Int. J Cancer: 43, 967-969 (1989)


UICC WORKSHOP ON EVALUATING INTERVENTIONS TO REDUCE CANCER RISK

Meeting held at Reykjavik, Iceland, September 21-24, 1988

Matti HAKAMA1, Valerie BERAL, Joseph CULLEN, and Max PARKIN


A UICC Workshop on evaluation of primary preventative interventions against cancer was held in Reyjavik, Iceland, on September 21-24, 1988. This report summarizes the topics covered and conclusions reached.

Since some major causes of cancer are known, the theoretical basis for the primary prevention of cancer is established. However, practicable ways of avoiding the onset of cancer are largely limited to reducing smoking cigarettes. smoking has been estimated to cause about one-third of all cancer deaths (Doll and Peto, 1981) and a similar proportion of male cancer cases (Higginson and Muir, 1979;Wynder and Gori, 1977) in Western countries.

Approaches to the prevention of cancer vary. Traditionally the most common method has been health education on an individual community basis via the mass media or otherwise. In several countries other means have also been applied: legislative action - e.g. banning advertisements - price policy, and policy decisions in relation to health services. A reduction in tobacco use has been the principle objective in such actions. Other environmental changes have also been attempted, for example the development of dietary policies and counselling, the elimination or reduction of chemical carcinogens in the working environment, and enhancements in radiation protection. Physical exercise and chemoprevention have been suggested as a means of reducing the adverse effects of unhealthy lifestyles.

Preventative efforts are mainly evaluated in terms of follow up and monitoring for changes in causative exposures-risk factors of cancer. Much less is known about the relationship between the preventive intervention and the final outcome, i.e., occurrence of cancer or otherwise. Efforts to evaluate the effectiveness of primary preventative measures are hampered by the paucity of reliable data and of well-established analytical methods.

However, the effectiveness of primary preventative measures has been demonstrated in several circumstances. Sometimes a carcinogenic hazard has been removed from the environment, and the effectiveness of this action can be evaluated from the study of subsequent incidence or mortality rates, with no attempt to involve a control group. Specific examples are in industry, where removal of hazards has resulted in the virtual elimination of some occupational cancers. Clinical treatment or diagnostic practices have sometimes had carcinogenic side effects; when these practices were changed after recognition, rates of cancer at the population level showed a substantial decrease. In interventions such as these, however, the exact design and method of intervention is rarely known.

Time trends and geographical differences in cancer rates indicate substantial changes and differences in the risk factors. Studies of time trends in cancer in relation to time trends in exposure to risk factors-particularly tobacco smoking and alcohol consumption- are the most illuminating. International trends in lung cancer mortality, in particular, can be convincingly related to cohort-specific trends in smoking, particularly if allowance is made for the introduction of low-tar and filter cigarettes in the last 30 years. Studies of migrant populations are a well established method of demonstrating the effect which environmental change can play in the risk of cancer; however, the extent to which the change in cancer risk in migrants can be related to changes in their external or socio-cultural environment has been poorly investigated.

Epidemiological studies of individuals who have adopted potentially healthier habits or lifestyles are relatively few, and, for practical purposes, are confined to the investigation of cancer in ex-smokers. Such studies show a marked decline in risk, which is proportional in time since quitting, and they constitute the most powerful evidence for the effectiveness of stopping smoking in preventing cancer.

Little is known, however, about the causes of changes in environment and life-styles at the population or individual level. Such changes may be due to efforts of health authorities or due to health conscious changes in the behaviour of individuals. From the health related point of view they may depict a more or less spontaneous, i.e., not health-conscious decrease in the prevalence and intensity of exposures in the environment and in the behaviour, just as the appearance of such exposures was independent of health related actions.

The main purpose of the workshop was to accumulate results on preventative interventions. relatively few studies have been performed in which specific, documented preventative interventions can be evaluated with respect to their effort on cancer risk. Those available are reviewed in this paper. Most such studies involved preventative programmes against cardiovascular diseases. Because of this background, the studies reviewed were mainly aimed at reduction in smoking and changes in diet, the most important causes of cancer known at present.

WHO European Collaborative Trial in the Multifactorial Prevention of Coronary Heart Disease

The study was based on group randomization of 80 factories in several european countries employing 63,732 men. Measures adopted were education on diet, smoking and physical exercise, and supplying of appropriate menus at workplace canteens. The change in the combined risk factor score ranged from 0.4 to 28.2% in different countries. Results on cancer mortality after 6 years of follow-up were available from the UK, Poland, Italy and Spain. O)nly in Spain was the total cancer mortality rate lower in the intervention group than in the controls (RR = 0.5, p = 0.02) in spite of the fact that total mortality and incidence of CHD were smaller in the intervention than in the control group.

The US Multiple Risk Factor Intervention Trial (MRFIT)

More than 12,000 men were individually randomized to either an active intervention or usual care control group. The intervention consisted of individual and group sessions to persuade cigarette smokers to give up smoking, give dietary advice to achieve lower blood cholesterol levels, and the use of anti- hypertensive medication to control high blood pressure. data on mortality over 10.5 years of follow-up is available. The clinical trial data on the impact of multiple preventions, especially stopping smoking and dietary changes, on cancer mortality showed no significant differences but the material was too limited to allow firm conclusions to be drawn.

The Finland North Karelia Study

This study is a non-randomized community trial. The intervention in smoking and diet was achieved by health education of the public, by reorganizing the community and health services and by changing the environment. A 14-year follow-up for cancer after the start of the programme was carried through the Finnish Cancer Registry and comparisons were made of cancer morbidity among the North Karelian population, in the population adjacent to North Karelia and in the rest of Finland. Lung cancer among males was about 10% less common than expected, as was breast cancer among women. The number of cancers of the stomach and oesophagus was somewhat greater than expected. The incidence of other cancers related a priori to smoking and diet was approximately as expected.

The Indian Oral Precancer Study

This was a cohort study with non-concurrent controls based on some 18,000 individuals, 12,000 of whom were subjected to health education against tobacco use. Over a period of at least 8 years, substantial differences in the use of tobacco persisted in favour of the intervention group. the incidence rate of leucoplakia was much lower in the intervention group (RR = 0.4) but no differences have so far emerged between the observed and expected number of oral cancer.

Other Published Studies

In the UK Whitehall study (Rose et al., 1982) about 1500 men were individually randomized; the intervention group was subjected to anti-smoking advice. After one year of follow-up, in the intervention group consumption of cigarettes was one quarter of the level among controls. The difference persisted up to ten years but at a lower level. According to the results published in 1982 (Rose et al.) no differences were observed in the total number of cancer deaths. There were fewer cases of lung cancer in the intervention (18) than in the control (24) group, but non-lung cancer deaths were higher (28 vs. 12) in the same group. This unexpected finding was attributed to chance by the authors.

Several other intervention studies mainly aiming at lowering cholesterol dietary levels or reducing smoking failed to show any consistent pattern (Ederer et al.,1971; Hjermann et al., 1981; Pearce and Dayton, 1971) Most of such studies reported only a few cases of cancer.

The studies, Collaborative WHO, MRFIT,Finnish North Karelia and Indian Oral Precancer, reviewed more carefully in the workshop, were based on about 1800 cancers during a follow-up ranging from 6-14 years. All these studies have demonstrated a substantial to moderate change in intensity and prevalence of risk factors-smoking and some dietary practices-in the target populations indicating the success of such intervention programmes in achieving changes in behaviour. Smoking and inappropriate diet are likely to be the most important causes of total cancer, since up to two-thirds of all cancer deaths were attributed to these risks by Doll and Peto (1981). It might therefore be expected that changes in cancer risk would follow an intervention which was successful on the basis of process indicators. However, the efficacy of the preventative trials in reducing the risk of cancer seem s to be low. Several explanations from the negative results should be taken into consideration, however, before such a conclusion is drawn.

None of the studies referred to was designed to study the prevention of cancer. Nevertheless, it is not certain that the interventions would have been very different if this had been the case. Smoking is a risk factor for several cancers and because of the high attributable risk for lung cancer, the smoking-related interventions should have led to a reduction of lung cancer risk if they were effective. Less is known about the dietary factors affecting cancer risk, although, for cancer, the elements of diet suspected of producing an increased risk are similar to those for cardiovascular diseases. Only for cancer of the upper gastrointestinal tract may risk be elevated by attempts to reduce fat and sugar consumption and to lower the general intake of calories.Some of the trials were designed to promote a diet high in polyunsaturated fat and low in saturated fat and cholesterol. Relatively little is known of the effects of such dietary changes on the risk of cancer.

Because the main focus of the studies was on the prevention of cardiovascular diseases, the accuracy of follow-up and recording of incident cases of cancer or cancer deaths was not necessarily of the same quality as for the cardiovascular events. For the community trial, however, the cancers were identified by a nation- wide, population-based registration system of high reliability. Furthermore, even if less accurate, the comparability of the cancer death certificates is probably valid between the cases and controls, and cancer diagnoses may be even less biassed(But more affected by random errors) than the diagnoses for cardiovascular deaths. It is therefore unlikely that poor quality of cancer diagnosis accounts for the limited effect.

Cancers are a group of diseases with long latent period between first exposure and diagnosis. When the risk factor in question acts late in the carcinogenic process, reduction of exposure might be expected to have a relatively rapid effect. For example,risk of lung cancer declines quite rapidly towards levels found in non-smokers within 15 years of stopping smoking (Doll and Peto, 1981). Migrant studies suggest that colorectal cancer risk also changes relatively rapidly after a change in the environment (Haenzel, 1970). The risk may therefore have declined in the intervention groups after the follow-up of 6-14 years available for the intervention trials, but because evaluation is based on all cases or deaths within the follow-up period, the results will be greatly diluted by cases with short latent periods during the first years of follow-up. Probably only the Finnish North Karelia study had a long enough follow-up for small effects of intervention to be detected, and from this study a total reduction of about 10% in the risk of lung cancer emerged.

It should also be noted that health education in some studies was of relatively short duration and of low intensity. Furthermore, most of them were undertaken in countries where health consciousness tends to spread relatively rapidly. It is likely that in the long run the control populations would also be exposed to similar information as offered in the health education interventions for the experimental group, and sometimes to similar services. In fact, the Indian study showed a substantial and long lasting effect of a fairly limited health education campaign. Dilution of the information in the control group imply that such information is neither effective nor necessary. In recent years, several community-based programmes have been started-again primarily aimed at cardiovascular disease-with continuous efforts to reduce smoking and change dietary habits by a variety of strategies. Unfortunately, no data are yet available on risk of cancer or on other diseases. The potential for evaluation of such programmes, in terms of reduction or cancer incidence and other outcome indicators, should not be neglected. Although the study design may be a poor one for inferring cause and effect, existing data collection systems, particularly cancer registries can be used at little extra cost.

The designs and populations were variable and one should not expect an identical effect in all studies. the studies were in principle similar: an intervention-admittedly from simple to sophisticated-was made for health purposes. A substantial to moderate effect was obtained in the risk factor which the intervention aimed at changing. However, it seems valid to conclude that the effects on cancer risk of interventions mainly based on health education on smoking and diet have not so far resulted in any substantial effect on cancer risk. This does not imply that such efforts were ineffective. The exposure and follow-up have been relatively short and similar changes in behaviour may have been diluted in the control groups.

ACKNOWLEDGEMENTS

We gratefully acknowledge the financial and other support of the Icelandic Cancer Society, the International Agency for Research on Cancer, the National Cancer Institute and the Nordic Cancer Union.

REFERENCES

DOLL, R. and PETO, R. The causes of cancer, Oxford University Press, Oxford, New York (1981).

EDERER, F., LEREN, P., TURPEINEN, O. and FRANTZ,I., Cancer among men on cholesterol lowering diets, Lancet, II, 203-206 (1971).

HAENSZEL, W., Studies of migrant populations. J. chron. Dis., 23, 289-291 (1970).

HIGGINSON,J. and MUIR, C.S., Environmental carcinogenesis: misconceptions and limitations to cancer control. J. Nat. Cancer Inst., 63, 1291-1298 (1979).

HJERMANN, I., HOLME,I., VELVE BYRE, K. and LEREN, P., Effect of diet and smoking intervention in the incidence of coronary heart disease: report from the Oslo study group of a randomized trial in healthy men. Lancet, II, 1303-1310 (1981).

PEARCE, M. and DAYTON, S., Incidence of cancer in men on a diet high in polyunsaturated fat. Lancet, I, 464-467 (1971)

ROSE, G., HAMILTON, P., COLWELL, L. and SHIPLEY, M., A randomised controlled trial of anti-smoking advice: 10 year results. J. Epidem. Comm. Hlth, 36, 102-108 (1982).

WYNDER, E.L. and GORI, G.B., Contribution of the environment to cancer incidence: an epidemiologic exercise. J. nat. Cancer inst., 58, 825-832 (1977).

LIST OF PARTICIPANTS

ARNESEN, E. ISM, P.B. 417, N-9001 Tromso, Norway.

BERAL, V., The Radcliffe Infirmary, Oxford 0X2 6HE, UK.

BURING, J., Brigham and Women's Hospital, Harvard Medical School, 55 Pond Ave, Brookline, MA 02146, USA.

BYAR, D., Division of Cancer Prevention and Control, National Cancer Institute, Bethesda, MD 20892-4200 USA

CARLETON, R., Pawtucket Heart Health Program, memorial Hospital, 126 Prospect Street, Pawtucket, RI 02865, USA.

CHENEY, J., International Agency for Research on Cancer, 150, Cours Albert-Thomas, 69372, Lyon Cedex 08, France.

COLEMAN, M., International Agency for Research on Cancer, 150, Cours Albert-Thomas, 69372, Lyon Cedex 08, France.

CULLEN, J., Division of Cancer Prevention and Control, National Cancer Institute, Bethesda, MD 20892-3100, USA.

DOBROSSY, L., Cancer unit, World Health Organization, Regional Office for Europe,8, Scherfigsvej, DK-2100 Copenhagen O/, Denmark.

FRIEDEWALD,W., Disease Prevention, National Institutes of Health, Bethesda, MD 20892, USA.

GUPTA, P., Tata Institute of Fundamental Research, National Centre of the Government of India for Nuclear Science and Mathematics, Home Bhabha Road, Bombay 400 005, India.

HAKAMA, M., University of Tampere, Department of Public Health, Box 607, SF-33101 Tampere, Finland.

HAKULINEN, T., Finnish Cancer Registry, LIISANKATU 21, SF-00170 Helsinki, Finland.

KALDOR, J., International Agency for Research on Cancer, 150, Cours Albert-Thomas,69372, Lyon Cedex 08, France.

LAMM, G., Klinikum der Universitat Heidelberg, Abteilung Klinische Soczialmedizin, Bergheimer Strasse 58, D-6900 Heidleberg 7, FRG

LAW, M., The Medical College of St. Bartholomew's Hospital, University of London, Charterhouse Square, London ECIM 6BQ, UK.

LOPEZ, A., Globel Epidemiological Surveillance and Health Situation Assessment, World Health Organization, 1211 Geneva 27, Switzerland.

PARKIN, M., International Agency for Research on Cancer, 150, Cours Albert-Thomas, 69372, Lyon Cedex 08, France.

SASAKI, R., Department of Preventative Medicine, School of Medicine, Nagoya University, Tsurumai-cho, Showa-ku, Nagoya 466, Japan.

SWERDLOW, A., London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK.

TULINIUS, H., Icelandic Cancer Society, PO Box 523, 121 Reykjavik, Iceland.

_________________________

(1) To whom reprint requests should be sent.