1 European Institute of Oncology, Milan, Italy; 2 Institut Gustave Roussy, Villejuif, France
Received 30 April 2003; accepted 23 May 2003
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Against a background of increasing cancer rates in the mid-1980s, Europe Against Cancer launched an ambitious programme aiming to reduce cancer mortality by 15% by the year 2000. A programme of activities and research, focussing on three major themes [prevention (particularly tobacco control), screening, and education and training], was developed together with the European Code Against Cancer.
Methods:
To evaluate the outcome of the programme, all cancer mortality data were abstracted for each member state of the European Union (EU) until the most recent year available. For each gender group in each member state, age-specific rates were estimated for 2000. For each countrygender grouping, the standardized mortality rate (SMR) and expected numbers of deaths in 2000 were calculated based on the age-specific rates for 1985 and the population estimates for 2000. The ratio of the SMR in 2000 to that of 1985, was used as the measure to compare mortality rates.
Results:
In 1985, there were 850 194 cancer deaths in the EU. Based solely on demographic considerations, this number was expected to rise to 1 033 083 in the year 2000. Between 1985 and 2000, the number of cancer deaths increased in both men (+12%) and women (+9%). The target of a 15% reduction in the expected numbers of cancer deaths in the EU was not met, although the 10% reduction in number of deaths expected in men and 8% in women, along with a 11% reduction in risk of cancer death in men and a 10% reduction in women, was noteworthy. Only Austria and Finland achieved the 15% reductions in deaths in both men and women. The UK and Luxembourg (where the small population and annual number of deaths make interpretation difficult) came close to meeting these targets, as did Italy. Portugal and Greece had the poorest performance, with increases in each gender group.
Conclusions:
Cancer deaths in the EU were expected to rise from 850 194 in 1985 to 1 033 083 in 2000. It is estimated that there will be 940 510 cancer deaths that year, due to the decline in risk observed since 1985. The Europe Against Cancer programme appears to have been associated with the avoidance of 92 573 cancer deaths in the year 2000. With few exceptions, most countries are experiencing declining trends in cancer death rates, which seem set to continue, at least in the near future. Renewed tobacco control efforts are clearly needed for women, and there is a strong case for the introduction of organized breast and cervix screening programmes in all member states. Continuing to emphasize prevention within cancer control will help to promote the continuing decline in death rates in the future.
Key words: cancer, epidemiology, Europe, mortality, survival
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
To achieve this ambitious goal the European Commission adopted a partnership approach, aimed at involving everyone concerned with the fight against cancer at national level. The partners were: (i) the Committee of Cancer Experts (the scientific soul of the programme); (ii) the cancer associations and leagues, and anti-smoking organizations of the European Community (the spearhead of the programme); (iii) producers of television medical programmes who help spread the message of cancer prevention; (iv) representatives of general practitioners, who play a central role in early detection and screening for cancer; and (v) senior officials in the Health, Education and Research Ministries. This European Commission programme was formulated by a High Level Cancer Experts Committee and has seen four 5-year cancer plans evaluated and introduced. The activities of the Committee focussed on three areas: prevention, screening, and education and training. At around the same time, the United States National Cancer Institute promoted a similar goal.
Comparison of cancer mortality rates between countries, and even over time within countries, is not without potential difficulties which must be kept in mind when interpreting the results. To be counted as a death from a particular cancer, at least three independent steps must take place: the cancer must (i) be diagnosed correctly; (ii) be entered onto the death certificate correctly; and (iii) be correctly selected as the underlying cause of death. This process does have inherent potential errors, although these vary according to different types of cancer. However, the time period covered by this study is relatively short and unlikely to be greatly affected by such variations. In addition, one impact of the European Union (EU) has been the coordination of efforts to have a closer standardization of death certification practices within the EU.
In order to evaluate the effect of the Europe Against Cancer programme on its target of reducing cancer mortality in Europe by 15% by the year 2000, all mortality data were assembled from all member states of the EU. Sweden, Austria and Finland were not member states when the target was initially developed, but have been included for completeness since many of the programmes launched were taken up by applicant member states.
![]() |
Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The mean numbers of deaths in each age class, from all forms of cancer, and the five leading forms of cancer (lung, stomach, colorectum, breast and prostate) were calculated over the period 19841986. These numbers are frequently referred to as the number of deaths in 1985. In all calculations, the baseline year for the SMR was chosen to be 1991 since this was the census year in most countries and, hence, the population numbers would be more accurately known [5].
The SMR for 2000 was estimated in two ways. First, using a linear regression of the SMR for the last 10 years of data availability and predicting the value for 2000. For a country with data available until 1998, the SMRs from 19891998 were used and the prediction was based on the linear model. Secondly, a generalized additive model based upon a smoothed cubic spline with six degrees of freedom (df) was fitted to all available data on SMRs. For most countries this covered the years 1950 until 1997/1998. This statistical model is a flexible procedure for modelling non-linear trends. The estimated value for 2000 is a linear prediction based on this model. The number of degrees of freedom controls the smoothness of the relationship, and six was chosen to ensure that the predictions were based upon the relatively more recent data. The predictions based on 58 df were all similar. The fit of the model was assessed graphically.
The age-specific population for 2000 in each gender group in each member state was estimated using both methods, with similar results, and the linear regression-predicted value was used. The 1985 (i.e. 19841986) age-, sex- and country-specific mortality rates were applied to these population estimates to obtain the expected number of cancer deaths assuming nothing other than the effect of demographic changes taking place.
The age-, sex- and country-specific rates of deaths for 2000 were estimated using the statistical methods outlined above. After evaluation, it was clear that the spline method, with 6 df, was the most appropriate. This was based on the similarity of the estimates of the methods and took into consideration that the spline would underestimate decreases and increases. These rates were then combined with the 2000 population by age group to give an estimate of the total number of deaths in 2000. The total number of deaths in the EU in 2000 was calculated by summing the predicted deaths in each country. The SMR for the EU in 1985 and 2000 was obtained using a weighted average of the SMR in all the countries, using the 1985 deaths in each country as the weights.
The SMR was calculated (referent 1985), as was the relative risk (RR), as the ratio of the SMR in 2000 to the SMR in 1985.
Confidence intervals for the estimated deaths in 2000, the SMR in 2000 and the RR in 2000 compared with 1985 were obtained using bootstrap methods. As the distribution of the bootstrap estimates were symmetric the confidence intervals reported were calculated using a normal distribution. these were similar to the bias corrected percentile limits. Both non-parametric and parametric bootstraps procedures were carried out with the results for the parametric bootstrap reported. For each country, age, year and sex the mortality rate was simulated from a Poisson distribution using the observed number of deaths and the observed populations. No bootstrapping was carried out on the population estimations. The spline model was fitted to the simulated data and projections obtained. The bootstrap variances were calculated from 1000 simulations.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
In women, the average number of cancer deaths around 1985 was 380 182 (Table 2). This was expected to rise to 450 918 in 2000 for demographic reasons. In 2000, there were an estimated 414 083 deaths, which represents 36 835 deaths avoided (Table 2). Overall, the number of cancer deaths observed compared with those expected in 2000 was reduced by 8% (RR = 0.92) in the EU overall, with the greatest reductions in Austria (15%), Finland (15%), Luxembourg (13%), the UK (10%), Italy, France and Germany (9%). Increases were observed in Greece (+2%) and Portugal (+3%), while no changes were observed in Denmark and Spain (Table 2). The risk of death was reduced by 10% (RR = 0.90) in the EU, with the greatest reductions found in Luxembourg (18%), Austria (16%), Finland (14%), Italy (11%), the UK (10%), Germany (9%) and France (9%). Increases were observed in Greece (+2%), Portugal (+2%) and Denmark (+2%) (Table 2). Favourable trends in cancer death rates are evident in most countries for women (Figure 2), with few exceptions. The declines appear to be very slow in Portugal, Spain and Sweden, and are unfavourable in Greece and Denmark (Figure 1).
|
|
|
|
|
|
In men, there were 53 746 deaths from colorectal cancer in the EU in 1985 (Supplementary data, table 5) and this total was expected to rise to 67 297 in 2000. In 2000, it is estimated that there will be 63 762 deaths from colorectal cancer, representing 3535 deaths avoided (Supplementary data, table 5). Favourable trends in risk are present in the majority of member states (Figure 7), although there are strongly increasing trends in risk present in Spain, Portugal and Greece (Figure 7).
|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Secondly, predictions for 2000 will tend to be conservative in that the true changes, which will be calculable when all data become available at some point in the future, may be greater. This can be explained by observing that over the period 19852000, the trends in most cancer types considered herein were reaching a peak. Both estimation methods are based on models that tend to under-predict. Under-prediction should be less with the spline method.
Thirdly, and of practical importance, the estimates give an over-precise indication of the accuracy of the statistical procedures. However, these over-precise estimates have continued to be used to help issues such as the total number of deaths approximating to the sum of all the component parts.
The Europe Against Cancer programme was established in the 1980s against a background of increasing cancer incidence and death rates, particularly in men, virtually worldwide. The high-level European Cancer Experts Committee (the Cancer Experts Committee) was created to direct and to supervise the programme. The Cancer Experts Committee, chaired initially by Professor Maurice Tubiana (Paris, France) and later by Professor Umberto Veronesi (Milan Italy), was almost exclusively comprised of clinicians involved in the treatment of cancer patients. The Committee took the strategic decision to concentrate on prevention rather than to focus on cancer treatment. A European Code Against Cancer was devised and later subjected to scientific scrutiny and revised [7], and a target to reduce cancer mortality in Europe by 15% by the year 2000 was launched. This ambitious target was widely publicized.
The main areas of activity of the Europe Against Cancer programme focussed on prevention, with an emphasis on tobacco control, screening, and education and training. The European Directive on Tobacco Advertising emerged from this Committee, and it was its report [8] that led to the wider ranging Tobacco Directive recently approved by the European Parliament. The programme also funded the establishment of tobacco control networks throughout the EU.
Demonstration programmes were established in breast and cervix cancer screening throughout the member states via the creation of networks. A series of guidelines for quality control were established for cervical [9] and breast cancer [10] screening, which have become accepted European standards.
In general terms, the Europe Against Cancer programme stressed prevention in all cancer control activities and created an environment within the member states, as well as neighbouring countries, where cancer control activities could flourish.
Overall, the number of cancer deaths observed in 2000 compared with that expected based on mid-1980s age-specific mortality rates, was reduced by 10% in men in the EU, while in women it decreased by 8%. Although the target of a 15% reduction was not met, the effects of the programme should by no means be viewed as a failure. A more detailed consideration gives some clear insights.
At first glance, the outcome of all this activity is not overly encouraging. Between 1985 and 2000, the numbers of cancer deaths in the EU increased by 12% in men and 9% in women. In total this means that there were 90 316 more deaths from cancer in 2000 than in 1985 (56 415 in men and 33 901 in women). Based solely on demographic factors however, applying the age-specific rates of 1985 to the estimated population of 2000, an increase of 182 889 deaths was expected by the year 2000. Thus, during the Europe Against Cancer programme, 92 573 cancer deaths were avoided in the EU, about half of the expected increase.
Aside from Luxembourg, whose population of only 500 000 makes calculation and comparison of annual rates statistically fraught, the countries that experience large declines in all forms of cancer in men also experience large declines in lung cancer. Successful tobacco control actions, started earlier than the Europe Against Cancer activities but continued throughout, have made a major contribution to the declines in cancer death rates in men. A good example of this is the situation described in the UK [11]. In fact, since the risk of cancer at several sites other than the lung is associated with tobacco smoking (including oral cavity, larynx, oesophagus, kidney, bladder, etc. [8]), those countries that experience declines in lung cancer also experience declines in all other forms of cancer combined. In reality, these declines are smaller than they are for lung cancer and reflect the smaller attributable fraction for smoking to this combined group of cancers.
In women the situation is, in many respects, the exact opposite. Nine (of 15) countries experienced declines of >10% in risk of death from all forms of cancer when lung cancer was excluded. However, the risk of dying from lung cancer in women increased substantially in every country considered: the weighted increase is 30%. The failure of tobacco control in women is a great disappointment, particularly since it has been known for >50 years [1216] that smoking cigarettes causes lung cancer. The fact that women are increasingly smoking, and smoking more, represents a great failure of public health in the recent past and its major challenge for the near future.
The influence of different patterns of cigarette smoking on incidence rates in men and women, and in different countries, is emphasized once more. If women in each country had experienced the same decline in lung cancer as men, then the overall target of the Europe Against Cancer programme may have been achieved in women and also overall. Tobacco control must continue to be a number-one priority, and women, alongside deprived populations of both sexes [17], must be a priority target.
Women around the world have taken up cigarette smoking with gusto. For many years it appeared that their lung cancer rates were low and that tobacco was not having the same effect on men. This complacency, which crept in during the two decades from the mid-1960s in particular, is now exposed as false, and neither is there evidence that the effect of cigarette smoking on lung cancer risk is greater in women than in men [18]. The dominance of the effect of smoking duration [19] means that a long period of time will pass from exposure (large numbers of women smoking) to effect (high levels of lung cancer). Lung cancer now exceeds breast cancer as the leading cause of cancer death in women in the United States, Canada, Scotland and several other countries. In Canada, breast cancer mortality has remained at least constant for nearly four decades while lung cancer death rates have increased between 3- and 4-fold during the same period [20]. While the higher case-fatality of lung cancer may be one factor in the mortality rates overtaking breast cancer, there is increasing evidence that there are regions of the world where the gap in the incidence rate is now closing. In Glasgow for example, an area where lung cancer incidence has historically been high, by 1990 the incidence rate for lung cancer (115 per 100 000) exceeded that for breast cancer (105 per 100 000) [21].
The largest contribution to the decline in cancer deaths came from stomach cancer in each gender (Table 3). Even over a relatively short period of time (19852000), this decline is remarkable. It is also unexplained, not only in terms of Europe Against Cancer, and unsustainable since the mortality rates are getting lower and there are increasingly fewer chances for prevention.
|
A major failure in Europe over the past 15 years has been the inability to bring about decreases in the mortality rate for breast cancer in many countries. Controversy still surrounds the question of whether tamoxifen or mammographic screening is the cause of the reduction in mortality found in the UK [25]. It is difficult to imagine tamoxifen being systematically used more in the UK than in the other member states. Only the UK (RR = 0.76), Sweden (RR = 0.90), The Netherlands (RR = 0.95), Finland (RR = 0.95) and Luxembourg (RR = 0.62) have national organized breast screening programmes. The case for organized national breast cancer screening programmes remains stronger then ever.
Only Finland and Austria achieved reductions of 15% in cancer mortality in each gender group; the UK and Luxembourg achieved it in men and came close in women, and Italy came close in each group (Figure 11). These countries can be considered to have the most effective overall programmes to reduce cancer mortality, whether directly or indirectly associated with the Europe Against Cancer programme. In contrast, cancer control seems to be failing in Spain, Portugal and Greece, and needs a boost in Denmark and Ireland where new efforts are surely needed.
|
|
The success of such a public health-based programme, as compared with a programme centred on basic research or drug development, should also act as a stimulus to review current funding priorities for cancer research and cancer control.
![]() |
Acknowledgements |
---|
![]() |
Footnotes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Decision of 7 November 1987. Official Journal of the European Union. L 334, 24, 11, 1987.
3. Decision of 17 May 1988. Official Journal of the European Union. L 160, 28, 6, 1988.
4. Cancer Plan 19871989. Official Journal of the European Union. C 50, 26, 2, 1987.
5. Boyle P, Parkin DM. Statistical methods for registries. In Jensen OM, Parkin DM, McLennan R et al. (eds): Cancer Registration: Principles and Methods. IARC Scientific Publication No. 95: Lyon, France: IARC: 1991; 126158.
6. Boyle P, Zaridze DG, Smans M. Descriptive epidemiology of colorectal cancer. Int J Cancer 1985; 36: 918.[ISI][Medline]
7. Boyle P, Veronesi U, Tubiana M et al. Special Paper. European School of Oncology Advisory Report to the European Commission for the Europe Against Cancer Programme European Code Against Cancer. Eur J Cancer 1995; 9: 13951405.[CrossRef]
8. Boyle P. Cancer, cigarette smoking and premature death in Europe. A review including the Recommendations of European Cancer Experts Consensus Meeting, Helsinki, October 1996. Lung Cancer 1997; 17: 160.[CrossRef][ISI][Medline]
9. Coleman D, Day NE, Douglas G et al. European Guidelines for Quality Assurance in Cervix Cancer Screening. Eur J Cancer 1993; 29 (Suppl 4): S1S38.
10. De Wolf CJM, Perry NM (eds). European Guidelines for Quality Assurance in Mammography Screening, 2nd edition. Luxembourg: European Commission 1996.
11. Peto R, Darby S, Deo H et al. Smoking, smoking cessation, and lung cancer in the UK since 1950: combination of national statistics with two casecontrol studies. Br Med J 2000; 321: 323329.
12. Schrek R, Baker LA, Ballard GP, Dolgoff S. Tobacco smoking as an etiologic factor in disease. Cancer Res 1950; 10: 4958.[ISI]
13. Levin ML, Goldstein H, Gerhardt PR. Cancer and tobacco smoking. J Am Med Assoc 1950; 143: 336338.[ISI]
14. Mills CA, Porter MM. Tobacco smoking habits and cancer of the mouth and respiratory system. Cancer Res 1950; 10: 539542.[ISI]
15. Wynder EL, Graham EA. Tobacco smoking as a possible etiologic factor in bronchiogenic carcinoma. J Am Med Assoc 1950: 143: 329336.[ISI]
16. Doll R, Hill AB. Smoking and carcinoma of the lung. Br Med J 1950; 2: 739748.
17. McLaren G, Bain M. Deprivation and Health in Scotland: Insights from NHS data. Edinburgh, UK: ISD Scotland Publications 1998.
18. Prescott E, Osler M, Hein HO et al. Gender and smoking-related risk of lung cancer. The Copenhagen Center for Prospective Population Studies. Epidemiology 1998; 9: 7983.[ISI][Medline]
19. Peto R. Influence of dose and duration of smoking on lung cancer rates. In Zaridze DG, Peto R (eds): Smoking: an International Problem. Lyon, France: IARC 1986.
20. National Cancer Institute of Canada. Canadian Cancer Statistics 1997. Toronto, Canada: Canadian Cancer Society 1997.
21. Gillis CR, Hole DJ, Lamont DW et al. The incidences of lung cancer and breast cancer in women in Glasgow. Br Med J 1992; 305: 1331.[ISI][Medline]
22. Boyle P, Langman MJS. ABC of colorectal cancer: Epidemiology. Br Med J 2000; 321: 805808.
23. Beral V, Banks E, Reeves G, Appleby P. Use of HRT and the subsequent risk of cancer. J Epidemiol Biostat 1999; 4: 191210.[Medline]
24. Fernandez E, La Vecchia C, Balducci A et al. Oral contraceptives and colorectal cancer risk. Br J Cancer 2001; 84: 722727.[CrossRef][ISI][Medline]
25. Peto R, Boreham J, Clarke M et al. UK and USA breast cancer deaths down 25% in year 2000 at ages 2069 years. Lancet 2000; 355: 18221830.[CrossRef][ISI][Medline]
26. Berrino F, Sant M, Verdacchia A et al. Survival of cancer patients in Europe: the EUROCARE study. IARC Scientific Publications. Lyon, France: No. 132 IARC 132, 1995.
27. Berrino F, Capocaccia R, Esteve J et al. Survival of cancer patients in Europe: the EUROCARE-2 study. IARC Scientific Publications No. 151. Lyon, France: IARC 1999.
28. Welch HG, Schwartz LM, Woloshin S. Are increasing 5-year survival rates evidence of success against Cancer? J Am Med Assoc 2000; 283: 29752978.
29. Threlfall AG, Collins S, Woodman CBJ. Avoided deaths may not be useful for predicting mortality reductions from cancer. Br Med J 2000; 321: 1470.
30. Richards MA, Stockton D, Babb P, Coleman MP. How many deaths have been avoided through improvements in cancer survival? Br Med J 2000; 320: 895898.
31. Welch G. Diagnostic practice in the United States is different. Br Med J 2000; 321: 1470.
32. Reynolds TM, Wierzbicki AS. Improvements in survival may be an illusion. Br Med J 2000; 321: 1470.