Cancer survival in five continents: a worldwide population-based study (CONCORD).

BACKGROUND: Cancer survival varies widely between countries. The CONCORD study provides survival estimates for 1.9 million adults (aged 15-99 years) diagnosed with a first, primary, invasive cancer of the breast (women), colon, rectum, or prostate during 1990-94 and followed up to 1999, by use of individual tumour records from 101 population-based cancer registries in 31 countries on five continents. This is, to our knowledge, the first worldwide analysis of cancer survival, with standard quality-control procedures and identical analytic methods for all datasets. METHODS: To compensate for wide international differences in general population (background) mortality by age, sex, country, region, calendar period, and (in the USA) ethnic origin, we estimated relative survival, the ratio of survival noted in the patients with cancer, and the survival that would have been expected had they been subject only to the background mortality rates. 2800 life tables were constructed. Survival estimates were also adjusted for differences in the age structure of populations of patients with cancer. FINDINGS: Global variation in cancer survival was very wide. 5-year relative survival for breast, colorectal, and prostate cancer was generally higher in North America, Australia, Japan, and northern, western, and southern Europe, and lower in Algeria, Brazil, and eastern Europe. CONCORD has provided the first opportunity to estimate cancer survival in 11 states in USA covered by the National Program of Cancer Registries (NPCR), and the study covers 42% of the US population, four-fold more than previously available. Cancer survival in black men and women was systematically and substantially lower than in white men and women in all 16 states and six metropolitan areas included. Relative survival for all ethnicities combined was 2-4% lower in states covered by NPCR than in areas covered by the Surveillance Epidemiology and End Results (SEER) Program. Age-standardised relative survival by use of the appropriate race-specific and state-specific life tables was up to 2% lower for breast cancer and up to 5% lower for prostate cancer than with the census-derived national life tables used by the SEER Program. These differences in population coverage and analytical method have both contributed to the survival deficit noted between Europe and the USA, from which only SEER data have been available until now. INTERPRETATION: Until now, direct comparisons of cancer survival between high-income and low-income countries have not generally been available. The information provided here might therefore be a useful stimulus for change. The findings should eventually facilitate joint assessment of international trends in incidence, survival, and mortality as indicators of cancer control.

Patient survival for all cancers combined as indicator of cancer control in Europe.

BACKGROUND: EUROCARE found marked differences in cancer survival across European populations, provoking extensive discussion as to the cause. We investigated the influence of socioeconomic indicators on survival, making use of the indicator population-based age-standardized and cancer site-standardized relative survival for all cancers combined (all cancer survival). METHODS: Bivariate correlation and multivariate regression analyses investigated relations between 1995 socioeconomic variables and all cancer survival in EUROCARE-3 patients from 19 European countries diagnosed 1990-94 and followed to 1999. RESULTS: Gross domestic product (GDP) and total national expenditure on health (TNEH) correlated highly with all cancer survival. Wealthy northern and western European countries had high survival; eastern European countries had low all cancer survival. GDP, TNEH, and number of computed tomography scanners per million--proxy of technological investment in cancer care--explained most survival differences. Low all cancer survival in the UK and Denmark compared to countries of similar wealth was closely related to fewer computed tomography scanners. Low all cancer survival in Poland compared to countries of similar wealth was also related to low TNEH. CONCLUSIONS: All cancer survival appears a useful and important indicator for monitoring countries' performance in cancer control. The most direct way for poorer European countries to improve all cancer survival would be to get richer; for richer countries more investment in health technology is important. However the sharply increasing costs of cancer care may render this impossible suggesting the need to radically rethink cancer control strategies.

Methodology for estimation of cancer incidence, survival and prevalence in Italian regions.

BACKGROUND: The Italian health care system is based on a regional responsibility and organization. Incidence, survival and prevalence of cancer patients were estimated for major cancer sites by single regions in Italy within a collaborative project "I TUMORI IN ITALIA" aimed at providing epidemiological information in a web site, www.tumori.net, for health operators, health programmers, epidemiologists and the general public. The aim of this paper is to present the methodology used to derive regional estimates of cancer burden indicators in Italy for major cancer sites. METHODS: Estimates require a complex methodology to be used. We present herein the methodology to provide regional estimates that involves a combined use of multiple methods to model and extrapolate patient survival to derive incidence and prevalence estimates and future projections. RESULTS: Regional patient survival estimates in Italy are presented and discussed. Cancer survival has greatly improved over the years and differences between northern-central and southern regions have persisted. The fraction of patients cured from cancer is today over 50% for young men and women aged 15-44 and declines to 15%-25% at older ages (75-99). DISCUSSION: We integrated different methods to derive estimates of cancer burden in Italy, at a regional level, in order to take advantage of all information available and to obtain the most reliable estimates. The value of producing regional estimates of cancer burden indicators was motivated by the lack of such information systematically on the Italian national territory.

Regional estimates of colorectal cancer burden in Italy.

AIMS AND BACKGROUND: In terms of new diagnoses, colorectal cancer is one of the most important cancers in Italy and worldwide. The aim of this paper is to present estimates of the mortality, incidence and prevalence of colorectal cancer in Italy at a national and regional scale over the period 1970-1999, with projections up to 2010. METHODS: The estimates were obtained by applying the MIAMOD method, a statistical back-calculation approach to derive incidence and prevalence estimates from mortality and relative survival data. Published data from the Italian cancer registries were modelled to obtain regional and national estimates of colorectal cancer survival. RESULTS: Different incidence patterns were observed for men and women, especially in the projection period: the national age-standardized rate is estimated to increase throughout the study period 1970-2010 for men from 30 to 70 per 100,000, and to stabilize from the end of the 1990s for women at around 38 per 100,000. A stabilization or a slight decrease in age-standardized incidence rates is expected in most regions for women and in most northern-central regions for men. The most critical situation is estimated among men for southern regions, where the rise in incidence is accompanied by a dramatic increase in mortality. About 46,000 incident cases, 267,000 prevalent cases, and 16,000 deaths from colorectal cancer are estimated in Italy for the year 2005. CONCLUSIONS: Despite the risk reduction estimated in most northern-central regions among men and in the large majority of regions among women, the colorectal cancer burden in Italy is expected to remain relevant in the next years. Prospects for reducing this burden appear mainly connected to the adoption of prevention policies aimed at increasing the awareness of the risk related to dietary habits and lifestyles and at promoting colorectal cancer screening.

Regional estimates of breast cancer burden in Italy.

AIMS AND BACKGROUND: Breast cancer is the most common cancer and the leading cause of cancer death among women. Knowledge of the present and future burden of the disease at a regional and national scale is a major issue in Italy, where the frequency and coverage of screening programs vary considerably across the country. This study presents estimates and projections of the female breast cancer incidence, prevalence and mortality for Italy and all Italian regions in the period 1970-2010. METHODS: The estimates were obtained by applying the MIAMOD method, a statistical back-calculation approach to derive incidence and prevalence figures from mortality and relative survival data. Published data from the Italian cancer registries were modelled to obtain regional and national estimates of breast cancer survival. RESULTS: Breast cancer mortality has been declining from the late 1980s in the northern-central regions and from the mid 1990s in the southern regions Puglia, Sicilia and Sardegna. Stable mortality rates are estimated for the other southern regions in the 2000's first decade. The incidence rate in Italy is estimated as increasing until the late 1990s, and stable thereafter (93 per 100,000). The incidence curve is also estimated to flatten in many northern-central regions from the late 1990s or later. Rising incidence trends are estimated in all southern regions, with the exception of Puglia. About 8,500 deaths, 37,000 new diagnoses and 416,000 prevalent cases for breast cancer are estimated among Italian women in 2005. In the same year, the proportion of prevalent cases in the northern area (1221 per 100,000) is about twice that estimated in the South (685 per 100,000). DISCUSSION: The geographical variation in female breast cancer burden can be explained by the unequal distribution of screening. A more widespread screening activity in the southern regions would help to bridge the gap between northern-central and southern regions. Continuous monitoring of regional epidemiological indicators for breast cancer is crucial to evaluate the effect of different health measures taken to control breast cancer in Italy.

Regional estimates of all cancer malignancies in Italy.

AIMS AND BACKGROUND: The aim of this paper is to present regional and national estimates of mortality, incidence and prevalence for all cancers in Italy over the period 1970-1999, with projections up to 2010. METHODS: The estimates were obtained by applying the MIAMOD method, a statistical back-calculation approach, to derive incidence and prevalence starting from mortality and relative survival data. Published data from the Italian Cancer Registries were modeled in order to estimate regional and national cancer survival. RESULTS: Cancer time trends resulted more favorable in northern-central regions than in southern regions, both for men and women. Mortality started to decrease in the northern-central area approximately from the mid 1980s, whereas it was expected to slightly decline only after the year 2000 in the southern area. Incidence was estimated to decrease in men from 1995 in northern and central areas only; no incidence reduction is expected for women during the study period. Overall, 130,000 cancer deaths, 250,000 new cancer cases and 1,700,000 prevalent cancer cases are estimated in Italy in the year 2005. CONCLUSIONS: This up-to-date picture of cancer risk and burden in the Italian regions shows as a relevant epidemiological change is ongoing in Italy. Although a clear geographical variability in mortality and morbidity levels still exists across the country, the historical North-to-South gap appears smaller than in the past. This change is particularly remarkable for men, as a consequence of trends that are favorable in northern-central regions but not yet in southern regions.

Regional estimates of lung cancer burden in Italy.

AIMS AND BACKGROUND: The aim of this paper is to present the Italian regional and national estimates of the mortality, incidence and prevalence over the period 1970-2010 for lung cancer, one of the most common cancers in Italy and Europe, especially among men. METHODS: The estimates were obtained with the MIAMOD method, a statistical back-calculation approach to derive incidence and prevalence figures using mortality and relative survival data. Published survival data from the Italian cancer registries were appropriately modelled in order to estimate survival at regional and national level. RESULTS: For men, the Italian lung cancer mortality and incidence rates reached their maximum values during the late 1980s and steeply decreased thereafter. This pattern is quite uniform across all regions, even though the decrease was delayed and less steep in the South of the country. For women, both indicators appear to be increasing at the national level, even though a certain regional variability in incidence trends remains within the country: some regions (such as Lazio, Umbria, Marche, Campania, Sardegna and Friuli Venezia Giulia) present increasing trends, while in others a clear levelling off is visible. A total of 33,500 incident cases, 72,000 prevalent cases and 28,000 deaths are estimated in Italy in 2005. CONCLUSIONS: This study produced an up-to-date description of lung cancer epidemiology both at national and regional level. The marked difference in lung cancer figures and trends between sexes suggests that prevention policies have led to changes in smoking habits for men, while for women they were lacking in efficacy. The attention should now focus on women, whose risk trend is the least favorable.

Regional estimates of stomach cancer burden in Italy.

AIMS AND BACKGROUND: Stomach cancer still remains one of the most frequent tumors in Italy and Europe. The aim of this paper is to present estimates for stomach cancer mortality, incidence and prevalence over the period 1970-2010 for the Italian regions and for Italy as a whole. METHODS: Estimated figures for incidence, prevalence and mortality were obtained by using the MIAMOD method. Starting from the knowledge of mortality in the period 1970-1999 and of relative survival in the period of diagnosis 1978-1994, we derived incidence and prevalence estimates and projections up to the year 2010 by means of a statistical back-calculation approach. Survival at the regional and national levels was modelled on the basis of published survival data from the Italian cancer registries. RESULTS: Incidence and mortality trends for both sexes decrease by about 60% during the estimation period 1970-2010. Both indicators show a 2-fold male/female ratio all over the country, and a similar gender time trend. The incidence and mortality in the North and Center of the country are estimated to be higher and to decrease more steeply than those in the South, both for men and women. A total of around 13,000 incident cases, 57,000 prevalent cases, and 8,000 deaths are estimated to have occurred in Italy in 2005. CONCLUSIONS: The incidence and mortality trends are estimated to decline during the entire period 1970-2010, with different slopes between northern-central and southern regions. The incidence and mortality are quite similar among Italian regions, showing that the risk of developing the disease diminishes and is becoming more homogeneous than in the past decades all over the country.

Regional estimates of prostate cancer burden in Italy.

AIMS AND BACKGROUND: Prostate cancer is one of the most common cancers in developed countries and the most common among men in industrialized countries. The introduction of new diagnostic procedures caused an increase in new diagnoses in Italy starting from the early 1990s, while the prognosis of prostate cancer improved due to the use of hormonal treatments. The aim of this paper is to present estimates of prostate cancer mortality, incidence and prevalence over the period 1970-2005 for the Italian regions and for Italy as a whole, and to assess the changes that opportunistic screening and the diffusion of more effective treatments introduced. METHODS: Estimated figures for incidence, prevalence and mortality were obtained with the MIAMOD method. Starting from the knowledge of mortality in the period 1970-1999 and of the relative survival in the period of diagnosis 1978-1994, we derived incidence and prevalence estimates up to the year 2005 by means of a statistical back-calculation approach. Survival at regional and national levels was modelled on the basis of published survival data from the Italian cancer registries. RESULTS: The incidence trend showed a steep increase all over the country during the entire estimation period 1970-2005 with a more pronounced increase in the Center-North then in the South of Italy. Incidence of northern and central regions was about twice as high as that of southern regions. Mortality trends were however constant or declining in the majority of northern-central regions, while they still increased in the South. A total of around 43,000 incident cases, 174,000 prevalent cases and 9,000 deaths were estimated for Italy in 2005. DISCUSSION: The effects of opportunistic screening are reflected in an earlier diagnosis for many patients. The existing North-South gradient in incidence seems to be associated with the different spread of the PSA test in different parts of the country. Prostate cancer remains a great health problem in terms of both incidence and prevalence.

Cancer prevalence estimates in Italy from 1970 to 2010.

AIMS AND BACKGROUND: The growing number of cancer survivors in Italy is expected to continue to increase as a consequence of population aging and survival improvements, but few estimates are currently available, particularly on the national and regional scale. The purpose of this work is to present detailed and updated prevalence estimates in Italy over the period 1970-2010 by cancer site (all cancers combined, stomach, colon and rectum, lung, breast and prostate) and gender. METHODS: Prevalence was derived with the MIAMOD statistical method, using cancer-specific mortality and relative survival as input data. Survival data from the Italian cancer registries were modeled to derive a national estimate for each cancer site and sex. To estimate prevalence trends, survival was assumed to improve in the future with the same rate observed in the period 1978-1994. A double scenario for survival--increasing or stationary--was considered to decompose the prevalence growth from 1995 to 2005 by its determinants: demographic changes, incidence and survival dynamics. The prevalence estimates were also decomposed by disease duration (2, 5 and 10 years) and by age (0-44, 45-59, 60-74 and 75-99). RESULTS: The proportion of cancer survivors in 2010 is expected to be about 4% in women and 3% in men, about twice the values attained in 1990. The highest dynamics was observed for prostate cancer, with a three-fold increase just in the 1995-2005 period (from 212 to 623 per 100,000), whereas in absolute terms breast cancer presented the highest levels (1700 per 100,000 in 2010). The overall number of cancer prevalent cases is expected to rise by about 48% in the decennium 1995-2005 (from 1,152,000 to 1,709,000), and this growth is mainly attributable to incidence dynamics (+21%), then to survival improvements (+14%) and population aging (13%). In 2005, the 2-year prevalent cases were estimated to be 20% of all cancer survivors, 21% between 2 and 5 years from the diagnosis, 23% between 5 and 10 years, with 36% surviving for more than 10 years. Prevalence proportion was very high in the elderly (12.6% for 75-84 years and 8% for 60-74 years). CONCLUSIONS: Updated prevalence data with appropriate coverage of the national territory are essential to define priorities in health care management and to develop cancer control programs. Prevalence by disease duration and by age should be the basis for planning research on the quality of life of cancer survivors, as long as cancer continues to become an even more chronic disease.

Italian cancer burden by broad geographical area.

AIMS AND BACKGROUND: Cancer burden estimates in Italian regions are available for the period 1970-2010 as a result of the project "I TUMORI IN ITALIA" connected with EUROCHIP, the European project on cancer control. The Italian health-care system is organized at a regional level, so regional estimates of cancer indicators are useful to identify priorities for cancer plans. We compared cancer site-specific epidemiological estimates by 3 macro-areas (obtained by grouping regions) to suggest priorities for Italian cancer control plans, both at national and regional levels. METHODS: Mortality and incidence estimates for all cancers combined and for stomach, colorectal, lung, breast and prostate cancers were downloaded from the website www.tumori.net and aggregated in broad age classes (0-54, 55-74 and 75-84 years) and macro-areas (northern, central and southern Italy). RESULTS: Historically, Southern Italy had a lower cancer risk than the Center and North. After 2000 this epidemiological picture disappeared and the incidence and mortality rates in the Center are reaching those of the North. Also the weight of various cancer sites on all cancers has changed in Italy in the last decades. Lung cancer is still the most frequent cancer in the male population in the South, while in the Center-North it has been surpassed by prostate cancer and colorectal cancer. The lung cancer weight on all cancer deaths is increasing in women. Prostate cancer has become the most frequent male cancer in the Center-North in the age class 55-84. Breast cancer is the most frequent cancer in the female population and its incidence rates in the North are higher than those in the Center-South for all age classes. Colorectal cancer incidence rates have dramatically increased in men and colorectal cancer is nowadays the second cancer diagnosed in women in all age classes and macro-areas. DISCUSSION: From the epidemiological data here presented we derived the following suggestions and observations for cancer control plans: (a) tobacco prevention should focus on the male population in the South, and on female populations in the country as a whole; (b) prevention concerning diet and physical activity (risk factors for colorectal cancer) should be considered mainly for men at a national level; (c) the coverage of breast cancer screening programs should be increased in the Center-South; (d) colorectal cancer screening should be promoted at a national level; (e) PSA testing (that is not actually included among the screening programs recommended) for prostate cancer is probably more widespread in the Center-North, resulting in an increased incidence without any evident decline in mortality as yet.

Reproductive disorders and pregnancy outcomes among female flight attendants.

BACKGROUND: Occupational exposures in female aircrew may cause adverse pregnancy outcomes and menstrual disturbances. We studied reproductive health among female flight attendants. METHODS: We conducted a cross-sectional health survey among an occupational cohort of current and former flight attendants using a postal questionnaire including items on pregnancy outcome, menstrual characteristics, and infertility. We investigated these factors by occupational status (in service or not). RESULTS: The questionnaire was sent to 3036 women with a response rate of 64% (74% for current and 48% for former flight attendants). Spontaneous abortion rates were similar for pregnancies of women in service and not (12.6 vs. 11.4%; p = 0.58). Induced abortion rates were lower for in-service pregnancies (7.9%) compared with pregnancies of women not in service (21.1%) (p < 0.001). Menstrual irregularities in the year preceding the survey for women under 40 yr were more frequent among current than former flight attendants (20.6% vs. 10.4%, p = 0.02). Fertility problems were reported by 20.6% of respondents. An association between infertility and irregular menstrual cycles was found: odds ratio 1.6; 95% confidence limits 1.1, 2.4. CONCLUSIONS: Active flight attendants had a lower rate of induced abortions than former flight attendants, but the spontaneous abortion rate was similar between the two groups. Active flight attendants reported more menstrual irregularities, which are a risk factor for infertility. Lack of comparison with working women in other occupations precludes a conclusion that flight attendants are at greater occupational risk of reproductive disorders. Future studies of reproductive health in flight attendants should address personal and work-related risk factors in more detail and include comparison occupational groups.

National estimates of cancer patients survival in Italy: a model-based method.

AIMS AND BACKGROUND: To provide model-based estimates of all cancers patient survival in Italy and in Italian large geographical areas (North-West, North-East, Center, South), where only partial coverage of cancer registries data is available, and to describe them in terms of time trends. Moreover, to measure the degree of representativeness of cancer patient survival obtained from Italian cancer registries data. METHODS: Relative survival in the four main Italian geographical areas was estimated by a parametric mixture model belonging to the class of "cure" survival models. Data used are from Italian cancer registries, stratified by sex, period of diagnosis and age. The Italian national survival was obtained as a weighted average of these area-specific estimates, with weights proportional to the number of estimated incident cases in every area. The model takes into account also differences in survival temporal trends between the areas. RESULTS: Relative survival for all cancers combined in Italian patients diagnosed in 1990-1994 was estimated to be higher in women (53%) than in men (38%) at 5 years from the diagnosis. The survival trend is increasing by period and decreasing by age, both for men and women. The greatest gain in terms of survival was obtained by the elderly, with annual mean growth rates in the period 1978-1994 equal to 3.5% and 3.2% for men and women, respectively. More than 50% of the youngest cancer patients were "cured", whereas for the elderly this proportion dropped to 15% and 25% for men and women, respectively. The South of Italy had the lowest survival and the North the most pronounced increase. CONCLUSIONS: The obtained national survival estimates are similar, but not identical, to previously published estimates, in which Italian registries' data were pooled without any adjustment for geographical representativeness. The four Italian areas have different survival levels and trends, showing variability within the country. The differences in survival between men and women may be explained by the different proportion of lethal cancers. Among males, most cases had a poor prognosis (lung and stomach cancers), whereas among females the largest proportion was made up of curable and less lethal cancers (breast cancer).

A retrospective cohort mortality study of Italian commercial airline cockpit crew and cabin attendants, 1965-96.

A retrospective cohort mortality study was conducted among Italian commercial flight personnel for the period 1965-1996. The cohort was composed of 3,022 male cockpit crew members and 3,418 male and 3,428 female cabin attendants. Cause-specific standardized mortality ratios (SMRs) and exact 95% confidence intervals (CIs) were calculated as estimates of the relative risk. Mortality from all cancers was less than expected for all categories (SMRs of 0.58 for male cockpit crew, 0.67 for male cabin attendants, and 0.90 for female cabin attendants). Among male flight personnel, the SMR for leukemia was somewhat elevated (SMR 1.73; 95% CI: 0.75-3.41) based on eight deaths, with a positive trend by length of employment (p = 0.046). Additionally, an excess of death by suicide was seen among female cabin attendants (SMR 3.38; 95% CI: 1.24-7.35). Other Italian studies of flight personnel are under way, including a detailed assessment of cosmic radiation exposure and investigations of non-radiation occupational risk factors and prevalence of nonfatal outcomes.

What reasons lie behind long-term survival differences for gastric cancer within Europe?

BACKGROUND: Wide geographic variations in survival for gastric cancer in Europe have been reported. The aim of this study was to analyse the effect of stage at diagnosis, treatment and cancer characteristics on long-term survival for gastric cancer in populations covered by cancer registries. METHODS: We analysed survival in 4620 cases of gastric cancer from 17 European population-based cancer registries from 8 countries. Univariate and multivariate regression of relative survival were performed. RESULTS: Five-year relative survival varied between 10.6% and 24.0%, while 10-year survival ranged from 7.7% to 23.0%. After adjustment for age and sex, the regional excess hazard ratio (EHR) of death was significantly higher in Ragusa, Granada, Yorkshire, Slovakia, Slovenia and Poland than in France, Northern Italy, The Netherlands and the Basque Country. After further adjustment for surgical resection versus no resection (a proxy of stage), the EHR of death remained significantly higher only in Granada and Yorkshire than in the reference country (France). After adjustment for stage, the EHR was significantly higher only in Yorkshire (EHR: 1.51; 95% confidence interval (CI): 1.29-1.77). The EHR in this area was limited to the first year following diagnosis. CONCLUSION: Differences across Europe in gastric cancer survival depend to a large extent on differences in stage at diagnosis. However they do not explain all variations. Quality of management and treatment can explain some differences.

Survival for cancer patients in Europe.

Comparing cancer survival among the European countries is important to evaluate the performance of Health Care Systems and reduce disparities in access to diagnostic and treatment facilities. The EUROCARE project compares survival in Europe since the nineties. The EUROCARE- 4 analysed 2 690 922 adult cancer cases from 83 cancer registries in 22 European countries, diagnosed in 1995-1999, and followed to December 2003. For each cancer site, the European area weighted mean and age-standardised country-specific observed and relative survival by age and sex is computed. Within-country variation in survival is analysed for selected cancers. Survival for most solid cancers, whose prognosis depends largely on stage at diagnosis (breast, colorectal, stomach cancers, and skin melanoma), was highest in Finland, Sweden, Norway, and Iceland, lower in the UK and Denmark, and lowest in the Czech Republic, Poland and Slovenia. France, Switzerland and Italy generally had good survival, slightly below that of the Northern countries. For all cancers, five-year survival was very variable also for the different sites mix. Continuing to monitoring cancer survival in Europe is important to reduce differences in access to diagnostic and therapeutic facilities. After publication of EUROCARE results, UK and Denmark developed a National cancer plan to improve time of diagnosis and treatment.

Survival trends in European cancer patients diagnosed from 1988 to 1999.

We analysed data from 49 cancer registries in 18 European countries over the period 1988-1999 to delineate time trends in cancer survival. Survival increased in Europe over the study period for all cancer sites that were considered. There were major survival increases in 5 year age-adjusted relative survival for prostate (from 58% to 79%), colon and rectum (from 48% to 54% men and women), and breast (from 74% to 83%). Improvements were also significant for stomach (from 22% to 24%), male larynx (from 62% to 64%), skin melanoma (from 78% to 83%), Hodgkin disease (from 77% to 83%), non-Hodgkin lymphoma (from 49% to 56%), leukaemias (from 37% to 42%), and for all cancers combined (from 34% to 39% in men, and from 52% to 59% in women). Survival did not change significantly for female larynx, lung, cervix or ovary. The largest increases in survival typically occurred in countries with the lowest survival, and contributed to the overall reduction of survival disparities across Europe over the study period. Differences in the extent of PSA testing and mammographic screening, and increasing use of colonoscopy and faecal blood testing together with improving cancer care are probably the major underlying reasons for the improvements in survival for cancers of prostate, breast, colon and rectum. The marked survival improvements in countries with poor survival may indicate that these countries have made efforts to adopt the new diagnostic procedures and the standardised therapeutic protocols in use in more affluent countries.