Search

EUROCARE-3 analysed the survival of 1815584 adult cancer patients diagnosed from 1990 to 1994 in 22 European countries. The results are reported in tables, one per cancer site, coded according to the International Classification of Diseases (ICD)-9 classification. The main findings of the tables are summarised and commented on in this article. For most solid cancers, wide differences in survival between different European populations were found, as also reported by EUROCARE-1 and EUROCARE-2, despite a remarkable (10%) overall increase in cancer survival from 1985 to 1994. Survival was highest in northern Europe (Sweden, Norway, Finland and Iceland), and fairly good in central-southern Europe (France, Switzerland, Austria and Spain). Survival was particularly low in eastern Europe, low in Denmark and the UK, and fairly low in Portugal and Malta. The mix of tumour stage at diagnosis explains much of the survival differences for cancers of the digestive tract, female reproductive system, breast, thyroid, and also skin melanoma. For tumours of the urinary tract and prostate, the differences were explained mainly by differences in diagnostic criteria and procedures. The case mix by anatomic subsite largely explains differences in survival for head and neck cancers. For oesophagus, pancreas, liver and brain cancer, with poor prognoses, survival differences were limited. Tumours, for which highly effective treatments are available, such as testicular cancer, Hodgkin's lymphoma and some haematological malignancies, had fairly uniform survival across Europe. Survival for all tumours combined (an indicator of the overall cancer care performance of a nation's health system) was better in young than old patients, and better in women than men. The affluence of countries influenced overall cancer survival through the availability of adequate diagnostic and treatment procedures, and screening programmes.

Summary International differences and trends in cancer survival withinEurope are larger than can reasonably be accounted for by arte-fact, bias or chance. The geographical patterns and trends in sur-vival are often broadly consistent with geographical differences ortrends in the type of cancer, diagnostic investigations or overallinvestment in health care, and for several major cancers, support-ing evidence is available from population-based studies of clinicalinformation. Incomplete ascertainment of cancer cases, particu-larly of long-term survivors, may contribute to some regional andinternational differences in survival, however, and more system-atic information on completeness is required. We may concludethat large international differences in survival do exist for manycancers, but we should be cautious in drawing quantitative orcausal conclusions from observational survival data.We do not yet have a fully satisfactory interpretation of thesedifferences, but we have few alternatives to this type of study if weare to understand the determinants of improved outcome for allcancer patients, and to enable better planning of their health care.The EUROCARE Working Group has developed several strategiesto disentangle the various possible explanations [73]. These includefurther development of high-resolution studies to examine theimpact on survival differences of disease stage, staging techniquesand treatment; and further development of mathematical modelsof cure. Extension of systematic international survival compari-sons to other regions of the world, such as Australia, Canada, Japanand the USA, is also in progress (the CONCORD study) [22].Oncologists and epidemiologists may provide insight into thegeographic differences and trends in survival reported by thisstudy, and may suggest further lines of enquiry. Do we need morerefined studies of survival to monitor progress against cancer andto plan future cancer care? Will such analyses help us quantify theeffect of new treatments arising from recent progress in the basicsciences and genomics on population cancer survival rates? Sub-stantial human and financial resources are required to improve theoutcome of cancer treatment. Will future investments in cancerservices include matching investment to monitor their impact onsurvival and mortality?Earlier diagnosis and prompt, universal access to optimal treat-ment would be expected to reduce international differences incancer survival in Europe. To achieve this, oncologists and healthcare planners will need better information on the comparativeperformance of their health systems. Population-based cancer reg-istries provide some of the information for such comparisons, buttheir traditional output may no longer be sufficient to evaluate theeffectiveness of health systems, and especially to explain geo-graphical differences in survival. In some countries, their role isalso under threat. Confidentiality constraints recently inhibitedthe collection of cancer registration data in the UK [90], and thelinkage of cancer registrations and deaths is currently illegal inEstonia [91]. Both activities are essential for internationally com-parable survival rates. Legal protection for cancer registrationacross Europe will be required.The mission of cancer registries should be reconsidered, and thepriority shifted from classical descriptive epidemiology and geo-graphical pathology toward more analytical monitoring ofprogress against cancer, including the probability of survival andcure, the burden of cancer prevalence, and the late effects oftherapy. Several European studies of this type have been reportedrecently [3, 36, 9294] and others are in progress. Many cancer–registries are developing closer relationships with cancer clini-cians and general practitioners, and some now systematicallycollect detailed clinical information that was collected eitherirregularly or not at all in the past. These developments willimprove the power of population-based cancer data to explain dif-ferences in cancer survival, and should enhance their relevance toclinical practice.European average survival rates are useful for comparativepurposes, but they should not become the goal for cancer controlprogrammes: the benchmark should always be the highest achiev-able survival rates.The aim of exploring geographic differences in cancer survivalis not to establish international league tables or to excite nationalrivalries, but to estimate the range of survival rates, and to identifyregions or countries in which survival could be improved.

International audience; We explored the influence of morphology on geographic differences in 5-year survival for non-Hodgkin lymphoma (NHL) diagnosed in 1990-1994 and followed for 5years: 16,955 cases from 27 EUROCARE-3 cancer registries, and 22,713 cases from 9 US SEER registries. Overall 5-year relative survival was 56.1% in EUROCARE west, 47.1% in EUROCARE east and 56.3% in SEER. Relative excess risk (RER) of death was 1.05 (95% confidence interval (CI) 1.01-1.10) in EUROCARE west, 1.52 (95% CI 1.44-1.60) in EUROCARE east (SEER reference). Excess risk of death was significantly above reference (diffuse B lymphoma) for Burkitt's and NOS lymphoma; not different for lymphoblastic and other T-cell; significantly below reference (in the order of decreasing relative excess risk) for NHL NOS, mantle cell/centrocytic, lymphoplasmacytic, follicular, small lymphocytic/chronic lymphocytic leukaemia, other specified NHL and cutaneous morphologies. Interpretation of marked variation in survival with morphology is complicated by classification inconsistencies. The completeness and standardisation of cancer registry morphology data needs to be improved.

The EUROCARE database contains data on 6.5 million cancer patients diagnosed from 1978 to 1994 in populations covered by 67 cancer registries in 22 European countries. The quality-checked entries specify age, sex, diagnosis date, cancer site, morphology, microscopic confirmation and vital status, as well as containing broad indicators of stage. For EUROCARE-3, which refers to diagnoses from 1990 to 1994, 3389 cases with major data problems and 142,525 second or subsequent cancers were removed, leaving more than 2 million cases for analysis. From these data, observed and relative survival for each cancer site and country were calculated at 1, 3 and 5 years from diagnosis. Overall European survival for each cancer site and for all cancers combined were calculated combining country-specific survival figures. Overall, 1.1% of cases were lost to follow-up, 4.2% were known from death certificates only and 1.2% were known at autopsy only. The percentage of microscopically confirmed cases varied with cancer site and country, and was always higher in northern European countries. Comparison of quality indicators for the EUROCARE-3 database with earlier EUROCARE databases indicates that data quality and standardisation have improved.