994 ; 8 : 716 – 722 . Google Scholar Crossref Search ADS PubMed WorldCat 57 Flondell-Sité D , Lindblad B, Kölbel T, Gottsäter A. Cytokines and systemic biomarkers are related to the size of abdominal aortic aneurysms . Cytokine 2009 ; 46 : 211 – 215 . Google Scholar Crossref Search ADS PubMed WorldCat 58 Kadoglou NP , Papadakis I, Moulakakis KG, Ikonomidis I, Alepaki M, Moustardas P et al. Arterial stiffness and novel biomarkers in patients with abdominal aortic aneurysms . Regul Pept 2012 ; 179 : 50 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat 59 Pan JP , Cheng TM, Shih CC, Chiang SC, Chou SC, Mao SJ et al. Haptoglobin phenotypes and plasma haptoglobin levels in patients with abdominal aortic aneurysm . J Vasc Surg 2011 ; 53 : 1189 – 1194 . Google Scholar Crossref Search ADS PubMed WorldCat 60 Golledge J , Muller R, Clancy P, McCann M, Norman PE. Evaluation of the diagnostic and prognostic value of plasma D-dimer for abdominal aortic aneurysm . Eur Heart J 2011 ; 32 : 354 – 364 . Google Scholar Crossref Search ADS PubMed WorldCat 61 Golledge J , Biros E, Cooper M, Warrington N, Palmer LJ, Norman PE. Apolipoprotein E genotype is associated with serum C-reactive protein but not abdominal aortic aneurysm . Atherosclerosis 2010 ; 209 : 487 – 491 . Google Scholar Crossref Search ADS PubMed WorldCat 62 Parry DJ , Al-Barjas HS, Chappell L, Rashid ST, Ariëns RA, Scott DJ. Markers of inflammation in men with small abdominal aortic aneurysm . J Vasc Surg 2010 ; 52 : 145 – 151 . Google Scholar Crossref Search ADS PubMed WorldCat 63 Badger SA , Soong CV, O'Donnell ME, Mercer C, Young IS, Hughes AE. C-reactive protein (CRP) elevation in patients with abdominal aortic aneurysm is independent of the most important CRP genetic polymorphism . J Vasc Surg 2009 ; 49 : 178 – 184 . Google Scholar Crossref Search ADS PubMed WorldCat 64 Powell JT , Greenhalgh RM. Multifactorial inheritance of abdominal aortic aneurysm . Eur J Vasc Surg 1987 ; 1 : 29 – 31 . Google Scholar Crossref Search ADS PubMed WorldCat 65 Ramos-Mozo P , Rodriguez C, Pastor-Vargas C, Blanco-Colio LM, Martinez-Gonzalez J, Meilhac O et al. Plasma profiling by a protein array approach identifies IGFBP-1 as a novel biomarker of abdominal aortic aneurysm . Atherosclerosis 2012 ; 221 : 544 – 550 . Google Scholar Crossref Search ADS PubMed WorldCat 66 Vega de Céniga M , Esteban M, Quintana JM, Barba A, Estallo L, de la Fuente N et al. Search for serum biomarkers associated with abdominal aortic aneurysm growth – a pilot study . Eur J Vasc Endovasc Surg 2009 ; 37 : 297 – 299 . Google Scholar Crossref Search ADS PubMed WorldCat 67 Dawson J , Cockerill GW, Choke E, Belli AM, Loftus I, Thompson MM. Aortic aneurysms secrete interleukin-6 into the circulation . J Vasc Surg 2007 ; 45 : 350 – 356 . Google Scholar Crossref Search ADS PubMed WorldCat 68 Forester ND , Cruickshank SM, Scott DJ, Carding SR. Increased natural killer cell activity in patients with an abdominal aortic aneurysm . Br J Surg 2006 ; 93 : 46 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat 69 Wanhainen A , Bergqvist D, Boman K, Nilsson TK, Rutegård J, Björck M. Risk factors associated with abdominal aortic aneurysm: a population-based study with historical and current data . J Vasc Surg 2005 ; 41 : 390 – 396 . Google Scholar Crossref Search ADS PubMed WorldCat 70 Muehling BM , Paintner A, Marx N, Barth TF, Babiak C, Orend KH. In vivo study on the expression pattern of resistin in patients with abdominal aortic aneurysm . Vasc Endovascular Surg 2011 ; 45 : 63 – 68 . Google Scholar Crossref Search ADS PubMed WorldCat 71 Vainas T , Lubbers T, Stassen FR, Herngreen SB, van Dieijen-Visser MP, Bruggeman CA et al. Serum C-reactive protein level is associated with abdominal aortic aneurysm size and may be produced by aneurysmal tissue . Circulation 2003 ; 107 : 1103 – 1105 . Google Scholar Crossref Search ADS PubMed WorldCat 72 De Haro J , Acin F, Bleda S, Varela C, Medina FJ, Esparza L. Prediction of asymptomatic abdominal aortic aneurysm expansion by means of rate of variation of C-reactive protein plasma levels . J Vasc Surg 2012 ; 56 : 45 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat 73 Medina F , de Haro J, Florez A, Acin F. Relationship between endothelial dependent vasodilation and size of abdominal aortic aneurysms . Ann Vasc Surg 2010 ; 24 : 752 – 757 . Google Scholar Crossref Search ADS PubMed WorldCat 74 Speelman L , Hellenthal FA, Pulinx B, Bosboom EM, Breeuwer M, van Sambeek MR et al. The influence of wall stress on AAA growth and biomarkers . Eur J Vasc Endovasc Surg 2010 ; 39 : 410 – 416 . Google Scholar Crossref Search ADS PubMed WorldCat 75 Flondell-Sité D , Lindblad B, Gottsäter A. High levels of endothelin (ET)-1 and aneurysm diameter independently predict growth of stable abdominal aortic aneurysms . Angiology 2010 ; 61 : 324 – 328 . Google Scholar Crossref Search ADS PubMed WorldCat 76 Norman P , Spencer CA, Lawrence-Brown MM, Jamrozik K. C-reactive protein levels and the expansion of screen-detected abdominal aortic aneurysms in men . Circulation 2004 ; 110 : 862 – 866 . Google Scholar Crossref Search ADS PubMed WorldCat 77 Golledge J , Muller J, Shephard N, Clancy P, Smallwood L, Moran C et al. Association between osteopontin and human abdominal aortic aneurysm . Arterioscler Thromb Vasc Biol 2007 ; 27 : 655 – 660 . Google Scholar Crossref Search ADS PubMed WorldCat 78 Juvonen J , Surcel HM, Satta J, Teppo AM, Bloigu A, Syrjala H et al. Elevated circulating levels of inflammatory cytokines in patients with abdominal aortic aneurysm . Arterioscler Thromb Vasc Biol 1997 ; 17 : 2843 – 2847 . Google Scholar Crossref Search ADS PubMed WorldCat 79 Wallinder J , Bergqvist D, Henriksson AE. Proinflammatory and anti-inflammatory cytokine balance in patients with abdominal aortic aneurysm and the impact of aneurysm size . Vasc Endovascular Surg 2009 ; 43 : 258 – 261 . Google Scholar Crossref Search ADS PubMed WorldCat 80 Ahnström J , Gottsäter A, Lindblad B, Dahlbäck B. Plasma concentrations of apolipoproteins A-I, B and M in patients with abdominal aortic aneurysms . Clin Biochem 2010 ; 43 : 407 – 410 . Google Scholar Crossref Search ADS PubMed WorldCat 81 Szekanecz Z , Shah MR, Pearce WH, Koch AE. Human atherosclerotic abdominal aortic aneurysms produce interleukin (IL)-6 and interferon-gamma but not IL-2 and IL-4: the possible role for IL-6 and interferon-gamma in vascular inflammation . Agents Actions 1994 ; 42 : 159 – 162 . Google Scholar Crossref Search ADS PubMed WorldCat 82 Kanse SM , Benzakour O, Kanthou C, Kost C, Lijnen HR, Preissner KT. Induction of vascular SMC proliferation by urokinase indicates a novel mechanism of action in vasoproliferative disorders . Arterioscler Thromb Vasc Biol 1997 ; 17 : 2848 – 2854 . Google Scholar Crossref Search ADS PubMed WorldCat 83 Lindeman JH , Abdul-Hussien H, Schaapherder AF, Van Bockel JH, Von der Thüsen JH, Roelen DL et al. Enhanced expression and activation of pro-inflammatory transcription factors distinguish aneurysmal from atherosclerotic aorta: IL-6- and IL-8-dominated inflammatory responses prevail in the human aneurysm . Clin Sci (Lond) 2008 ; 114 : 687 – 697 . Google Scholar Crossref Search ADS PubMed WorldCat 84 Newman KM , Jean-Claude J, Li H, Ramey WG, Tilson MD. Cytokines that activate proteolysis are increased in abdominal aortic aneurysms . Circulation 1994 ; 90 : II224 – II227 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 85 Middleton RK , Lloyd GM, Bown MJ, Cooper NJ, London NJ, Sayers RD. The pro-inflammatory and chemotactic cytokine microenvironment of the abdominal aortic aneurysm wall: a protein array study . J Vasc Surg 2007 ; 45 : 574 – 580 . Google Scholar Crossref Search ADS PubMed WorldCat 86 Schönbeck U , Sukhova GK, Gerdes N, Libby P. T(H)2 predominant immune responses prevail in human abdominal aortic aneurysm . Am J Pathol 2002 ; 161 : 499 – 506 . Google Scholar Crossref Search ADS PubMed WorldCat 87 Davis VA , Persidskaia RN, Baca-Regen LM, Fiotti N, Halloran BG, Baxter BT. Cytokine pattern in aneurysmal and occlusive disease of the aorta . J Surg Res 2001 ; 101 : 152 – 156 . Google Scholar Crossref Search ADS PubMed WorldCat 88 Witkowska AM , Borawska MH, Gacko M. Relationship among TNF-alpha, sICAM-1, and selenium in presurgical patients with abdominal aortic aneurysms . Biol Trace Elem Res 2006 ; 114 : 31 – 40 . Google Scholar Crossref Search ADS PubMed WorldCat 89 Hamano K , Li TS, Takahashi M, Kobayashi T, Shirasawa B, Ito H et al. Enhanced tumor necrosis factor-alpha expression in small sized abdominal aortic aneurysms . World J Surg 2003 ; 27 : 476 – 480 . Google Scholar Crossref Search ADS PubMed WorldCat 90 Satoh H , Nakamura M, Satoh M, Nakajima T, Izumoto H, Maesawa C et al. Expression and localization of tumour necrosis factor-alpha and its converting enzyme in human abdominal aortic aneurysm . Clin Sci (Lond) 2004 ; 106 : 301 – 306 . Google Scholar Crossref Search ADS PubMed WorldCat 91 Shteinberg D , Halak M, Shapiro S, Kinarty A, Sobol E, Lahat N et al. Abdominal aortic aneurysm and aortic occlusive disease: a comparison of risk factors and inflammatory response . Eur J Vasc Endovasc Surg 2000 ; 20 : 462 – 465 . Google Scholar Crossref Search ADS PubMed WorldCat 92 Moran CS , McCann M, Karan M, Norman P, Ketheesan N, Golledge J. Association of osteoprotegerin with human abdominal aortic aneurysm progression . Circulation 2005 ; 111 : 3119 – 3125 . Google Scholar Crossref Search ADS PubMed WorldCat 93 Liu X , Winrow VR, Horrocks M, Stevens CR. Differential expression of TRAIL and its receptors relative to calcification in AAA . Biochem Biophys Res Commun 2007 ; 358 : 18 – 23 . Google Scholar Crossref Search ADS PubMed WorldCat 94 Koole D , Hurks R, Schoneveld A, Vink A, Golledge J, Moran CS et al. Osteoprotegerin is associated with aneurysm diameter and proteolysis in abdominal aortic aneurysm disease . Arterioscler Thromb Vasc Biol 2012 ; 32 : 1497 – 1504 . Google Scholar Crossref Search ADS PubMed WorldCat 95 Moran CS , Clancy P, Biros E, Blanco-Martin B, McCaskie P, Palmer LJ et al. Association of PPARgamma allelic variation, osteoprotegerin and abdominal aortic aneurysm . Clin Endocrinol (Oxf) 2010 ; 72 : 128 – 132 . Google Scholar Crossref Search ADS PubMed WorldCat 96 Blann AD , Devine C, Amiral J, McCollum CN. Soluble adhesion molecules, endothelial markers and atherosclerosis risk factors in abdominal aortic aneurysm: a comparison with claudicants and healthy controls . Blood Coagul Fibrinolysis 1998 ; 9 : 479 – 484 . Google Scholar Crossref Search ADS PubMed WorldCat 97 Davis CA III, Pearce WH, Haines GK, Shah M, Koch AE. Increased ICAM-1 expression in aortic disease . J Vasc Surg 1993 ; 18 : 875 – 880 . Google Scholar Crossref Search ADS PubMed WorldCat 98 Szekanecz Z , Shah MR, Pearce WH, Koch AE. Intercellular adhesion molecule-1 (ICAM-1) expression and soluble ICAM-1 (sICAM-1) production by cytokine-activated human aortic endothelial cells: a possible role for ICAM-1 and sICAM-1 in atherosclerotic aortic aneurysms . Clin Exp Immunol 1994 ; 98 : 337 – 343 . Google Scholar Crossref Search ADS PubMed WorldCat 99 Nakai K , Itoh C, Kawazoe K, Miura Y, Sotoyanagi H, Hotta K et al. Concentration of soluble vascular cell adhesion molecule-1 (VCAM-1) correlated with expression of VCAM-1 mRNA in the human atherosclerotic aorta . Coron Artery Dis 1995 ; 6 : 497 – 502 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 100 Singh K , Bønaa KH, Jacobsen BK, Bjork L, Solberg S. Prevalence of and risk factors for abdominal aortic aneurysms in a population-based study : the Tromso Study . Am J Epidemiol 2001 ; 154 : 236 – 244 . Google Scholar Crossref Search ADS PubMed WorldCat 101 Pleumeekers HJ , Hoes AW, van der Does E, van Urk H, Hofman A, de Jong PT et al. Aneurysms of the abdominal aorta in older adults. The Rotterdam Study . Am J Epidemiol 1995 ; 142 : 1291 – 1299 . Google Scholar Crossref Search ADS PubMed WorldCat 102 Golledge J , van Bockxmeer F, Jamrozik K, McCann M, Norman PE. Association between serum lipoproteins and abdominal aortic aneurysm . Am J Cardiol 2010 ; 105 : 1480 – 1484 . Google Scholar Crossref Search ADS PubMed WorldCat 103 McConathy WJ , Alaupovic P, Woolcock N, Laing SP, Powell J, Greenhalgh R. Lipids and apolipoprotein profiles in men with aneurysmal and stenosing aorto-iliac atherosclerosis . Eur J Vasc Surg 1989 ; 3 : 511 – 514 . Google Scholar Crossref Search ADS PubMed WorldCat 104 Naydeck BL , Sutton-Tyrrell K, Schiller KD, Newman AB, Kuller LH. Prevalence and risk factors for abdominal aortic aneurysms in older adults with and without isolated systolic hypertension . Am J Cardiol 1999 ; 83 : 759 – 764 . Google Scholar Crossref Search ADS PubMed WorldCat 105 Louwrens HD , Adamson J, Powell JT, Greenhalgh RM. Risk factors for atherosclerosis in men with stenosing or aneurysmal disease of the abdominal aorta . Int Angiol 1993 ; 12 : 21 – 24 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 106 Simoni G , Gianotti A, Ardia A, Baiardi A, Galleano R, Civalleri D. Screening study of abdominal aortic aneurysm in a general population: lipid parameters . Cardiovasc Surg 1996 ; 4 : 445 – 448 . Google Scholar Crossref Search ADS PubMed WorldCat 107 Franks PJ , Edwards RJ, Greenhalgh RM, Powell JT. Risk factors for abdominal aortic aneurysms in smokers . Eur J Vasc Endovasc Surg 1996 ; 11 : 487 – 492 . Google Scholar Crossref Search ADS PubMed WorldCat 108 Hobbs SD , Claridge MW, Quick CR, Day NE, Bradbury AW, Wilmink AB. LDL cholesterol is associated with small abdominal aortic aneurysms . Eur J Vasc Endovasc Surg 2003 ; 26 : 618 – 622 . Google Scholar Crossref Search ADS PubMed WorldCat 109 Rizzo M , Krayenbuhl PA, Pernice V, Frasheri A, Battista Rini G, Berneis K. LDL size and subclasses in patients with abdominal aortic aneurysm . Int J Cardiol 2009 ; 134 : 406 – 408 . Google Scholar Crossref Search ADS PubMed WorldCat 110 Jones GT , van Rij AM, Cole J, Williams MJ, Bateman EH, Marcovina SM et al. Plasma lipoprotein(a) indicates risk for 4 distinct forms of vascular disease . Clin Chem 2007 ; 53 : 679 – 685 . Google Scholar Crossref Search ADS PubMed WorldCat 111 Schillinger M , Domanovits H, Ignatescu M, Exner M, Bayegan K, Sedivy R et al. Lipoprotein (a) in patients with aortic aneurysmal disease . J Vasc Surg 2002 ; 36 : 25 – 30 . Google Scholar Crossref Search ADS PubMed WorldCat 112 Sofi F , Marcucci R, Giusti B, Pratesi G, Lari B, Sestini I et al. High levels of homocysteine, lipoprotein (a) and plasminogen activator inhibitor-1 are present in patients with abdominal aortic aneurysm . Thromb Haemost 2005 ; 94 : 1094 – 1098 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 113 Yeap BB , Chubb SA, McCaul KA, Flicker L, Ho KK, Golledge J et al. Associations of IGF1 and its binding proteins with abdominal aortic aneurysm and aortic diameter in older men . Eur J Endocrinol 2012 ; 166 : 191 – 197 . Google Scholar Crossref Search ADS PubMed WorldCat 114 Panek B , Gacko M, Palka J. Metalloproteinases, insulin-like growth factor-I and its binding proteins in aortic aneurysm . Int J Exp Pathol 2004 ; 85 : 159 – 164 . Google Scholar Crossref Search ADS PubMed WorldCat 115 Golledge AL , Walker P, Norman PE, Golledge J. A systematic review of studies examining inflammation associated cytokines in human abdominal aortic aneurysm samples . Dis Markers 2009 ; 26 : 181 – 188 . Google Scholar Crossref Search ADS PubMed WorldCat 116 Golledge J , Tsao PS, Dalman RL, Norman PE. Circulating markers of abdominal aortic aneurysm presence and progression . Circulation 2008 ; 118 : 2382 – 2392 . Google Scholar Crossref Search ADS PubMed WorldCat 117 Hellenthal FA , Buurman WA, Wodzig WK, Schurink GW. Biomarkers of AAA progression. Part 1: extracellular matrix degeneration . Nat Rev Cardiol 2009 ; 6 : 464 – 474 . Google Scholar Crossref Search ADS PubMed WorldCat 118 Hellenthal FA , Buurman WA, Wodzig WK, Schurink GW. Biomarkers of abdominal aortic aneurysm progression. Part 2: inflammation . Nat Rev Cardiol 2009 ; 6 : 543 – 552 . Google Scholar Crossref Search ADS PubMed WorldCat 119 Brophy CM , Reilly JM, Smith GJ, Tilson MD. The role of inflammation in nonspecific abdominal aortic aneurysm disease . Ann Vasc Surg 1991 ; 5 : 229 – 233 . Google Scholar Crossref Search ADS PubMed WorldCat 120 Levine DM , Sloan BJ, Donner JE, Lorenz JD, Heinzerling RH. Automated measurement of lipoprotein(a) by immunoturbidimetric analysis . Int J Clin Lab Res 1992 ; 22 : 173 – 178 . Google Scholar Crossref Search ADS PubMed WorldCat 121 Koch AE , Kunkel SL, Pearce WH, Shah MR, Parikh D, Evanoff HL et al. Enhanced production of the chemotactic cytokines interleukin-8 and monocyte chemoattractant protein-1 in human abdominal aortic aneurysms . Am J Pathol 1993 ; 142 : 1423 – 1431 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 122 Murphy G , Cockett MI, Ward RV, Docherty AJ. Matrix metalloproteinase degradation of elastin, type IV collagen and proteoglycan. A quantitative comparison of the activities of 95 kDa and 72 kDa gelatinases, stromelysins-1 and -2 and punctuated metalloproteinase (PUMP) . Biochem J 1991 ; 277 : 277 – 279 . Google Scholar Crossref Search ADS PubMed WorldCat 123 Kadoglou NP , Gerasimidis T, Golemati S, Kapelouzou A, Karayannacos PE, Liapis CD. The relationship between serum levels of vascular calcification inhibitors and carotid plaque vulnerability . J Vasc Surg 2008 ; 47 : 55 – 62 . Google Scholar Crossref Search ADS PubMed WorldCat 124 Schroder K , Hertzog PJ, Ravasi T, Hume DA. Interferon-gamma: an overview of signals, mechanisms and functions . J Leukoc Biol 2004 ; 75 : 163 – 189 . Google Scholar Crossref Search ADS PubMed WorldCat 125 Poston RN , Haskard DO, Coucher JR, Gall NP, Johnson-Tidey RR. Expression of intercellular adhesion molecule-1 in atherosclerotic plaques . Am J Pathol 1992 ; 140 : 665 – 673 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 126 Tang WH , Wu Y, Hartiala J, Fan Y, Stewart AF, Roberts R et al. Clinical and genetic association of serum ceruloplasmin with cardiovascular risk . Arterioscler Thromb Vasc Biol 2012 ; 32 : 516 – 522 . Google Scholar Crossref Search ADS PubMed WorldCat 127 Takagi H , Manabe H, Kawai N, Goto SN, Umemoto T. Serum high-density and low-density lipoprotein cholesterol is associated with abdominal aortic aneurysm presence: a systematic review and meta-analysis . Int Angiol 2010 ; 29 : 371 – 375 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 128 Takagi H , Manabe H, Kawai N, Goto SN, Umemoto T. Circulating lipoprotein(a) concentrations and abdominal aortic aneurysm presence . Interact Cardiovasc Thorac Surg 2009 ; 9 : 467 – 470 . Google Scholar Crossref Search ADS PubMed WorldCat 129 Teppala S , Shankar A. Association between serum IGF-1 and diabetes among U.S. adults . Diabetes Care 2010 ; 33 : 2257 – 2259 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Sanders, D L; Nienhuijs, S; Ziprin, P; Miserez, M; Gingell-Littlejohn, M; Smeds, S

doi: 10.1002/bjs.9598pmid: 25146918

Abstract Background Postoperative pain is an important adverse event following inguinal hernia repair. The aim of this trial was to compare postoperative pain within the first 3 months and 1 year after surgery in patients undergoing open mesh inguinal hernia repair using either a self-gripping lightweight polyester mesh or a polypropylene lightweight mesh fixed with sutures. Methods Adult men undergoing Lichtenstein repair for primary inguinal hernia were randomized to ProGrip™ self-gripping mesh or standard sutured lightweight polypropylene mesh. Results In total 557 men were included in the final analysis (self-gripping mesh 270, sutured mesh 287). Early postoperative pain scores were lower with self-gripping mesh than with sutured lightweight mesh: mean visual analogue pain score relative to baseline +1·3 and +8·6 respectively at discharge (P = 0·033), and mean surgical pain scale score relative to baseline +4·2 and +9·7 respectively on day 7 (P = 0·027). There was no significant difference in mid-term (1 month) and long-term (3 months and 1 year) pain scores between the groups. Surgery was significantly quicker with self-gripping mesh (mean difference 7·6 min; P < 0·001). There were no significant differences in reported mesh handling, analgesic consumption, other wound complications, patient satisfaction or hernia recurrence between the groups. Conclusion Self-gripping mesh for open inguinal hernia repair was well tolerated and reduced early postoperative pain (within the first week), without increasing the risk of early recurrence. It did not reduce chronic pain. Registration number: NCT00827944 (http://www.clinicaltrials.gov). Introduction Advances in surgical practice and mesh technology have reduced recurrence rates after inguinal hernia repair to below 5 per cent1–3. As recurrence rates have declined, more attention has been placed on other important outcome measures, particularly quality of life, chronic pain and surgical-site infection (SSI). The true incidence of chronic pain (defined as postoperative pain persisting for more than 3 months)4 after inguinal hernia repair is unknown, and until fairly recently was only a secondary consideration. Grant and colleagues5 noted that 9·7 per cent of patients reported chronic pain 5 years after open inguinal hernia repair. Other studies3,6–8 have put the proportion with chronic pain as high as 34 per cent. The cause of postoperative pain is also uncertain, but is likely to relate to preoperative variables such as patient age3,7,9, hernia size3 and preoperative symptoms3,10–11. Certain aspects of the repair technique, which are under the surgeon's control, may also affect postoperative pain12; these include the handling of cutaneous nerves12–18, the type of mesh19–22 and the method of fixation23–26. The role of nerve identification and/or elective neurectomy is unclear; however, recent systematic reviews have suggested that better results are obtained if the nerves are identified and either protected from injury or divided systematically if there is a risk of incidental damage or entrapment14–18. Fixation of synthetic mesh in inguinal hernia repair requires a balance between strength of fixation and minimization of the risk of inadvertent damage to cutaneous nerves. Since the introduction of the synthetic mesh in the 1950s, non-absorbable sutures have been the most commonly used method for mesh fixation in open hernia surgery. Lower rates of chronic pain have been reported when the mesh is fixed with glue, fibrin sealant or absorbable sutures compared with permanent sutures23–26. An alternative is to secure the mesh to the posterior inguinal wall using a self-gripping mesh, which adheres to the underlying tissues like Velcro® (Velcro, Middlewich, UK). These lightweight meshes incorporate resorbable polylactic acid (PLA) microgrips, which provide self-gripping fixation during the first few months after implantation. After reabsorption of the PLA microgrips, the remaining lightweight mesh (38 g/m2) is secured by tissue ingrowth27. Small randomized trials and case series of self-gripping mesh have reported promising results in terms of recurrence and chronic pain28–32. However, there is no appropriately powered randomized trial comparing suture fixation and self-gripping mesh. This multicentre, patient-blinded, randomized clinical trial was undertaken to compare postoperative pain within the first 3 months and 1 year after surgery in patients undergoing open inguinal hernia mesh repair using either a self-gripping lightweight polyester mesh (Parietex™ ProGrip™; Sofadrim Production, Trevoux, France) or a sutured polypropylene lightweight mesh (less than 40 g/m2) (Parietene™ Light; Sofadrim Production). Methods The study protocol was posted on the http://www.clinicaltrials.gov protocol registration system (identifier NCT00827944). The trial was performed within the framework of the standard EN ISO 14155, in accordance with the Declaration of Helsinki and the recommendations of Good Clinical Practice. Each participating centre obtained approval from the local or regional medical ethics committees. Patients were provided with information about the research project and were enrolled only after appropriate signed informed consent. Prosthetic meshes The test material was Parietex™ ProGrip™ mesh. This is an isoelastic large-pore knitted fabric of monofilament polyester that incorporates resorbable PLA microgrips which provide self-gripping properties on the entire surface of the mesh. It has a density of 74 g/m2 at implantation and 38 g/m2 after absorption. The control mesh was Parietene™ Light, a lightweight (38 g/m2) polypropylene mesh that does not have microgrips and requires fixation with sutures. Study centres and patients Between October 2008 and April 2011, nine European hospitals recruited consecutive men, between the ages of 30 and 75 years, with a primary, uncomplicated inguinal hernia, and with a defect diameter of less than 4 cm. Patients with bilateral inguinal hernias, recurrent hernia and those involved in other trials were excluded. The nine recruiting hospitals were Derriford Hospital (Plymouth, UK), Western Infirmary (Glasgow, UK), Medicinskt Centrum i Linköping (Linköping, Sweden), Catharina Ziekenhuis (Eindhoven, The Netherlands), Klinikum Bremen-Mitte (Bremen, Germany), Chirurgische Klinik (Fulda, Germany), St Mary's Hospital (London, UK), Universitaire Ziekenhuizen Leuven (Leuven, Belgium) and Central Hospital (Västerås, Sweden). Randomization In the operating theatre, after the skin incision, patients were randomized using sealed envelopes containing a computer-generated randomization to either Parietex™ ProGrip™ mesh (P group) or suture fixation of Parietene™ Light mesh (L group). Patients were blinded to the treatment allocation. Follow-up The follow-up schedule was as follows: at 7 days (questionnaire and telephone call); 1 month (physical examination by surgeon); 3 months (questionnaire and telephone call); and a final evaluation 1 year after surgery (physical examination by surgeon). Outcome measures and sample size The primary outcome measure was postoperative pain within the first 3 months and at 1 year after surgery. This was assessed using a 0–150-mm visual analogue scale (VAS) for current pain, and a surgical pain scale (SPS) for pain in the preceding 24 h. Results were reported as the change from baseline (preoperative) pain. Subgroup analysis of postoperative pain (measured using VAS and SPS) in working versus retired patients was also performed. Subgroup analyses were also carried out to assess the effects of nerve handling and resection on postoperative pain in each group. Secondary outcomes were: duration of surgery; ease of mesh handling; wound complications (surgical-site infection (SSI) as defined by the internationally accredited Centers for Disease Control and Prevention33, haematomas – any evidence of a bruise in the postoperative site, seromas – any clinically apparent fluid collection at the site of mesh placement); recurrence; analgesic consumption; patient satisfaction; scores on an activity assessment scale (AAS) to assess the social consequences of pain; and any other adverse events. The effects of resecting or preserving the ilioinguinal and iliohypogastric nerves on postoperative pain were analysed. Analysis of preservation versus resection of the genital branch of the genitofemoral nerve was not done because of the low resection numbers (only 4·5 per cent in total). Surgical technique A qualified surgeon experienced in the technique performed the procedures, with nerve handling as proposed by Wijsmuller and colleagues14. The type of anaesthesia (local or general) was determined by each investigator to meet individual patient needs, in accordance with protocol. A standard surgical technique was used for both mesh types (Appendix S1, supporting information). Statistical analysis The sample size was based on mean pain (VAS score) at 3 months as described in the literature20,29,34. Based on the assumption of a mean pain score of 7 on a scale of 0–100 mm in the control group (lightweight sutured mesh) and 2 in the treatment group (self-gripping mesh) with a two-sided statistical test, α error of 5 per cent and power of 80 per cent, and allowing 15 per cent loss to follow-up, it was calculated that 300 patients were needed in each group. All statistical tests were performed on per-protocol populations. Pain score results are expressed as a change from the baseline score, as previously validated32, and presented with 95 per cent confidence intervals (c.i.); other quantitative data are shown as mean(s.d.). Quantitative variables were compared using ANOVA, t test or Mann–Whitney U test as appropriate. Categorical variables were compared using the χ2 test or Fisher's exact test. Univariable and multivariable logistic regression analyses were performed to identify any preoperative or surgical variables that influenced the VAS scores. All tests were two-sided and P < 0·050 was considered statistically significant. Statistical analysis was done using Minitab® version 15.0 (Minitab, Coventry, UK) and SAS/STAT® version 9.2 for Windows® (SAS Institute, Cary, North Carolina USA) (multivariable analysis for pain was done using proc glm ANCOVA). Results Between October 2008 and April 2011, 603 consecutive men with a unilateral primary inguinal hernia were assessed in the participating hospitals (Fig. 1). Thirty-nine patients were excluded from the analysis because of major protocol violations. Some 275 patients underwent open anterior repair with ProGrip™ self-gripping mesh (P group) and 289 patients had sutured repair with Parietene™ Light mesh (L group). A further five patients in the P group and two in the L group were lost to follow-up, leaving 270 and 287 patients respectively for data analysis. As VAS and SPS scores were not completed at the 3-month follow-up for 21 patients in the P group and 29 in the L group, these patients could not be included in the primary outcome analysis. Demographic and clinical data (age, body mass index, occupation, smoking, co-morbidity and American Society of Anesthesiologists fitness grade) were well matched between the two groups (Table 1). Fig. 1 Open in new tabDownload slide CONSORT diagram for the trial. SDV, site data verification Table 1 Baseline demographic data . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P† . Age (years)* 56·9(12·2) 57·4(10·9) 0·577‡ Body mass index (kg/m2)* 25·4(3·0) 25·5(2·9) 0·537‡ Occupation 0·197 Non-manual work 85 (31·5) 94 (32·8) Manual work 67 (24·8) 70 (24·4) Retired 107 (39·6) 104 (36·2) Unemployed 6 (2·2) 18 (6·3) Unknown 5 (1·9) 1 (0·3) Smoker 46 (17·0) 57 (19·9) 0·391 Co-morbidity COPD 13 (4·8) 21 (7·3) 0·222 Diabetes mellitus 15 (5·6) 17 (5·9) 0·861 Cardiac disease 45 (16·7) 54 (18·8) 0·520 Previous surgery 171 (63·3) 181 (63·1) 0·899 ASA grade n.a. I 134 (49·6) 138 (48·1) II 112 (41·5) 121 (42·2) III 22 (8·1) 26 (9·1) IV 0 (0) 1 (0·3) Unknown 2 (0·7) 1 (0·3) . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P† . Age (years)* 56·9(12·2) 57·4(10·9) 0·577‡ Body mass index (kg/m2)* 25·4(3·0) 25·5(2·9) 0·537‡ Occupation 0·197 Non-manual work 85 (31·5) 94 (32·8) Manual work 67 (24·8) 70 (24·4) Retired 107 (39·6) 104 (36·2) Unemployed 6 (2·2) 18 (6·3) Unknown 5 (1·9) 1 (0·3) Smoker 46 (17·0) 57 (19·9) 0·391 Co-morbidity COPD 13 (4·8) 21 (7·3) 0·222 Diabetes mellitus 15 (5·6) 17 (5·9) 0·861 Cardiac disease 45 (16·7) 54 (18·8) 0·520 Previous surgery 171 (63·3) 181 (63·1) 0·899 ASA grade n.a. I 134 (49·6) 138 (48·1) II 112 (41·5) 121 (42·2) III 22 (8·1) 26 (9·1) IV 0 (0) 1 (0·3) Unknown 2 (0·7) 1 (0·3) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). COPD, chronic obstructive pulmonary disease; ASA, American Society of Anesthesiologists; n.a., not applicable. † χ2 test, except ‡ Student's t test. Open in new tab Table 1 Baseline demographic data . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P† . Age (years)* 56·9(12·2) 57·4(10·9) 0·577‡ Body mass index (kg/m2)* 25·4(3·0) 25·5(2·9) 0·537‡ Occupation 0·197 Non-manual work 85 (31·5) 94 (32·8) Manual work 67 (24·8) 70 (24·4) Retired 107 (39·6) 104 (36·2) Unemployed 6 (2·2) 18 (6·3) Unknown 5 (1·9) 1 (0·3) Smoker 46 (17·0) 57 (19·9) 0·391 Co-morbidity COPD 13 (4·8) 21 (7·3) 0·222 Diabetes mellitus 15 (5·6) 17 (5·9) 0·861 Cardiac disease 45 (16·7) 54 (18·8) 0·520 Previous surgery 171 (63·3) 181 (63·1) 0·899 ASA grade n.a. I 134 (49·6) 138 (48·1) II 112 (41·5) 121 (42·2) III 22 (8·1) 26 (9·1) IV 0 (0) 1 (0·3) Unknown 2 (0·7) 1 (0·3) . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P† . Age (years)* 56·9(12·2) 57·4(10·9) 0·577‡ Body mass index (kg/m2)* 25·4(3·0) 25·5(2·9) 0·537‡ Occupation 0·197 Non-manual work 85 (31·5) 94 (32·8) Manual work 67 (24·8) 70 (24·4) Retired 107 (39·6) 104 (36·2) Unemployed 6 (2·2) 18 (6·3) Unknown 5 (1·9) 1 (0·3) Smoker 46 (17·0) 57 (19·9) 0·391 Co-morbidity COPD 13 (4·8) 21 (7·3) 0·222 Diabetes mellitus 15 (5·6) 17 (5·9) 0·861 Cardiac disease 45 (16·7) 54 (18·8) 0·520 Previous surgery 171 (63·3) 181 (63·1) 0·899 ASA grade n.a. I 134 (49·6) 138 (48·1) II 112 (41·5) 121 (42·2) III 22 (8·1) 26 (9·1) IV 0 (0) 1 (0·3) Unknown 2 (0·7) 1 (0·3) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). COPD, chronic obstructive pulmonary disease; ASA, American Society of Anesthesiologists; n.a., not applicable. † χ2 test, except ‡ Student's t test. Open in new tab Operative data There were no significant differences in operative variables between the groups (Table 2). Hernia side, type35 and diameter of the defect were similar both groups. There were no significant differences in nerve identification or resection rates between the two groups (Table 3). Table 2 Operative data . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P‡ . Hernia side 0·472 Right 163 (60·4) 166 (57·8) Left 104 (38·5) 120 (41·8) Unknown 3 (1·1) 1 (0·3) Hernia classification† 0·927 PL (indirect) 170 (63·0) 178 (62·0) PM (direct) 86 (31·9) 96 (33·4) Combined (direct + indirect) 13 (4·8) 13 (4·5) Unknown 1 (0·4) 0 (0) Defect diameter (cm)* 2·43(0·78) 2·40(0·85) 0·709§ Anaesthetic type 0·750 General anaesthetic 177 (65·6) 185 (64·5) Local anaesthetic 46 (17·0) 56 (19·5) Locoregional 46 (17·0) 46 (16·0) Unknown 1 (0·4) 0 (0) . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P‡ . Hernia side 0·472 Right 163 (60·4) 166 (57·8) Left 104 (38·5) 120 (41·8) Unknown 3 (1·1) 1 (0·3) Hernia classification† 0·927 PL (indirect) 170 (63·0) 178 (62·0) PM (direct) 86 (31·9) 96 (33·4) Combined (direct + indirect) 13 (4·8) 13 (4·5) Unknown 1 (0·4) 0 (0) Defect diameter (cm)* 2·43(0·78) 2·40(0·85) 0·709§ Anaesthetic type 0·750 General anaesthetic 177 (65·6) 185 (64·5) Local anaesthetic 46 (17·0) 56 (19·5) Locoregional 46 (17·0) 46 (16·0) Unknown 1 (0·4) 0 (0) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). † European Hernia Society classification. P, primary hernia; M, medial; L, lateral. ‡ χ2 test, except § Student's t test. Open in new tab Table 2 Operative data . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P‡ . Hernia side 0·472 Right 163 (60·4) 166 (57·8) Left 104 (38·5) 120 (41·8) Unknown 3 (1·1) 1 (0·3) Hernia classification† 0·927 PL (indirect) 170 (63·0) 178 (62·0) PM (direct) 86 (31·9) 96 (33·4) Combined (direct + indirect) 13 (4·8) 13 (4·5) Unknown 1 (0·4) 0 (0) Defect diameter (cm)* 2·43(0·78) 2·40(0·85) 0·709§ Anaesthetic type 0·750 General anaesthetic 177 (65·6) 185 (64·5) Local anaesthetic 46 (17·0) 56 (19·5) Locoregional 46 (17·0) 46 (16·0) Unknown 1 (0·4) 0 (0) . Self-gripping mesh (n = 270) . Lightweight sutured mesh (n = 287) . P‡ . Hernia side 0·472 Right 163 (60·4) 166 (57·8) Left 104 (38·5) 120 (41·8) Unknown 3 (1·1) 1 (0·3) Hernia classification† 0·927 PL (indirect) 170 (63·0) 178 (62·0) PM (direct) 86 (31·9) 96 (33·4) Combined (direct + indirect) 13 (4·8) 13 (4·5) Unknown 1 (0·4) 0 (0) Defect diameter (cm)* 2·43(0·78) 2·40(0·85) 0·709§ Anaesthetic type 0·750 General anaesthetic 177 (65·6) 185 (64·5) Local anaesthetic 46 (17·0) 56 (19·5) Locoregional 46 (17·0) 46 (16·0) Unknown 1 (0·4) 0 (0) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). † European Hernia Society classification. P, primary hernia; M, medial; L, lateral. ‡ χ2 test, except § Student's t test. Open in new tab Table 3 Nerve identification and resection . All . Self-gripping mesh . Lightweight sutured mesh . P* . Ilioinguinal nerve Identified 492 of 553 (89·0) 243 of 269 (90·3) 249 of 284 (87·7) 0·319 Resected 142 of 551 (25·8) 70 of 268 (26·1) 72 of 283 (25·4) 0·856 Iliohypogastric nerve Identified 369 of 552 (66·8) 182 of 269 (67·7) 187 of 283 (66·1) 0·693 Resected 124 of 549 (22·6) 65 of 268 (24·3) 59 of 281 (21·0) 0·362 Genital branch of genitofemoral nerve Identified 264 of 553 (47·7) 126 of 269 (46·8) 138 of 284 (48·6) 0·680 Resected 25 of 550 (4·5) 9 of 269 (3·3) 16 of 281 (5·7) 0·186 . All . Self-gripping mesh . Lightweight sutured mesh . P* . Ilioinguinal nerve Identified 492 of 553 (89·0) 243 of 269 (90·3) 249 of 284 (87·7) 0·319 Resected 142 of 551 (25·8) 70 of 268 (26·1) 72 of 283 (25·4) 0·856 Iliohypogastric nerve Identified 369 of 552 (66·8) 182 of 269 (67·7) 187 of 283 (66·1) 0·693 Resected 124 of 549 (22·6) 65 of 268 (24·3) 59 of 281 (21·0) 0·362 Genital branch of genitofemoral nerve Identified 264 of 553 (47·7) 126 of 269 (46·8) 138 of 284 (48·6) 0·680 Resected 25 of 550 (4·5) 9 of 269 (3·3) 16 of 281 (5·7) 0·186 Values in parentheses are percentage of patients for whom data were available in the per-protocol population. * χ2 test. Open in new tab Table 3 Nerve identification and resection . All . Self-gripping mesh . Lightweight sutured mesh . P* . Ilioinguinal nerve Identified 492 of 553 (89·0) 243 of 269 (90·3) 249 of 284 (87·7) 0·319 Resected 142 of 551 (25·8) 70 of 268 (26·1) 72 of 283 (25·4) 0·856 Iliohypogastric nerve Identified 369 of 552 (66·8) 182 of 269 (67·7) 187 of 283 (66·1) 0·693 Resected 124 of 549 (22·6) 65 of 268 (24·3) 59 of 281 (21·0) 0·362 Genital branch of genitofemoral nerve Identified 264 of 553 (47·7) 126 of 269 (46·8) 138 of 284 (48·6) 0·680 Resected 25 of 550 (4·5) 9 of 269 (3·3) 16 of 281 (5·7) 0·186 . All . Self-gripping mesh . Lightweight sutured mesh . P* . Ilioinguinal nerve Identified 492 of 553 (89·0) 243 of 269 (90·3) 249 of 284 (87·7) 0·319 Resected 142 of 551 (25·8) 70 of 268 (26·1) 72 of 283 (25·4) 0·856 Iliohypogastric nerve Identified 369 of 552 (66·8) 182 of 269 (67·7) 187 of 283 (66·1) 0·693 Resected 124 of 549 (22·6) 65 of 268 (24·3) 59 of 281 (21·0) 0·362 Genital branch of genitofemoral nerve Identified 264 of 553 (47·7) 126 of 269 (46·8) 138 of 284 (48·6) 0·680 Resected 25 of 550 (4·5) 9 of 269 (3·3) 16 of 281 (5·7) 0·186 Values in parentheses are percentage of patients for whom data were available in the per-protocol population. * χ2 test. Open in new tab Baseline scores The mean VAS scores before surgery were 25·7 (95 per cent c.i. 21·8 to 29·6) in the P group and 23·9 (20·4 to 27·4) in the L group (P = 0·497). There was no significant difference in SPS scores, analgesic consumption or AAS scores between the groups before operation. Primary outcome Mean VAS pain scores at 1 year were 5·1 (3·4 to 6·8) in patients treated with self-gripping mesh and 4·7 (3·3 to 6·1) in those who had sutured mesh repair. The mean change from the preoperative baseline value was −20·9 (−25·0 to −16·8) and −19·7 (−23·5 to −16·0) respectively (Fig. 2). There was no significant difference between the groups in VAS scores, SPS scores (Fig. 3), AAS scores (Fig. S1, supporting information) or analgesic consumption at either 3 months or 1 year. Fig. 2 Open in new tabDownload slide Mean (95 per cent confidence interval) change from baseline in pain scores measured on a visual analogue scale in self-gripping mesh and lightweight sutured mesh groups. *P < 0·050 versus lightweight sutured mesh (ANOVA) Fig. 3 Open in new tabDownload slide Mean (95 per cent confidence interval) change from baseline in surgical pain scale scores measured at rest in self-gripping mesh and lightweight sutured mesh groups. *P < 0·050 versus lightweight sutured mesh (ANOVA) Pain within the first 3 months The mean VAS pain score at discharge was 26·7 (23·4 to 30·0) and 31·5 (27·7 to 35·4) in the P and L groups respectively. Patients who received self-gripping mesh had consistently lower VAS scores at all follow-up points, but the difference was not statistically significant. When the change in pain scores from the preoperative baseline was analysed, patients who had self-gripping mesh experienced significantly less pain at discharge: mean VAS relative to baseline +1·3 (−3·3 to +5·9) versus +8·6 (+3·7 to +13·5) for those with sutured mesh (P = 0·033). The difference was still present 1 week after surgery but did not reach statistical significance: mean VAS relative to baseline +0·2 (−4·2 to +4·6) in the P group and +6·6 (+1·9 to +11·4) in the L group (P = 0·051). There was no significant difference between the groups in mean difference from baseline VAS scores at 1 and 3 months (Fig. 2). There were significant differences in SPS scores at rest at 1 week, with significantly less pain in patients treated with self-gripping mesh: mean SPS score relative to baseline +4·2 (95 per cent c.i. +4·1 to +11·2) versus +9·7 (+4·9 to +23·7) (P = 0·027) (Fig. 3). However, there was no significant difference in SPS scores at 1 month. Similarly, there was no significant difference in AAS scores at 1 month: mean 91·9 (95 per cent c.i. 90·5 to 93·4) in the P group and 93·1 (91·7 to 94·4) in the L group (P = 0·258). Analgesic consumption was similar. There was no difference in mean VAS scores between the retired and working population. Impact of nerve resection Resection of the ilioinguinal nerve did not have significant effect on postoperative pain (Fig. 4a). In contrast, resection of the iliohypogastric nerve significantly reduced postoperative pain at all follow-up times (P ≤ 0·001) (Fig. 4b). Subgroup analysis of the fixation method showed that, when the iliohypogastric nerve was preserved, postoperative pain was significantly lower in the P group than the L group at all follow-up points from discharge to 1 year (Fig. 5). However, there was no significant difference between the groups when the iliohypogastric nerve was resected. Multivariable analysis revealed that resection of the iliohypogastric nerve was the only variable that consistently (across all follow-up points) had a significant effect in reducing VAS pain scores. Fig. 4 Open in new tabDownload slide Effect of resection versus preservation of a ilioinguinal and b iliohypogastric nerves on mean(s.d.) pain scores measured on a visual analogue scale. *P < 0·050 versus nerve preserved (ANOVA) Fig. 5 Open in new tabDownload slide Subgroup analysis of pain scores measured on a visual analogue scale among patients in whom the iliohypogastric nerve was preserved. Values are mean (95 per cent confidence interval) change from baseline. *P < 0·050 versus lightweight sutured mesh (ANOVA) Secondary outcomes Duration of surgery, ease of dissection and mesh handling The mean(s.d.) duration of surgery was 35·4(15·2) min for hernia repair with self-gripping mesh and 43·0(14·9) for sutured mesh repair (mean difference 7·6 min; P < 0·001). Surgeons reported no difference in the difficulty of dissection (P = 0·851) or handling of the mesh (P = 0·126) between the groups. Wound complications There were fewer SSIs in the P group than in the L group (7 of 270 versus 14 of 287 respectively), but the difference was not significant (P = 0·255). The majority were superficial SSIs (6 of 7 and 13 of 14 respectively), all of which were treated successfully with antibiotics. In both groups there was one deep incisional SSI, which was treated successfully by opening the skin wound and giving antibiotic therapy. No mesh had to be removed owing to infection. There was no significant difference in seroma or haematoma rates between the groups (Table 4). Table 4 Secondary outcome measures . Self-gripping mesh . Lightweight sutured mesh . P† . Duration of surgery (min)* 35·4(15·2) 43·0(14·9) < 0·001‡ Difficulty of dissection 0·851 Normal 232 (86·6) 250 (87·1) More difficult than normal 36 (13·4) 37 (12·9) Ease of mesh handling 0·126 Bad 9 (3·3) 11 (3·8) Good 107 (39·8) 137 (47·9) Very good 153 (56·9) 138 (48·3) Wound complications Seroma 5 (1·9) 7 (2·4) 0·773§ Haematoma 12 (4·4) 7 (2·4) 0·244§ Surgical-site infection 7 (2·6) 14 (4·9) 0·255§ Superficial 6 13 0·182§ Deep incisional 1 1 Hospital stay (days)* 0·38(0·78) 0·44(0·87) 0·452‡ Patient satisfaction at 3 months 0·707 Very satisfied 152 (60·1) 166 (62·4) Satisfied 76 (30·0) 71 (26·7) Neutral 19 (7·5) 20 (7·5) Dissatisfied 6 (2·4) 9 (3·4) . Self-gripping mesh . Lightweight sutured mesh . P† . Duration of surgery (min)* 35·4(15·2) 43·0(14·9) < 0·001‡ Difficulty of dissection 0·851 Normal 232 (86·6) 250 (87·1) More difficult than normal 36 (13·4) 37 (12·9) Ease of mesh handling 0·126 Bad 9 (3·3) 11 (3·8) Good 107 (39·8) 137 (47·9) Very good 153 (56·9) 138 (48·3) Wound complications Seroma 5 (1·9) 7 (2·4) 0·773§ Haematoma 12 (4·4) 7 (2·4) 0·244§ Surgical-site infection 7 (2·6) 14 (4·9) 0·255§ Superficial 6 13 0·182§ Deep incisional 1 1 Hospital stay (days)* 0·38(0·78) 0·44(0·87) 0·452‡ Patient satisfaction at 3 months 0·707 Very satisfied 152 (60·1) 166 (62·4) Satisfied 76 (30·0) 71 (26·7) Neutral 19 (7·5) 20 (7·5) Dissatisfied 6 (2·4) 9 (3·4) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). SSI, surgical-site infection. † χ2 test, except ‡ Student's t test and § Fisher's exact test. Open in new tab Table 4 Secondary outcome measures . Self-gripping mesh . Lightweight sutured mesh . P† . Duration of surgery (min)* 35·4(15·2) 43·0(14·9) < 0·001‡ Difficulty of dissection 0·851 Normal 232 (86·6) 250 (87·1) More difficult than normal 36 (13·4) 37 (12·9) Ease of mesh handling 0·126 Bad 9 (3·3) 11 (3·8) Good 107 (39·8) 137 (47·9) Very good 153 (56·9) 138 (48·3) Wound complications Seroma 5 (1·9) 7 (2·4) 0·773§ Haematoma 12 (4·4) 7 (2·4) 0·244§ Surgical-site infection 7 (2·6) 14 (4·9) 0·255§ Superficial 6 13 0·182§ Deep incisional 1 1 Hospital stay (days)* 0·38(0·78) 0·44(0·87) 0·452‡ Patient satisfaction at 3 months 0·707 Very satisfied 152 (60·1) 166 (62·4) Satisfied 76 (30·0) 71 (26·7) Neutral 19 (7·5) 20 (7·5) Dissatisfied 6 (2·4) 9 (3·4) . Self-gripping mesh . Lightweight sutured mesh . P† . Duration of surgery (min)* 35·4(15·2) 43·0(14·9) < 0·001‡ Difficulty of dissection 0·851 Normal 232 (86·6) 250 (87·1) More difficult than normal 36 (13·4) 37 (12·9) Ease of mesh handling 0·126 Bad 9 (3·3) 11 (3·8) Good 107 (39·8) 137 (47·9) Very good 153 (56·9) 138 (48·3) Wound complications Seroma 5 (1·9) 7 (2·4) 0·773§ Haematoma 12 (4·4) 7 (2·4) 0·244§ Surgical-site infection 7 (2·6) 14 (4·9) 0·255§ Superficial 6 13 0·182§ Deep incisional 1 1 Hospital stay (days)* 0·38(0·78) 0·44(0·87) 0·452‡ Patient satisfaction at 3 months 0·707 Very satisfied 152 (60·1) 166 (62·4) Satisfied 76 (30·0) 71 (26·7) Neutral 19 (7·5) 20 (7·5) Dissatisfied 6 (2·4) 9 (3·4) Values in parentheses are percentages unless indicated otherwise; * values are mean(s.d.). SSI, surgical-site infection. † χ2 test, except ‡ Student's t test and § Fisher's exact test. Open in new tab Length of hospital stay Most operations were done as a day case (over 75 per cent in both groups). There was no significant difference in the mean(s.d.) length of stay, which was 0·38(0·78) h in the P group and 0·44(0·87) h in the L group (P = 0·452). Recurrence Hernia recurrence rates at 1 year were similar: 1·5 per cent (4 patients) after repair with self-gripping mesh and 2·8 per cent (8 patients) in the sutured repair group (P = 0·289). Other adverse events There were few adverse events other than pain or wound complications (detailed above). Five serious adverse events were reported, two in the L group (ischaemic testicle; oedema, poor kidney function and pneumonia) and three in the P group (pulmonary sarcoidosis; prostate adenocarcinoma; chronic renal insufficiency). Patient satisfaction Overall, both groups of patients were satisfied with their surgical experience. At 1 year, 90·1 per cent of patients who had self-gripping mesh and 89·1 per cent with sutured mesh were ‘very satisfied’ or ‘satisfied’ (P = 0·707). Discussion This trial is one of the largest randomized clinical trials in open inguinal hernia surgery. It demonstrated improvements in early postoperative pain after repair with ProGrip™ self-gripping mesh compared with suture fixation of lightweight mesh. However, there was no significant difference in mid-term (1 month) and long-term (3 months and 1 year) pain scores between the groups. The procedure was quicker with self-gripping mesh, and there were fewer SSIs, albeit the difference was not statistically significant. There were no differences in mesh handling, other wound complications, analgesic consumption or patient satisfaction between the groups. The study was based on the hypothesis that suture fixation may potentiate postoperative pain, by compression or inadvertent damage to the inguinal nerves. The findings here of reduced early postoperative pain support the results of smaller trials and observational studies assessing self-gripping meshes36,37. A recently published randomized clinical trial38 comparing self-gripping mesh with suture fixation in 398 patients reported no significant difference in early VAS pain scores between the procedures. Identification of the inguinal nerves and their handling has been a topic of vigorous debate in hernia surgery. In 2008 an International Consensus Conference recommended identification of all three inguinal nerves, and documentation and resection of any nerves at risk39. The recommendations were based largely on two long-term studies40,41 which concluded that identification and preservation of all three inguinal nerves reduced chronic incapacitating groin pain to less than 1 per cent. Furthermore, the working group estimated that all three nerves could be identified in 70–90 per cent of inguinal hernia repairs. In the present study nerve identification rates were well below this target (ilioinguinal nerve 89·0 per cent, iliohypogastric nerve 66·8 per cent, genital branch of the genitofemoral nerve 47·7 per cent), despite the participating surgeons all being hernia specialists. The present results are in keeping with a recent publication42 documenting nerve identification rates in 244 primary inguinal hernias. Nerve identification may not be as easy as previously thought. Furthermore, the reported rates are probably an overestimation, as there may be an element of incorrect identification during surgery. Iliohypogastric nerve resection significantly reduced postoperative pain compared with nerve preservation at all follow-up points in the present study. Yet resection or preservation of the ilioinguinal nerve had no significant effect on postoperative pain. Multivariable analysis revealed that resection of the iliohypogastric nerve was the only variable that consistently reduced the VAS pain score. Although resection of the iliohypogastric nerve reduced postoperative pain, this method of nerve management is far from ideal because it may result in local paraesthesia. Subgroup analysis of the fixation method suggested that, when the iliohypogastric nerve was preserved, postoperative pain was significantly lower with self-gripping mesh up to 1 year. Although subgroup analyses must always be interpreted with caution, these results suggest that damage to the iliohypogastric nerve may have an important role in postoperative pain. The reduction in pain with self-gripping mesh suggests that entrapment of this nerve by suture fixation could be the cause of chronic pain. In support of this theory, lower rates of postoperative pain have been reported with fibrin sealants23–26 and surgical glues43 compared with sutures or staples. These conclusions may understandably receive some scrutiny and it is acknowledged that identification of the nerves must be accompanied by proper technique and nerve handling, which in theory should minimize the risk of inadvertent injury to the nerve or its protective fascia. A further limitation is that, despite the significant statistical differences in VAS pain scores between resection versus preservation of the iliohypogastric nerve, the clinical significance of a small difference in VAS score is less clear. The role of the iliohypogastric nerve in postoperative pain requires clarification. Overall 1-year recurrence rates were very low (1·5 per cent in P group and 2·8 per cent in L group) and compare favourably with those in the wider literature1–3. The low recurrence rate demonstrates the efficacy of the self-gripping mesh, at least in the short term, and addresses concerns regarding the strength of fixation in small and medium-sized hernias. There was a significant reduction in operating time of 7·6 min with self-gripping mesh, in keeping with other randomized trials comparing non-suture fixation with suture fixation36,38,44–46. Although these time savings were statistically significant, the cost efficiency is questionable. Acknowledgements Meshes were provided by Sofradim Production/Covidien. The research programme was funded by an educational grant from Sofradim Production/Covidien (Trevoux, France). Sofradim Production/Covidien were not involved directly with the trial design, analysis or reporting of results. Medical writing and editing support was provided by H. Belalit, C. Martinella, M. Lourd and P. Becker (Sofradim Production/Covidien); and freelance editorial support by K. Bass, funded by Sofradim Production/Covidien. The authors acknowledge E. Kullman, S. Schulenberg, A. Eklund, P. Appel and A. Kingsnorth for their participation in the trial. Disclosure: The authors declare no conflict of interest. Editor's comments References 1 Frey DM , Wildisen A, Hamel CT, Zuber M, Oertli D, Metzger J. Randomized clinical trial of Lichtenstein's operation versus mesh plug for inguinal hernia repair . Br J Surg 2007 ; 94 : 36 – 41 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Vironen J , Nieminen J, Eklund A, Paavolainen P. Randomized clinical trial of Lichtenstein patch or Prolene Hernia System for inguinal hernia repair . Br J Surg 2006 ; 93 : 33 – 39 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Franneby U , Sandblom G, Nordin P, Nyrén O, Gunnarsson U. Risk factors for long-term pain after hernia surgery . Ann Surg 2006 ; 244 : 212 – 219 . Google Scholar Crossref Search ADS PubMed WorldCat 4 International Association for the Study of Pain. Classification of chronic pain . Descriptions of chronic pain syndromes and definitions of pain terms. Prepared by the International Association for the Study of Pain, Subcommittee on Taxonomy . Pain Suppl 1999 ; 3 : S1 – S226 . OpenURL Placeholder Text WorldCat 5 Grant AM , Scott NW, O'Dwyer PJ; MRC Laparoscopic Groin Hernia Trial Group . Five-year follow-up of a randomized trial to assess pain and numbness after laparoscopic or open repair of groin hernia . Br J Surg 2004 ; 91 : 1570 – 1574 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Bay-Nielsen M , Perkins FM, Kehlet H; Danish Hernia Database . Pain and functional impairment 1 year after inguinal herniorrhaphy: a nationwide questionnaire study . Ann Surg 2001 ; 233 : 1 – 7 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Nienhuijs SW , Boelens OB, Strobbe LJ. Pain after anterior mesh hernia repair . J Am Coll Surg 2005 ; 200 : 885 – 889 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Callesen T , Bech K, Kehlet H. Prospective study of chronic pain after groin hernia repair . Br J Surg 1999 ; 86 : 1528 – 1531 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Kalliomaki ML , Meyerson J, Gunnarsson U, Gordh T, Sandblom G. Long-term pain after inguinal hernia repair in a population-based cohort; risk factors and interference with daily activities . Eur J Pain 2008 ; 12 : 214 – 225 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Nienhuijs S , Staal E, Strobbe L, Rosman C, Groenewoud H, Bleichrodt R. Chronic pain after mesh repair of inguinal hernia: a systematic review . Am J Surg 2007 ; 194 : 394 – 400 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Poobalan AS , Bruce J, Smith WC, King PM, Krukowski ZH, Chambers WA. A review of chronic pain after inguinal herniorrhaphy . Clin J Pain 2003 ; 19 : 48 – 54 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Yeo AE , Berney CR. Carolinas Comfort Scale for mesh repair of inguinal hernia . ANZ J Surg 2012 ; 82 : 285 – 286 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Aasvang E , Kehlet H. Chronic postoperative pain: the case of inguinal herniorrhaphy . Br J Anaesth 2005 ; 95 : 69 – 76 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Wijsmuller AR , van Veen RN, Bosch JL, Lange JF, Kleinrensink GJ, Jeekel J et al. Nerve management during open hernia repair . Br J Surg 2007 ; 94 : 17 – 22 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Wijsmuller AR , Lange JF, Kleinrensink GJ, van Geldere D, Simons MP, Huygen FJ et al. Nerve-identifying inguinal hernia repair: a surgical anatomical study . World J Surg 2007 ; 31 : 414 – 420 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Bartlett DC , Porter C, Kingsnorth AN. A pragmatic approach to cutaneous nerve division during open inguinal hernia repair . Hernia 2007 ; 11 : 243 – 246 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Smeds S , Löfström L, Eriksson O. Influence of nerve identification and the resection of nerves ‘at risk’ on postoperative pain in open inguinal hernia repair . Hernia 2010 ; 14 : 265 – 270 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Alfieri S , Rotondi F, Di Giorgio A, Fumagalli U, Salzano A, Di Miceli D et al. Influence of preservation versus division of ilioinguinal, iliohypogastric, and genital nerves during open mesh herniorrhaphy: prospective multicentric study of chronic pain . Ann Surg 2006 ; 243 : 553 – 558 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Post S , Weiss B, Willer M, Neufang T, Lorenz D. Randomized clinical trial of lightweight composite mesh for Lichtenstein inguinal hernia repair . Br J Surg 2004 ; 91 : 44 – 48 . Google Scholar Crossref Search ADS PubMed WorldCat 20 O'Dwyer PJ , Kingsnorth AN, Molloy RG, Small PK, Lammers B, Horeyseck G. Randomized clinical trial assessing impact of a lightweight or heavyweight mesh on chronic pain after inguinal hernia repair . Br J Surg 2005 ; 92 : 166 – 170 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Deysine MX . Hernia Infections: Pathophysiology, Diagnosis, Treatment, Prevention . Marcel Dekker : New York , 2004 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 22 Simons MP , Aufenacker T, Bay-Nielsen M, Bouillot JL, Campanelli G, Conze J et al. European Hernia Society guidelines on the treatment of inguinal hernia in adult patients . Hernia 2009 ; 13 : 343 – 403 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Lau H Fibrin sealant versus mechanical stapling for mesh fixation during endoscopic extraperitoneal inguinal hernioplasty: a randomized prospective trial . Ann Surg 2005 ; 242 : 670 – 675 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Hidalgo M , Castillo MJ, Eymar JL, Hidalgo A. Lichtenstein inguinal hernioplasty: sutures versus glue . Hernia 2005 ; 9 : 242 – 244 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Lovisetto F , Zonta S, Rota E, Mazzilli M, Bardone M, Bottero L et al. Use of human fibrin glue (Tissucol) versus staples for mesh fixation in laparoscopic transabdominal preperitoneal hernioplasty: a prospective, randomized study . Ann Surg 2007 ; 245 : 222 – 231 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Campanelli G , Pascual MH, Hoeferlin A, Rosenberg J, Champault G, Kingsnorth A et al. Randomized, controlled, blinded trial of Tisseel/Tissucol for mesh fixation in patients undergoing Lichtenstein technique for primary inguinal hernia repair: results of the TIMELI trial . Ann Surg 2012 ; 255 : 650 – 657 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Champault G , Polliand C, Dufour F, Ziol M, Behr L. A ‘self adhering’ prosthesis for hernia repair: experimental study . Hernia 2009 ; 13 : 49 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat 28 Champault G , Torcivia A, Paolino L, Chaddad W, Lacaine F, Barrat C. A self-adhering mesh for inguinal hernia repair: preliminary results of a prospective, multicenter study . Hernia 2011 ; 15 : 635 – 641 . Google Scholar Crossref Search ADS PubMed WorldCat 29 Chastan P Tension free open inguinal hernia repair using an innovative self gripping semi-resorbable mesh . J Minim Access Surg 2006 ; 2 : 139 – 143 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Chastan P Tension-free open hernia repair using an innovative self-gripping semi-resorbable mesh . Hernia 2009 ; 13 : 137 – 142 . Google Scholar Crossref Search ADS PubMed WorldCat 31 García Ureña MÁ , Hidalgo M, Feliu X, Velasco MÁ, Revuelta S, Gutiérrez R et al. Multicentric observational study of pain after the use of a self-gripping lightweight mesh . Hernia 2011 ; 15 : 511 – 515 . Google Scholar Crossref Search ADS PubMed WorldCat 32 Kingsnorth A , Gingell-Littlejohn M, Nienhuijs S, Schüle S, Appel P, Ziprin P et al. Randomized controlled multicenter international clinical trial of self-gripping Parietex ProGrip polyester mesh versus lightweight polypropylene mesh in open inguinal hernia repair: interim results at 3 months . Hernia 2012 ; 16 : 287 – 294 . Google Scholar Crossref Search ADS PubMed WorldCat 33 Horan TC , Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections . Infect Control Hosp Epidemiol 1992 ; 13 : 606 – 608 . Google Scholar Crossref Search ADS PubMed WorldCat 34 Nienhuijs S , Staal E, Keemers-Gels M, Rosman C, Strobbe L. Pain after open preperitoneal repair versus Lichtenstein repair: a randomized trial . World J Surg 2007 ; 31 : 1751 – 1757 . Google Scholar Crossref Search ADS PubMed WorldCat 35 Miserez M , Alexandre JH, Campanelli G, Corcione F, Cuccurullo D, Pascual MH et al. The European Hernia Society groin hernia classification: simple and easy to remember . Hernia 2007 ; 11 : 113 – 116 . Google Scholar Crossref Search ADS PubMed WorldCat 36 Kapischke M , Schulze H, Caliebe A. Self-fixating mesh for the Lichtenstein procedure – a prestudy . Langenbecks Arch Surg 2010 ; 395 : 317 – 322 . Google Scholar Crossref Search ADS PubMed WorldCat 37 Gallinella Muzi M , Andreoloi F, Ciangola I, Nigro C. Lichtenstein vs ProGrip repair: preliminary result of a multicentric randomized controlled trial . Hernia 2010 ; 14 ( Suppl 1 ): S13 . Google Scholar OpenURL Placeholder Text WorldCat 38 Pierides G , Scheinin T, Remes V, Hermunen K, Vironen J. Randomized comparison of self-fixating and sutured mesh in open inguinal hernia repair . Br J Surg 2012 ; 99 : 630 – 636 . Google Scholar Crossref Search ADS PubMed WorldCat 39 Chen DC , Amid PK. Questioning the favorable influence of nerve identification during open inguinal herniorrhaphy is misleading and may increase the risk of surgeons' liability: Invited Commentary: ‘Does nerve identification during open inguinal herniorrhaphy reduce the risk of persistent pain?’ . Hernia 2012 ; 16 : 579 – 581 . Google Scholar Crossref Search ADS PubMed WorldCat 40 Izard G , Gailleton R, Randrianasolo S, Houry R. [ Treatment of inguinal hernias by McVay's technique. Apropos of 1332 cases .] Ann Chir 1996 ; 50 : 755 – 766 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 41 Lange JF , Wijsmuller AR, van Geldere D, Simons MP, Swart R, Oomen J et al. Feasibility study of three-nerve-recognizing Lichtenstein procedure for inguinal hernia . Br J Surg 2009 ; 96 : 1210 – 1214 . Google Scholar Crossref Search ADS PubMed WorldCat 42 Bischoff JM , Aasvang EK, Kehlet H, Werner MU. Does nerve identification during open inguinal herniorrhaphy reduce the risk of nerve damage and persistent pain? Hernia 2012 ; 16 : 573 – 577 . Google Scholar Crossref Search ADS PubMed WorldCat 43 Testini M , Lissidini G, Poli E, Gurrado A, Lardo D, Piccinni G. A single-surgeon randomized trial comparing sutures, N-butyl-2-cyanoacrylate and human fibrin glue for mesh fixation during primary inguinal hernia repair . Can J Surg 2010 ; 53 : 155 – 160 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 44 Shen YM , Sun WB, Chen J, Liu SJ, Wang MG. NBCA medical adhesive (n-butyl-2-cyanoacrylate) versus suture for patch fixation in Lichtenstein inguinal herniorrhaphy: a randomized controlled trial . Surgery 2012 ; 151 : 550 – 555 . Google Scholar Crossref Search ADS PubMed WorldCat 45 Kim-Fuchs C , Angst E, Vorburger S, Helbling C, Candinas D, Schlumpf R. Prospective randomized trial comparing sutured with sutureless mesh fixation for Lichtenstein hernia repair: long-term results . Hernia 2012 ; 16 : 21 – 27 . Google Scholar Crossref Search ADS PubMed WorldCat 46 Lionetti R , Neola B, Dilillo S, Bruzzese D, Ferulano GP. Sutureless hernioplasty with light-weight mesh and fibrin glue versus Lichtenstein procedure: a comparison of outcomes focusing on chronic postoperative pain . Hernia 2012 ; 16 : 127 – 131 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Kasivisvanathan, R; Abbassi-Ghadi, N; Kumar, S; Mackenzie, H; Thompson, K; James, K; Mallett, S V

doi: 10.1002/bjs.9592pmid: 25088505

Abstract Background Patients often fail to stop clopidogrel appropriately before non-cardiac surgery. Thromboelastography platelet mapping (TEG-PM) can be used to measure the percentage adenosine 5′-diphosphate platelet receptor inhibition (ADP-PRI) by clopidogrel in these patients. Methods This prospective case–control study investigated the risk of bleeding in patients who had taken clopidogrel within 7 days of scheduled operation. Patients underwent TEG-PM to stratify their bleeding risk. Low-risk (ADP-PRI below 30 per cent) and urgent priority high-risk (ADP-PRI 30 per cent or more) patients proceeded to surgery. The outcomes of these patients were compared with those of matched controls. Regression analysis, with bootstrapping validation, was used to identify independent risk factors for bleeding and an optimal cut-off value of ADP-PRI for cancellation of surgery. Results From May 2008 to October 2013, 182 patients failed to discontinue clopidogrel. No correlation was observed between duration of clopidogrel omission and percentage ADP-PRI; 112 low-risk and 19 high-risk patients proceeded to surgery. High-risk patients had significantly greater intraoperative packed red blood cell (PRBC) transfusion in comparison with their matched controls, and a strong positive correlation between percentage ADP-PRI and units of intraoperative PRBCs transfused (r = 0·749, 95 per cent confidence interval (c.i.) 0·410 to 0·940; P < 0·001). Percentage ADP-PRI was the only independent risk factor for intraoperative PRBC transfusion (odds ratio 1·07, 95 per cent c.i. 1·02 to 1·13; P = 0·005). Conclusion An objective measure of platelet inhibition with TEG-PM, using an ADP-PRI cut-off of 34 per cent, can be used to prevent unnecessary cancellations, while minimizing patient risk. Introduction Clopidogrel is the mainstay of antiplatelet therapy following coronary stenting and it is estimated that nearly 25 000 patients are being treated with this medication in the UK1. Approximately 5 per cent of patients started on clopidogrel following coronary stent insertion will undergo non-cardiac surgery within 1 year2. The risk of perioperative bleeding associated with clopidogrel use3–6 needs to be balanced against the thrombotic risks of therapy cessation7,8. Clopidogrel is usually discontinued for 7–10 days before elective surgery to allow platelet function to recover and reduce the risk of perioperative bleeding. However, clopidogrel is often taken beyond the requested stop date, which may result in cancellation of planned surgery9. Clopidogrel has significant interindividual variability in the extent of platelet inhibition; a significant proportion of patients have little or no platelet effect10,11. Measurement of platelet inhibition could allow stratification of bleeding risk and aid decision-making with regard to the cancellation of surgery. The standard for measuring platelet inhibition is light transmission aggregometry (LTA)12. However, this is laboratory-based, time-consuming and labour-intensive, and not practical for rapid assessment of platelet function. More appropriate are several point-of-care platelet function analysers, which allow immediate evaluation of platelet inhibition, including: VerifyNow® (Accumetrics, San Diego, California, USA)13; the Multiplate® analyser (Roche, Mannheim, Germany)14; and Thrombelastograph® Hemostasis Analyzer (Haemoscope, Niles, Illinois, USA), used for thromboelastography platelet mapping (TEG-PM)15. TEG-PM has been validated against LTA as a measure of platelet inhibition16. There are currently no validated data concerning the level of platelet inhibition that predicts bleeding in non-cardiac surgery. Adenosine 5′-diphosphate (ADP) platelet receptor inhibition (PRI) reduced to 30 per cent or less, measured by TEG-PM, has been shown to increase the risk of stent thrombosis and ischaemic events in patients with coronary artery disease17. Therefore, the value of 30 per cent was used as a surrogate for adequate platelet function in surgical patients taking clopidogrel. The aim of this study was to identify the value of TEG-PM testing in patients taking clopidogrel within 7 days of non-cardiac surgery. The specific objectives were: to study the risk of bleeding among patients with an ADP-PRI < or ≥ 30% in comparison with that of matched controls, and to identify the optimal cut-off of ADP-PRI for the cancellation of surgery. Methods A prospective cohort study was conducted between May 2008 and October 2013. All patients who had taken clopidogrel within 7 days of scheduled non-cardiac surgery underwent TEG-PM to measure percentage ADP-PRI. TEG-PM allowed stratification of these patients into two groups according to bleeding risk: low risk (ADP-PRI below 30 per cent) and high risk (ADP-PRI 30 per cent or more). All low-risk patients and those who required urgent surgery in the high-risk group proceeded to surgery, and were included in the study cohort. For each patient enrolled in the study a matched control, not treated with clopidogrel, was identified prospectively from specific operative codes (Classification of Interventions and Procedures version 4, OPCS-4) using an in-hospital database. Cases were matched for surgeon, anaesthetist, age (within 5 years), sex, surgical priority, type of surgery, operative complexity and approach. The study was approved by the local institutional review board. Patients adhering to preoperative advice to discontinue clopidogrel 7 days before surgery were identified prospectively and omitted from the analysis. Exclusion criteria also included: equipment failure of the TEG-PM test; known acquired and congenital coagulopathy; patients on treatment-dose low molecular weight and intravenous heparin, or glycoprotein IIb/IIIa inhibitors; and abnormal preoperative activated partial thromboplastin time (APTT), prothrombin time (PT) and haemoglobin measurements. Patients on dual antiplatelet therapy were advised to continue taking aspirin, unless they were having urological or neurosurgical procedures, for which the advice was to stop aspirin too for 7 days. High-risk patients who underwent surgery also received a perioperative transfusion of 1 pool of platelets. TEG-PM was performed by a trained clinical scientist in the operating theatre on the day of planned surgery. The tests were done using one of two TEG® 5000 Thrombelastograph® Hemostasis Analyzers, which were calibrated daily, in accordance with the manufacturer's instructions. The reproducibility of TEG-PM measurement of ADP-PRI has been reported by the US Food and Drug Administration, with a coefficient of variance of 10 per cent17. Whole venous blood samples of between 10 and 15 ml, retrieved through a clean venous stab from the antecubital fossa, were used for analysis in accordance with the manufacturer's guidelines18,19. Testing allowed quantification of PRI, at the ADP receptor for the effect of clopidogrel, and at the thromboxane (Tx) A2 receptor for aspirin, by measuring clot strength18. PRI was measured as a percentage, with 100 per cent representing complete platelet inhibition and 0 per cent indicating no platelet inhibition at the specific receptor sites. Inhibition of these receptors is directly related to clot strength. Clinical outcomes Patient demographics (age, sex, body mass index (BMI)), antiplatelet therapy (clopidogrel or dual therapy), predicted morbidity and mortality, preoperative haemoglobin level, platelet count, PT and APTT ratio, case specifications (surgical complexity – major complex, major, intermediate, minor20; surgical priority – elective, urgent) were recorded prospectively for all patients in an anonymized database. The primary clinical outcome was intraoperative packed red blood cell (PRBC) transfusion. Secondary outcomes included: other intraoperative blood product transfusions, total 30-day postoperative blood product transfusion, reintervention secondary to bleeding (return to theatre/radiological intervention), major adverse cardiac events, postoperative morbidity according to the Clavien–Dindo classification21, specific complications and 30-day mortality. The predicted risk of morbidity and mortality was calculated using a risk prediction model based on the Portsmouth Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM)22. Intraoperative transfusion of PRBCs was done if the haemoglobin level fell below 8 g/dl (9 g/dl in patients with ischaemic heart disease), as measured by an automated haematology analyser (pOCH-100i™; Sysmex, Milton Keynes, UK) at the time of the operation. If blood loss exceeded 1000 ml or the haemoglobin level dropped by more than 4 g/dl, the institutional major haemorrhage transfusion protocol was followed, with tranexamic acid also administered as a single intravenous dose of 1 g. Autologous transfusion of PRBCs through cell salvage was also factored into the transfusion requirement, with 300 ml being equivalent to 1 unit of allogeneic PRBCs. Statistical analysis Statistical analysis was done using SPSS® version 20 (IBM, Armonk, New York, USA) and R version 3.1.0 (http://www.r-project.org/). P < 0·050 was used to define statistical significance. Case–control analysis The groups at low and high risk of bleeding were compared with their respective case-matched controls for patient demographics, case characteristics, interventions secondary to bleeding and postoperative outcomes. Mann–Whitney U test and χ2 test were used to compare continuous and categorical variables respectively. Relative risk, comparing the low- and high-risk groups with their matched controls, was calculated for the following outcomes: intraoperative PRBC transfusion (at least 1 unit, at least 2 units and more than 6 units), severe morbidity (at least Clavien–Dindo grade IV) and mortality. The Pearson product–moment correlation coefficient was used to study the relationship between number of days that clopidogrel was stopped and the percentage ADP-PRI, and the percentage TxA2-PRI or ADP-PRI and units of intraoperative PRBC transfusion in the groups that underwent surgery. Identification of causal factors for intraoperative packed red blood cell transfusion All patients in the study group who underwent surgery were included in a binary logistic regression analysis to identify causal risk factors for intraoperative PRBC transfusion. The dependent variable was the need for intraoperative PRBC transfusion (yes or no); potential causal factors (independent variables) included: age, sex, BMI, surgical priority, surgical complexity, predicted morbidity and mortality, preoperative PT, platelet count, haemoglobin level, and percentage ADP-PRI and TxA2-PRI. Predictive capability of percentage ADP-PRI for adverse outcomes To investigate the ability of percentage ADP-PRI to predict intraoperative PRBC transfusion (at least 1 unit, at least 2 units, more than 6 units), severe morbidity and mortality, logistic regression models were created with each of these outcomes as the dependent variable and ADP-PRI as the independent variable. Internal validation for these models was done by using bootstrapping with resampling. The average receiver operating characteristic (ROC) curve of 2000 bootstrapped samples was plotted for each predictive model. To identify the optimal percentage ADP-PRI cut-off for cancelling surgery, the maximum sum of the combined sensitivity and specificity of these curves was calculated. Results During the study, 368 patients booked for surgery were being treated with clopidogrel. Of these, 182 (49·5 per cent) did not stop clopidogrel treatment, as identified on the day of surgery, and were enrolled in the study. Twenty-three (12·6 per cent) of the 182 patients did not have sufficient time to stop clopidogrel owing to the urgent nature of their operation. The TEG-PM test failed in five patients, leaving 177 in the study group. Some 115 patients had ADP-PRI of less than 30 per cent and 62 had ADP-PRI of 30 per cent or more. A total of 112 patients underwent surgery in the low-risk group (including 4 urgent operations), and 19 patients in the high-risk group because of urgent clinical need. Three patients in the low-risk group did not have an operation for logistical reasons. Case–control analysis Demographics of patients and case-matched controls are compared in Table 1. In the study group of 177 patients, clopidogrel had been stopped for a median of 1 (i.q.r. 0–2) days before surgery. No correlation was observed between number of days that clopidogrel was stopped and the percentage ADP-PRI (r = 0·008, 95 per cent confidence interval (c.i.) –0·141 to 0·158; P = 0·991). Table 1 Demographics of cases and controls in the study . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Age (years)* 69 (62–76) 68 (62–73·8) 0·243‡ 65 (58–78) 64 (55–72) 0·412‡ Sex ratio (M : F) 71 : 41 70 : 42 0·890 13 : 6 13 : 6 1·000 Body mass index (kg/m2)* 27·4 (25·0–34·9) 27·5 (23·9–34·8) 0·737‡ 23·2 (20·2–26·6) 27·0 (22·0–31·9) 0·199‡ Predicted morbidity (%)* 21·3 (11·4–34·5) 20·3 (11·1–34·3) 0·352‡ 32 (23·3–44·5) 32·3 (23·3–45) 0·748‡ Predicted mortality (%)* 1·8 (0·23–3·5) 1·7 (0·23–3·3) 0·481‡ 3·0 (1·2–4·4) 3·2 (1·2–4·5) 0·770‡ Type of surgery 1·000 1·000  Gastrointestinal 14 14 5 5 Hepatobiliary 11 11 5 5 Vascular 33 33 4 4 Neurosurgery 16 16 2 2 Orthopaedics 12 12 1 1 Ear, nose and throat 7 7 1 1 Plastics 3 3 1 1 Radiological 3 3 0 0 Renal 7 7 0 0 Urology 6 6 0 0 Surgical complexity 1·000 1·000 Minor 4 4 0 0 Intermediate 24 24 1 1 Major 37 37 3 3 Major – complex 47 47 15 15 Antiplatelet agent Clopidogrel 112 0 19 0 Aspirin 107 96 0·012 17 15 0·374 Preop. blood tests* Haemoglobin (g/dl) 11·2 (9·9–12·2) 11·2 (10·0–12·2) 0·771‡ 10·7 (9·5–11·9) 11·2 (9·8–12·2) 0·349‡ Platelets (×109/l) 200 (150–301·3) 230 (170–345) 0·058‡ 167 (145–345) 200 (167–230) 0·526‡ Prothrombin time (s) 10·6 (10·4–10·6) 10·5 (10·5–10·6) 0·091‡ 10·5 (10·5–10·6) 10·5 (10·5–10·6) 0·501‡ APTT (s) 34·0 (32·0–45·0) 35·0 (30·0–43·0) 0·852‡ 42 (32–45) 40 (31–32) 0·169‡ . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Age (years)* 69 (62–76) 68 (62–73·8) 0·243‡ 65 (58–78) 64 (55–72) 0·412‡ Sex ratio (M : F) 71 : 41 70 : 42 0·890 13 : 6 13 : 6 1·000 Body mass index (kg/m2)* 27·4 (25·0–34·9) 27·5 (23·9–34·8) 0·737‡ 23·2 (20·2–26·6) 27·0 (22·0–31·9) 0·199‡ Predicted morbidity (%)* 21·3 (11·4–34·5) 20·3 (11·1–34·3) 0·352‡ 32 (23·3–44·5) 32·3 (23·3–45) 0·748‡ Predicted mortality (%)* 1·8 (0·23–3·5) 1·7 (0·23–3·3) 0·481‡ 3·0 (1·2–4·4) 3·2 (1·2–4·5) 0·770‡ Type of surgery 1·000 1·000  Gastrointestinal 14 14 5 5 Hepatobiliary 11 11 5 5 Vascular 33 33 4 4 Neurosurgery 16 16 2 2 Orthopaedics 12 12 1 1 Ear, nose and throat 7 7 1 1 Plastics 3 3 1 1 Radiological 3 3 0 0 Renal 7 7 0 0 Urology 6 6 0 0 Surgical complexity 1·000 1·000 Minor 4 4 0 0 Intermediate 24 24 1 1 Major 37 37 3 3 Major – complex 47 47 15 15 Antiplatelet agent Clopidogrel 112 0 19 0 Aspirin 107 96 0·012 17 15 0·374 Preop. blood tests* Haemoglobin (g/dl) 11·2 (9·9–12·2) 11·2 (10·0–12·2) 0·771‡ 10·7 (9·5–11·9) 11·2 (9·8–12·2) 0·349‡ Platelets (×109/l) 200 (150–301·3) 230 (170–345) 0·058‡ 167 (145–345) 200 (167–230) 0·526‡ Prothrombin time (s) 10·6 (10·4–10·6) 10·5 (10·5–10·6) 0·091‡ 10·5 (10·5–10·6) 10·5 (10·5–10·6) 0·501‡ APTT (s) 34·0 (32·0–45·0) 35·0 (30·0–43·0) 0·852‡ 42 (32–45) 40 (31–32) 0·169‡ * Values are median (i.q.r.). ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; APTT, activated partial thromboplastin time. † χ2 test, except ‡ Mann–Whitney U test. Open in new tab Table 1 Demographics of cases and controls in the study . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Age (years)* 69 (62–76) 68 (62–73·8) 0·243‡ 65 (58–78) 64 (55–72) 0·412‡ Sex ratio (M : F) 71 : 41 70 : 42 0·890 13 : 6 13 : 6 1·000 Body mass index (kg/m2)* 27·4 (25·0–34·9) 27·5 (23·9–34·8) 0·737‡ 23·2 (20·2–26·6) 27·0 (22·0–31·9) 0·199‡ Predicted morbidity (%)* 21·3 (11·4–34·5) 20·3 (11·1–34·3) 0·352‡ 32 (23·3–44·5) 32·3 (23·3–45) 0·748‡ Predicted mortality (%)* 1·8 (0·23–3·5) 1·7 (0·23–3·3) 0·481‡ 3·0 (1·2–4·4) 3·2 (1·2–4·5) 0·770‡ Type of surgery 1·000 1·000  Gastrointestinal 14 14 5 5 Hepatobiliary 11 11 5 5 Vascular 33 33 4 4 Neurosurgery 16 16 2 2 Orthopaedics 12 12 1 1 Ear, nose and throat 7 7 1 1 Plastics 3 3 1 1 Radiological 3 3 0 0 Renal 7 7 0 0 Urology 6 6 0 0 Surgical complexity 1·000 1·000 Minor 4 4 0 0 Intermediate 24 24 1 1 Major 37 37 3 3 Major – complex 47 47 15 15 Antiplatelet agent Clopidogrel 112 0 19 0 Aspirin 107 96 0·012 17 15 0·374 Preop. blood tests* Haemoglobin (g/dl) 11·2 (9·9–12·2) 11·2 (10·0–12·2) 0·771‡ 10·7 (9·5–11·9) 11·2 (9·8–12·2) 0·349‡ Platelets (×109/l) 200 (150–301·3) 230 (170–345) 0·058‡ 167 (145–345) 200 (167–230) 0·526‡ Prothrombin time (s) 10·6 (10·4–10·6) 10·5 (10·5–10·6) 0·091‡ 10·5 (10·5–10·6) 10·5 (10·5–10·6) 0·501‡ APTT (s) 34·0 (32·0–45·0) 35·0 (30·0–43·0) 0·852‡ 42 (32–45) 40 (31–32) 0·169‡ . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Age (years)* 69 (62–76) 68 (62–73·8) 0·243‡ 65 (58–78) 64 (55–72) 0·412‡ Sex ratio (M : F) 71 : 41 70 : 42 0·890 13 : 6 13 : 6 1·000 Body mass index (kg/m2)* 27·4 (25·0–34·9) 27·5 (23·9–34·8) 0·737‡ 23·2 (20·2–26·6) 27·0 (22·0–31·9) 0·199‡ Predicted morbidity (%)* 21·3 (11·4–34·5) 20·3 (11·1–34·3) 0·352‡ 32 (23·3–44·5) 32·3 (23·3–45) 0·748‡ Predicted mortality (%)* 1·8 (0·23–3·5) 1·7 (0·23–3·3) 0·481‡ 3·0 (1·2–4·4) 3·2 (1·2–4·5) 0·770‡ Type of surgery 1·000 1·000  Gastrointestinal 14 14 5 5 Hepatobiliary 11 11 5 5 Vascular 33 33 4 4 Neurosurgery 16 16 2 2 Orthopaedics 12 12 1 1 Ear, nose and throat 7 7 1 1 Plastics 3 3 1 1 Radiological 3 3 0 0 Renal 7 7 0 0 Urology 6 6 0 0 Surgical complexity 1·000 1·000 Minor 4 4 0 0 Intermediate 24 24 1 1 Major 37 37 3 3 Major – complex 47 47 15 15 Antiplatelet agent Clopidogrel 112 0 19 0 Aspirin 107 96 0·012 17 15 0·374 Preop. blood tests* Haemoglobin (g/dl) 11·2 (9·9–12·2) 11·2 (10·0–12·2) 0·771‡ 10·7 (9·5–11·9) 11·2 (9·8–12·2) 0·349‡ Platelets (×109/l) 200 (150–301·3) 230 (170–345) 0·058‡ 167 (145–345) 200 (167–230) 0·526‡ Prothrombin time (s) 10·6 (10·4–10·6) 10·5 (10·5–10·6) 0·091‡ 10·5 (10·5–10·6) 10·5 (10·5–10·6) 0·501‡ APTT (s) 34·0 (32·0–45·0) 35·0 (30·0–43·0) 0·852‡ 42 (32–45) 40 (31–32) 0·169‡ * Values are median (i.q.r.). ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; APTT, activated partial thromboplastin time. † χ2 test, except ‡ Mann–Whitney U test. Open in new tab Median (i.q.r.) ADP-PRI in the low- and high-risk groups before surgery was 23·0 (11·2–26·4) and 68·2 (35–87) respectively. Intraoperative transfusion requirements, postoperative intervention secondary to bleeding, and postoperative morbidity and mortality are reported in Table 2. There were no differences between patients in the low-risk group and their respective controls. However, patients in the high-risk group who proceeded to surgery had significantly greater intraoperative PRBC transfusion than their matched controls (P < 0·001). They also had worse secondary outcomes: intraoperative fresh frozen plasma (FFP) (P < 0·001), platelet (P = 0·046) and total blood product (P < 0·001) transfusion; 30-day postoperative PRBCs (P = 0·012), platelet (P = 0·025) and total blood product (P = 0·006) transfusion; and return to theatre (P = 0·034). However, there were no significant differences in postoperative morbidity or mortality. The relative risk of intraoperative PRBC transfusion and adverse postoperative outcomes are shown in Table 3. Table 2 Interventions secondary to bleeding, and postoperative outcomes in case and control groups . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Intraoperative transfusion* PRBCs (units) 0·3(0·9); 0 0·3(1·0); 0 0·851‡ 5·3(6·0); 4 (2–8) 0·6(1·1); 0 (0–1) < 0·001‡ FFP (units) 0·1(0·3); 0 0·1(0·3); 0 0·537‡ 2·6(2·8); 2 (0–3) 0; 0 < 0·001‡ Platelets (pools) 0·1(0·3); 0 0·1(0·4); 0 0·702‡ 1·4(2·1); 0 (0–2) 0·1(0·5); 0 0·046‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·702‡ 1·1(1·9); 0 (0–2) 0·1(0·5); 0 0·096‡ Total (units/pools) 0·5(1·3); 0 0·5(1·4); 0 0·894‡ 10·4(12·1); 6 (2–15) 0·7(1·3); 0 (0–1) < 0·001‡ 30-day postoperative transfusion* PRBCs (units) 0·2(0·6); 0 0·3(0·7); 0 0·440‡ 2·4(2·5); 2 (0–4) 0·6(1·0); 0 (0–2) 0·012‡ FFP (units) 0·1(0·4); 0 0·1(0·4); 0 0·794‡ 0·9(1·3); 0 (0–2) 0·1(0·5); 0 0·091‡ Platelets (pools) 0·1(0·3); 0 0·1(0·3); 0 0·356‡ 0·7(0·9); 0 (0–2) 0; 0 0·025‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·694‡ 0·5(0·8); 0 (0–1) 0·1(0·3); 0 0·234‡ Total (units/pools) 0·4(1·3); 0 0·5(1·4); 0 0·192‡ 4·5(5·0); 2 (0–8) 0·8(1·3); 0 (0–2) 0·006‡ Return to theatre secondary to bleeding 3 5 0·471 4 0 0·034 Postoperative radiological intervention for bleeding 2 0 0·155 6 2 0·111 Major adverse cardiac events 10 10 1·000 4 2 0·374 Clavien–Dindo classification 0·866 0·282 0 76 73 2 4 I 17 17 0 1 II 15 17 1 5 IIIa 1 1 1 2 IIIb 2 1 4 3 IVa 1 1 3 2 IVb 0 0 4 1 V 0 2 4 1 Specific postoperative complications Myocardial infarction 3 2 0·651 2 2 1·000 Pulmonary embolism 3 2 0·651 0 0 1·000 Venous thromboembolism 3 5 0·471 2 0 0·146 Stroke/TIA 2 3 0·651 0 0 1·000 30-day mortality 0 2 0·155 4 1 0·150 . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Intraoperative transfusion* PRBCs (units) 0·3(0·9); 0 0·3(1·0); 0 0·851‡ 5·3(6·0); 4 (2–8) 0·6(1·1); 0 (0–1) < 0·001‡ FFP (units) 0·1(0·3); 0 0·1(0·3); 0 0·537‡ 2·6(2·8); 2 (0–3) 0; 0 < 0·001‡ Platelets (pools) 0·1(0·3); 0 0·1(0·4); 0 0·702‡ 1·4(2·1); 0 (0–2) 0·1(0·5); 0 0·046‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·702‡ 1·1(1·9); 0 (0–2) 0·1(0·5); 0 0·096‡ Total (units/pools) 0·5(1·3); 0 0·5(1·4); 0 0·894‡ 10·4(12·1); 6 (2–15) 0·7(1·3); 0 (0–1) < 0·001‡ 30-day postoperative transfusion* PRBCs (units) 0·2(0·6); 0 0·3(0·7); 0 0·440‡ 2·4(2·5); 2 (0–4) 0·6(1·0); 0 (0–2) 0·012‡ FFP (units) 0·1(0·4); 0 0·1(0·4); 0 0·794‡ 0·9(1·3); 0 (0–2) 0·1(0·5); 0 0·091‡ Platelets (pools) 0·1(0·3); 0 0·1(0·3); 0 0·356‡ 0·7(0·9); 0 (0–2) 0; 0 0·025‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·694‡ 0·5(0·8); 0 (0–1) 0·1(0·3); 0 0·234‡ Total (units/pools) 0·4(1·3); 0 0·5(1·4); 0 0·192‡ 4·5(5·0); 2 (0–8) 0·8(1·3); 0 (0–2) 0·006‡ Return to theatre secondary to bleeding 3 5 0·471 4 0 0·034 Postoperative radiological intervention for bleeding 2 0 0·155 6 2 0·111 Major adverse cardiac events 10 10 1·000 4 2 0·374 Clavien–Dindo classification 0·866 0·282 0 76 73 2 4 I 17 17 0 1 II 15 17 1 5 IIIa 1 1 1 2 IIIb 2 1 4 3 IVa 1 1 3 2 IVb 0 0 4 1 V 0 2 4 1 Specific postoperative complications Myocardial infarction 3 2 0·651 2 2 1·000 Pulmonary embolism 3 2 0·651 0 0 1·000 Venous thromboembolism 3 5 0·471 2 0 0·146 Stroke/TIA 2 3 0·651 0 0 1·000 30-day mortality 0 2 0·155 4 1 0·150 * Values are mean(s.d.); median (i.q.r.). ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell; FFP, fresh frozen plasma; RF7, recombinant factor 7; TIA, transient ischaemic attack. † χ2 test, except ‡ Mann–Whitney U test. Open in new tab Table 2 Interventions secondary to bleeding, and postoperative outcomes in case and control groups . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Intraoperative transfusion* PRBCs (units) 0·3(0·9); 0 0·3(1·0); 0 0·851‡ 5·3(6·0); 4 (2–8) 0·6(1·1); 0 (0–1) < 0·001‡ FFP (units) 0·1(0·3); 0 0·1(0·3); 0 0·537‡ 2·6(2·8); 2 (0–3) 0; 0 < 0·001‡ Platelets (pools) 0·1(0·3); 0 0·1(0·4); 0 0·702‡ 1·4(2·1); 0 (0–2) 0·1(0·5); 0 0·046‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·702‡ 1·1(1·9); 0 (0–2) 0·1(0·5); 0 0·096‡ Total (units/pools) 0·5(1·3); 0 0·5(1·4); 0 0·894‡ 10·4(12·1); 6 (2–15) 0·7(1·3); 0 (0–1) < 0·001‡ 30-day postoperative transfusion* PRBCs (units) 0·2(0·6); 0 0·3(0·7); 0 0·440‡ 2·4(2·5); 2 (0–4) 0·6(1·0); 0 (0–2) 0·012‡ FFP (units) 0·1(0·4); 0 0·1(0·4); 0 0·794‡ 0·9(1·3); 0 (0–2) 0·1(0·5); 0 0·091‡ Platelets (pools) 0·1(0·3); 0 0·1(0·3); 0 0·356‡ 0·7(0·9); 0 (0–2) 0; 0 0·025‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·694‡ 0·5(0·8); 0 (0–1) 0·1(0·3); 0 0·234‡ Total (units/pools) 0·4(1·3); 0 0·5(1·4); 0 0·192‡ 4·5(5·0); 2 (0–8) 0·8(1·3); 0 (0–2) 0·006‡ Return to theatre secondary to bleeding 3 5 0·471 4 0 0·034 Postoperative radiological intervention for bleeding 2 0 0·155 6 2 0·111 Major adverse cardiac events 10 10 1·000 4 2 0·374 Clavien–Dindo classification 0·866 0·282 0 76 73 2 4 I 17 17 0 1 II 15 17 1 5 IIIa 1 1 1 2 IIIb 2 1 4 3 IVa 1 1 3 2 IVb 0 0 4 1 V 0 2 4 1 Specific postoperative complications Myocardial infarction 3 2 0·651 2 2 1·000 Pulmonary embolism 3 2 0·651 0 0 1·000 Venous thromboembolism 3 5 0·471 2 0 0·146 Stroke/TIA 2 3 0·651 0 0 1·000 30-day mortality 0 2 0·155 4 1 0·150 . ADP-PRI < 30% (n = 112) . Control (n = 112) . P† . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . P† . Intraoperative transfusion* PRBCs (units) 0·3(0·9); 0 0·3(1·0); 0 0·851‡ 5·3(6·0); 4 (2–8) 0·6(1·1); 0 (0–1) < 0·001‡ FFP (units) 0·1(0·3); 0 0·1(0·3); 0 0·537‡ 2·6(2·8); 2 (0–3) 0; 0 < 0·001‡ Platelets (pools) 0·1(0·3); 0 0·1(0·4); 0 0·702‡ 1·4(2·1); 0 (0–2) 0·1(0·5); 0 0·046‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·702‡ 1·1(1·9); 0 (0–2) 0·1(0·5); 0 0·096‡ Total (units/pools) 0·5(1·3); 0 0·5(1·4); 0 0·894‡ 10·4(12·1); 6 (2–15) 0·7(1·3); 0 (0–1) < 0·001‡ 30-day postoperative transfusion* PRBCs (units) 0·2(0·6); 0 0·3(0·7); 0 0·440‡ 2·4(2·5); 2 (0–4) 0·6(1·0); 0 (0–2) 0·012‡ FFP (units) 0·1(0·4); 0 0·1(0·4); 0 0·794‡ 0·9(1·3); 0 (0–2) 0·1(0·5); 0 0·091‡ Platelets (pools) 0·1(0·3); 0 0·1(0·3); 0 0·356‡ 0·7(0·9); 0 (0–2) 0; 0 0·025‡ Other (cryoprecipitate/RF7) (units) 0(0·2); 0 0(0·2); 0 0·694‡ 0·5(0·8); 0 (0–1) 0·1(0·3); 0 0·234‡ Total (units/pools) 0·4(1·3); 0 0·5(1·4); 0 0·192‡ 4·5(5·0); 2 (0–8) 0·8(1·3); 0 (0–2) 0·006‡ Return to theatre secondary to bleeding 3 5 0·471 4 0 0·034 Postoperative radiological intervention for bleeding 2 0 0·155 6 2 0·111 Major adverse cardiac events 10 10 1·000 4 2 0·374 Clavien–Dindo classification 0·866 0·282 0 76 73 2 4 I 17 17 0 1 II 15 17 1 5 IIIa 1 1 1 2 IIIb 2 1 4 3 IVa 1 1 3 2 IVb 0 0 4 1 V 0 2 4 1 Specific postoperative complications Myocardial infarction 3 2 0·651 2 2 1·000 Pulmonary embolism 3 2 0·651 0 0 1·000 Venous thromboembolism 3 5 0·471 2 0 0·146 Stroke/TIA 2 3 0·651 0 0 1·000 30-day mortality 0 2 0·155 4 1 0·150 * Values are mean(s.d.); median (i.q.r.). ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell; FFP, fresh frozen plasma; RF7, recombinant factor 7; TIA, transient ischaemic attack. † χ2 test, except ‡ Mann–Whitney U test. Open in new tab Table 3 Relative risk of intraoperative packed red blood cell transfusion and adverse postoperative outcomes . ADP-PRI < 30% (n = 112) . Control (n = 112) . Relative risk* . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . Relative risk* . Intraoperative PRBC transfusion (units) ≥ 1 15 (13·4) 14 (12·5) 1·07 (0·54, 2·11) 15 (79) 5 (26) 3·00 (1·37, 6·59) ≥ 2 11 (9·8) 11 (9·8) 1·00 (0·45, 2·21) 15 (79) 4 (21) 3·75 (1·52, 9·23) > 6 0 (0) 1 (0·9) – 6 (32) 0 (0) – Severe morbidity 3 (2·7) 4 (3·6) 0·75 (0·17, 3·28) 15 (79) 7 (37) 2·14 (1·14, 4·04) Mortality 0 (0) 2 (1·8) – 4 (21) 1 (5) 4·00 (0·49, 32·57) . ADP-PRI < 30% (n = 112) . Control (n = 112) . Relative risk* . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . Relative risk* . Intraoperative PRBC transfusion (units) ≥ 1 15 (13·4) 14 (12·5) 1·07 (0·54, 2·11) 15 (79) 5 (26) 3·00 (1·37, 6·59) ≥ 2 11 (9·8) 11 (9·8) 1·00 (0·45, 2·21) 15 (79) 4 (21) 3·75 (1·52, 9·23) > 6 0 (0) 1 (0·9) – 6 (32) 0 (0) – Severe morbidity 3 (2·7) 4 (3·6) 0·75 (0·17, 3·28) 15 (79) 7 (37) 2·14 (1·14, 4·04) Mortality 0 (0) 2 (1·8) – 4 (21) 1 (5) 4·00 (0·49, 32·57) Values in parentheses are percentages unless indicated otherwise; * values in parentheses are 95 per cent confidence intervals. ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell. Open in new tab Table 3 Relative risk of intraoperative packed red blood cell transfusion and adverse postoperative outcomes . ADP-PRI < 30% (n = 112) . Control (n = 112) . Relative risk* . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . Relative risk* . Intraoperative PRBC transfusion (units) ≥ 1 15 (13·4) 14 (12·5) 1·07 (0·54, 2·11) 15 (79) 5 (26) 3·00 (1·37, 6·59) ≥ 2 11 (9·8) 11 (9·8) 1·00 (0·45, 2·21) 15 (79) 4 (21) 3·75 (1·52, 9·23) > 6 0 (0) 1 (0·9) – 6 (32) 0 (0) – Severe morbidity 3 (2·7) 4 (3·6) 0·75 (0·17, 3·28) 15 (79) 7 (37) 2·14 (1·14, 4·04) Mortality 0 (0) 2 (1·8) – 4 (21) 1 (5) 4·00 (0·49, 32·57) . ADP-PRI < 30% (n = 112) . Control (n = 112) . Relative risk* . ADP-PRI ≥ 30% (n = 19) . Control (n = 19) . Relative risk* . Intraoperative PRBC transfusion (units) ≥ 1 15 (13·4) 14 (12·5) 1·07 (0·54, 2·11) 15 (79) 5 (26) 3·00 (1·37, 6·59) ≥ 2 11 (9·8) 11 (9·8) 1·00 (0·45, 2·21) 15 (79) 4 (21) 3·75 (1·52, 9·23) > 6 0 (0) 1 (0·9) – 6 (32) 0 (0) – Severe morbidity 3 (2·7) 4 (3·6) 0·75 (0·17, 3·28) 15 (79) 7 (37) 2·14 (1·14, 4·04) Mortality 0 (0) 2 (1·8) – 4 (21) 1 (5) 4·00 (0·49, 32·57) Values in parentheses are percentages unless indicated otherwise; * values in parentheses are 95 per cent confidence intervals. ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell. Open in new tab No correlation was observed between the percentage ADP-PRI and the amount of intraoperative PRBCs transfused (r = 0·098, 95 per cent c.i. –0·090 to 0·286; P = 0·303) in the group with a low bleeding risk. However, a strong positive correlation was observed in the high-risk group (r = 0·749, 0·410 to 0·940; P < 0·001) (Fig. 1). The percentage TxA2-PRI did not correlate with intraoperative PRBC transfusion requirements in the low-risk (r = 0·041, −0·156 to 0·239; P = 0·676) or high-risk (r = −0·039, −0·458 to 0·398; P = 0·883) groups. Fig. 1 Open in new tabDownload slide Correlation between adenosine 5′-diphosphate platelet receptor inhibition (ADP-PRI) and intraoperative packed red blood cell (PRBC) transfusion. Dotted line indicates threshold value for stratification of patients into low-risk (ADP-PRI less than 30 per cent) and high-risk (ADP-PRI 30 per cent or more) groups. Low risk (112 patients): r = 0·098, P = 0·303; high risk (19 patients): r = 0·749, P < 0·001 There were four and zero deaths within 30 days in the high- and low-risk groups respectively. The first patient (ADP-PRI 78 per cent) died during laparotomy for an obstructing rectal tumour. The patient received a massive transfusion during the operation as a consequence of severe blood loss/coagulopathy and eventually suffered a cardiac arrest secondary to hypovolaemia. The second patient (ADP-PRI 72 per cent), undergoing urgent endovascular aortic aneurysm repair, had a considerable perioperative haemorrhage and received transfusion of 15 units PRBCs, 6 units FFP and 4 pools of platelets. He developed suspected transfusion-related acute lung injury after surgery and died from multiple organ failure (MOF) on day 16. The third patient (ADP-PRI 83 per cent) required a massive transfusion during femoropopliteal bypass and died from MOF on day 6. The fourth high-risk patient died from medical complications unrelated to bleeding. Regression analysis Identification of causal factors for intraoperative packed red blood cell transfusion The only apparent causal factor for intraoperative PRBC transfusion was percentage ADP-PRI (odds ratio 1·07, 95 per cent c.i. 1·02 to 1·13; P = 0·005) (Table S1, supporting information). Predictive capability of percentage ADP-PRI for adverse outcomes From the 2000 bootstrapped samples the average area under the ROC curve (95 per cent c.i.) for ADP-PRI and intraoperative PRBC transfusion of at least 1 unit, at least 2 units or over 6 units, severe morbidity (at least grade IV Clavien–Dindo) and mortality were: 0·744 (0·627 to 0·860), 0·817 (0·713 to 0·920), 0·997 (0·991 to 1·000), 0·956 (0·907 to 1·000) and 0·962 (0·918 to 1·000) respectively (Fig. S1, supporting information). When the maximum sum of the sensitivity and specificity of all these curves was combined, the optimal cut-off for ADP-PRI was 34 per cent. The sensitivities and specificities of each predictive model at this cut-off value are recorded in Table 4. Table 4 Predictive values of 34 per cent ADP-PRI cut-off in 2000 bootstrapped samples . Intraoperative PRBC transfusion (units) . Severe morbidity . Mortality . . ≥ 1 . ≥ 2 . > 6 . Total no. with outcome 30 26 6 18 4 No. of true positives 15 (10, 21) 15 (10, 20) 6 (6, 6) 14 (10, 17) 4 (4, 4) No. of true negatives 100 (98, 101) 104 (102, 105) 115 (109, 121) 111 (108, 113) 115 (108, 121) Sensitivity (%) 50 (33, 70) 58 (38, 77) 100 (100, 100) 78 (56, 94) 100 (100, 100) Specificity (%) 99 (97, 100) 99 (97, 100) 92 (88, 97) 98 (96, 100) 91 (85, 95) PPV (%) 94 (80, 100) 94 (80, 100) 38 (28, 60) 88 (71, 100) 25 (17, 40) NPV (%) 87 (83, 92) 90 (87, 95) 100 (100, 100) 97 (93, 99) 100 (100, 100) . Intraoperative PRBC transfusion (units) . Severe morbidity . Mortality . . ≥ 1 . ≥ 2 . > 6 . Total no. with outcome 30 26 6 18 4 No. of true positives 15 (10, 21) 15 (10, 20) 6 (6, 6) 14 (10, 17) 4 (4, 4) No. of true negatives 100 (98, 101) 104 (102, 105) 115 (109, 121) 111 (108, 113) 115 (108, 121) Sensitivity (%) 50 (33, 70) 58 (38, 77) 100 (100, 100) 78 (56, 94) 100 (100, 100) Specificity (%) 99 (97, 100) 99 (97, 100) 92 (88, 97) 98 (96, 100) 91 (85, 95) PPV (%) 94 (80, 100) 94 (80, 100) 38 (28, 60) 88 (71, 100) 25 (17, 40) NPV (%) 87 (83, 92) 90 (87, 95) 100 (100, 100) 97 (93, 99) 100 (100, 100) Values in parentheses are 95 per cent confidence intervals. ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell; PPV, positive predictive value; NPV, negative predictive value. Open in new tab Table 4 Predictive values of 34 per cent ADP-PRI cut-off in 2000 bootstrapped samples . Intraoperative PRBC transfusion (units) . Severe morbidity . Mortality . . ≥ 1 . ≥ 2 . > 6 . Total no. with outcome 30 26 6 18 4 No. of true positives 15 (10, 21) 15 (10, 20) 6 (6, 6) 14 (10, 17) 4 (4, 4) No. of true negatives 100 (98, 101) 104 (102, 105) 115 (109, 121) 111 (108, 113) 115 (108, 121) Sensitivity (%) 50 (33, 70) 58 (38, 77) 100 (100, 100) 78 (56, 94) 100 (100, 100) Specificity (%) 99 (97, 100) 99 (97, 100) 92 (88, 97) 98 (96, 100) 91 (85, 95) PPV (%) 94 (80, 100) 94 (80, 100) 38 (28, 60) 88 (71, 100) 25 (17, 40) NPV (%) 87 (83, 92) 90 (87, 95) 100 (100, 100) 97 (93, 99) 100 (100, 100) . Intraoperative PRBC transfusion (units) . Severe morbidity . Mortality . . ≥ 1 . ≥ 2 . > 6 . Total no. with outcome 30 26 6 18 4 No. of true positives 15 (10, 21) 15 (10, 20) 6 (6, 6) 14 (10, 17) 4 (4, 4) No. of true negatives 100 (98, 101) 104 (102, 105) 115 (109, 121) 111 (108, 113) 115 (108, 121) Sensitivity (%) 50 (33, 70) 58 (38, 77) 100 (100, 100) 78 (56, 94) 100 (100, 100) Specificity (%) 99 (97, 100) 99 (97, 100) 92 (88, 97) 98 (96, 100) 91 (85, 95) PPV (%) 94 (80, 100) 94 (80, 100) 38 (28, 60) 88 (71, 100) 25 (17, 40) NPV (%) 87 (83, 92) 90 (87, 95) 100 (100, 100) 97 (93, 99) 100 (100, 100) Values in parentheses are 95 per cent confidence intervals. ADP-PRI, adenosine 5′-diphosphate platelet receptor inhibition; PRBC, packed red blood cell; PPV, positive predictive value; NPV, negative predictive value. Open in new tab Discussion This study assessed the risk of intraoperative bleeding in non-cardiac surgery, through measurement of preoperative ADP-PRI in patients taking clopidogrel. Most importantly, there did not appear to be any correlation between the duration of omission of clopidogrel and percentage ADP-PRI. However, increasing percentage ADP-PRI increased the need for, and was the sole independent risk factor for, intraoperative PRBC transfusion. An optimal ADP-PRI cut-off of 34 per cent was identified to assist with decision-making with regard to proceeding with, or cancelling elective surgery. The use of TEG-PM testing could prevent significant bleeding complications and unnecessary cancellations. The poor correlation between the duration of clopidogrel omission and ADP-PRI is consistent with the significant interindividual variability observed with clopidogrel treatment10–11,23–24. It has been reported previously that up to 30 per cent of patients taking clopidogrel have minimal or no platelet inhibtion25. These findings are important in view of the recommendations to omit clopidogrel, between 5 and 10 days before surgery, to minimize intraoperative bleeding26–28. A study29 of patients requiring urgent cardiac surgery stratified patients' risk by measuring ADP-PRI daily; fewer than 20 per cent of patients had to wait the recommended 5 days of clopidogrel omission before surgery. If a safe preoperative ADP-PRI cut-off could be identified, this could be used for operative decision-making resulting in an individualized approach. A cut-off of 30 per cent was used formerly at this institution, based on risk of coronary stent thrombosis from a previous study in cardiac patients18. Patients in the present study who underwent surgery above this threshold needed greater intraoperative and postoperative transfusion than their matched controls, despite preoperative platelet transfusion in all patients. An increased risk of bleeding and transfusion requirement was determined with an ADP inhibition of 70 per cent in off-pump cardiac surgery30. Recent guidelines from the Society of Thoracic Surgeons31 also recommended using point-of-care platelet function analysis to assess patients on antiplatelets presenting for both cardiac and non-cardiac surgery because of its high negative predictive value (NPV)31. In the present study, logistic regression for pertinent clinical outcomes was used to identify a 34 per cent cut-off for ADP-PRI. At this level, half of patients requiring intraoperative PRBC transfusion, 58 per cent of those requiring at least 2 units, and all those requiring more than 6 units were predicted. The 34 per cent cut-off had very high NPVs; only 15 patients (NPV 87 per cent) below this level required transfusion, 11 patients required transfusion of at least 2 units (NPV 90 per cent) and no patient required more than 6 units (NPV 100 per cent). These high NPVs suggest that, with respect to bleeding, it is safe to operate below this threshold. The positive predictive values are weaker and so it is more important not to operate on these higher-risk patients than to prevent unnecessary cancellations. The major limitation of this study is that everyone in the high-risk group who underwent surgery had, by definition, an urgent operation. To overcome this, the study cohort was first compared with case-matched controls, and then regression analysis of the study cohort was undertaken. There was a significant difference in bleeding risk, even though all of the matched controls in the high-risk group had urgent operations. Moreover, the only independent risk factor identified in the regression analysis was the percentage ADP-PRI, and not surgical priority. The ideal study design would be to operate on all patients, irrespective of their ADP-PRI; however, it is not ethical to plan elective surgery in such high-risk patients. There were comparatively fewer patients in the high-risk group who underwent surgery, limiting the ability to identify the correct threshold; therefore, further studies are necessary to confirm that it is accurate. In this study, the 34 per cent ADP-PRI threshold identified the majority of severe complications, all PBRC transfusions of more than 6 units and all deaths, with relatively few false-positives. Disclosure The authors declare no conflict of interest. References 1 Douglas IJ , Evans SJ, Hingorani AD, Grosso AM, Timmis A, Hemingway H et al. Clopidogrel and interaction with proton pump inhibitors: comparison between cohort and within person study designs . BMJ 2012 ; 345 : e4388 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Howard-Alpe GM , de Bono J, Hudsmith L, Orr WP, Foex P, Sear JW. Coronary artery stents and non-cardiac surgery . Br J Anaesth 2007 ; 98 : 560 – 574 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Eberli D , Chassot PG, Sulser T, Samama CM, Mantz J, Delabay A et al. Urological surgery and antiplatelet drugs after cardiac and cerebrovascular accidents . J Urol 2010 ; 183 : 2128 – 2136 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Chernoguz A , Telem DA, Chu E, Ozao-Choy J, Tammaro Y, Divino CM. Cessation of clopidogrel before major abdominal procedures . Arch Surg 2011 ; 146 : 334 – 339 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Maltais S , Perrault LP, Do QB. Effect of clopidogrel on bleeding and transfusions after off-pump coronary artery bypass graft surgery: impact of discontinuation prior to surgery . Eur J Cardiothorac Surg 2008 ; 34 : 127 – 131 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Firanescu CE , Martens EJ, Schönberger JP, Soliman Hamad MA, van Straten AH. Postoperative blood loss in patients undergoing coronary artery bypass surgery after preoperative treatment with clopidogrel. A prospective randomised controlled study . Eur J Cardiothorac Surg 2009 ; 36 : 856 – 862 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Spahn DR , Howell SJ, Delabays A, Chassot PG. Coronary stents and perioperative anti-platelet regimen: dilemma of bleeding and stent thrombosis . Br J Anaesth 2006 ; 96 : 675 – 677 . Google Scholar Crossref Search ADS PubMed WorldCat 8 McFadden EP , Stabile E, Regar E, Cheneau E, Ong AT, Kinnaird T et al. Late thrombosis in drug-eluting coronary stents after discontinuation of antiplatelet therapy . Lancet 2004 ; 364 : 1519 – 1521 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Harty JA , McKenna P, Moloney D, D'Souza L, Masterson E. Anti-platelet agents and surgical delay in elderly patients with hip fractures . J Orthop Surg (Hong Kong) 2007 ; 15 : 270 – 272 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Ferreiro JL , Angiolillo DJ. Clopidogrel response variability: current status and future directions . Thromb Haemost 2009 ; 102 : 7 – 14 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Angiolillo DJ , Fernandez-Ortiz A, Bernardo E, Alfonso F, Macaya C, Bass TA et al. Variability in individual responsiveness to clopidogrel: clinical implications, management, and future perspectives . J Am Coll Cardiol 2007 ; 49 : 1505 – 1516 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Frontroth JP . Light transmission aggregometry . Methods Mol Biol 2013 ; 992 : 227 – 240 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Malinin A , Pokov A, Swaim L, Kotob M, Serebruany V. Validation of a VerifyNow-P2Y12 cartridge for monitoring platelet inhibition with clopidogrel . Methods Find Exp Clin Pharmacol 2006 ; 28 : 315 – 322 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Ranucci M , Baryshnikova E, Soro G, Ballotta A, De Benedetti D, Conti D; Surgical and Clinical Outcome Research (SCORE) Group. Multiple electrode whole-blood aggregometry and bleeding in cardiac surgery patients receiving thienopyridines . Ann Thorac Surg 2011 ; 91 : 123 – 129 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Cattano D , Altamirano AV, Kaynak HE, Seitan C, Paniccia R, Chen Z et al. Perioperative assessment of platelet function by thromboelastograph platelet mapping in cardiovascular patients undergoing non-cardiac surgery . J Thromb Thrombolysis 2012 ; 35 : 25 – 30 . Google Scholar OpenURL Placeholder Text WorldCat 16 Agarwal S , Coakley M, Coakely M, Reddy K, Riddell A, Mallett S. Quantifying the effect of antiplatelet therapy: a comparison of the platelet function analyzer (PFA-100) and modified thromboelastography (mTEG) with light transmission platelet aggregometry . Anesthesiology 2006 ; 105 : 676 – 683 . Google Scholar Crossref Search ADS PubMed WorldCat 17 US Food and Drug Administration . 510(k) Substantial Equivalence, Determination, Decision Summary and Device and Instrument Template for TEG-PMA . http://www.accessdata.fda.gov/cdrh_docs/reviews/K041502.pdf [accessed 1 February 2014]. 18 Bliden KP , DiChiara J, Tantry US, Bassi AK, Chaganti SK, Gurbel PA. Increased risk in patients with high platelet aggregation receiving chronic clopidogrel therapy undergoing percutaneous coronary intervention: is the current antiplatelet therapy adequate? J Am Coll Cardiol 2007 ; 49 : 657 – 666 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Craft RM , Chavez JJ, Bresee SJ, Wortham DC, Cohen E, Carroll RC. A novel modification of the Thrombelastograph assay, isolating platelet function, correlates with optical platelet aggregation . J Lab Clin Med 2004 ; 143 : 301 – 309 . Google Scholar Crossref Search ADS PubMed WorldCat 20 National Institute of Clinical Excellence . Preoperative Tests: the Use of Routine Preoperative Tests for Elective Surgery . Clinical Guideline 3; 2003 . http://www.nice.org.uk/nicemedia/live/10920/29090/29090.pdf [accessed 1 February 2014]. 21 Clavien PA , Barkun J, de Oliveira ML, Vauthey JN, Dindo D, Schulick RD et al. The Clavien–Dindo classification of surgical complications: five-years experience . Ann Surg 2009 ; 250 : 187 – 196 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Prytherch DR , Whiteley MS, Higgins B, Weaver PC, Prout WG, Powell SJ. POSSUM and Portsmouth POSSUM for predicting mortality. Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity . Br J Surg 1998 ; 85 : 1217 – 1220 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Gurbel PA , Bliden KP, Hiatt BL, O'Connor CM. Clopidogrel for coronary stenting: response variability, drug resistance, and the effect of pretreatment platelet reactivity . Circulation 2003 ; 107 : 2908 – 2913 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Gurbel PA , Bliden KP, Hayes KM, Yoho JA, Herzog WR, Tantry US. The relation of dosing to clopidogrel responsiveness and the incidence of high post-treatment platelet aggregation in patients undergoing coronary stenting . J Am Coll Cardiol 2005 ; 45 : 1392 – 1396 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Nguyen TA , Diodati JG, Pharand C. Resistance to clopidogrel: a review of the evidence . J Am Coll Cardiol 2005 ; 45 : 1157 – 1164 . Google Scholar Crossref Search ADS PubMed WorldCat 26 British National Formulary . British Medical Journal Publishing Group/Royal Pharmaceutical Society Publishing : London , 2007 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 27 Albaladejo P , Marret E, Piriou V, Samama CM; French Society of Anesthesiology and Intensive Care. Perioperative management of antiplatelet agents in patients with coronary stents: recommendations of a French Task Force . Br J Anaesth 2006 ; 97 : 580 – 582 . Google Scholar Crossref Search ADS PubMed WorldCat 28 Grines CL , Bonow RO, Casey DE, Gardner TJ, Lockhart PB, Moliterno DJ et al. Prevention of premature discontinuation of dual antiplatelet therapy in patients with coronary artery stents: a science advisory from the American Heart Association, American College of Cardiology, Society for Cardiovascular Angiography and Interventions, American College of Surgeons, and American Dental Association, with representation from the American College of Physicians . Circulation 2007 ; 117 : 813 – 818 . Google Scholar OpenURL Placeholder Text WorldCat 29 Mahla E , Suarez TA, Bliden KP, Rehak P, Metzler H, Sequeira AJ et al. Platelet function measurement-based strategy to reduce bleeding and waiting time in clopidogrel-treated patients undergoing coronary artery bypass graft surgery: the timing based on platelet function strategy to reduce clopidogrel-associated bleeding related to CABG (TARGET-CABG) study . Circ Cardiovasc Interv 2012 ; 5 : 261 – 269 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Kwak YL , Kim JC, Choi YS, Yoo KJ, Song Y, Shim JK. Clopidogrel responsiveness regardless of the discontinuation date predicts increased blood loss and transfusion requirement after off-pump coronary artery bypass graft surgery . J Am Coll Cardiol 2010 ; 56 : 1994 – 2002 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Ferraris VA , Saha SP, Oestreich JH, Song HK, Rosengart T, Reece TB et al.; Society of Thoracic Surgeons. 2012 update to the Society of Thoracic Surgeons guideline on use of antiplatelet drugs in patients having cardiac and noncardiac operations . Ann Thorac Surg 2012 ; 94 : 1761 – 1781 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

Mihara, M; Hara, H; Furniss, D; Narushima, M; Iida, T; Kikuchi, K; Ohtsu, H; Gennaro, P; Gabriele, G; Murai, N

doi: 10.1002/bjs.9588pmid: 25116167

Abstract Background One of the complications of lymphoedema is recurrent cellulitis. The aim was to determine whether lymphaticovenous anastomosis (LVA) was effective at reducing cellulitis in patients with lymphoedema. Methods This was a retrospective review of patients with arm/leg lymphoedema who underwent LVA. The frequency of cellulitis was compared before and after surgery. The diagnostic criteria for cellulitis were a fever of 38·5°C or higher, and warmth/redness in the affected limb(s). Results A total of 95 patients were included. The mean number of episodes of cellulitis in the year preceding surgery was 1·46, compared with 0·18 in the year after surgery (P < 0·001). Conclusion LVA reduced the rate of cellulitis in these patients with lymphoedema. Introduction Lymphoedema consists of a subcutaneous accumulation of protein-rich fluid caused by dysfunction of the lymphatic system. Primary lymphoedema is caused by congenital hypoplasia or aplasia of lymphatic vessels1,2. Secondary lymphoedema is caused by constriction, stenosis or obstruction of the lymphatic vessels associated with cancer therapy, direct tumour infiltration or infection3,4. Secondary lymphoedema affects 6–63 per cent of patients who have undergone lymph node removal during cancer surgery, or who have received radiotherapy5. Worldwide, the most common cause of secondary lymphoedema is infection with a parasitic filarial nematode Wuchereria bancrofti, leading to filariasis. This affects nearly 90 million people in 81 countries worldwide, primarily in Africa and South-East Asia6. Although lymphoedema itself is not life-threatening, externally visible changes cause psychological distress7. If untreated, significant worsening of oedema can interfere with quality of life, and rarely can lead to lymphangiosarcoma (Stewart–Treves syndrome), which has a poor prognosis8. Another severe complication of lymphoedema is repeated cellulitis of the affected limb, which often requires admission to hospital9. Inflammation increases damage to the lymph vessels and causes worsening of the lymphoedema, rendering the patient ever more susceptible to recurrent cellulitis10. Approximately 23–35 per cent of patients with lymphoedema have this pattern of recurrent, progressive cellulitis10,11. Effective prevention has yet to be determined. Prophylactic antibiotics may protect against cellulitis, but it is not uncommon for cellulitis to reappear the moment they are discontinued, and in some instances cellulitis may occur despite their use12,13. Efficient lymphatic drainage is considered an important measure for preventing cellulitis. Complex physical therapy (CPT) is now a routine treatment for lymphoedema of the extremities; it not only reduces limb circumference, but also may prevent cellulitis14,15. A surgical treatment that improves lymphatic drainage is lymphaticovenous anastomosis (LVA), where collecting lymphatic vessels are anastomosed to a cutaneous vein under surgical microscopy using 11/0 or 12/0 nylon sutures. LVA is reported to reduce limb diameter and avoid dermal sclerosis16–20. Few reports have addressed the effect of LVA on the rate of cellulitis17. The concomitant use of CPT and LVA may have a synergistic effect on improving lymphatic stasis. The present study compared the frequency of cellulitis before and after LVA in patients with arm/leg lymphoedema. Methods This was a retrospective review of clinical data from the medical records, and information from telephone interviews, concerning the number of preoperative and postoperative episodes of cellulitis among consecutive patients who underwent LVA for lymphoedema at the University of Tokyo Hospital or Saiseikai Kawaguchi General Hospital from September 2005 to April 2012. The diagnostic criteria for lymphoedema were based on the findings from indocyanine green lymphography, as well as limb girth21,22. Lymphoedema was staged using the lymphoedema staging classification recommended by the International Society of Lymphology23. The diagnostic criteria for cellulitis were a fever of 38·5°C or higher, and warmth/redness in the affected limb(s). Exclusion criteria were: follow-up of less than 1 year; previous LVA; and use of additional specialized compression therapy (for example multilayer bandaging), but not simple compression after surgery. Where clinically necessary, deep vein thrombosis was excluded by ultrasonography. LVA was conducted using a previously reported method16. Briefly, several 0·2−5-cm skin incisions were made in the affected limb, to identify the collecting lymphatic vessels and cutaneous veins of approximately 0·3−1 mm in diameter beneath the skin, followed by end-to-end or end-to-side anastomosis (Fig. 1). Based on previous patency data24, at least three anastomoses were made in each limb. The duration of the operation is about 3–4 h. Fig. 1 Open in new tabDownload slide a Lymphaticovenous anastomosis (LVA) for leg lymphoedema is usually done at a minimum of three sites. b A collecting lymphatic vessel (Ly) is sutured to a cutaneous vein (V) under microscopy. The arrow indicates the direction of lymphatic flow Standard CPT14,15 for lymphoedema, involving skin care, manual lymph drainage and compression garments, was conducted until just before surgery, and was continued after the operation, depending on the severity of lymphoedema. CPT was discontinued in mild lymphoedema if there was improvement after surgery, but was continued in severe cases. The same antibiotic schedules were used for cellulitis before and after LVA. Written informed consent was obtained from all patients following the approval of each facility's ethics committee regarding indocyanine green lymphography. Statistical analysis The frequency of cellulitis in the year immediately before surgery was compared with that in the following year. A paired two-way Student's t test was performed to compare the rates of cellulitis. The significance level was set at 0·050. Patients with longer follow-up were also studied for the mean number of episodes of cellulitis per year thereafter. Results Of 124 patients who underwent LVA, 24 had less than 1 year of follow-up and five received additional specialized compression therapy after surgery, resulting in 95 patients in the present study (Table 1). The study included six men and 89 women with a mean age of 57·8 (range 31–90) years; mean follow-up was 27·3 (range 12–57) months. There were 84 patients with leg lymphoedema and 11 with arm lymphoedema; 88 had secondary lymphoedema following cancer treatment, and seven had primary lymphoedema. Three, 14, 38, 27 and 13 patients had lymphoedema stages 0, 1, 2a, 2b and 3 respectively. Three patients underwent simultaneous resection of excess tissue and LVA. Table 1 Patient characteristics and results of lymphaticovenous anastomosis for lymphoedema . No. of patients . Mean no. of episodes of cellulitis in the year before LVA . Mean no. of episodes of cellulitis in the year after LVA . P* . All patients 95 1·46 0·18 < 0·001 Sex M 6 0·67 0·67 – F 89 1·52 0·15 < 0·001 Location Upper limb 11 0·82 0·09   0·100 Lower limb 84 1·55 0·19 < 0·001 International Society of Lymphology stage 0 3 0·00 0·00 – 1 14 0·00 0·00 – 2a 38 1·39 0·08   0·007 2b 27 1·26 0·22   0·016 3 13 4·00 0·62   0·020 Aetiology Primary 7 0·57 0·57 – Secondary 88 1·53 0·15 < 0·001 Underlying cause Cervical cancer 44 2·34 0·18 < 0·001 Uterine cancer 23 0·57 0·09   0·020 Breast cancer 10 0·8  0·10   0·152 Ovarian cancer 4 1·25 0·25   0·250 Other cancer 7 0·86 0·14   0·280 Radiotherapy Yes 29 2·72 0·28   0·001 No 59 1·89 0·15   0·003 . No. of patients . Mean no. of episodes of cellulitis in the year before LVA . Mean no. of episodes of cellulitis in the year after LVA . P* . All patients 95 1·46 0·18 < 0·001 Sex M 6 0·67 0·67 – F 89 1·52 0·15 < 0·001 Location Upper limb 11 0·82 0·09   0·100 Lower limb 84 1·55 0·19 < 0·001 International Society of Lymphology stage 0 3 0·00 0·00 – 1 14 0·00 0·00 – 2a 38 1·39 0·08   0·007 2b 27 1·26 0·22   0·016 3 13 4·00 0·62   0·020 Aetiology Primary 7 0·57 0·57 – Secondary 88 1·53 0·15 < 0·001 Underlying cause Cervical cancer 44 2·34 0·18 < 0·001 Uterine cancer 23 0·57 0·09   0·020 Breast cancer 10 0·8  0·10   0·152 Ovarian cancer 4 1·25 0·25   0·250 Other cancer 7 0·86 0·14   0·280 Radiotherapy Yes 29 2·72 0·28   0·001 No 59 1·89 0·15   0·003 LVA, lymphaticovenous anastomosis. * Paired t test. Open in new tab Table 1 Patient characteristics and results of lymphaticovenous anastomosis for lymphoedema . No. of patients . Mean no. of episodes of cellulitis in the year before LVA . Mean no. of episodes of cellulitis in the year after LVA . P* . All patients 95 1·46 0·18 < 0·001 Sex M 6 0·67 0·67 – F 89 1·52 0·15 < 0·001 Location Upper limb 11 0·82 0·09   0·100 Lower limb 84 1·55 0·19 < 0·001 International Society of Lymphology stage 0 3 0·00 0·00 – 1 14 0·00 0·00 – 2a 38 1·39 0·08   0·007 2b 27 1·26 0·22   0·016 3 13 4·00 0·62   0·020 Aetiology Primary 7 0·57 0·57 – Secondary 88 1·53 0·15 < 0·001 Underlying cause Cervical cancer 44 2·34 0·18 < 0·001 Uterine cancer 23 0·57 0·09   0·020 Breast cancer 10 0·8  0·10   0·152 Ovarian cancer 4 1·25 0·25   0·250 Other cancer 7 0·86 0·14   0·280 Radiotherapy Yes 29 2·72 0·28   0·001 No 59 1·89 0·15   0·003 . No. of patients . Mean no. of episodes of cellulitis in the year before LVA . Mean no. of episodes of cellulitis in the year after LVA . P* . All patients 95 1·46 0·18 < 0·001 Sex M 6 0·67 0·67 – F 89 1·52 0·15 < 0·001 Location Upper limb 11 0·82 0·09   0·100 Lower limb 84 1·55 0·19 < 0·001 International Society of Lymphology stage 0 3 0·00 0·00 – 1 14 0·00 0·00 – 2a 38 1·39 0·08   0·007 2b 27 1·26 0·22   0·016 3 13 4·00 0·62   0·020 Aetiology Primary 7 0·57 0·57 – Secondary 88 1·53 0·15 < 0·001 Underlying cause Cervical cancer 44 2·34 0·18 < 0·001 Uterine cancer 23 0·57 0·09   0·020 Breast cancer 10 0·8  0·10   0·152 Ovarian cancer 4 1·25 0·25   0·250 Other cancer 7 0·86 0·14   0·280 Radiotherapy Yes 29 2·72 0·28   0·001 No 59 1·89 0·15   0·003 LVA, lymphaticovenous anastomosis. * Paired t test. Open in new tab The mean number of episodes of cellulitis in the year preceding surgery was 1·46 (range 0–12), whereas that in the year after surgery was 0·18 (0–3) (P < 0·001) (Table 1). The number of episodes of cellulitis following LVA was significantly reduced in women more than in men, the leg compared with the arm, more advanced lymphoedema, secondary compared with primary lymphoedema, and was unaffected by whether or not radiotherapy had been used (Table 1). Other subgroups did not reach statistical significance, chiefly owing to smaller sample sizes. The results of surgery appeared to be sustained, as there was a trend towards reduced cellulitis rates with longer follow-up, although the numbers were small (data not shown). Discussion LVA was first reported as a treatment for chyluria in 196219. Since then, it has been reported as a treatment for extremity lymphoedema16–20, facial lymphoedema25, lymphocyst26 and lymphorrhoea27. The procedure is technically challenging, even for surgeons trained in conventional microsurgery. Lymphatic intraoperative fluorescence angiography using indocyanine green has made it possible to find lymph vessels that have retained function28. In addition, newly developed surgical instruments have made possible reliable anastomosis of the lymph vessels and veins16–17,29. The present study demonstrates the effectiveness of LVA for lymphoedema, where surgery resulted in a significant decrease in the rate of cellulitis (Fig. 2). Fig. 2 Open in new tabDownload slide Clinical appearance a,c before and b,d 1 year after lymphaticovenous anastomosis (LVA) for lymphoedema of the left leg. Changes in limb diameter are shown in b. This patient had 12 episodes of cellulitis in the year before LVA, and none in the year afterwards Definitive prevention of cellulitis in lymphoedema remains uncertain. Standard clinical practice has been to offer prophylactic antibiotics, yet some patients continue to suffer from cellulitis despite the use of penicillin12,13. Cellulitis often recurs after prophylactic penicillin is discontinued. Reduction of oedema through physiotherapy may reduce the frequency of cellulitis15, but relief may be temporary or limited. In the present study, the number of episodes of cellulitis per year was reduced from a mean 1·46 before LVA to 0·18 after LVA. Initial concerns that the invasiveness of the procedure, or redirecting the lymph into the veins peripherally, might increase the rate of cellulitis or cause systemic bacterial infection seem unfounded. Bacteria invading the skin are normally detected by dendritic cells and macrophages, leading to their activation and transport along lymphatics to initiate an immune response. In lymphoedema, there is a diminished transport function of the lymph vessels, leading to inefficient priming of the immune system30. LVA may decrease the frequency of cellulitis by restoring lymph circulation in the peripheral tissue. Cellulitis provokes fibrosis of the lymph vessels31 and subcutaneous tissue, thus triggering a vicious cycle in which the lymphoedema worsens and cellulitis becomes even more likely to develop; LVA can interrupt this cycle. Limitations of the present study include its retrospective nature. Further investigation with a controlled study should be conducted. The effects of this treatment for lymphoedema caused by filariasis, which affects a greater number of patients worldwide, will also need to be validated. Acknowledgements M.M. and H.H. contributed equally to this publication. The authors thank T. Okitsu, M. Zhou and S. Tange for discussion of these data, and daily discussions regarding the effects of LVA. They also thank I. Koshima (University of Tokyo) and S. Harasawa and H. Tsubaki (Saiseikai Kawaguchi General Hospital) for support in helping to deliver the best lymphatic treatment for their patients. Disclosure: The authors declare no conflict of interest. Open in new tabDownload slide References 1 Hara H , Mihara M, Okuda I, Okuda I, Hirota A, Narushima M et al. Presence of thoracic duct abnormalities in patients with primary lymphoedema of the extremities . J Plast Reconstr Aesthet Surg 2012 ; 65 : e305 – e310 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Hara H , Koshima I, Okuda I, Narushima M, Mihara M, Todokoro T et al. Assessment of configuration of thoracic duct using magnetic resonance thoracic ductography in idiopathic lymphoedema . Ann Plast Surg 2012 ; 68 : 300 – 302 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Olszewski W On the pathomechanism of development of postsurgical lymphoedema . Lymphology 1973 ; 6 : 35 – 51 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 4 Witte CL , Witte MH. Disorders of lymph flow . Acad Radiol 1995 ; 2 : 324 – 334 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Cormier JN , Askew RL, Mungovan KS, Xing Y, Ross MI, Armer JM. Lymphoedema beyond breast cancer: a systematic review and meta-analysis of cancer-related secondary lymphoedema . Cancer 2010 ; 116 : 5138 – 5149 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Szuba A , Shin WS, Strauss HW, Rockson S. The third circulation: radionuclide lymphoscintigraphy in the evaluation of lymphoedema . J Nucl Med 2003 ; 44 : 43 – 57 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 7 McWayne J , Heiney SP. Psychologic and social sequelae of secondary lymphoedema: a review . Cancer 2005 ; 104 : 457 – 466 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Cozen W , Bernstein L, Wang F, Press MF, Mack TM. The risk of angiosarcoma following primary breast cancer . Br J Cancer 1999 ; 81 : 532 – 536 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Simon MS , Cody RL. Cellulitis after axillary lymph node dissection for carcinoma of the breast . Am J Med 1992 ; 93 : 543 – 548 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Soran A , D'Angelo G, Begovic M, Ardic F, Harlak A, Samuel Wieand H et al. Breast cancer-related lymphoedema – what are the significant predictors and how they affect the severity of lymphoedema? Breast J 2006 ; 12 : 536 – 543 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Schirger A , Martin WJ, Spittell JA Jr. Acute lymphangitis and cellulitis. Diagnosis and therapy . Minn Med 1965 ; 48 : 191 – 194 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 12 Olszewski WL . Episodic dermatolymphangioadenitis (DLA) in patients with lymphoedema of the lower extremities before and after administration of benzathine penicillin: a preliminary study . Lymphology 1996 ; 29 : 126 – 131 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 13 Vignes S , Dupuy A. Recurrence of lymphoedema-associated cellulitis (erysipelas) under prophylactic antibiotherapy: a retrospective cohort study . J Eur Acad Dermatol Venereol 2006 ; 20 : 818 – 822 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 14 Ko DS , Lerner R, Klose G, Cosimi AB. Effective treatment of lymphoedema of the extremities . Arch Surg 1998 ; 133 : 452 – 458 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Földi E Prevention of dermatolymphangioadenitis by combined physiotherapy of the swollen arm after treatment for breast cancer . Lymphology 1996 ; 29 : 48 – 49 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 16 Mihara M , Hara H, Kikuchi K, Yamamoto T, Iida T, Narushima M et al. Scarless lymphatic venous anastomosis for latent and early-stage lymphoedema using indocyanine green lymphography and non-invasive instruments for visualising subcutaneous vein . J Plast Reconstr Aesthet Surg 2012 ; 65 : 1551 – 1558 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Campisi C , Eretta C, Pertile D, Da Rin E, Campisi C, Macciò A et al. Microsurgery for treatment of peripheral lymphoedema: long-term outcome and future perspectives . Microsurgery 2007 ; 27 : 333 – 338 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Koshima I , Nanba Y, Tsutsui T, Takahashi Y, Itoh S, Fujitsu M et al. Minimal invasive lymphaticovenular anastomosis under local anesthesia for leg lymphoedema: is it effective for stage III and IV? Ann Plast Surg 2004 ; 53 : 261 – 266 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Cockett AT , Goodwin WE. Chyluria: attempted surgical treatment by lymphatic–venous anastomosis . J Urol 1962 ; 88 : 566 – 568 . Google Scholar Crossref Search ADS PubMed WorldCat 20 O'Brien BM , Mellow CG, Khazanchi RK, Dvir E, Kumar V, Pederson WC. Long-term results after microlymphaticovenous anastomoses for the treatment of obstructive lymphoedema . Plast Reconstr Surg 1990 ; 85 : 562 – 572 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Mihara M , Hara H, Araki J, Kikuchi K, Narushima M, Yamamoto T et al. Indocyanine green (ICG) lymphography is superior to lymphoscintigraphy for diagnostic imaging of early lymphoedema of the upper limbs . PLoS One 2012 ; 7 : e38182 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Unno N , Nishiyama M, Suzuki M, Yamamoto N, Inuzuka K, Sagara D et al. Quantitative lymph imaging for assessment of lymph function using indocyanine green fluorescence lymphography . Eur J Vasc Endovasc Surg 2008 ; 36 : 230 – 236 . Google Scholar Crossref Search ADS PubMed WorldCat 23 International Society of Lymphology . The diagnosis and treatment of peripheral lymphoedema. Consensus document of the International Society of Lymphology . Lymphology 2003 ; 36 : 84 – 91 . PubMed OpenURL Placeholder Text WorldCat 24 Maegawa J , Mikami T, Yamamoto Y, Satake T, Kobayashi S. Types of lymphoscintigraphy and indications for lymphaticovenous anastomosis . Microsurgery 2010 ; 30 : 437 – 442 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Mihara M , Uchida G, Hara H, Hayashi Y, Narushima M, Iida T et al. Lymphaticovenous anastomosis for facial lymphoedema after multiple courses of therapy for head-and-neck cancer . J Plast Reconstr Aesthet Surg 2011 ; 64 : 1221 – 1225 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Todokoro T , Furniss D, Oda K, Kawana K, Narushima M, Mihara M et al. Effective treatment of pelvic lymphocele by lymphaticovenular anastomosis . Gynecol Oncol 2013 ; 128 : 209 – 214 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Mihara M , Hara H, Narushima M, Mitsui K, Murai N, Koshima I. Low-invasive lymphatic surgery and lymphatic imaging for completely healed intractable pudendal lymphorrhea after gynecologic cancer treatment . J Minim Invasive Gynecol 2012 ; 19 : 658 – 662 . Google Scholar Crossref Search ADS PubMed WorldCat 28 Hara H , Mihara M, Seki Y, Todokoro T, Iida T, Koshima I. Comparison of indocyanine green lymphographic findings with the conditions of collecting lymphatic vessels of limbs in patients with lymphoedema . Plast Reconstr Surg 2013 ; 132 : 1612 – 1618 . Google Scholar Crossref Search ADS PubMed WorldCat 29 Becker C , Vasile JV, Levine JL, Batista BN, Studinger RM, Chen CM et al. Microlymphatic surgery for the treatment of iatrogenic lymphoedema . Clin Plast Surg 2012 ; 39 : 385 – 398 . Google Scholar Crossref Search ADS PubMed WorldCat 30 Gashev AA , Zawieja DC. Hydrodynamic regulation of lymphatic transport and the impact of aging . Pathophysiology 2010 ; 17 : 277 – 287 . Google Scholar Crossref Search ADS PubMed WorldCat 31 Mihara M , Hara H, Hayashi Y, Narushima M, Yamamoto T, Todokoro T et al. Pathological steps of cancer-related lymphedema: histological changes in the collecting lymphatic vessels after lymphadenectomy . PLoS One 2012 ; 7 : e41126 . Google Scholar Crossref Search ADS PubMed WorldCat © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd

de Glas, N A; Jonker, J M; Bastiaannet, E; de Craen, A J M; van de Velde, C J H; Siesling, S; Liefers, G-J; Portielje, J E A; Hamaker, M E

doi: 10.1002/bjs.9616pmid: 25088709

Abstract Background Older patients with breast cancer are often not treated in accordance with guidelines. With the emergence of endocrine therapy, omission of surgery can be considered in some patients. The aim of this population-based study was to investigate time trends in surgical treatment between 1995 and 2011, and to evaluate the effects of omitting surgery on overall and relative survival in older patients with resectable breast cancer. Methods Patients aged 75 years and older with stage I–III breast cancer diagnosed between 1995 and 2011 were selected from the Netherlands Cancer Registry. Time trends of all treatment modalities were evaluated using linear regression models. Changes in overall survival were calculated by Cox regression. Relative survival was calculated using the Ederer II method. Results Overall, 26 292 patients were included. The proportion of patients receiving surgical treatment decreased significantly, from 90·8 per cent in 1995 to 69·9 per cent in 2011 (P < 0·001). Multivariable analysis showed that overall survival did not change over time (hazard ratio 1·00 (95 per cent confidence interval (c.i.) 0·99 to 1·00) per year); nor did relative survival (relative excess risk 1·00 (0·98 to 1·02) per year). Conclusion Omission of surgery has become more common in older patients with breast cancer during the past 15 years in the Netherlands, but this has not altered overall or relative survival. Introduction Owing to demographic changes an increasing proportion of patients with breast cancer will be elderly1,2. The most recent guideline of the International Society of Geriatric Oncology3,4 states that all older patients with resectable breast cancer should receive either a mastectomy or breast-conserving surgery with whole-breast radiotherapy. However, older patients are often not treated in accordance with these guidelines5,6. A previous study7 showed that an increasing number of older patients with breast cancer in the Netherlands did not receive surgical treatment. The most important reasons for omission of surgery were the patient's preference, age, general health status and co-morbidity7. With the emergence of endocrine therapy, omission of surgery might be considered in some patients because the risk of surgical complications as well as the risk of dying from non-cancer-related causes increases with age and number of co-morbidities8–10. Although several clinical trials have investigated tamoxifen as monotherapy compared with surgery in older patients with breast cancer11, extrapolation of the findings of clinical trials to the general breast cancer population, especially older patients, is difficult. Inclusion in trials is often restricted based on co-morbidity and functional status5. However, observational data may provide insight into the effect of omission of surgery among older patients with breast cancer. The aim of this study was to investigate time trends in surgical treatment between 1995 and 2011 in a population-based cohort, and to evaluate the effects of different treatment strategies on overall and relative survival in older patients with resectable breast cancer. Methods Patients were selected from the Netherlands Cancer Registry. This registry identifies patients through the central pathology database. Trained personnel subsequently obtain patient, tumour and treatment characteristics from the patient's medical file. Follow-up status is available through linkage with municipal population registries. All women with stage I–III breast cancer, aged 75 years and older, and diagnosed between 1995 and 2011, were included in the study. Patients with a previous malignancy were excluded, with the exception of those who had a history of basal cell carcinoma of the skin or cervical intraepithelial neoplasia. Oestrogen receptor (ER) and progesterone receptor (PgR) status were available from 2005 onwards. Stage was determined according to the current pathological TNM classification in the year of diagnosis. If histopathological tumour stage was missing, clinical tumour stage was used. Morphology was defined as ductal, lobular or other. The most extensive local and axillary procedures (sentinel node procedure or axillary lymph node dissection) were used for the analyses. Endocrine therapy, chemotherapy and radiotherapy were recoded as yes or no, because details about the specific therapies were lacking. Primary endocrine therapy was defined as endocrine therapy without surgery. Statistical analysis If data were missing, patients were analysed as a separate unknown group within the same variable. The main variable of interest was surgical treatment, and the main outcomes were overall survival and relative survival. Patients and treatment characteristics were assessed using χ2 tests. Time trends of different treatment strategies were assessed by means of linear regression analyses, with the percentage of patients receiving the specific treatment as the outcome and the year of diagnosis as the independent variable. Overall survival was calculated using Cox proportional hazard models. The year of diagnosis was assessed as the independent variable, with adjustments for patient characteristics. As cause of death is not available in the Netherlands Cancer Registry, relative survival over time was assessed by calculating the relative expected risk (RER) using the Ederer II method. This method calculates the ratio of the survival observed among patients with cancer divided by the survival of the corresponding general population, matched by age, sex and year of diagnosis. National life-tables were used to estimate the expected survival. Assessing the effect of omitting surgery by directly comparing treated and not-treated patients using straightforward Cox proportional hazard models would result in bias due to confounding by indication12. Therefore, overall survival was calculated in Cox proportional hazard models using individual patient data with hospital as instrumental variable. This is a factor that determines treatment allocation but is presumably unrelated to the outcome13. The cohort was divided into six groups based on the percentage of patients operated on in the hospital of diagnosis (less than 70, 70–74, 75–79, 80–84, 85–89 and at least 90 per cent). All analyses were adjusted for age, stage, ER status, PgR status and hospital of treatment to account for possible differences in case mix between hospitals. In addition, the analyses were adjusted for type of hospital (university hospital, teaching hospital or other). Differences in relative survival between hospitals were assessed using the Ederer II method as described above. To compare differences in treatment strategies in different age groups, additional sensitivity analyses were performed for patients aged 75–79 years and 80 years or older. All survival analyses were done within the different stages (I, II and III), and a sensitivity analysis was carried out using the percentage of operated patients as a continuous variable. All statistical tests were two-sided, and P < 0·050 was considered significant. All analyses were performed in SPSS® version 20.0 (IBM, Armonk, New York, USA). Results Overall, 26 292 patients aged 75 years or older with breast cancer were included in the study. The age distribution changed over time, reflecting the ageing of the Dutch population; 3401 (48·9 per cent) of 6953 patients were aged 75–79 years between 1995 and 1999, compared with only 1138 (32·8 per cent) of 3473 in 2010–2011 (P < 0·001). Surgery was omitted in 4522 (17·2 per cent) of 26 292 women. Patients who did not receive surgical treatment were generally older, had more advanced tumour stage and more frequently had hormone receptor-positive disease (Table 1). Of 4522 patients who did not have surgery, 3925 (86·8 per cent) received endocrine therapy. Only 509 patients (1·9 per cent) did not receive any form of oncological treatment. Among 11 403 patients who were diagnosed from 2005 onwards with known ER status, radiotherapy was more frequently omitted in patients with ER-positive tumours: 7002 (71·2 per cent) of 9828 compared with 1001 (63·6 per cent) of 1575 patients with ER-negative tumours (P < 0·001). Table 1 Patient characteristics . All patients (n = 26 292) . Operated (n = 21 770) . Not operated (n = 4522) . P† . Time interval < 0·001 1995–1999   6953 (26·4) 6389 (29·3) 564 (12·5) 2000–2004   7757 (29·5) 6705 (30·8) 1052 (23·3) 2005–2009   8109 (30·8) 6234 (28·6) 1875 (41·5) 2010–2011 3473 (13·2) 2442 (11·2) 1031 (22·8) Age (years) < 0·001 75–79 10 425 (39·7) 9876 (45·4) 549 (12·1) 80–84   8890 (33·8) 7621 (35·0) 1269 (28·1) 85–89   5056 (19·2) 3441 (15·8) 1615 (35·7) ≥ 90 1921 (7·3)  832 (3·8)  1089 (24·1) Tumour stage* < 0·001 I 7363 (28·0) 6339 (29·1) 1024 (22·6) II 14 218 (54·1) 12 072 (55·5) 2146 (47·5) III 4393 (16·7) 3308 (15·2) 1085 (24·0) Unknown 318 (1·2)  51 (0·2)  267 (5·9)  ER status < 0·001 Negative 1870 (7·1)  1687 (7·7)  183 (4·0)  Positive 11 633 (44·2) 8839 (40·6) 2794 (61·8) Unknown 12 789 (48·6) 11 244 (51·6) 1545 (34·2) PgR status < 0·001 Negative 4370 (16·6) 3567 (16·4) 803 (17·8) Positive 8780 (33·4) 6735 (30·9) 2045 (45·2) Unknown 13 142 (50·0) 11 468 (52·7) 1674 (37·0) Type of hospital < 0·001 University hospital 1246 (4·7)  1022 (4·7)  224 (5·0)  Teaching hospital 17 327 (65·9) 14 262 (65·5) 3065 (67·8) Other 7702 (29·3) 6479 (29·8) 1223 (27·0) Unknown 17 (0·1)  7 (0·0)  10 (0·2)  . All patients (n = 26 292) . Operated (n = 21 770) . Not operated (n = 4522) . P† . Time interval < 0·001 1995–1999   6953 (26·4) 6389 (29·3) 564 (12·5) 2000–2004   7757 (29·5) 6705 (30·8) 1052 (23·3) 2005–2009   8109 (30·8) 6234 (28·6) 1875 (41·5) 2010–2011 3473 (13·2) 2442 (11·2) 1031 (22·8) Age (years) < 0·001 75–79 10 425 (39·7) 9876 (45·4) 549 (12·1) 80–84   8890 (33·8) 7621 (35·0) 1269 (28·1) 85–89   5056 (19·2) 3441 (15·8) 1615 (35·7) ≥ 90 1921 (7·3)  832 (3·8)  1089 (24·1) Tumour stage* < 0·001 I 7363 (28·0) 6339 (29·1) 1024 (22·6) II 14 218 (54·1) 12 072 (55·5) 2146 (47·5) III 4393 (16·7) 3308 (15·2) 1085 (24·0) Unknown 318 (1·2)  51 (0·2)  267 (5·9)  ER status < 0·001 Negative 1870 (7·1)  1687 (7·7)  183 (4·0)  Positive 11 633 (44·2) 8839 (40·6) 2794 (61·8) Unknown 12 789 (48·6) 11 244 (51·6) 1545 (34·2) PgR status < 0·001 Negative 4370 (16·6) 3567 (16·4) 803 (17·8) Positive 8780 (33·4) 6735 (30·9) 2045 (45·2) Unknown 13 142 (50·0) 11 468 (52·7) 1674 (37·0) Type of hospital < 0·001 University hospital 1246 (4·7)  1022 (4·7)  224 (5·0)  Teaching hospital 17 327 (65·9) 14 262 (65·5) 3065 (67·8) Other 7702 (29·3) 6479 (29·8) 1223 (27·0) Unknown 17 (0·1)  7 (0·0)  10 (0·2)  Values in parentheses are percentages. * Stage was determined according to the current pathological TNM classification in the year of diagnosis. ER, oestrogen receptor; PgR, progesterone receptor. † χ2 test. Open in new tab Table 1 Patient characteristics . All patients (n = 26 292) . Operated (n = 21 770) . Not operated (n = 4522) . P† . Time interval < 0·001 1995–1999   6953 (26·4) 6389 (29·3) 564 (12·5) 2000–2004   7757 (29·5) 6705 (30·8) 1052 (23·3) 2005–2009   8109 (30·8) 6234 (28·6) 1875 (41·5) 2010–2011 3473 (13·2) 2442 (11·2) 1031 (22·8) Age (years) < 0·001 75–79 10 425 (39·7) 9876 (45·4) 549 (12·1) 80–84   8890 (33·8) 7621 (35·0) 1269 (28·1) 85–89   5056 (19·2) 3441 (15·8) 1615 (35·7) ≥ 90 1921 (7·3)  832 (3·8)  1089 (24·1) Tumour stage* < 0·001 I 7363 (28·0) 6339 (29·1) 1024 (22·6) II 14 218 (54·1) 12 072 (55·5) 2146 (47·5) III 4393 (16·7) 3308 (15·2) 1085 (24·0) Unknown 318 (1·2)  51 (0·2)  267 (5·9)  ER status < 0·001 Negative 1870 (7·1)  1687 (7·7)  183 (4·0)  Positive 11 633 (44·2) 8839 (40·6) 2794 (61·8) Unknown 12 789 (48·6) 11 244 (51·6) 1545 (34·2) PgR status < 0·001 Negative 4370 (16·6) 3567 (16·4) 803 (17·8) Positive 8780 (33·4) 6735 (30·9) 2045 (45·2) Unknown 13 142 (50·0) 11 468 (52·7) 1674 (37·0) Type of hospital < 0·001 University hospital 1246 (4·7)  1022 (4·7)  224 (5·0)  Teaching hospital 17 327 (65·9) 14 262 (65·5) 3065 (67·8) Other 7702 (29·3) 6479 (29·8) 1223 (27·0) Unknown 17 (0·1)  7 (0·0)  10 (0·2)  . All patients (n = 26 292) . Operated (n = 21 770) . Not operated (n = 4522) . P† . Time interval < 0·001 1995–1999   6953 (26·4) 6389 (29·3) 564 (12·5) 2000–2004   7757 (29·5) 6705 (30·8) 1052 (23·3) 2005–2009   8109 (30·8) 6234 (28·6) 1875 (41·5) 2010–2011 3473 (13·2) 2442 (11·2) 1031 (22·8) Age (years) < 0·001 75–79 10 425 (39·7) 9876 (45·4) 549 (12·1) 80–84   8890 (33·8) 7621 (35·0) 1269 (28·1) 85–89   5056 (19·2) 3441 (15·8) 1615 (35·7) ≥ 90 1921 (7·3)  832 (3·8)  1089 (24·1) Tumour stage* < 0·001 I 7363 (28·0) 6339 (29·1) 1024 (22·6) II 14 218 (54·1) 12 072 (55·5) 2146 (47·5) III 4393 (16·7) 3308 (15·2) 1085 (24·0) Unknown 318 (1·2)  51 (0·2)  267 (5·9)  ER status < 0·001 Negative 1870 (7·1)  1687 (7·7)  183 (4·0)  Positive 11 633 (44·2) 8839 (40·6) 2794 (61·8) Unknown 12 789 (48·6) 11 244 (51·6) 1545 (34·2) PgR status < 0·001 Negative 4370 (16·6) 3567 (16·4) 803 (17·8) Positive 8780 (33·4) 6735 (30·9) 2045 (45·2) Unknown 13 142 (50·0) 11 468 (52·7) 1674 (37·0) Type of hospital < 0·001 University hospital 1246 (4·7)  1022 (4·7)  224 (5·0)  Teaching hospital 17 327 (65·9) 14 262 (65·5) 3065 (67·8) Other 7702 (29·3) 6479 (29·8) 1223 (27·0) Unknown 17 (0·1)  7 (0·0)  10 (0·2)  Values in parentheses are percentages. * Stage was determined according to the current pathological TNM classification in the year of diagnosis. ER, oestrogen receptor; PgR, progesterone receptor. † χ2 test. Open in new tab In general, most operated patients did not receive axillary surgery: 19 749 (75·1 per cent) of 26 292 (Table S1, supporting information). Some 5427 women (20·6 per cent) had breast-conserving surgery and 586 had both breast-conserving surgery and mastectomy; 4523 (75·2 per cent) of these 6013 women underwent adjuvant radiotherapy. Most patients were treated with adjuvant endocrine therapy (15 305, 58·2 per cent), and 272 (1·0 per cent) received adjuvant chemotherapy. There was a large variation in surgery rates between the three hospital types: 69·3–92·6 per cent in university hospitals, 68·1–94·3 per cent in teaching hospitals and 70·8–93·1 per cent in other hospital types. Time trends of treatment Fig. 1 shows time trends for all treatments between 1995 and 2011. The percentage of patients receiving surgical treatment decreased significantly; 1186 (90·8 per cent) of 1306 patients were treated surgically in 1995 compared with 1214 (69·9 per cent) of 1737 in 2011 (P < 0·001). The decrease was similar in all three hospital types (data not shown). Fig. 1 Open in new tabDownload slide Time trends of different treatments according to age group: a surgical treatment, b primary endocrine therapy, c radiotherapy after breast-conserving surgery (BCS) and d any axillary surgery among operated patients. P < 0·001 for all treatments and age groups (linear regression analysis) The proportion of patients having primary endocrine therapy increased in the interval evaluated, from 93 (7·1 per cent) of 1306 in 1995 to 475 (27·3 per cent) of 1737 in 2011 (P < 0·001). This increase was more pronounced among the oldest patients. In patients aged 75–79 years the use of adjuvant endocrine therapy increased from 252 (39·6 per cent) of 636 patients in 1995 to 311 (56·1 per cent) of 554 in 2011 (P < 0·001); for patients aged 80–84 years it rose from 160 (37·6 per cent) of 426 in 1995 to 266 (47·8 per cent) of 557 in 2011 (P = 0·022). The percentage of patients receiving chemotherapy did not change over time for any age group (data not shown). The use of radiotherapy after breast-conserving surgery increased in all patients: 165 (66·0 per cent) of 250 women received radiotherapy in 1995, compared with 353 (81·3 per cent) of 434 in 2011 (P < 0·001). Furthermore, the percentage of operated patients who received axillary surgery increased over time, from none of 217 in 1995 to 330 (81·3 per cent) of 406 in 2011 (P < 0·001). Survival analyses Follow-up was complete until 31 December 2011. Median follow-up was 3·7 (range 0–16·9) years for women diagnosed between 1995 and 2011, and 4·0 (0–16·9) years for those who were still alive at the end of 2011. Multivariable analysis showed that overall survival did not change between 1995 and 2011 (hazard ratio (HR) 1·00 (95 per cent confidence interval (c.i.) 0·99 to 1·00) per year; P = 0·171) (Fig. 2). Overall survival improved slightly in patients aged 75–79 years (HR 0·99 (0·98 to 1·00) per year; P = 0·030), and decreased somewhat in patients aged 90 years or more (HR 1·03 (1·00 to 1·05) per year; P = 0·026). Fig. 2 Open in new tabDownload slide Five-year overall survival over time according to age group. Hazard ratio (95 per cent confidence interval) per year: 1·00 (0·99 to 1·00) for all patients (P = 0·171), 0·99 (0·98 to 1·00) for patients aged 75–79 years (P = 0·030), 0·99 (0·98 to 1·00) for patients aged 80–84 years (P = 0·089), 1·00 (0·99 to 1·02) for patients aged 85–89 years (P = 0·686) and 1·03 (1·00 to 1·05) for patients aged at least 90 years (P = 0·026) (Cox proportional hazards model) Multivariable analysis revealed that relative survival did not change during the included years (RER 1·00 (95 per cent c.i. 0·98 to 1·02) per year; P = 0·830). Similarly, stratified analyses showed no changes in relative survival for any age group (Fig. 3). Fig. 3 Open in new tabDownload slide Five-year relative survival over time according to age group. Relative excess risk (95 per cent confidence interval) per year: 1·00 (0·98 to 1·02) for all patients (P = 0·830), 0·99 (0·97 to 1·02) for patients aged 75–79 years (P = 0·524), 1·00 (0·96 to 1·03) for patients aged 80–84 years (P = 0·797) and 1·02 (0·98 to 1·07) for patients aged at least 85 years (P = 0·345) (Ederer II method) The percentage of operated patients in the treating hospital did not influence overall survival; in multivariable analysis the HR was 0·86 (95 per cent c.i. 0·66 to 1·12) for hospitals that operated on at least 90 per cent of patients compared with hospitals that operated on less than 70 per cent (P = 0·088) (Table 2). Additional sensitivity analyses comparing overall survival rates between the hospital types within different age groups showed similar results for both age groups (data not shown). Relative mortality did not differ between hospitals in the multivariable analysis; the RER was 1·02 (95 per cent c.i. 0·75 to 1·41) for hospitals that operated on 90 per cent or more of patients compared with hospitals that operated on less than 70 per cent (P = 0·347) (Table 3). Table 2 Overall survival analyses using hospital as instrumental variable Hospital (% operated) . No. of events . Univariable analysis . Multivariable adjusted analysis* . Hazard ratio . P . Hazard ratio . P . < 70  593 1·00 (reference)  0·089 1·00 (reference)  0·088 70–74 1220 1·02 (0·93, 1·13) 0·97 (0·75, 1·25) 75–79 2853 0·93 (0·85, 1·02) 1·11 (0·87, 1·42) 80–84 5584 0·96 (0·88, 1·05) 0·89 (0·69, 1·17) 85–89 3259 0·98 (0·90, 1·07) 0·94 (0·69, 1·27) ≥ 90 1770 0·95 (0·87, 1·05) 0·86 (0·66, 1·12) Hospital (% operated) . No. of events . Univariable analysis . Multivariable adjusted analysis* . Hazard ratio . P . Hazard ratio . P . < 70  593 1·00 (reference)  0·089 1·00 (reference)  0·088 70–74 1220 1·02 (0·93, 1·13) 0·97 (0·75, 1·25) 75–79 2853 0·93 (0·85, 1·02) 1·11 (0·87, 1·42) 80–84 5584 0·96 (0·88, 1·05) 0·89 (0·69, 1·17) 85–89 3259 0·98 (0·90, 1·07) 0·94 (0·69, 1·27) ≥ 90 1770 0·95 (0·87, 1·05) 0·86 (0·66, 1·12) Values in parentheses are 95 per cent confidence intervals. * Adjusted for age, tumour stage, oestrogen receptor and progesterone receptor status, and hospital of treatment. Cox proportional hazards models were used for analysis. Open in new tab Table 2 Overall survival analyses using hospital as instrumental variable Hospital (% operated) . No. of events . Univariable analysis . Multivariable adjusted analysis* . Hazard ratio . P . Hazard ratio . P . < 70  593 1·00 (reference)  0·089 1·00 (reference)  0·088 70–74 1220 1·02 (0·93, 1·13) 0·97 (0·75, 1·25) 75–79 2853 0·93 (0·85, 1·02) 1·11 (0·87, 1·42) 80–84 5584 0·96 (0·88, 1·05) 0·89 (0·69, 1·17) 85–89 3259 0·98 (0·90, 1·07) 0·94 (0·69, 1·27) ≥ 90 1770 0·95 (0·87, 1·05) 0·86 (0·66, 1·12) Hospital (% operated) . No. of events . Univariable analysis . Multivariable adjusted analysis* . Hazard ratio . P . Hazard ratio . P . < 70  593 1·00 (reference)  0·089 1·00 (reference)  0·088 70–74 1220 1·02 (0·93, 1·13) 0·97 (0·75, 1·25) 75–79 2853 0·93 (0·85, 1·02) 1·11 (0·87, 1·42) 80–84 5584 0·96 (0·88, 1·05) 0·89 (0·69, 1·17) 85–89 3259 0·98 (0·90, 1·07) 0·94 (0·69, 1·27) ≥ 90 1770 0·95 (0·87, 1·05) 0·86 (0·66, 1·12) Values in parentheses are 95 per cent confidence intervals. * Adjusted for age, tumour stage, oestrogen receptor and progesterone receptor status, and hospital of treatment. Cox proportional hazards models were used for analysis. Open in new tab Table 3 Relative survival analyses using hospital as instrumental variable Hospital (% operated) . 5-year relative survival . Univariable analysis . Multivariable adjusted analysis* . RER . P . RER . P . < 70 84·8 1·00 (reference)  0·168 1·00 (reference)  0·347 70–74 81·1 1·32 (0·91, 1·91) 1·28 (0·93, 1·76) 75–79 83·9 0·99 (0·70, 1·41) 1·10 (0·81, 1·49) 80–84 82·5 1·09 (0·78, 1·53) 1·11 (0·83, 1·56) 85–89 82·4 1·13 (0·80, 1·60) 1·16 (0·86, 1·56) ≥ 90 84·8 1·00 (0·69, 1·45) 1·02 (0·75, 1·41) Hospital (% operated) . 5-year relative survival . Univariable analysis . Multivariable adjusted analysis* . RER . P . RER . P . < 70 84·8 1·00 (reference)  0·168 1·00 (reference)  0·347 70–74 81·1 1·32 (0·91, 1·91) 1·28 (0·93, 1·76) 75–79 83·9 0·99 (0·70, 1·41) 1·10 (0·81, 1·49) 80–84 82·5 1·09 (0·78, 1·53) 1·11 (0·83, 1·56) 85–89 82·4 1·13 (0·80, 1·60) 1·16 (0·86, 1·56) ≥ 90 84·8 1·00 (0·69, 1·45) 1·02 (0·75, 1·41) Values in parentheses are 95 per cent confidence intervals. * Adjusted for age, tumour stage, oestrogen receptor and progesterone receptor status, and hospital of treatment. Relative excess risk (RER) was calculated using the Ederer II method. Open in new tab Table 3 Relative survival analyses using hospital as instrumental variable Hospital (% operated) . 5-year relative survival . Univariable analysis . Multivariable adjusted analysis* . RER . P . RER . P . < 70 84·8 1·00 (reference)  0·168 1·00 (reference)  0·347 70–74 81·1 1·32 (0·91, 1·91) 1·28 (0·93, 1·76) 75–79 83·9 0·99 (0·70, 1·41) 1·10 (0·81, 1·49) 80–84 82·5 1·09 (0·78, 1·53) 1·11 (0·83, 1·56) 85–89 82·4 1·13 (0·80, 1·60) 1·16 (0·86, 1·56) ≥ 90 84·8 1·00 (0·69, 1·45) 1·02 (0·75, 1·41) Hospital (% operated) . 5-year relative survival . Univariable analysis . Multivariable adjusted analysis* . RER . P . RER . P . < 70 84·8 1·00 (reference)  0·168 1·00 (reference)  0·347 70–74 81·1 1·32 (0·91, 1·91) 1·28 (0·93, 1·76) 75–79 83·9 0·99 (0·70, 1·41) 1·10 (0·81, 1·49) 80–84 82·5 1·09 (0·78, 1·53) 1·11 (0·83, 1·56) 85–89 82·4 1·13 (0·80, 1·60) 1·16 (0·86, 1·56) ≥ 90 84·8 1·00 (0·69, 1·45) 1·02 (0·75, 1·41) Values in parentheses are 95 per cent confidence intervals. * Adjusted for age, tumour stage, oestrogen receptor and progesterone receptor status, and hospital of treatment. Relative excess risk (RER) was calculated using the Ederer II method. Open in new tab Discussion This study shows that the proportion of older patients with stage I–III breast cancer who did not receive surgical treatment has increased considerably over the past 15 years, as has the proportion who received primary endocrine therapy. However, overall and relative survival did not change over this interval. No differences in overall and relative survival were observed between treatment strategies of different hospitals. Currently, the Dutch guideline14 recommends either a mastectomy or breast-conserving surgery followed by radiotherapy in all patients with resectable breast cancer. This study shows that this guideline is not being adhered to in an increasing number of patients. These findings may be explained by increased patient participation in clinical decision-making, as a previous study7 showed that surgery is most often omitted owing to the patient's preference. A previous meta-analysis11, which evaluated clinical trials investigating tamoxifen as monotherapy compared with surgical treatment, showed no difference in overall survival despite somewhat poorer locoregional control11. This finding may have influenced daily clinical practice. Interestingly, overall and relative survival rates have not changed in the past 15 years, except in patients aged 75–79 years for whom overall survival has improved marginally. This is most likely due to an increased life expectancy of the general population, as the relative survival of this group did not improve and the findings are thus of dubious clinical significance. Analyses that compared different treatment strategies between hospitals showed no differences in overall and relative survival. Thus the findings of this study imply that the current therapies, including the omission of surgery, use of primary and adjuvant endocrine therapy and the increase in axillary surgery, have not influenced overall and relative survival in this patient group. Several factors may have contributed to negating the effects of omitting surgical treatment. The most important factor is most likely the increased use of adjuvant endocrine therapy, even though only 40–60 per cent of patients received such therapy. Further, the use of axillary surgery has increased over time, most likely as a result of the gradual adoption of the sentinel node procedure. The additional information on tumour stage that this provides may have contributed to improved decision-making concerning adjuvant therapy. An increasing number of patients who received primary endocrine treatment were probably more likely to be treated with aromatase inhibitors instead of tamoxifen, although this was not evaluated in the present study. Aromatase inhibitors have been demonstrated to be superior to tamoxifen in a neoadjuvant setting15,16, and could potentially be superior as monotherapy. However, the ESTEeM trial17,18, which set out to compare surgery with aromatase inhibitors as primary treatment for patients aged 75 years and older, was closed prematurely owing to poor patient recruitment. Another likely explanation for the absence of differences in survival rates between different treatment strategies may be that the likelihood of dying from causes other than breast cancer increases with age and co-morbidity8,19. Nonetheless, the median life expectancy for an 85-year-old woman in the Netherlands is 6·6 years20, leaving sufficient time to suffer the consequences of inadequately treated breast cancer. As overall survival has improved for younger patients with breast cancer in recent years21, one might argue that older patients with breast cancer are generally undertreated compared with their younger counterparts; in a large proportion of the older patients who did undergo surgery, axillary staging and adjuvant endocrine treatment were omitted. This is a worrying alternative conclusion based on the results of the present investigation. In agreement, a large proportion of the oldest patients who were treated with breast-conserving surgery did not receive subsequent radiotherapy, possibly because of fear of toxicity. Functional outcomes should be taken into consideration when making treatment decisions for older patients22. Omission of surgery may lead to poor locoregional control11, which may result in poor quality of life and a decline in functional status. Quality-of-life assessments may help to improve individualized clinical decision-making in older patients23–25. Ideally, guidelines for treatment of older patients should be based on results from clinical trials. However, clinical trials generally include only fit older patients5, and as a result extrapolation of their findings to the general older population is not possible. This observational study assessed the effects of surgery on overall and relative survival by evaluating treatment strategies and their effects on survival over time, and subsequently by using the different hospitals as instrumental variable. This methodological approach is a major strength of this paper. Another strength of the study is the use of data from the Netherlands Cancer Registry, as it provides well registered, reliable data for large numbers of patients. This study has some important limitations. First, the proportion of patients who received surgery did not differ substantially between the different hospitals. However, this study involves a very large number of patients, meaning that even small differences in survival would have emerged. By evaluating overall and relative survival over time, it was possible to investigate the effects of omitting surgery on these outcomes. A more important weakness is that hospital was arguably not the most reliable instrumental variable, as it might be related to patient outcomes. It was not possible to assess breast cancer-specific survival as causes of death were not available. However, it is well known from previous studies that relative survival can be used as a valid proxy for cancer-specific survival26. This method is particularly useful in the older population as determining cause of death can be notoriously difficult in this group, as unexpected or unexplained deaths may be attributed incorrectly to the cancer27,28. An important consideration is that increased omission of surgery was not the only treatment strategy that changed over time, which makes it impossible to attribute survival trends to the omission of surgery alone. Data on specific tumour biology were not available for patients diagnosed before 2005, but it is unlikely that the tumour biology of the entire population has changed appreciably over the years. Finally, detailed patient and treatment characteristics, such as co-morbidities, functional status, secondary treatments, disease recurrence and locoregional disease control, were not available for analysis. In the context of these limitations, this study showed that omission of surgery and the use of primary endocrine therapy have increased in older patients with resectable breast cancer in the past 15 years, while both relative and overall survival have remained unchanged. Acknowledgements The authors thank the registration teams of the Comprehensive Cancer Centres for the collection of data for the Netherlands Cancer Registry and the scientific staff of the Netherlands Cancer Registry. Disclosure: The authors declare no conflict of interest. Snapshot quiz 14/14 References 1 Cancer Research UK . Breast Cancer Incidence Statistics . 2013 . http://www.cancerresearchuk.org/ [accessed 13 March 2013]. 2 DeSantis C , Siegel R, Bandi P, Jemal A. Breast cancer statistics, 2011 . CA Cancer J Clin 2011 ; 61 : 409 – 418 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Wildiers H , Kunkler I, Biganzoli L, Fracheboud J, Vlastos G, Bernard-Marty C et al. Management of breast cancer in elderly individuals: recommendations of the International Society of Geriatric Oncology . Lancet Oncol 2007 ; 8 : 1101 – 1115 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Biganzoli L , Wildiers H, Oakman C, Marotti L, Loibl S, Kunkler I et al. Management of elderly patients with breast cancer: updated recommendations of the International Society of Geriatric Oncology (SIOG) and European Society of Breast Cancer Specialists (EUSOMA) . Lancet Oncol 2012 ; 13 : e148 – e160 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Zulman DM , Sussman JB, Chen X, Cigolle CT, Blaum CS, Hayward RA. Examining the evidence: a systematic review of the inclusion and analysis of older adults in randomized controlled trials . J Gen Intern Med 2011 ; 26 : 783 – 790 . Google Scholar Crossref Search ADS PubMed WorldCat 6 van de Water W , Bastiaannet E, Dekkers OM, de Craen AJ, Westendorp RG, Voogd AC et al. Adherence to treatment guidelines and survival in patients with early-stage breast cancer by age at diagnosis . Br J Surg 2012 ; 99 : 813 – 820 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Hamaker ME , Bastiaannet E, Evers D, Wv Water, Smorenburg CH, Maartense E et al. Omission of surgery in elderly patients with early stage breast cancer . Eur J Cancer 2013 ; 49 : 545 – 552 . Google Scholar Crossref Search ADS PubMed WorldCat 8 van de Water W , Markopoulos C, van de Velde CJ, Seynaeve C, Hasenburg A, Rea D et al. Association between age at diagnosis and disease-specific mortality among postmenopausal women with hormone receptor-positive breast cancer . JAMA 2012 ; 307 : 590 – 597 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 9 Dehal A , Abbas A, Johna S. Comorbidity and outcomes after surgery among women with breast cancer: analysis of nationwide in-patient sample database . Breast Cancer Res Treat 2013 ; 139 : 469 – 476 . Google Scholar Crossref Search ADS PubMed WorldCat 10 de Glas NA , Kiderlen M, Bastiaannet E, de Craen AJ, van de Water W, van de Velde CJ et al. Postoperative complications and survival of elderly breast cancer patients: a FOCUS study analysis . Breast Cancer Res Treat 2013 ; 138 : 561 – 569 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Hind D , Wyld L, Reed MW. Surgery, with or without tamoxifen, vs tamoxifen alone for older women with operable breast cancer: Cochrane review . Br J Cancer 2007 ; 96 : 1025 – 1029 . Google Scholar Crossref Search ADS PubMed WorldCat 12 de Glas NA , Kiderlen M, Liefers GJ. Survival after lumpectomy and mastectomy for early stage invasive breast cancer: the effect of age and hormone receptor status . Cancer 2013 ; 119 : 3253 – 3254 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Vandenbroucke JP . When are observational studies as credible as randomised trials? Lancet 2004 ; 363 : 1728 – 1731 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Nederland Nationaal Borstkanker Overleg . Breast Cancer. Dutch Guideline, Version 2.0 ; 2012 . http://www.oncoline.nl/uploaded/docs/mammacarcinoom/Dutch%20Breast%20Cancer%20Guideline%202012.pdf [accessed 29 May 2013]. 15 Eiermann W , Paepke S, Appfelstaedt J, Llombart-Cussac A, Eremin J, Vinholes J et al. Preoperative treatment of postmenopausal breast cancer patients with letrozole: a randomized double-blind multicenter study . Ann Oncol 2001 ; 12 : 1527 – 1532 . Google Scholar Crossref Search ADS PubMed WorldCat 16 Cataliotti L , Buzdar AU, Noguchi S, Bines J, Takatsuka Y, Petrakova K et al. Comparison of anastrozole versus tamoxifen as preoperative therapy in postmenopausal women with hormone receptor-positive breast cancer: the Pre-Operative ‘Arimidex’ Compared to Tamoxifen (PROACT) trial . Cancer 2006 ; 106 : 2095 – 2103 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Reed MW , Wyld L, Ellis P, Bliss J, Leonard R. Breast cancer in older women: trials and tribulations . Clin Oncol (R Coll Radiol) 2009 ; 21 : 99 – 102 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Cancer Research UK . A Trial Looking at Anastrozole With or Without Surgery for Older Women with Breast Cancer (ESTEeM) ; 2013 . http://www.cancerresearchuk.org/ [accessed 14 May 2013]. 19 Land LH , Dalton SO, Jensen MB, Ewertz M. Impact of comorbidity on mortality: a cohort study of 62 591 Danish women diagnosed with early breast cancer, 1990–2008 . Breast Cancer Res Treat 2012 ; 131 : 1013 – 1020 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Statistics Netherlands . Levensverwachting; geslacht en leeftijd, vanaf 1950 (per jaar) ; 2013 . http://statline.cbs.nl/StatWeb/publication/?DM=SLNL&PA=37360ned&D1=3&D2=a&D3=0,11,21,31,41,61,81&D4=0&VW=T [accessed 1 July 2013]. 21 Bastiaannet E , Portielje JE, van de Velde CJ, de Craen AJ, van der Velde S, Kuppen PJ et al. Lack of survival gain for elderly women with breast cancer . Oncologist 2011 ; 16 : 415 – 423 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Yellen SB , Cella DF, Leslie WT. Age and clinical decision making in oncology patients . J Natl Cancer Inst 1994 ; 86 : 1766 – 1770 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Puts MT , Hardt J, Monette J, Girre V, Springall E, Alibhai SM. Use of geriatric assessment for older adults in the oncology setting: a systematic review . J Natl Cancer Inst 2012 ; 104 : 1133 – 1163 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Kenis C , Bron D, Libert Y, Decoster L, Van PK, Scalliet P et al. Relevance of a systematic geriatric screening and assessment in older patients with cancer: results of a prospective multicentric study . Ann Oncol 2013 ; 24 : 1306 – 1312 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Pallis AG , Fortpied C, Wedding U, Van Nes MC, Penninckx B, Ring A et al. EORTC elderly task force position paper: approach to the older cancer patient . Eur J Cancer 2010 ; 46 : 1502 – 1513 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Dignam JJ , Huang L, Ries L, Reichman M, Mariotto A, Feuer E. Estimating breast cancer-specific and other-cause mortality in clinical trial and population-based cancer registry cohorts . Cancer 2009 ; 115 : 5272 – 5283 . Google Scholar Crossref Search ADS PubMed WorldCat 27 Hu CY , Xing Y, Cormier JN, Chang GJ. Assessing the utility of cancer-registry-processed cause of death in calculating cancer-specific survival . Cancer 2013 ; 119 : 1900 – 1907 . Google Scholar Crossref Search ADS PubMed WorldCat 28 Goldoni CA , Bonora K, Ciatto S, Giovannetti L, Patriarca S, Sapino A et al. Misclassification of breast cancer as cause of death in a service screening area . Cancer Causes Control 2009 ; 20 : 533 – 538 . Google Scholar Crossref Search ADS PubMed WorldCat Author notes Presented to the Annual Meeting of the International Society of Geriatric Oncology, Copenhagen, Denmark, October 2013; published in abstract form as J Geriatr Oncol 2013; 4(Suppl 1): S17 © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2014 BJS Society Ltd. Published by John Wiley & Sons Ltd