Effect of Screening in the Nordic Cancer Control up to the Year 2017

Transcription

1 Acta Oncologica ISSN: X (Print) X (Online) Journal homepage: Effect of Screening in the Nordic Cancer Control up to the Year 2017 Matti Hakama & Lora Hristova To cite this article: Matti Hakama & Lora Hristova (1997) Effect of Screening in the Nordic Cancer Control up to the Year 2017, Acta Oncologica, 36:2, , DOI: / To link to this article: Published online: 08 Jul Submit your article to this journal Article views: 189 View related articles Citing articles: 8 View citing articles Full Terms & Conditions of access and use can be found at

2 ORIGINAL ARTICLE Effect of Screening in the Nordic Cancer Control Up to the Year 2017 Matti Hakama and Lora Hristova Finnish Cancer Registry (M. Hakama, L. Hristova), University of Tampere School of Public Health (M. Hakama, L. Hristova) and Bulgarian Cancer Registry (L. Hristova) Correspondence to: Professor Matti Hakama, Tampere School of Public Health, University of Tampere, PL 607, FIN Tampere, Finland Acta Oncologica Vol. 36, No. 2, pp , 1997 The purpose of this study was to estimate the effect of screening for cancer on mortality, quality of life and cost in the Nordic countries up to Data from the Nordic Cancer Registries were used to predict, by means of the age-period cohort models, cancer mortality assuming no screening, the screening actually practised and most effective screening observed. Cost of screening was assessed assuming the cost of breast, cervical and colorectal cancer screening per person as well as treatment, follow-up and terminal care by stage. The results were adjusted for quality of life directly associated with screening and indirectly induced as dementia by the prolongation of life. Ultimately, in the year 2017 screening will prevent about 4000 deaths in the Nordic countries, most (91Y0) cervical cancer and a smaller proportion of breast cancer (1 8%) and colorectal cancer ( I 8%) deaths are preventable. This is equal to 5.7% among all potential cancer deaths in 2017, which corresponds to about one life year gained due to all cancer screening per I 000 years lived by the Nordic population in Adjustment for quality of life may at most reduce the benefit by about 20%. The cost of screening for breast cancer is expensive, that for cervical cancer is likely to be cost saving, and cost of screening for colorectal cancer may occupy an intermediate position. It is concluded that about 6% of the cancer deaths in the Nordic countries can be prevented by screening. This figure is fully realistic given the Nordic type of organized screening programmes, whereas many other predictions affecting cancer burden are more speculative. Received 27 Nouember 1996 Accepted 6 December 1996 Screening is practised for cancer at many sites using several different screening tests (1). However, most of these lack scientific evidence as to the effectiveness, i.e., reduction in mortality. This applies even more to the effects of screening on quality of life and cost of health services. Only for breast cancer, cervical cancer and colorectal cancer can the evidence be regarded sufficient for the screening to be run as a public health policy. The aim of this study was to evaluate the potential effects of screening for cancer in the Nordic countries. The effects were measured in terms of mortality reduction, costs for the society and quality of life and they were predicted up to the year This report is a summary of a monograph (2) which is included in the series of reports on Cancer in the Nordic Countries (3, 4) by the Nordic Cancer Registries. MATERIAL AND METHODS The basic material was extracted from the cancer registries in the Nordic countries, who provided age-sex-specific data for new cases of and deaths from cancer. Population data, actual and predicted, were from the statistical offices of the Nordic countries. The predictions were made under the assumptions of no screening and screening as a nationwide public health policy, and by the age-cohort-period models. The effects of screening on the predictions were estimated on the basis of the Nordic experience and that of the randomized trials world-wide. The cost of PAP smear and mammography was the average charge per woman in the agreements on organizing the screening made between the individual municipalities and the Cancer Society of Finland, $10 per PAP smear and $40 per mammography. The cost per FOBT was assumed at $5 (5). Cost estimates of breast cancer treatment and follow-up including terminal care in Tampere University Hospital by clinical stage (6) were applied. Treatment costs for cervical and colorectal cancer were approximated on the basis of ratio of breast cancer treatment costs to the costs of cervical and colorectal cancer by stage as described by Eddy (5) and Koopmanschap et al. (7). Terminal care costs were included in the estimates as a 0 Scandinavian University Press ISSN X Acta Oncologica

3 120 M. Hakama and L. Hristova Acta Onrologica 36 (1997) part of the costs by stage. Cost estimates assume the numbers of cases known, the actual Nordic data and incidence predictions to the year 2012 (3) were available. For assessment of the results of a screening programme, not only the costs and duration of life (life years gained, LYG) accounted for by screening were estimated but also the quality of the possibly gained life years (8-10). Effect of screening on the quality-adjusted life years (QUALY) was derived from published research, mainly from Belgium and The Netherlands. In this study quality of life not directly related to cancer was also assessed, because not all the life years gained by screening will be active ones and of good quality. Years with active life are called dementiafree life years gained (DFLYG). Three levels of dementia were distinguished: mild, moderate and severe. Moderate dementia was defined as living independently is risky and some supervision is necessary. Severe dementia was defined as health status requiring continual supervision. It was assumed that 7% of the remaining life at age 65 would be lived in moderate or severe dementia, and at the age of 85 this proportion would be 26% with a difference by sex (11). Finally, life years gained with good quality (GQ- LYG) were those life years gained which were adjusted both for quality in the traditional sense (QUALY) and for dementia (DFLYG). Cost utility was found by estimating the costs per life years gained, where the life years were either crude or adjusted for quality and/or dementia. Details of the materials and methods can be found in Hristova (12) with a similar analysis for Finland and from the full length monograph published simultaneously with this report (2). Assumptions on the effect were based on the screening trials for breast cancer in Sweden (1 3) and colorectal cancer in Minnesota (14) and on the service screening for cervical cancer as practised in Finland (1 5) with the largest effect on incidence of invasive disease in the Nordic countries. For breast cancer, biannual screening at ages from 50 to 69 with 30% reduction in mortality in the target population was assumed. Screening for cervical cancer takes place in Finland every 5 years for women between the ages of 30 and 55. The participation rate is about 70%. For colorectal cancer, annual FOBT at ages was assumed. According to the Minnesota trial, such a programme was assumed to result in a 20% reduction in mortality if routinely applied. RESULTS At present there are about annual deaths from cancer in the Nordic countries. Because of changes in population size, age distribution and risk of cancer, this number will increase to annual deaths in the years assuming no screening. Since cancer registration in the Nordic countries was established, mortality from breast cancer has been increas- ing in Denmark and Finland, but has remained relatively stable in Iceland, Norway and Sweden. The prediction of what the mortality rate would be if mass-screening for breast cancer, covering the ages from 50 to 69, had been established in 1987, and had reached the optimal effect, is shown in Fig. 1. A total of deaths from breast cancer in the Nordic countries was estimated as avoidable in the period The reduction in the annual number of deaths from breast cancer as a result of screening was expected to be 900 in the years , when the screening programme was predicted to have its ultimate effect. In the age-adjusted mortality from cervical cancer was between 11 per woman-years in Denmark and 6 in Sweden. During the subsequent calendar periods mortality from cervical cancer decreased gradually in all the Nordic countries. During the last observed period ( ) the highest mortality figure from the disease was only 5 (Denmark) and the lowest was 1.7 in Finland (Fig. 2). These trends would have resulted in predictions termed observed in Fig. 2. The age-adjusted mortality would have increased up to the 1980s and would have decreased thereafter, if no screening were established (Fig. 2). If the Finnish screening policy were applied in the other Nordic countries too, a decreasing trend stronger than that observed since the 1960s would have taken place (Fig. 2). The total difference between the estimated numbers of deaths with no screening and if the Finnish screening had been applied throughout the Nordic countries is in the period The reduction in the annual number of deaths from cervical cancer as a result of screening was in the period The age-adjusted mortality from colorectal cancer in males has been increasing during the observed period in all the Nordic countries with the exception of Sweden (Fig. 3). In , a decrease in mortality from colorectal cancer in men was predicted in Denmark and Sweden, with no changes in the mortality rates in Finland and Iceland, while an increase was predicted in Norway (Fig. 3). In the last period the highest mortality rate was expected in Norway, 28 per man-years, and the lowest was in Iceland, 12. The age-adjusted mortality from colorectal cancer in females was highest in Denmark and lowest in Finland (Fig. 4). Mortality was predicted to decrease in all the Nordic countries. The predicted mortality rates from colorectal cancer, assuming that screening started in 1993, are presented in Fig. 3 for males and in Fig. 4 for females. When the ultimate effect of screening was reached, the mortality was predicted to be between 23 (Norwegian males) and 6 (Swedish females). A total of deaths from colorectal cancer in males and in females could have been prevented during the study period in the Nordic countries. The reduction in

4 Acta Oncologica 36 (1 997) Ejfkcts of screening in the Nordic cancer control Dimnark 01 1, I, I, I 1 I I,, Frland I, ( I I, I I I I % b d i % Y o l l l, l,, l,, ls95 Mos Without screening i- With screening Sweden Fig. I. Age-adjusted mortality rates for breast cancer with and without screening in the Nordic countries (2) from (Denmark and Finland), (Iceland) and (Norway and Sweden) to (Norway) and (other Nordic countries).

5 122 M. Hakama and L. Hrislova Acta Oncologica 36 (1997) l a Denmark 01 I 1 I I I I, W Finland tgs Mo , H lo f, B Mo I l l Iceland Norway g m % Without screening f Observed *With optimal screening Fig. 2. Age-adjusted mortality rates for cervical cancer without screening, observed and with the optimal screening in the Nordic countries (2) from (Denmark and Finland), (Iceland) and (Norway and Sweden) to (Norway) and (other Nordic countries).

6 Acta Oncologica 36 (1997) 301 Effects of screening in the Nordic cancer control ).... c z q/e 1s r" o Mos 2015 Denmark Finland \ Iceland Noway Without screening +-With screening g20 $A - 0 _E - e p lo Sweden Fig. 3. Age-adjusted mortality rates for colorectal cancer with and without screening in the Nordic countries in males from (Denmark and Finland), (Iceland) and (Norway and Sweden) to (Norway) and (other Nordic countries) (2).

7 124 M. Hakama and L. Hristova Acfa Oncologica 36 (1997) = - m :(o... I Denmark , 1 1 1, 1 1, Finland I----== g ! fg15 l D g , 6 lo Bs MOS 2015 Iceland Norway Without screening -I- With screening Sweden Fig. 4. Age-adjusted mortality rates for colorectal cancer with and without screening in the Nordic countries in females from (Denmark and Finland), (Iceland) and (Norway and Sweden) to (Norway) and (other Nordic countries) (2).

8 Actu Oncologica 36 (1997) Effects of screening in the Nordic cancer control 125 Table 1 Acerage annual number of deaths from breast, cervical and colorectul cancer in (1995) and (2015) in the Nordic countries with und without screening Site Screening Number Screening Number prevented prevented No Yes % No Yes '%I Breast Cervix Colorectum Total the annual number of deaths from colorectal cancer as a direct result of screening was (700 in females and 700 in males) in the period , when the screening programme was predicted to reach its ultimate effect. In the absence of screening, a total of deaths was predicted to occur in the Nordic countries in the period from these three types of cancer from breast cancer, from cervical cancer and from colorectal cancer. This number would be 21% smaller ( fewer cancer deaths) with screening as described in this study for cervical and breast cancer, and assumed for colorectal cancer. In the period the annual number of deaths from cancer in these three sites was predicted to be (5 300 from breast cancer, from cervical cancer and from colorectal cancer) assuming no screening to have taken place (Table 1). If the full effect of screening had started in for cervical cancer, in 1988 for breast cancer and in 1993 for colorectal cancer, the predicted number of deaths would be (4400 from breast cancer, 130 from cervical cancer and from colorectal cancer), which is 26%) less than the number assuming no screening. Those percentages are 18, 91 and 18 for breast, cervical and colorectal cancer screenings, respectively. The reduction in the risk of total cancer amounted to 5.7%, 2.0% among males and 9.7%) among females (Table 2). The estimates of the screening costs for 1995 and 2010 are given for the Nordic countries combined in Table 3. Screening for colorectal cancer would be the most expensive, $157 m. in The difference between cost with screening and without screening was the greatest for breast cancer, $47 m., however. After full maturation of the programmes, as estimated for the year 2010, the relative position of colorectal screening becomes less favourable: the differences between screening and no screening in terms of cost are the same, more than $60 m., for both breast and colorectal cancer. Colorectal cancer screening will cover both sexes, however. Screening for cervical cancer is cost saving, about 17 m. a year and the savings are stable over time. In the year 2010 screening for these three primary sites would assume $111 m. more than would be the resources without screening in the Nordic countries combined. Table 4 gives the summary of life years gained as a result of screening for all the Nordic countries in the year The LYGs varied from 4300 for breast cancer to for cervical cancer. The total number of LYG was which decreased to GQLYG after adjusting for quality of life and dementia. The additional cost of screening per death avoided by means of the three screening programmes was about $ and about $5 300 per GQLYG for all the Nordic countries combined in 2010 (Table 5). Screening for breast cancer was the most expensive, $77000 per death prevented and $18700 per GQLYG. The cost saved by screening for cervical cancer was $ per death prevented and $2000 per GQLYG. DISCUSSION The aim of this study was to evaluate the effects of screening for cancer in the Nordic countries. There is sufficient scientific evidence to indicate that screening for cervical cancer, breast cancer and colorectal cancer will result in a reduction in mortality. The effects on mortality were predicted for the future up to the year 2017, assuming that the Nordic countries are covered by screening as a nation-wide, population-based public health policy and comparing the predicted mortality trends with those assuming no screening programmes. For cervical cancer, the programme as practised in Finland (1 5) was used as a point of reference. For breast cancer (13) and for colorectal cancer (14) the results of randomized preventive trials were assumed in the absence of detailed results based on any public health policy, i.e., a reduction of 30% in the mortality from breast cancer and 20% in the mortality from colorectal cancer. The assumed ages and frequencies of screening were from 25 to 59 with 5-year intervals for cervical cancer, from 50 to 69 with 2-year intervals for breast cancer and annual screening from 50 to 74 for colorectal cancer.

9 126 M. Hakama and L. Hristova Acta Oncologica 36 (1997) Table 2 Estimated mortality reduction in merage annual number of deaths due to screening for breast, cervical and colorectal cancers in the Nordic countries in by sex Sex Total annual number Reduction in Proportional of deaths from cancer* annual numbers** reduction (%) Female Male Total * Predicted on basis of Engeland et al. (4) and this study. ** Annual number of deaths from breast, cervical and colorectal cancers predicted in this study. The screening policy assumed would result in 1600 annual deaths prevented out of the potential deaths (1 1%) from cancer in the three primary sites in the Nordic countries in Only after the year 2010 will such a screening policy have its ultimate effect. In 2017 the number of cancer deaths prevented will be (26%) out of potential deaths in the three primary sites. Most (91%) of the cervical cancers can be prevented, whereas the proportion for breast cancer (1 8%) and colorectal cancer (1 8%) will be much smaller. Engeland et al. (4) predicted the mortality figures from cancer in the Nordic countries. These are somewhat higher than ours (2). The Engeland et al. (4) prediction on total cancer mortality in the year 2010 if corrected for effect of no screening and extrapolated for 2017 will be about Therefore, the proportion of screen-prevented cancer deaths is equal to 5.7% of all cancer deaths in the Nordic countries in 2017 (2.0% for males and 9.7% for females). Costs of the screening programmes were estimated taking into consideration direct costs of screening and savings from advanced disease treatment and terminal care. The total estimated cost of screening for cervical, breast and Table 3 Sumtnary of estimated costs of breast, cervical and colorectal cancer treatment, Jollow-up and rerminal care in millions of dollars. Screening compared with no screening in 1995 and 2010 by primary site. All Nordic countries combined Year, Primary site Year 1995 Breast Cervix Colorectum, females Colorectum, males Total Year 2010 Breast Cervix Colorectum, females Colorectum, males Total Screening No Yes Difference I I colorectal cancer in the Nordic countries in 2010 is $111 m. Screening for cervical cancer is approaching a phase when both the effect and costs are relatively stable and it was estimated to be cost saving. The effect of screening for breast and colorectal cancers is expected to appear gradually during the predicted period because of the increasing number (and percentage) of patients diagnosed by screening. When the screening programmes are assumed to achieve the optimal effect, the mortality reduction increases and treatment costs fall, resulting in a substantial decrease in the cost-effectiveness ratio. In the last period considered the costs per life year gained are approximately one half of those at the onset of screening. The differences in the costs per LYG were relatively small between the Nordic countries (2) and mainly dependent on the differences in baseline risk of cancer. The total cost of the three screening programmes was estimated at $4 400 per life year gained in Cost estimates have wide variations (2) and there are substantial differences in the methodology and in the estimates even after application of similar methodologies. Our estimates are within the range of results in other studies. Cost savings for cervical cancer are at the low end and costs for breast cancer screening at the high end of the cost spectrum. For colorectal cancer screening the experience is limited. The relative position of the three screening modalities is likely to remain unchanged under wide variation in the assumptions, because of the considerable differences in the cost estimates between the three screening programmes. Table 4 Summary of estimated crude (LYG) and a4usted (QALYG, DFLYG, GQLYG) life years gained in 2010 due to screening by primary site. AN Nordic countries combined Life years gained Breast Cervix Colorectum Total Females Males Deaths avoided LYG QALYG ' DFLYG GQLYG

10 Acra Oncologica 36 (1997) Eflects of'.screerzrng in rhe Nordic cancer control 127 Table 5 Summary of cost (in US dollars) uiility analysis in 2010 due 10 screening by primary site. All Nordic countries combined Additional cost ($1 per Primary site Breast Cervix Colorectum Total Females Males Death avoided I LY G QALYG I DFLYG GQLYG The impact of mass screening on the quality of life at population level was estimated including the psychological consequences of the screening test, the adverse effect of false positives and the advantages for those who would avoid radical treatment and advanced stage disease. The reduction in the number of life years gained (LYG) after adjustment for quality of life (QALYG) was relatively small. Not all the life years gained are of good quality and the prolongation of life will not take place without medical cost. The costs due to ill health during the life years gained were not estimated. Instead, a further adjustment for dementia during the life years gained was carried out. Dementia adjustment had a greater effect on LYG than the traditional adjustment for quality of life. Such adjustments for dementia have not been carried out in other studies. The difference between quality- and dementiaadjusted LYG is small in men and smaller than that for women, because men spend a smaller part of their remaining life in dementia and because they have a shorter life expectancy than women. The life years gained attributable to screening after adjusting both for the traditional quality of life and for dementia are called good quality life years gained (GQLYG). The total number of GQLYG in the year 2010 was instead of life years gained. The estimated costs per GQLYG were higher for breast cancer than for colorectal cancer screening and there were savings estimated for screening for cervical cancer. Because the population of the Nordic countries is likely to be about in the early 2000s, about one life year will be gained per I000 Nordic years lived as a result of full application of cancer screening. CONCLUSION Two screening programmes are run in most of the Nordic countries as organized public health policy, screening for cervical cancer and screening for breast cancer. Screening based on PAP smear tests is inexpensive and effective in reducing cervical cancer mortality. Its positive and negative effects on quality of life are relatively small. Mammography is a more expensive technology, screening is more intensive (with a shorter interval between the rounds), and the effect in numbers of deaths prevented and especially as a proportion of all breast cancer is smaller than that of screening for cervical cancer. Screening for colorectal cancer may occupy an intermediate position even if experience is limited. For cancers at other primary sites the evidence is not sufficient to make such evaluations, because no valid studies exist on the effect of screening on mortality despite the multitude of tests. Only lung cancer screening has been repeatedly shown to be ineffective in reduceing mortality. Compared to cancer prevention or treatment of cancer, the effect of screening may seem modest in total cancer control (5.7%) of all cancer deaths). This estimate is similar to that previously proposed for a typical European population (16) but more than that (3%) predicted by the year 2000 for the USA (17). Higher potential was assumed in the US study for other modes of control (prevention about 16% or more, treatment about lo'%, or more). However, the effectiveness of screening is tried and tested, while many of the other health service activities were never subjected to rigorous scientific evaluation. Therefore the predicted effects of screening are likely to be more realistic than the predictions for cancer control in general. Any screening programme has an effect on length of life, i.e. mortality, quality of life and cost. Anyone of these components may be the basis for the decision on whether to screen or not to screen. Unfortunately, the effects (on length, quality, cost) may be contradictory and weighting or finding of utilities of each of them should be attempted. The relative importance of these components cannot be scientifically proven. Therefore there is no unique combination, i.e. relative weighting or utilities which would be universally true, nor is there any scientific proof on the decision whether to run or not to run an organized programme. In such a decision value judgements are also needed on the relative importance of the different effects and, hence, in deciding whether to screen or not to screen for cancer. The same applies to the relative importance of the various components of cancer control, prevention, early diagnosis and treatment as well as to the cancer control contrasted to other health service activities. ACKNOWLEDGEMENTS The helpful cooperation of the Nordic Cancer Registries and the financial support by the Nordic Council of Ministers and Finnish Cancer Institute are gratefully acknowledged. REFERENCES 1. Miller AB, Chamberlain J, Day NE, Hakama M, Prorok PC (eds). Cancer screening. UICC Project on Evaluation of Screening for Cancer. Cambridge: International Union Against Cancer, Hristova L, Hakama M. Effect of screening for cancer in the Nordic countries on deaths, cost and quality of life up to year Acta Oncol 1997; 36 (Suppl. 9): 1-60,

11 128 M. Hakarna and L. Hristova Acta Oncologica 36 (I 997) 3. Engeland A, Haldorsen T, Tretli S, et al. Prediction of cancer incidence in the Nordic countries up to the years 2000 and APMIS 1993: 101 (Suppl. 38). 4. Engeland A, Haldorsen T, Tretli S, et al. Prediction of cancer mortality in the Nordic countries up to the years 2000 and APMIS 1995; 103 (Suppl. 49). 5. Eddy DM. Common screening tests. Philadelphia: American College of Physicians, Holli K, Hakama M, Hakala T. Use of hospital services by breast cancer patients by stage of the disease: implications on the costs of cancer control. Breast Cancer Res Treat 1996; 37: Koopmanschap MA, van Roijen L, Bonneux L, Barendregt JJ. Current and future costs of cancer. Eur J Cancer 1994; 30A: Aaronson NK, Bullinger M, Ahmedzai S. A modular approach to quality-of-life assessment in cancer clinical trials. Recent Results Cancer Res 1988; I 1 I: Heyink J. Adding years to your life or adding life to your years. Int J Health Sciences 1990; I: Nord E. Methods for quality adjustment of life years. SOC Sci Med 1992; 34: Roelands M, van Oyen H, Baro F. Dementia-free life expectancy in Belgium. Eur J Public Health 1994; 4: Hristova L. Effect of screening for cancer on mortality, costs and quality of life in Finland. (Doctoral thesis). Tampere: Acta Universitatis Tamperensis, ser A vol 456, Tabar L, Fagerberg CJG, South MC, Day NE, Duffy SW. The Swedish two-county trial of mammographic screening for breast cancer: recent results on mortality and tumour characteristics. In: Miller AB, et al., eds. Cancer screening. Cambridge: IUAC, Cambridge University Press, 1991 : Mandel JS, Bond JH, Church T, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. New Engl J Med 1993; 328: Hakama M, Magnus K, Pettersson F, Storm H, Tulinius H. Effect of organized screening on the risk of cervical cancer in the Nordic countries. In: Miller AB, et al., eds. Cancer screening. UlCC Project on Evaluation of Screening for Cancer. Cambridge: IUAC, Cambridge University Press, 1991: Hakama M. Potential contribution of screening to cancer mortality reduction. Cancer Detect Prev 1993; 17: Greenwald P, Sondik E, eds. NCI Monographs. Cancer control objectives for the nation: NIH Publication NO , 1986.