Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Cancer survival in England and the influence of early diagnosis: what can we learn from recent EUROCARE results?

Cancer survival in England and the influence of early diagnosis: what can we learn from recent... British Journal of Cancer (2009) 101, S102 – S109 & 2009 Cancer Research UK All rights reserved 0007 – 0920/09 $32.00 www.bjcancer.com Discussion Paper Cancer survival in England and the influence of early diagnosis: what can we learn from recent EUROCARE results? ,1 2,3 CS Thomson and D Forman 1 2 Cancer Research UK, Statistical Information Team, 61 Lincoln’s Inn Fields, London WC2A 3PX, UK; Centre for Epidemiology and Biostatistics, University of Leeds, Beckett Street, Leeds, UK; National Cancer Intelligence Network, Queens House, 56 Lincoln’s Inn Fields, London, UK BACKGROUND: This review of the EUROCARE-4 results attempts to separate out the early and late mortality effects contributing to the widely reported poorer 5-year survival rates for cancer patients in the United Kingdom compared with other European countries for 26 cancer sites. METHODS: Patients diagnosed with cancer in 1996–1999 in 23 European countries were included in the analyses. Comparison of 1-year, 5-year and 5|1-year (i.e. only including those patients who had survived to 1 year) survival estimates between data for England and the ‘European average’ was undertaken. This analysis was to highlight the relative contribution of early diagnosis, using 1-year survival as a proxy measure, on 5-year survival for the different sites of cancer. Three groups of cancer sites were identified according to whether the survival differences at 1, 5 and 5|1 years were statistically significant. RESULTS AND CONCLUSIONS: Breast cancer showed significantly poorer 1- and 5-year survival estimates in England, but the 5|1-year survival figure was not significantly different. Thus, successful initiatives around awareness and early detection could eradicate the survival gap. In contrast, the 5|1-year survival estimates remained significantly worse for lung, colorectal and prostate cancers, showing that although early detection could make some difference, late effects such as treatment and management of the patients were also influencing long-term outcome differences between England and Europe. British Journal of Cancer (2009) 101, S102 – S109. doi:10.1038/sj.bjc.6605399 www.bjcancer.com & 2009 Cancer Research UK Keywords: EUROCARE; cancer survival in England; National Awareness and Early Diagnosis Initiative (NAEDI) It has been widely reported that 5-year survival rates for cancer Initiative (NAEDI). One objective of NAEDI is to review the patients in the United Kingdom have, in general, lagged behind evidence on links between early diagnosis and survival and comparator European countries (and the ‘European average’) since evaluate the extent to which international differences in survival the late 1970s (Berrino et al, 1995b, 1999, 2003, 2007). The NHS are due to differences in delays in patient presentation; in Cancer Plan (Department of Health, 2000) highlighted this survival practitioner referrals; in the availability and quality of diagnostic ‘gap’ and ‘catching up with Europe’ has been a major driver in techniques used; and in the management and care of patients once formulating national cancer policy. diagnosed. High-resolution studies of a few specific sites of cancer Use of 1-year survival as a proxy for early/late diagnosis, and undertaken within some of the registries included in EURO- hence for stage data, has been advocated in the CRS and CARE-2, and in which additional data on stage and diagnostic comparison of this outcome measure across Europe may help techniques were available, suggested that later stage at presenta- benchmark the current disparities. However, the appropriateness tion among UK patients was a major factor explaining the poorer of this measure will vary by cancer site and early/late presentation survival. These data are, however, now quite old; they relate to is not the only possible reason for observed differences. For cancers diagnosed in the 1980s and have not been updated (Gatta example, for cancers such as oesophageal and pancreatic with a et al, 2000; Sant et al, 2003, 2007). generally poor prognosis, the introduction of chemotherapeutic The most recently published EUROCARE data, based on new regimens for advanced disease will have improved 1-year survival diagnoses between 1995 and 1999, indicate that the survival gap rates for patients irrespective of the proportion of early stage between England and the ‘European average’ had diminished but tumours in the populations (Rao and Cunningham, 2008; Starling had not been completely removed (Berrino et al, 2007; Richards, and Cunningham, 2008; Mitry et al, 2008a, b). Also factors, such as 2007; Verdecchia et al, 2007). As a consequence, the 2007 Cancer differences in underlying comorbidity, will impact on short-term Reform Strategy (CRS) maintained an objective of bringing about survival but are rarely adjusted for in comparative survival service improvements to reduce further the difference in survival estimates. between England and Europe (Department of Health, 2007). Use of 5-year survival data conditional on surviving 1 year (i.e. The CRS set out several initiatives to improve survival including 1–5 year or 5|1 survival) has been presented in the most recent establishment of the National Awareness and Early Diagnosis EUROCARE-4 publications (Sant et al, 2009). Examination of this along side 1-year survival for different cancer sites shows which sites had variation between the countries at 1 year, which was not present at 5 years after those who died within the first year had *Correspondence: CS Thomson; been excluded from the analyses. To some extent, this enables E-mail: [email protected] What can we learn from recent EUROCARE results? CS Thomson and D Forman S103 separation of the effects of stage distribution, significant comor- countries taking part in the study, and 30% of the total EU bidity and either peri-operative (or postoperative treatment- population (excluding Norway, Switzerland and Iceland, which are related) mortality, shown generally by the 1-year survival results, not in the EU). from the effects of treatment and patient management on survival, The methodology used to derive the ‘European average’ was shown by differences in longer-term survival. This assessment is neither a simple country-specific weighted average based on fundamental to the NAEDI objective of trying to identify cancer coverage, nor one based on population, because either of these sites where delays in diagnosis have had the biggest impact on averages would have been heavily biased. Instead, regional-specific survival and which could be targeted to reduce such delays in the survival estimates were obtained by splitting Europe into five areas future. and pooling the data available for the regions. This makes the This paper analyses the most recent EUROCARE data for the assumption that the survival for the population covered by more common sites of cancer, comparing outcomes in England registration was representative of the unknown survival of the with the EUROCARE-4 ‘European average’ at 5, 1 and 5|1 years whole region. Weighting was then applied to each of these five after diagnosis, and also looks at recent trends in 1-year survival regional estimates using the mean population for those countries data within England. This provides a means of beginning to included in EUROCARE-4 for each region. identify cancer sites for which the poor outcomes in England can The method of creating the ‘European average’ led to the United be attributed to later stage at diagnosis. Kingdom and Ireland region making up 14.6% of the ‘European average’. Thus, comparison of the survival estimates for England (or indeed any other country) with the survival estimate for the ‘European average’ is not strictly straightforward, due to the MATERIALS AND METHODS non-independence between them. However, it was assumed that The EUROCARE project has been running since 1990 (Berrino the covariance (a measure of the non-independence) between the et al, 1995a). The latest published data cover patients diagnosed England value and the ‘European average’ would not be large between 1996 and 1999 and are based on 2.8 million adults (aged enough to greatly affect interpretation of statistical significance, 15–99) who were diagnosed with cancer in 23 European countries and as such, the difference between the survival rates for England (Sant et al, 2009). Data for 26 of the more common sites or site and the ‘European average’ for each of the 5-year, 1-year and groups of cancer from the EUROCARE-4 study have been 5|1-year estimates was obtained, along with an estimate of its presented. The coding classification used in EUROCARE-4 is the standard error (s.e.). The standard error for the survival difference third revision of the International Classification of Diseases for was calculated by taking the square root of the sum of the squared Oncology (ICD-O3). standard errors obtained for each of the survival estimates for The data presented here from the EUROCARE-4 study are the England and the ‘European average’, ignoring the covariance figures for England and the ‘European average’, which have both between the two cohorts. been appropriately standardised for cancer survival analyses. This was to take into account any differences in age structures between the countries, and four different standards are chosen depending RESULTS on the cancer site of choice. The International Cancer Survival Standard (ICSS; Corazziari et al, 2004) was applied to the Table 1 shows results for England and the ‘European average’ for 5, EUROCARE-4 data, except for prostate cancer (De Angelis et al, 1 and 5|1-year survival results for 26 of the more common sites or 2009; pp 20). This differs from previous EUROCARE studies, site groups of cancer included in the EUROCARE-4 study. Five- which used internally weighted average age structure for Europe year survival was significantly worse in England for 14 of these for that study. The ICSS gives standard weights, which should cancers, including the four major sites, breast, lung, colorectal and make future comparisons easier. prostate, together with cancers of the oesophagus, stomach, liver, Data from recent analyses within England (Rachet et al, 2009) pancreas, cervix, ovary, kidney, brain, thyroid and multiple for 20 cancer sites or site groupings have also been included. The myeloma. Five-year survival was significantly better in England classification used for this study was the tenth revision of the ICD- for head and neck cancers (including the group of oral cancers) 10, which is not entirely comparable with those coded using ICD- and malignant melanoma; and not significantly different for the O3. These data were not originally age standardised although remaining 10 sites. equivalent age-standardised rates were made available to us All of the 14 cancers showing a statistically significant worse (personal communication, B Rachet, 2009). The age-standardised outcome in England at 5 years also showed a significantly worse data were not reported in the original paper because the authors outcome at 1 year. A further five cancers – bone and cartilage, soft had found that the age distributions of the cancer patients had tissue, uterus, bladder and non-Hodgkin lymphoma – also showed a changed little over the 11 years they examined. However, the significantly worse 1-year survival, but this was not maintained at 5 weights used in the age standardisation for the recent study in years. Of the 14 cancers with significantly worse 5-year outcomes, England (Rachet et al, 2009) were different to those used in the eight also showed a significantly worse 5|1-year survival – lung, EUROCARE-4 publications. Instead of the ICSS weighting system, colorectal and prostate together with stomach, ovary, kidney and Rachet et al (2009) used the ‘England Standard’, which is based on thyroid cancers and multiple myeloma. The other six cancers had the age distributions of cancer patients diagnosed between 1986 5|1-year survival that was no longer significantly different from the and 1990 in England (Coleman et al, 1999; pp 50). This was to ‘European average’. This group included breast cancer together with maintain comparability with previous survival analyses within oesophagus, liver, pancreas, cervix and brain cancers. England and used a different set of weights for each of the 20 Table 2 shows the 14 cancers with significantly worse 5-year cancer sites. outcomes ordered by the magnitude of the 5-year ‘survival gap’ The ‘European average’ survival rates reported are not strictly between England and the ‘European average’ together with the true averages across the whole of Europe. This is because only 13 corresponding survival gap at 1 year and 5|1 years. Kidney cancer of the 23 countries represented in EUROCARE-4 have national showed the largest survival gap, being 12.4% worse than the cancer registration and, hence, 100% coverage. The range of ‘European average’ at 5 years, while four other sites had gaps that coverage for the other countries was from 1.3% of the adult were larger than 5% worse (stomach (8.4%), prostate (6.7%), ovary population for Germany to 58% for Belgium. The data included in (6.3%) and thyroid (5.3%)). EUROCARE-4 covered a population of just over 150 million, which Of the sites for which a statistically significant gap was maintained represented about 35% of the population of the combined for 5|1-year survival, all except stomach cancer had a reduced gap in & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 What can we learn from recent EUROCARE results? CS Thomson and D Forman S104 Table 1 Patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival (5-year, 1-year and 5|1-year) % 5-year survival % 1-year survival % 5|1-year survival England– ‘European England EC4 England– ‘European England– ‘European ‘Europe a b b b Site average’ 1995–1999 ‘Europe Avg’ average’ England ‘Europe Avg’ average’ England Avg’ Head and neck 39.5 44.8 Sig higher 69.1 68.5 Lower 57.2 65.3 Sig higher Oesophagus 11.1 9.9 Sig lower 35.8 32.9 Sig lower 30.9 30.1 Lower Stomach 24.5 16.1 Sig lower 46.3 38.0 Sig lower 52.9 42.4 Sig lower Colorectal 54.0 50.5 Sig lower 75.8 72.2 Sig lower 71.2 70.0 Sig lower Liver 9.1 7.7 Sig lower 30.0 23.7 Sig lower 30.2 32.5 Higher Gallbladder and 14.4 15.6 Higher 35.9 37.7 Higher 40.1 41.2 Higher biliary tract Pancreas 5.7 4.4 Sig lower 20.9 16.2 Sig lower 27.1 27.0 Lower Larynx 62.8 63.9 Higher 84.7 83.7 Lower 74.1 76.4 Sig higher Lung 12.0 8.4 Sig lower 36.0 26.9 Sig lower 33.3 31.3 Sig lower Bone and cartilage 55.5 52.6 Lower 77.0 73.2 Sig lower 72.0 71.9 Lower Soft tissue 59.5 58.0 Lower 80.4 78.6 Sig lower 73.9 73.8 Lower Melanoma of skin 82.6 84.6 Sig higher 95.0 96.0 Sig higher 87.0 88.1 Sig higher Breast 79.4 77.3 Sig lower 93.8 91.8 Sig lower 84.6 84.2 Lower Cervix uteri 62.6 59.1 Sig lower 84.4 80.6 Sig lower 74.1 73.3 Lower Corpus uteri 76.2 75.2 Lower 89.5 87.9 Sig lower 85.2 85.6 Higher Ovary and uterine 36.5 30.2 Sig lower 66.7 59.9 Sig lower 54.7 50.5 Sig lower adnexa Prostate 76.4 69.7 Sig lower 92.7 90.2 Sig lower 82.4 77.3 Sig lower Testis 89.5 89.7 Higher 93.3 93.7 Higher 96.0 95.8 Lower Bladder 72.4 72.4 Lower 86.4 84.7 Sig lower 83.8 85.5 Sig higher Kidney 58.0 45.6 Sig lower 73.8 62.7 Sig lower 78.6 72.6 Sig lower Brain 19.7 17.6 Sig lower 42.7 37.1 Sig lower 46.0 47.4 Higher Thyroid 82.9 77.6 Sig lower 87.6 83.7 Sig lower 94.6 92.8 Sig lower Hodgkin 80.1 78.6 Lower 90.4 89.7 Lower 88.7 87.6 Lower NHL 51.5 50.7 Lower 71.8 69.4 Sig lower 71.8 73.1 Sig higher Multiple myeloma 35.1 30.6 Sig lower 72.3 67.1 Sig lower 48.6 45.6 Sig lower All leukaemias 42.4 42.3 Lower 64.7 64.6 Lower 65.6 65.4 Lower a b Site groups classified according to ICD-O3 coding. ‘Sig lower’¼ England had significantly poorer survival than ‘European average’; ‘Sig higher’¼ England had significantly better survival than ‘European average’. Non-significant differences were labelled as either ‘Lower’ or ‘Higher’ based on the point estimates for England compared with ‘European average’. Based on persons. Table 2 Patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, % difference in relative survival (5-year, 1-year and 5|1-year) for the 14 sites with significantly lower 5-year survival % difference in 5-year % difference in 1-year survival survival % difference in 5|1-year survival a b b Site England –‘Europe Avg’ England–‘Europe Avg’ Difference England–‘Europe Avg’ Difference Kidney 12.4 11.1 Sig Lower 6.0 Sig Lower Stomach 8.4 8.3 Sig Lower 10.5 Sig Lower Prostate 6.7 2.5 Sig Lower 5.1 Sig Lower Ovary and uterine adnexa 6.3 6.8 Sig Lower 4.2 Sig Lower Thyroid 5.3 3.9 Sig Lower 1.8 Sig Lower Multiple myeloma 4.5 5.2 Sig Lower 3.0 Sig Lower Lung 3.6 9.1 Sig Lower 2.0 Sig Lower Colorectal 3.5 3.6 Sig Lower 1.2 Sig Lower Cervix 3.5 3.8 Sig Lower 0.8 Lower Breast 2.1 2.0 Sig Lower 0.4 Lower Brain 2.1 5.6 Sig Lower 1.4 Higher Liver 1.4 6.3 Sig Lower 2.3 Higher Pancreas 1.3 4.7 Sig Lower 0.1 Lower Oesophagus 1.2 2.9 Sig Lower 0.8 Lower a b Site groups classified according to ICD-O3 coding. ‘Sig lower’¼ England had significantly poorer survival than ‘European average’; ‘Sig higher’¼ England had significantly better survival than ‘European average’. Non-significant differences were labelled as either ‘Lower’ or ‘Higher’ based on the point estimates for England compared with ‘European average’. Based on persons. comparison with that observed from the conventional 5-year survival only kidney (6.0%) and prostate cancer (5.1%) now had gaps of more analysis. For stomach cancer, survival in England was 10.5% worse than 5%. than the ‘European average’ in the 5|1-year analysis (compared Figures 1–3 provide graphical comparisons of the differences with 8.4% for the conventional 5-year outcome). Of the other sites, between England and the ‘European average’ at 5, 1 and 5|1 years. British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK What can we learn from recent EUROCARE results? CS Thomson and D Forman S105 Table 3 provides a crude comparison between the EUROCARE-4 between these two time periods. Due to the different methodo- data, for both England and the ‘European average’ for patients logies used, direct comparison between EUROCARE-4 and the diagnosed in 1995–1999, with the 1-year survival rates for England more recent England data is not strictly possible but, as expected, between 1996 and 2000 reported in 2009, together with data for for most of the cancers considered, the England results from 2004–2006 showing changes in 1-year survival rates in England EUROCARE-4 are fairly similar to the more recent analysis of European average’ England Significantly Significantly lower Lower Higher higher Figure 1 Five-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. European average’ England Significantly Significantly lower Lower Higher higher Figure 2 One-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 Relative survival (%) Survival (%) Breast Breast Prostate Prostate Corpus uteri Cervix uteri Thyroid Kidney Bladder Colorectal Cervix uteri Ovary and uterine adnexa Soft tissue Multiple myeloma Bone and cartilage Stomach Colorectal Brain Kidney Lung Multiple myeloma Oesophagus NHL Liver Ovary and uterine adnexa Pancreas Stomach Brain Hodgkin Lung Corpus uteri Oesophagus Bladder Liver Soft tissue Pancreas Bone and cartilage Hodgkin NHL Larynx All leukaemias Head and neck Testis All leukaemias Larynx Testis Gall bladder and biliary tract Gall bladder and biliary tract Melanoma of skin Melanoma of skin Head and neck ’ What can we learn from recent EUROCARE results? CS Thomson and D Forman S106 European average’ England Significantly lower Lower Higher Significantly higher Figure 3 5|1-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. Table 3 Age-standardised relative survival (1-year) for patients diagnosed 1995 – 1999 for England and the ‘European average’ followed up to the end of 2003 (EUROCARE-4 Study); and patients diagnosed 1996 – 2000 and 2004 – 2006 in England (Rachet p. comm) followed up to the end of 2007 ‘European average’ 1995–1999 England 1995 –1999 England 1996–2000 (Rachet) England 2004 –2006 (Rachet) Site EUROCARE-4 (ICSS) EUROCARE-4 (ICSS) (England Standard) (England Standard) b b c c Breast 93.8 91.8 93.9 95.5 d d Lung 36.0 26.9 27.0 28.9 d d Colon 74.2 69.9 69.5 71.8 d d Rectum 78.7 76.3 76.1 78.1 Prostate 92.7 90.2 91.5 93.3 d d Oesophagus 35.8 32.9 32.8 37.8 d d Stomach 46.3 38.0 35.8 41.1 d d Pancreas 20.9 16.2 14.7 16.8 e e Larynx 84.7 83.7 83.7 85.8 d d Melanoma 95.0 96.0 95.5 96.6 Cervix 84.4 80.6 83.0 82.9 Corpus uteri 89.5 87.9 88.3 90.1 Ovary and 66.7 59.9 69.1 71.3 uterine adnexa Testis 93.3 93.7 97.8 97.9 d d Kidney 73.8 62.7 67.2 69.7 d d Brain 42.7 37.1 33.7 38.1 d d NHL 71.8 69.4 71.6 75.2 d d Hodgkin 90.4 89.7 91.3 89.9 d d Multiple 72.3 67.1 65.4 66.3 myeloma d d All leukaemias 64.7 64.6 63.4 61.1 a b c Site groups classified according to ICD-O3 coding for EUROCARE-4 but using ICD-10 coding for the England analyses by Rachet et al, 2009. Based on persons. Based d e on females only. Estimates taken as simple averages of the male and female estimates. Based on males only. ICSS¼ The International Cancer Survival Standard; ICD-O3¼ International Classification of Diseases for Oncology. England data for 1996–2000 (the major exception being ovarian and leukaemia (2.3% decrease)). These improvements were cancer for which the EUROCARE-4 analysis provides a 1-year generally around 1–2% between the two time periods and the survival of 59.9% while the more recent analysis for 1996–2000 major cancers, breast, lung, colon, rectum and prostate all showed estimates 69.1%). survival increases of this magnitude. For testicular cancer, 1-year Of the 20 individual cancer sites or site groups considered in survival in 1996–2000 was already extremely high (97.8%) and the Table 3 for England, 17 have shown an improvement in 1-year improvement, by 0.1%, was marginal. The improvements in survival between 1996–2000 and 2004–2006 (the exceptions being oesophageal (5.0%), stomach (5.3%) and brain cancer (4.4%) were cervical cancer (0.1% decrease), Hodgkin disease (1.4% decrease) the most substantial. British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK Relative survival (%) Thyroid Prostate Kidney Colorectal Ovary and uterine adnexa Stomach Multiple myeloma Lung Testis Hodgkin Breast Cervix uteri Soft tissue Bone and cartilage All leukaemias Oesophagus Pancreas Corpus uteri Brain Gall bladder and biliary tract Liver Melanoma of skin Bladder Larynx NHL Head and neck ’ What can we learn from recent EUROCARE results? CS Thomson and D Forman S107 DISCUSSION average’, although this is not translated into a differential at 5 years. All of these cancers are associated with reasonable survival Until recently, the only evidence enabling an understanding of the rates overall and thus might benefit most from efforts towards reasons for the relatively poor 5-year survival outcomes in the earlier diagnosis. With bladder cancer, there is also the additional United Kingdom compared with elsewhere in Europe was the problem that changes in coding schemes defining the classification EUROCARE high-resolution studies (Gatta et al, 2000; Sant et al, of invasive tumours and introduced at different times in different 2003, 2007). While these indicated later stage at diagnosis as being countries, may introduce artefactual survival differences. an important explanation for part of the difference, the studies The final group of eight cancers are those that had significantly only covered a few sites of cancer and were based on relatively lower survival rates in England at 5 years, which were also evident small numbers of patients diagnosed during the early 1980s. The at both 1 year and 5|1 years. This group included the three samples of UK patients included in these were not necessarily common cancers of the lung, colorectum and prostate, as well as representative of either their regions of residence, or the country as cancers of the kidney, thyroid, ovary and stomach and multiple a whole. Relevance to the current situation is, therefore, myeloma. The results for these cancers suggest that improving the questionable. stage at diagnosis of these cancers would not have eradicated all of the variation observed in 5-year survival between England and The present review of the EUROCARE-4 results cannot provide an exhaustive evaluation of the relative contribution of stage Europe. There may also have been differences in the treatment at diagnosis to 5-year survival on a site-by-site basis but the given or the management of services that affected the longer-term analysis of 5, 1 and 5|1-year survival helps provide some overall outcome. However, the survival gap for all of these sites, except perspective. stomach, was smaller between 1–5 years than it was between Of the cancer sites considered, one important group is those diagnosis and 5 years (Table 2) indicating a role for earlier cancers for which survival in England was significantly worse than diagnosis. In addition, all of these cancers (except thyroid that was the ‘European average’ at 5 years, but separating this into not analysed by Rachet et al, 2009) have shown recent improve- outcomes after 1 year and between 1 and 5 years (i.e. having ments in 1-year survival in the England data (Table 3). Prostate survived for 1 year) indicated a statistically significant detriment cancer survival results are particularly difficult to interpret given associated only with the former, but not the latter. This the unknown impact of differing intensity of PSA testing in implies that, for these cancers, it is low survival in the first year different countries leading to overdiagnosis of non-fatal disease. after diagnosis, which gives rise exclusively to the relatively poor Coding changes for the classification of invasive behaviour for overall outcome. The major site falling into this category was ovarian cancer make comparisons over time problematic and are breast cancer, but others in this group were cancers of the most apparent when using the different coding classifications ICD- oesophagus, liver, pancreas, cervix and brain. It is noteworthy that O3 and ICD-10, as in Table 3. this group includes both breast and cervical cancers for which the This analysis dissects the EUROCARE-4 results to discriminate, benefit of population-based screening has long been established. among those cancers for which 5-year survival in England seemed Both of these cancers are associated with a generally good to be lagging behind Europe, those for which the explanation prognosis and the poor comparative survival in England is seems to be largely due to effects within the year after diagnosis probably best explained by relatively small subgroups of women (group one, not showing detrimental 5|1–year outcomes) and diagnosed with late stage disease, and not being detected by those for which the explanation is related to short- screening. and longer-term effects (group three, showing detrimental 1 and In contrast, the results for oesophageal, pancreatic and liver 5|1–year outcomes). There is also a group of cancers for which cancers, all with an extremely poor prognosis, may be explained in 5-year survival in England was not statistically different from part by differences in access to new therapeutic regimens for the ‘European average’, but for which there was a detrimental advanced disease, which may have started use in England later effect at 1 year (group two). It should also be noted that, for 10 of than elsewhere in Europe (Mitry et al, 2008a, b; Rao and the 26 cancer sites listed in Table 3, there was no statistical Cunningham, 2008; Starling and Cunningham, 2008; Rimassa difference between the England outcomes and the ‘European and Santoro, 2009). As a group, however, these cancers are often average’ at 5 years, and for only two cancers (head and neck and diagnosed at an advanced, incurable stage. If registration and malignant melanoma), England had significantly better survival follow-up of cases registered initially through death certificates than Europe. were more effective in England than elsewhere, this would result in Efforts and interventions directed towards increased cancer other countries having a higher proportion of cases that are more awareness and earlier detection, if successful, are likely to benefit likely to be registered only through death certification and/or as all forms of cancer. Relative improvements to 1-year survival in cancers with an unknown primary site. These cases have very poor England are likely to benefit those cancers in groups one and two outcomes but need to be excluded from survival analyses. This to such an extent that the survival gap with Europe could almost would result in the poorer 1-year survival observed in England, but completely be overcome. NAEDI could make this improvement to would not influence the 5|1-year survival. Finally brain cancer, most of the cancers in these two groups although, for the poor associated with moderate prognosis, is a type of cancer with a prognosis sites (oesophageal, stomach and liver), life-extending potential for early diagnosis if prompt attention is paid to therapies for advanced disease must also have a role. NAEDI could symptoms and appropriate diagnostic facilities are available. With also have an important function in the cancers in group three, the exception of cervical cancer (marginal decrease) and liver although the current survival gap between England and Europe for cancer (not included in the analysis by Rachet et al, (2009)), all of these cancers is not fully explained by short-term survival the cancers in this group have shown recent improvements in 1- differences and other factors will thus also have a role. year survival in England (Table 3). This study has several limitations. It is acknowledged that the A second group of cancers comprises those that showed non-independence between the survival estimates for England and significantly lower survival at 1 year in England compared with the ‘European average’ means that the method of testing the ‘European average’, which was no longer evident at either 5 or differences between the survival estimates, using only the pooled 5|1 years. The five cancers in this group were bone and cartilage, variance to highlight significant differences, is not perfect. It soft tissue, uterus, bladder and non-Hodgkin lymphoma and, as should, however, be sufficient to help identify which sites require with the preceding group, differences in early diagnosis and/or further efforts to separate out the early and late mortality effects initial management of advanced disease would explain the and where future international benchmarking projects should be disadvantage observed in England compared with the ‘European considered. & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 What can we learn from recent EUROCARE results? CS Thomson and D Forman S108 De Angelis et al (2009) describe the data quality and resource constraints involved in data collection, are not available completeness across the EUROCARE-4 data sets, presenting for contemporary comparisons. indicators of quality known to affect survival analyses. One of Another weakness currently faced is that though it is known how these measures is the percentage of registrations made solely on survival has improved in England during the past decade, how this the basis of death certificates (DCOs), which are excluded from the compares with improvements made across Europe is not known. It survival calculations thus biasing the results. High DCO rates are is possible that the gaps are closing but have not been fully associated with over-inflation of the survival estimates (Berrino eradicated, and without further studies this cannot be determined et al, 1995a), and the higher DCO rates in England compared with for certain (Berrino et al, 2007; Richards, 2007; Verdecchia et al, Europe (6.1% compared with 2.7%) suggest this is a factor which 2007). This study has focussed on comparing survival within needs consideration and a possible correction could be applied to England with the ‘European average’. It could be argued that as a survival estimates (Silcocks and Thomson, 2009). rich Western European country, England should be striving to be The concept of the ‘European average’ is itself problematic the equivalent of the ‘European best’ not just as good as the mainly because the EUROCARE-4 Study only covers a minority ‘European average’. This was the adopted approach for estimating (about 30%) of the EU population. From the variation within the the numbers of avoidable deaths in the United Kingdom relative to countries with partial coverage, it is likely that the national the ‘European best’ (Abdel-Rahman et al, 2009). estimates for some of these countries would change if 100% There is, therefore, a need for new international benchmarking population coverage were available, thus, in turn, affecting the studies to assess this. These studies will need to compare registry ‘European average’. Another problem with the ‘European average’ data within England with other countries with good quality is that it assumes that all the data sets used in its estimation are of registry data. The data collected will need to contain good staging equal quality in terms of accuracy and completeness of recording information, which has been robustly collected to agreed both incident cases and subsequent deaths. Further discussion of standards, along with information about the diagnostic techniques this is provided in both Møller et al (2009) and Abdel-Rahman used, the presence of any significant comorbidity and details of the et al (2009). treatment that was given. Only then may studies provide the real An alternative to using the ‘European average’ for comparison understanding as to how much stage at diagnosis and earlier with England would be to identify specific populations with which detection improve cancer survival. to compare the results from England; these are likely to be with countries where comparable quality of data to those in England was assured. This is the basis of the future proposed international ACKNOWLEDGEMENTS benchmarking studies. The authors thank Matt Wickenden for creating and formatting all These results also only give a crude assessment of the possible of the charts for this publication. effects of stage, early diagnosis and treatment on outcomes. A better way to examine these effects is to include this information, when it has been collected robustly, directly in the survival Conflict of interest analyses. The high-resolution studies, undertaken as part of the earlier EUROCARE studies, attempted this but, partly because of The authors declare no conflict of interest. REFERENCES Abdel-Rahman M, Stockton D, Rachet B, Hakulinen T, Coleman MP (2009) Corazziari I, Quinn M, Capocaccia R (2004) Standard cancer patient popu- What if cancer survival in Britain were the same as in Europe: how many lation for age standardising survival ratios. Eur J Cancer 40: 2307– 2316 deaths are avoidable? Br J Cancer 101(Suppl 2): S115–S124 De Angelis R, Francisci S, Baili P, Marchesi F, Roazzi P, Belot A, Crocetti E, Berrino F, Capocaccia R, Coleman MP, Este`ve J, Gatta G, Hakulinen T, Pury P, Knijn A, Coleman M, Capocaccia R, EUROCARE-4 Working Micheli A, Sant M, Verdecchia A (eds). (2003) EUROCARE-3: the Group (2009) The EUROCARE-4 database on cancer survival in Europe: survival of cancer patients diagnosed in Europe during 1990–1994. Ann data standardisation, quality control and methods of statistical analysis. Oncol 14(Suppl 5): 1 –155 Eur J Cancer 45: 909–930 Berrino F, Capocaccia R, Este`ve J, Gatta G, Hakulinen T, Micheli M, Sant M, Department of Health (2000) The NHS Cancer Plan. Department of Health: Verdecchia A (eds). (1999) Survival of Cancer Patients in Europe: the London EUROCARE-2 Study (IARC Scientific Publications No. 151). International Department of Health (2007) Cancer Reform Strategy. Department of Agency for Research on Cancer: Lyon Health: London Berrino F, De Angelis R, Sant M, Rosso S, Lasota MB, Coebergh JWW, Gatta G, Capocaccia R, Sant M, Bell CMJ, Coebergh JWW, Damhuis RAM, Santaquilani M, EUROCARE Working Group (2007) Survival for Faivre J, Martinez-Garcia C, Pawlega J, Ponz de Leon M, Pottier D, eight major cancers and all cancers combined for European adults Raverdy N, Williams EMI, Berrino F (2000) Understanding variations in diagnosed in 1995–1999: results of the EUROCARE-4 study. Lancet Oncol survival for colorectal cancer in Europe: a EUROCARE high resolution 8: 773–783 study. Gut 47: 533–538 Berrino F, Este`ve J, Coleman MP (1995a) Basic issues in the estimation and Mitry E, Rachet B, Quinn MJ, Cooper N, Coleman MP (2008a) Survival from comparison of cancer patient survival. In Survival of Cancer Patients in cancer of the oesophagus in England and Wales up to 2001 – survival Europe: the EUROCARE Study (IARC Scientific Publications No. 132), analysis. Br J Cancer 99(Suppl 1): S11 –S13 Berrino F, Sant M, Verdecchia A, Capocaccia R, Hakulinen T, Esteve J Mitry E, Rachet B, Quinn MJ, Cooper N, Coleman MP (2008b) Survival (eds), pp 1 –14. International Agency for Research on Cancer (WHO): from cancer of the pancreas in England and Wales up to 2001 – survival Lyon analysis. Br J Cancer 99(Suppl 1): S21 –S23 Berrino F, Sant M, Verdecchia A, Capocaccia R, Hakulinen T, Esteve J Møller H, Linklater KM, Robinson D (2009) A visual summary of the (1995b) (eds). Survival of Cancer Patients in Europe: the EUROCARE EUROCARE-4 results: a UK perspective. Br J Cancer 101(Suppl 2): S110–S114 Study (IARC Scientific Publications No. 132). International Agency for Rachet B, Maringe C, Nur U, Quaresma M, Shah A, Woods LM, Ellis L, Research on Cancer: Lyon Walters S, Forman D, Steward J, Coleman MP (2009) Population-based Coleman MP, Babb P, Damiecki P, Grosclaude PC, Honjo S, Jones J, Knerer cancer survival trends in England and Wales up to 2007: an assessment G, Pitard A, Quinn MJ, Sloggett A, De Stavola BL (1999) Cancer Survival of the NHS cancer plan for England. Lancet Oncol 10: 351–369 Trends in England and Wales 1971 –1995: Deprivation and NHS Region Rao S, Cunningham D (2008) Survival from cancer of the oesophagus in (Studies on Medical and Population Subjects No. 61). The Stationery England and Wales up to 2001 – clinical commentary. Br J Cancer Office: London 99(Suppl 1): S14 –S15 British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK What can we learn from recent EUROCARE results? CS Thomson and D Forman S109 Richards M (2007) EUROCARE-4 studies bring new data on cancer Sant M, Allemani C, Santaquilani M, Knijn A, Marchesi F, Capocaccia R, survival. Lancet Oncol 8(9): 752–753 EUROCARE Working Group (2009) EUROCARE-4. Survival of cancer Rimassa L, Santoro A (2009) Sorafenib in advanced hepatocellular patients diagnosed in 1995–1999. Results and commentary. Eur J Cancer carcinoma. Expert Rev Anticancer Ther 9(6): 739 –745 45: 931–991 Sant M, Aareleid T, Artiolic ME, Berrino F, Coebergh JW, Colonna M, Silcocks P, Thomson CS (2009) Correcting population-based survival for Forman D, Hedelin J, Rachtan J, Lutz JM, Otter R, Raverdy R, Plesko II, DCOs – why a simple method works and when to avoid it. Eur J Cancer Primic M-Z, Tagliabue G (2007) Ten-year survival and risk of relapse for DOI: 10.1016/j.ejca.2009.06.013 (in press) testicular cancer: A EUROCARE High Resolution Study. Eur J Cancer 43: Starling N, Cunningham D (2008) Survival from cancer of the pancreas in 585–592 England and Wales up to 2001 – clinical commentary. Br J Cancer Sant M, Allemani C, Capocaccia R, Hakulinen T, Aareleid T, Coebergh JW, 99(Suppl 1): S24 –S25 Coleman MP, Grosclaude P, Martinez C, Bell J, Youngson J, Berrino F, Verdecchia A, Francisci S, Brenner H, Gatta G, Micheli A, Mangone L, EUROCARE Working Group (2003) Stage at diagnosis is a key Kunkler I, EUROCARE-4 Working Group (2007) Recent cancer survival explanation of differences in breast cancer survival across Europe. Int in Europe: a period analysis 2000– 2002 of the EUROCARE-4 data. J Cancer 106: 416–422 Lancet Oncol 8: 784 –796 & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png British Journal of Cancer Springer Journals

Cancer survival in England and the influence of early diagnosis: what can we learn from recent EUROCARE results?

Thomson, C S; Forman, D
British Journal of Cancer , Volume 101 (2) – Dec 3, 2009
Download PDF
8 pages

Loading next page...
 
/lp/springer-journals/cancer-survival-in-england-and-the-influence-of-early-diagnosis-what-7Rhs0CEYX0

References (30)

Publisher
Springer Journals
Copyright
Copyright © 2009 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Cancer Research; Epidemiology; Molecular Medicine; Oncology; Drug Resistance
ISSN
0007-0920
eISSN
1532-1827
DOI
10.1038/sj.bjc.6605399
Publisher site
See Article on Publisher Site

Abstract

British Journal of Cancer (2009) 101, S102 – S109 & 2009 Cancer Research UK All rights reserved 0007 – 0920/09 $32.00 www.bjcancer.com Discussion Paper Cancer survival in England and the influence of early diagnosis: what can we learn from recent EUROCARE results? ,1 2,3 CS Thomson and D Forman 1 2 Cancer Research UK, Statistical Information Team, 61 Lincoln’s Inn Fields, London WC2A 3PX, UK; Centre for Epidemiology and Biostatistics, University of Leeds, Beckett Street, Leeds, UK; National Cancer Intelligence Network, Queens House, 56 Lincoln’s Inn Fields, London, UK BACKGROUND: This review of the EUROCARE-4 results attempts to separate out the early and late mortality effects contributing to the widely reported poorer 5-year survival rates for cancer patients in the United Kingdom compared with other European countries for 26 cancer sites. METHODS: Patients diagnosed with cancer in 1996–1999 in 23 European countries were included in the analyses. Comparison of 1-year, 5-year and 5|1-year (i.e. only including those patients who had survived to 1 year) survival estimates between data for England and the ‘European average’ was undertaken. This analysis was to highlight the relative contribution of early diagnosis, using 1-year survival as a proxy measure, on 5-year survival for the different sites of cancer. Three groups of cancer sites were identified according to whether the survival differences at 1, 5 and 5|1 years were statistically significant. RESULTS AND CONCLUSIONS: Breast cancer showed significantly poorer 1- and 5-year survival estimates in England, but the 5|1-year survival figure was not significantly different. Thus, successful initiatives around awareness and early detection could eradicate the survival gap. In contrast, the 5|1-year survival estimates remained significantly worse for lung, colorectal and prostate cancers, showing that although early detection could make some difference, late effects such as treatment and management of the patients were also influencing long-term outcome differences between England and Europe. British Journal of Cancer (2009) 101, S102 – S109. doi:10.1038/sj.bjc.6605399 www.bjcancer.com & 2009 Cancer Research UK Keywords: EUROCARE; cancer survival in England; National Awareness and Early Diagnosis Initiative (NAEDI) It has been widely reported that 5-year survival rates for cancer Initiative (NAEDI). One objective of NAEDI is to review the patients in the United Kingdom have, in general, lagged behind evidence on links between early diagnosis and survival and comparator European countries (and the ‘European average’) since evaluate the extent to which international differences in survival the late 1970s (Berrino et al, 1995b, 1999, 2003, 2007). The NHS are due to differences in delays in patient presentation; in Cancer Plan (Department of Health, 2000) highlighted this survival practitioner referrals; in the availability and quality of diagnostic ‘gap’ and ‘catching up with Europe’ has been a major driver in techniques used; and in the management and care of patients once formulating national cancer policy. diagnosed. High-resolution studies of a few specific sites of cancer Use of 1-year survival as a proxy for early/late diagnosis, and undertaken within some of the registries included in EURO- hence for stage data, has been advocated in the CRS and CARE-2, and in which additional data on stage and diagnostic comparison of this outcome measure across Europe may help techniques were available, suggested that later stage at presenta- benchmark the current disparities. However, the appropriateness tion among UK patients was a major factor explaining the poorer of this measure will vary by cancer site and early/late presentation survival. These data are, however, now quite old; they relate to is not the only possible reason for observed differences. For cancers diagnosed in the 1980s and have not been updated (Gatta example, for cancers such as oesophageal and pancreatic with a et al, 2000; Sant et al, 2003, 2007). generally poor prognosis, the introduction of chemotherapeutic The most recently published EUROCARE data, based on new regimens for advanced disease will have improved 1-year survival diagnoses between 1995 and 1999, indicate that the survival gap rates for patients irrespective of the proportion of early stage between England and the ‘European average’ had diminished but tumours in the populations (Rao and Cunningham, 2008; Starling had not been completely removed (Berrino et al, 2007; Richards, and Cunningham, 2008; Mitry et al, 2008a, b). Also factors, such as 2007; Verdecchia et al, 2007). As a consequence, the 2007 Cancer differences in underlying comorbidity, will impact on short-term Reform Strategy (CRS) maintained an objective of bringing about survival but are rarely adjusted for in comparative survival service improvements to reduce further the difference in survival estimates. between England and Europe (Department of Health, 2007). Use of 5-year survival data conditional on surviving 1 year (i.e. The CRS set out several initiatives to improve survival including 1–5 year or 5|1 survival) has been presented in the most recent establishment of the National Awareness and Early Diagnosis EUROCARE-4 publications (Sant et al, 2009). Examination of this along side 1-year survival for different cancer sites shows which sites had variation between the countries at 1 year, which was not present at 5 years after those who died within the first year had *Correspondence: CS Thomson; been excluded from the analyses. To some extent, this enables E-mail: [email protected] What can we learn from recent EUROCARE results? CS Thomson and D Forman S103 separation of the effects of stage distribution, significant comor- countries taking part in the study, and 30% of the total EU bidity and either peri-operative (or postoperative treatment- population (excluding Norway, Switzerland and Iceland, which are related) mortality, shown generally by the 1-year survival results, not in the EU). from the effects of treatment and patient management on survival, The methodology used to derive the ‘European average’ was shown by differences in longer-term survival. This assessment is neither a simple country-specific weighted average based on fundamental to the NAEDI objective of trying to identify cancer coverage, nor one based on population, because either of these sites where delays in diagnosis have had the biggest impact on averages would have been heavily biased. Instead, regional-specific survival and which could be targeted to reduce such delays in the survival estimates were obtained by splitting Europe into five areas future. and pooling the data available for the regions. This makes the This paper analyses the most recent EUROCARE data for the assumption that the survival for the population covered by more common sites of cancer, comparing outcomes in England registration was representative of the unknown survival of the with the EUROCARE-4 ‘European average’ at 5, 1 and 5|1 years whole region. Weighting was then applied to each of these five after diagnosis, and also looks at recent trends in 1-year survival regional estimates using the mean population for those countries data within England. This provides a means of beginning to included in EUROCARE-4 for each region. identify cancer sites for which the poor outcomes in England can The method of creating the ‘European average’ led to the United be attributed to later stage at diagnosis. Kingdom and Ireland region making up 14.6% of the ‘European average’. Thus, comparison of the survival estimates for England (or indeed any other country) with the survival estimate for the ‘European average’ is not strictly straightforward, due to the MATERIALS AND METHODS non-independence between them. However, it was assumed that The EUROCARE project has been running since 1990 (Berrino the covariance (a measure of the non-independence) between the et al, 1995a). The latest published data cover patients diagnosed England value and the ‘European average’ would not be large between 1996 and 1999 and are based on 2.8 million adults (aged enough to greatly affect interpretation of statistical significance, 15–99) who were diagnosed with cancer in 23 European countries and as such, the difference between the survival rates for England (Sant et al, 2009). Data for 26 of the more common sites or site and the ‘European average’ for each of the 5-year, 1-year and groups of cancer from the EUROCARE-4 study have been 5|1-year estimates was obtained, along with an estimate of its presented. The coding classification used in EUROCARE-4 is the standard error (s.e.). The standard error for the survival difference third revision of the International Classification of Diseases for was calculated by taking the square root of the sum of the squared Oncology (ICD-O3). standard errors obtained for each of the survival estimates for The data presented here from the EUROCARE-4 study are the England and the ‘European average’, ignoring the covariance figures for England and the ‘European average’, which have both between the two cohorts. been appropriately standardised for cancer survival analyses. This was to take into account any differences in age structures between the countries, and four different standards are chosen depending RESULTS on the cancer site of choice. The International Cancer Survival Standard (ICSS; Corazziari et al, 2004) was applied to the Table 1 shows results for England and the ‘European average’ for 5, EUROCARE-4 data, except for prostate cancer (De Angelis et al, 1 and 5|1-year survival results for 26 of the more common sites or 2009; pp 20). This differs from previous EUROCARE studies, site groups of cancer included in the EUROCARE-4 study. Five- which used internally weighted average age structure for Europe year survival was significantly worse in England for 14 of these for that study. The ICSS gives standard weights, which should cancers, including the four major sites, breast, lung, colorectal and make future comparisons easier. prostate, together with cancers of the oesophagus, stomach, liver, Data from recent analyses within England (Rachet et al, 2009) pancreas, cervix, ovary, kidney, brain, thyroid and multiple for 20 cancer sites or site groupings have also been included. The myeloma. Five-year survival was significantly better in England classification used for this study was the tenth revision of the ICD- for head and neck cancers (including the group of oral cancers) 10, which is not entirely comparable with those coded using ICD- and malignant melanoma; and not significantly different for the O3. These data were not originally age standardised although remaining 10 sites. equivalent age-standardised rates were made available to us All of the 14 cancers showing a statistically significant worse (personal communication, B Rachet, 2009). The age-standardised outcome in England at 5 years also showed a significantly worse data were not reported in the original paper because the authors outcome at 1 year. A further five cancers – bone and cartilage, soft had found that the age distributions of the cancer patients had tissue, uterus, bladder and non-Hodgkin lymphoma – also showed a changed little over the 11 years they examined. However, the significantly worse 1-year survival, but this was not maintained at 5 weights used in the age standardisation for the recent study in years. Of the 14 cancers with significantly worse 5-year outcomes, England (Rachet et al, 2009) were different to those used in the eight also showed a significantly worse 5|1-year survival – lung, EUROCARE-4 publications. Instead of the ICSS weighting system, colorectal and prostate together with stomach, ovary, kidney and Rachet et al (2009) used the ‘England Standard’, which is based on thyroid cancers and multiple myeloma. The other six cancers had the age distributions of cancer patients diagnosed between 1986 5|1-year survival that was no longer significantly different from the and 1990 in England (Coleman et al, 1999; pp 50). This was to ‘European average’. This group included breast cancer together with maintain comparability with previous survival analyses within oesophagus, liver, pancreas, cervix and brain cancers. England and used a different set of weights for each of the 20 Table 2 shows the 14 cancers with significantly worse 5-year cancer sites. outcomes ordered by the magnitude of the 5-year ‘survival gap’ The ‘European average’ survival rates reported are not strictly between England and the ‘European average’ together with the true averages across the whole of Europe. This is because only 13 corresponding survival gap at 1 year and 5|1 years. Kidney cancer of the 23 countries represented in EUROCARE-4 have national showed the largest survival gap, being 12.4% worse than the cancer registration and, hence, 100% coverage. The range of ‘European average’ at 5 years, while four other sites had gaps that coverage for the other countries was from 1.3% of the adult were larger than 5% worse (stomach (8.4%), prostate (6.7%), ovary population for Germany to 58% for Belgium. The data included in (6.3%) and thyroid (5.3%)). EUROCARE-4 covered a population of just over 150 million, which Of the sites for which a statistically significant gap was maintained represented about 35% of the population of the combined for 5|1-year survival, all except stomach cancer had a reduced gap in & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 What can we learn from recent EUROCARE results? CS Thomson and D Forman S104 Table 1 Patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival (5-year, 1-year and 5|1-year) % 5-year survival % 1-year survival % 5|1-year survival England– ‘European England EC4 England– ‘European England– ‘European ‘Europe a b b b Site average’ 1995–1999 ‘Europe Avg’ average’ England ‘Europe Avg’ average’ England Avg’ Head and neck 39.5 44.8 Sig higher 69.1 68.5 Lower 57.2 65.3 Sig higher Oesophagus 11.1 9.9 Sig lower 35.8 32.9 Sig lower 30.9 30.1 Lower Stomach 24.5 16.1 Sig lower 46.3 38.0 Sig lower 52.9 42.4 Sig lower Colorectal 54.0 50.5 Sig lower 75.8 72.2 Sig lower 71.2 70.0 Sig lower Liver 9.1 7.7 Sig lower 30.0 23.7 Sig lower 30.2 32.5 Higher Gallbladder and 14.4 15.6 Higher 35.9 37.7 Higher 40.1 41.2 Higher biliary tract Pancreas 5.7 4.4 Sig lower 20.9 16.2 Sig lower 27.1 27.0 Lower Larynx 62.8 63.9 Higher 84.7 83.7 Lower 74.1 76.4 Sig higher Lung 12.0 8.4 Sig lower 36.0 26.9 Sig lower 33.3 31.3 Sig lower Bone and cartilage 55.5 52.6 Lower 77.0 73.2 Sig lower 72.0 71.9 Lower Soft tissue 59.5 58.0 Lower 80.4 78.6 Sig lower 73.9 73.8 Lower Melanoma of skin 82.6 84.6 Sig higher 95.0 96.0 Sig higher 87.0 88.1 Sig higher Breast 79.4 77.3 Sig lower 93.8 91.8 Sig lower 84.6 84.2 Lower Cervix uteri 62.6 59.1 Sig lower 84.4 80.6 Sig lower 74.1 73.3 Lower Corpus uteri 76.2 75.2 Lower 89.5 87.9 Sig lower 85.2 85.6 Higher Ovary and uterine 36.5 30.2 Sig lower 66.7 59.9 Sig lower 54.7 50.5 Sig lower adnexa Prostate 76.4 69.7 Sig lower 92.7 90.2 Sig lower 82.4 77.3 Sig lower Testis 89.5 89.7 Higher 93.3 93.7 Higher 96.0 95.8 Lower Bladder 72.4 72.4 Lower 86.4 84.7 Sig lower 83.8 85.5 Sig higher Kidney 58.0 45.6 Sig lower 73.8 62.7 Sig lower 78.6 72.6 Sig lower Brain 19.7 17.6 Sig lower 42.7 37.1 Sig lower 46.0 47.4 Higher Thyroid 82.9 77.6 Sig lower 87.6 83.7 Sig lower 94.6 92.8 Sig lower Hodgkin 80.1 78.6 Lower 90.4 89.7 Lower 88.7 87.6 Lower NHL 51.5 50.7 Lower 71.8 69.4 Sig lower 71.8 73.1 Sig higher Multiple myeloma 35.1 30.6 Sig lower 72.3 67.1 Sig lower 48.6 45.6 Sig lower All leukaemias 42.4 42.3 Lower 64.7 64.6 Lower 65.6 65.4 Lower a b Site groups classified according to ICD-O3 coding. ‘Sig lower’¼ England had significantly poorer survival than ‘European average’; ‘Sig higher’¼ England had significantly better survival than ‘European average’. Non-significant differences were labelled as either ‘Lower’ or ‘Higher’ based on the point estimates for England compared with ‘European average’. Based on persons. Table 2 Patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, % difference in relative survival (5-year, 1-year and 5|1-year) for the 14 sites with significantly lower 5-year survival % difference in 5-year % difference in 1-year survival survival % difference in 5|1-year survival a b b Site England –‘Europe Avg’ England–‘Europe Avg’ Difference England–‘Europe Avg’ Difference Kidney 12.4 11.1 Sig Lower 6.0 Sig Lower Stomach 8.4 8.3 Sig Lower 10.5 Sig Lower Prostate 6.7 2.5 Sig Lower 5.1 Sig Lower Ovary and uterine adnexa 6.3 6.8 Sig Lower 4.2 Sig Lower Thyroid 5.3 3.9 Sig Lower 1.8 Sig Lower Multiple myeloma 4.5 5.2 Sig Lower 3.0 Sig Lower Lung 3.6 9.1 Sig Lower 2.0 Sig Lower Colorectal 3.5 3.6 Sig Lower 1.2 Sig Lower Cervix 3.5 3.8 Sig Lower 0.8 Lower Breast 2.1 2.0 Sig Lower 0.4 Lower Brain 2.1 5.6 Sig Lower 1.4 Higher Liver 1.4 6.3 Sig Lower 2.3 Higher Pancreas 1.3 4.7 Sig Lower 0.1 Lower Oesophagus 1.2 2.9 Sig Lower 0.8 Lower a b Site groups classified according to ICD-O3 coding. ‘Sig lower’¼ England had significantly poorer survival than ‘European average’; ‘Sig higher’¼ England had significantly better survival than ‘European average’. Non-significant differences were labelled as either ‘Lower’ or ‘Higher’ based on the point estimates for England compared with ‘European average’. Based on persons. comparison with that observed from the conventional 5-year survival only kidney (6.0%) and prostate cancer (5.1%) now had gaps of more analysis. For stomach cancer, survival in England was 10.5% worse than 5%. than the ‘European average’ in the 5|1-year analysis (compared Figures 1–3 provide graphical comparisons of the differences with 8.4% for the conventional 5-year outcome). Of the other sites, between England and the ‘European average’ at 5, 1 and 5|1 years. British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK What can we learn from recent EUROCARE results? CS Thomson and D Forman S105 Table 3 provides a crude comparison between the EUROCARE-4 between these two time periods. Due to the different methodo- data, for both England and the ‘European average’ for patients logies used, direct comparison between EUROCARE-4 and the diagnosed in 1995–1999, with the 1-year survival rates for England more recent England data is not strictly possible but, as expected, between 1996 and 2000 reported in 2009, together with data for for most of the cancers considered, the England results from 2004–2006 showing changes in 1-year survival rates in England EUROCARE-4 are fairly similar to the more recent analysis of European average’ England Significantly Significantly lower Lower Higher higher Figure 1 Five-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. European average’ England Significantly Significantly lower Lower Higher higher Figure 2 One-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 Relative survival (%) Survival (%) Breast Breast Prostate Prostate Corpus uteri Cervix uteri Thyroid Kidney Bladder Colorectal Cervix uteri Ovary and uterine adnexa Soft tissue Multiple myeloma Bone and cartilage Stomach Colorectal Brain Kidney Lung Multiple myeloma Oesophagus NHL Liver Ovary and uterine adnexa Pancreas Stomach Brain Hodgkin Lung Corpus uteri Oesophagus Bladder Liver Soft tissue Pancreas Bone and cartilage Hodgkin NHL Larynx All leukaemias Head and neck Testis All leukaemias Larynx Testis Gall bladder and biliary tract Gall bladder and biliary tract Melanoma of skin Melanoma of skin Head and neck ’ What can we learn from recent EUROCARE results? CS Thomson and D Forman S106 European average’ England Significantly lower Lower Higher Significantly higher Figure 3 5|1-year survival of patients diagnosed 1995 – 1999 from the EUROCARE-4 Study followed up to until the end of 2003, age-standardised % relative survival, ‘European average’ and England. Table 3 Age-standardised relative survival (1-year) for patients diagnosed 1995 – 1999 for England and the ‘European average’ followed up to the end of 2003 (EUROCARE-4 Study); and patients diagnosed 1996 – 2000 and 2004 – 2006 in England (Rachet p. comm) followed up to the end of 2007 ‘European average’ 1995–1999 England 1995 –1999 England 1996–2000 (Rachet) England 2004 –2006 (Rachet) Site EUROCARE-4 (ICSS) EUROCARE-4 (ICSS) (England Standard) (England Standard) b b c c Breast 93.8 91.8 93.9 95.5 d d Lung 36.0 26.9 27.0 28.9 d d Colon 74.2 69.9 69.5 71.8 d d Rectum 78.7 76.3 76.1 78.1 Prostate 92.7 90.2 91.5 93.3 d d Oesophagus 35.8 32.9 32.8 37.8 d d Stomach 46.3 38.0 35.8 41.1 d d Pancreas 20.9 16.2 14.7 16.8 e e Larynx 84.7 83.7 83.7 85.8 d d Melanoma 95.0 96.0 95.5 96.6 Cervix 84.4 80.6 83.0 82.9 Corpus uteri 89.5 87.9 88.3 90.1 Ovary and 66.7 59.9 69.1 71.3 uterine adnexa Testis 93.3 93.7 97.8 97.9 d d Kidney 73.8 62.7 67.2 69.7 d d Brain 42.7 37.1 33.7 38.1 d d NHL 71.8 69.4 71.6 75.2 d d Hodgkin 90.4 89.7 91.3 89.9 d d Multiple 72.3 67.1 65.4 66.3 myeloma d d All leukaemias 64.7 64.6 63.4 61.1 a b c Site groups classified according to ICD-O3 coding for EUROCARE-4 but using ICD-10 coding for the England analyses by Rachet et al, 2009. Based on persons. Based d e on females only. Estimates taken as simple averages of the male and female estimates. Based on males only. ICSS¼ The International Cancer Survival Standard; ICD-O3¼ International Classification of Diseases for Oncology. England data for 1996–2000 (the major exception being ovarian and leukaemia (2.3% decrease)). These improvements were cancer for which the EUROCARE-4 analysis provides a 1-year generally around 1–2% between the two time periods and the survival of 59.9% while the more recent analysis for 1996–2000 major cancers, breast, lung, colon, rectum and prostate all showed estimates 69.1%). survival increases of this magnitude. For testicular cancer, 1-year Of the 20 individual cancer sites or site groups considered in survival in 1996–2000 was already extremely high (97.8%) and the Table 3 for England, 17 have shown an improvement in 1-year improvement, by 0.1%, was marginal. The improvements in survival between 1996–2000 and 2004–2006 (the exceptions being oesophageal (5.0%), stomach (5.3%) and brain cancer (4.4%) were cervical cancer (0.1% decrease), Hodgkin disease (1.4% decrease) the most substantial. British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK Relative survival (%) Thyroid Prostate Kidney Colorectal Ovary and uterine adnexa Stomach Multiple myeloma Lung Testis Hodgkin Breast Cervix uteri Soft tissue Bone and cartilage All leukaemias Oesophagus Pancreas Corpus uteri Brain Gall bladder and biliary tract Liver Melanoma of skin Bladder Larynx NHL Head and neck ’ What can we learn from recent EUROCARE results? CS Thomson and D Forman S107 DISCUSSION average’, although this is not translated into a differential at 5 years. All of these cancers are associated with reasonable survival Until recently, the only evidence enabling an understanding of the rates overall and thus might benefit most from efforts towards reasons for the relatively poor 5-year survival outcomes in the earlier diagnosis. With bladder cancer, there is also the additional United Kingdom compared with elsewhere in Europe was the problem that changes in coding schemes defining the classification EUROCARE high-resolution studies (Gatta et al, 2000; Sant et al, of invasive tumours and introduced at different times in different 2003, 2007). While these indicated later stage at diagnosis as being countries, may introduce artefactual survival differences. an important explanation for part of the difference, the studies The final group of eight cancers are those that had significantly only covered a few sites of cancer and were based on relatively lower survival rates in England at 5 years, which were also evident small numbers of patients diagnosed during the early 1980s. The at both 1 year and 5|1 years. This group included the three samples of UK patients included in these were not necessarily common cancers of the lung, colorectum and prostate, as well as representative of either their regions of residence, or the country as cancers of the kidney, thyroid, ovary and stomach and multiple a whole. Relevance to the current situation is, therefore, myeloma. The results for these cancers suggest that improving the questionable. stage at diagnosis of these cancers would not have eradicated all of the variation observed in 5-year survival between England and The present review of the EUROCARE-4 results cannot provide an exhaustive evaluation of the relative contribution of stage Europe. There may also have been differences in the treatment at diagnosis to 5-year survival on a site-by-site basis but the given or the management of services that affected the longer-term analysis of 5, 1 and 5|1-year survival helps provide some overall outcome. However, the survival gap for all of these sites, except perspective. stomach, was smaller between 1–5 years than it was between Of the cancer sites considered, one important group is those diagnosis and 5 years (Table 2) indicating a role for earlier cancers for which survival in England was significantly worse than diagnosis. In addition, all of these cancers (except thyroid that was the ‘European average’ at 5 years, but separating this into not analysed by Rachet et al, 2009) have shown recent improve- outcomes after 1 year and between 1 and 5 years (i.e. having ments in 1-year survival in the England data (Table 3). Prostate survived for 1 year) indicated a statistically significant detriment cancer survival results are particularly difficult to interpret given associated only with the former, but not the latter. This the unknown impact of differing intensity of PSA testing in implies that, for these cancers, it is low survival in the first year different countries leading to overdiagnosis of non-fatal disease. after diagnosis, which gives rise exclusively to the relatively poor Coding changes for the classification of invasive behaviour for overall outcome. The major site falling into this category was ovarian cancer make comparisons over time problematic and are breast cancer, but others in this group were cancers of the most apparent when using the different coding classifications ICD- oesophagus, liver, pancreas, cervix and brain. It is noteworthy that O3 and ICD-10, as in Table 3. this group includes both breast and cervical cancers for which the This analysis dissects the EUROCARE-4 results to discriminate, benefit of population-based screening has long been established. among those cancers for which 5-year survival in England seemed Both of these cancers are associated with a generally good to be lagging behind Europe, those for which the explanation prognosis and the poor comparative survival in England is seems to be largely due to effects within the year after diagnosis probably best explained by relatively small subgroups of women (group one, not showing detrimental 5|1–year outcomes) and diagnosed with late stage disease, and not being detected by those for which the explanation is related to short- screening. and longer-term effects (group three, showing detrimental 1 and In contrast, the results for oesophageal, pancreatic and liver 5|1–year outcomes). There is also a group of cancers for which cancers, all with an extremely poor prognosis, may be explained in 5-year survival in England was not statistically different from part by differences in access to new therapeutic regimens for the ‘European average’, but for which there was a detrimental advanced disease, which may have started use in England later effect at 1 year (group two). It should also be noted that, for 10 of than elsewhere in Europe (Mitry et al, 2008a, b; Rao and the 26 cancer sites listed in Table 3, there was no statistical Cunningham, 2008; Starling and Cunningham, 2008; Rimassa difference between the England outcomes and the ‘European and Santoro, 2009). As a group, however, these cancers are often average’ at 5 years, and for only two cancers (head and neck and diagnosed at an advanced, incurable stage. If registration and malignant melanoma), England had significantly better survival follow-up of cases registered initially through death certificates than Europe. were more effective in England than elsewhere, this would result in Efforts and interventions directed towards increased cancer other countries having a higher proportion of cases that are more awareness and earlier detection, if successful, are likely to benefit likely to be registered only through death certification and/or as all forms of cancer. Relative improvements to 1-year survival in cancers with an unknown primary site. These cases have very poor England are likely to benefit those cancers in groups one and two outcomes but need to be excluded from survival analyses. This to such an extent that the survival gap with Europe could almost would result in the poorer 1-year survival observed in England, but completely be overcome. NAEDI could make this improvement to would not influence the 5|1-year survival. Finally brain cancer, most of the cancers in these two groups although, for the poor associated with moderate prognosis, is a type of cancer with a prognosis sites (oesophageal, stomach and liver), life-extending potential for early diagnosis if prompt attention is paid to therapies for advanced disease must also have a role. NAEDI could symptoms and appropriate diagnostic facilities are available. With also have an important function in the cancers in group three, the exception of cervical cancer (marginal decrease) and liver although the current survival gap between England and Europe for cancer (not included in the analysis by Rachet et al, (2009)), all of these cancers is not fully explained by short-term survival the cancers in this group have shown recent improvements in 1- differences and other factors will thus also have a role. year survival in England (Table 3). This study has several limitations. It is acknowledged that the A second group of cancers comprises those that showed non-independence between the survival estimates for England and significantly lower survival at 1 year in England compared with the ‘European average’ means that the method of testing the ‘European average’, which was no longer evident at either 5 or differences between the survival estimates, using only the pooled 5|1 years. The five cancers in this group were bone and cartilage, variance to highlight significant differences, is not perfect. It soft tissue, uterus, bladder and non-Hodgkin lymphoma and, as should, however, be sufficient to help identify which sites require with the preceding group, differences in early diagnosis and/or further efforts to separate out the early and late mortality effects initial management of advanced disease would explain the and where future international benchmarking projects should be disadvantage observed in England compared with the ‘European considered. & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109 What can we learn from recent EUROCARE results? CS Thomson and D Forman S108 De Angelis et al (2009) describe the data quality and resource constraints involved in data collection, are not available completeness across the EUROCARE-4 data sets, presenting for contemporary comparisons. indicators of quality known to affect survival analyses. One of Another weakness currently faced is that though it is known how these measures is the percentage of registrations made solely on survival has improved in England during the past decade, how this the basis of death certificates (DCOs), which are excluded from the compares with improvements made across Europe is not known. It survival calculations thus biasing the results. High DCO rates are is possible that the gaps are closing but have not been fully associated with over-inflation of the survival estimates (Berrino eradicated, and without further studies this cannot be determined et al, 1995a), and the higher DCO rates in England compared with for certain (Berrino et al, 2007; Richards, 2007; Verdecchia et al, Europe (6.1% compared with 2.7%) suggest this is a factor which 2007). This study has focussed on comparing survival within needs consideration and a possible correction could be applied to England with the ‘European average’. It could be argued that as a survival estimates (Silcocks and Thomson, 2009). rich Western European country, England should be striving to be The concept of the ‘European average’ is itself problematic the equivalent of the ‘European best’ not just as good as the mainly because the EUROCARE-4 Study only covers a minority ‘European average’. This was the adopted approach for estimating (about 30%) of the EU population. From the variation within the the numbers of avoidable deaths in the United Kingdom relative to countries with partial coverage, it is likely that the national the ‘European best’ (Abdel-Rahman et al, 2009). estimates for some of these countries would change if 100% There is, therefore, a need for new international benchmarking population coverage were available, thus, in turn, affecting the studies to assess this. These studies will need to compare registry ‘European average’. Another problem with the ‘European average’ data within England with other countries with good quality is that it assumes that all the data sets used in its estimation are of registry data. The data collected will need to contain good staging equal quality in terms of accuracy and completeness of recording information, which has been robustly collected to agreed both incident cases and subsequent deaths. Further discussion of standards, along with information about the diagnostic techniques this is provided in both Møller et al (2009) and Abdel-Rahman used, the presence of any significant comorbidity and details of the et al (2009). treatment that was given. Only then may studies provide the real An alternative to using the ‘European average’ for comparison understanding as to how much stage at diagnosis and earlier with England would be to identify specific populations with which detection improve cancer survival. to compare the results from England; these are likely to be with countries where comparable quality of data to those in England was assured. This is the basis of the future proposed international ACKNOWLEDGEMENTS benchmarking studies. The authors thank Matt Wickenden for creating and formatting all These results also only give a crude assessment of the possible of the charts for this publication. effects of stage, early diagnosis and treatment on outcomes. A better way to examine these effects is to include this information, when it has been collected robustly, directly in the survival Conflict of interest analyses. The high-resolution studies, undertaken as part of the earlier EUROCARE studies, attempted this but, partly because of The authors declare no conflict of interest. REFERENCES Abdel-Rahman M, Stockton D, Rachet B, Hakulinen T, Coleman MP (2009) Corazziari I, Quinn M, Capocaccia R (2004) Standard cancer patient popu- What if cancer survival in Britain were the same as in Europe: how many lation for age standardising survival ratios. Eur J Cancer 40: 2307– 2316 deaths are avoidable? Br J Cancer 101(Suppl 2): S115–S124 De Angelis R, Francisci S, Baili P, Marchesi F, Roazzi P, Belot A, Crocetti E, Berrino F, Capocaccia R, Coleman MP, Este`ve J, Gatta G, Hakulinen T, Pury P, Knijn A, Coleman M, Capocaccia R, EUROCARE-4 Working Micheli A, Sant M, Verdecchia A (eds). (2003) EUROCARE-3: the Group (2009) The EUROCARE-4 database on cancer survival in Europe: survival of cancer patients diagnosed in Europe during 1990–1994. Ann data standardisation, quality control and methods of statistical analysis. Oncol 14(Suppl 5): 1 –155 Eur J Cancer 45: 909–930 Berrino F, Capocaccia R, Este`ve J, Gatta G, Hakulinen T, Micheli M, Sant M, Department of Health (2000) The NHS Cancer Plan. Department of Health: Verdecchia A (eds). (1999) Survival of Cancer Patients in Europe: the London EUROCARE-2 Study (IARC Scientific Publications No. 151). International Department of Health (2007) Cancer Reform Strategy. Department of Agency for Research on Cancer: Lyon Health: London Berrino F, De Angelis R, Sant M, Rosso S, Lasota MB, Coebergh JWW, Gatta G, Capocaccia R, Sant M, Bell CMJ, Coebergh JWW, Damhuis RAM, Santaquilani M, EUROCARE Working Group (2007) Survival for Faivre J, Martinez-Garcia C, Pawlega J, Ponz de Leon M, Pottier D, eight major cancers and all cancers combined for European adults Raverdy N, Williams EMI, Berrino F (2000) Understanding variations in diagnosed in 1995–1999: results of the EUROCARE-4 study. Lancet Oncol survival for colorectal cancer in Europe: a EUROCARE high resolution 8: 773–783 study. Gut 47: 533–538 Berrino F, Este`ve J, Coleman MP (1995a) Basic issues in the estimation and Mitry E, Rachet B, Quinn MJ, Cooper N, Coleman MP (2008a) Survival from comparison of cancer patient survival. In Survival of Cancer Patients in cancer of the oesophagus in England and Wales up to 2001 – survival Europe: the EUROCARE Study (IARC Scientific Publications No. 132), analysis. Br J Cancer 99(Suppl 1): S11 –S13 Berrino F, Sant M, Verdecchia A, Capocaccia R, Hakulinen T, Esteve J Mitry E, Rachet B, Quinn MJ, Cooper N, Coleman MP (2008b) Survival (eds), pp 1 –14. International Agency for Research on Cancer (WHO): from cancer of the pancreas in England and Wales up to 2001 – survival Lyon analysis. Br J Cancer 99(Suppl 1): S21 –S23 Berrino F, Sant M, Verdecchia A, Capocaccia R, Hakulinen T, Esteve J Møller H, Linklater KM, Robinson D (2009) A visual summary of the (1995b) (eds). Survival of Cancer Patients in Europe: the EUROCARE EUROCARE-4 results: a UK perspective. Br J Cancer 101(Suppl 2): S110–S114 Study (IARC Scientific Publications No. 132). International Agency for Rachet B, Maringe C, Nur U, Quaresma M, Shah A, Woods LM, Ellis L, Research on Cancer: Lyon Walters S, Forman D, Steward J, Coleman MP (2009) Population-based Coleman MP, Babb P, Damiecki P, Grosclaude PC, Honjo S, Jones J, Knerer cancer survival trends in England and Wales up to 2007: an assessment G, Pitard A, Quinn MJ, Sloggett A, De Stavola BL (1999) Cancer Survival of the NHS cancer plan for England. Lancet Oncol 10: 351–369 Trends in England and Wales 1971 –1995: Deprivation and NHS Region Rao S, Cunningham D (2008) Survival from cancer of the oesophagus in (Studies on Medical and Population Subjects No. 61). The Stationery England and Wales up to 2001 – clinical commentary. Br J Cancer Office: London 99(Suppl 1): S14 –S15 British Journal of Cancer (2009) 101(S2), S102 – S109 & 2009 Cancer Research UK What can we learn from recent EUROCARE results? CS Thomson and D Forman S109 Richards M (2007) EUROCARE-4 studies bring new data on cancer Sant M, Allemani C, Santaquilani M, Knijn A, Marchesi F, Capocaccia R, survival. Lancet Oncol 8(9): 752–753 EUROCARE Working Group (2009) EUROCARE-4. Survival of cancer Rimassa L, Santoro A (2009) Sorafenib in advanced hepatocellular patients diagnosed in 1995–1999. Results and commentary. Eur J Cancer carcinoma. Expert Rev Anticancer Ther 9(6): 739 –745 45: 931–991 Sant M, Aareleid T, Artiolic ME, Berrino F, Coebergh JW, Colonna M, Silcocks P, Thomson CS (2009) Correcting population-based survival for Forman D, Hedelin J, Rachtan J, Lutz JM, Otter R, Raverdy R, Plesko II, DCOs – why a simple method works and when to avoid it. Eur J Cancer Primic M-Z, Tagliabue G (2007) Ten-year survival and risk of relapse for DOI: 10.1016/j.ejca.2009.06.013 (in press) testicular cancer: A EUROCARE High Resolution Study. Eur J Cancer 43: Starling N, Cunningham D (2008) Survival from cancer of the pancreas in 585–592 England and Wales up to 2001 – clinical commentary. Br J Cancer Sant M, Allemani C, Capocaccia R, Hakulinen T, Aareleid T, Coebergh JW, 99(Suppl 1): S24 –S25 Coleman MP, Grosclaude P, Martinez C, Bell J, Youngson J, Berrino F, Verdecchia A, Francisci S, Brenner H, Gatta G, Micheli A, Mangone L, EUROCARE Working Group (2003) Stage at diagnosis is a key Kunkler I, EUROCARE-4 Working Group (2007) Recent cancer survival explanation of differences in breast cancer survival across Europe. Int in Europe: a period analysis 2000– 2002 of the EUROCARE-4 data. J Cancer 106: 416–422 Lancet Oncol 8: 784 –796 & 2009 Cancer Research UK British Journal of Cancer (2009) 101(S2), S102 – S109

Journal

British Journal of Cancer – Springer Journals

Published: Dec 3, 2009

There are no references for this article.