Positive Health Online

How do we assess the risks of cancer from the various hazards we are exposed to in our environment? Estimates of 75 to 90% of all cancers are believed to be caused by agents in the environment. These agents are a mixed bag – from synthetic chemicals, natural toxins, diet, cultural habits, industrial processes, drugs, viruses, sunlight, and ionizing radiation. Which of these are most risky and how is risk defined?

Risk is usually defined quantitatively. It's the probability of cancer occurring as a result of exposure to cancer-causing or cancer-inhibiting agents. If the probability is low, then the risk is perceived to be low. A high probability is judged to be a high risk. These judgements are often based on uncertain evidence, requiring assumptions and scientific methods which have many limitations (an example being the ability to predict a chemical's likelihood of causing human cancer on the basis of animal studies).

What are some of the agents known to cause cancer? Much of the quantitative risk assessment of the causes of cancer has been carried out by the International Agency for Research on Cancer (IARC). IARC working groups evaluate carcinogenicity based on evidence from studies on experimental animals, laboratory studies and epidemiological studies. Since 1972, they have reviewed more than 800 agents, an impressive achievement but still only a small fraction of the thousands of agents in the environment.

Examples of the agents for which IARC found sufficient evidence of carcinogenicity are aluminium production, asbestos, vinyl chloride, tobacco smoke, hepatitis B virus and ionizing radiation. In total, 63 agents have been assessed as causing cancer in humans.

Another 50 agents are probably carcinogenic – there is sufficient evidence of carcinogenicity in animals and while it could cause cancer in humans, bias or chance could muddle the interpretation. These include petroleum refining, formaldehyde, diesel engine exhaust and the HIV virus.

A further 209 agents are classified as possibly carcinogenic, based only on animal and laboratory studies. These include DDT, a dioxin, saccharin, and petrol engine exhaust.

Whether or not an agent is carcinogenic is of little relevance unless people come in contact with that agent. Exposure is a necessary condition for cancer to develop. For this reason, epidemiologists (who study patterns of human disease) have attempted to rank these agents by their overall importance to the causes of cancer. In 1981, Richard Doll and Richard Peto (JNCI Vol 66 No 6 June 1981, p1192-1308) estimated the proportion of US cancers attributable to different environmental causes. Given the lack of reliability of some of the evidence and the limits of epidemiology, they proposed a best estimate and a range of acceptable estimates.

Doll and Peto's estimates are widely accepted though there is some dissension. A few scientists have argued that occupational exposures should account for 20 to 33% of cancers. However, the evidence for tobacco smoking is particularly reliable and all researchers agree that successful measures to reduce smoking will eventually reduce many kinds of cancer, especially lung cancer.

There is less certainty about what it is exactly in the diet which increases the risk of cancer and even disagreement that diet should have such a large share. Doll and Peto suggest that it is not carcinogens in food such as pesticide residues which make diet so risky. It is likely to be the amount of vitamins, minerals, protective factors, fibre and fat eaten in the diet.

By this ranking, only a small proportion of cancers are attributed to carcinogenic pollutants of air, water and food. Yet these are perceived as riskier than the lifestyle risks of smoking, eating and drinking. This is because quantitative risk assessment is not the only way people make judgements about risk. Scientists complain that if only people understood probability better, they would see risks as they really are. They accuse the public of wanting an impossible and unreasonable guarantee of life without even the smallest risk. But judgements about risk are ultimately about acceptability and the level of outrage and indignation a risk provokes. These are not only about the size of the problem. As well as the probability of harm, people assess riskiness by taking into account various qualitative aspects such as:

Lack of personal choice – passive smoking to a non-smoker is considered more alarming than smoking is to a smoker.

Lack of personal control – residence near a dioxin-polluting incinerator is less acceptable than smoking is to a smoker.

Unfair distribution of risks and benefits – the risks of fluoridating tap water affect all who drink it but the benefits are mainly to children under 12 while the risks of eating an unhealthy diet is borne only by the individual.

Whether the risk is due to human failure or natural causes – pesticide residues on food are judged more risky than natural carcinogens such as aflatoxins in peanut butter. Other aspects are the seriousness of the consequences, whether it affects children or future generations more, uncertainty about the evidence, and whether the effects are delayed as opposed to immediate.

The size of the risk is clearly important and must be a guide to setting public health policy. But when judging risks, we also have to include concepts such as acceptability, justifiability, fairness and level of control. These are value judgements to do with the power we want to have over our lives. Much of the disagreement about cancer risks is to do with different ways of defining risks. Yet both are valid.

We are still a long way from finding a common language with which to discuss judgements on risk.