Epidemiology in the context of changing society

During the last few decades (approximately since the late eighties) epidemiology has been used by both public and private sector for political purposes. Predefined political views have been scientifically legitimized by “proving” the correctness of the measures to be taken, using epidemiological “evidence”.

The problem is that epidemiology is not a medical science, but a mathematical one. It is purely a specialization of statistics. Epidemiology is therefore no more than a machine to crunch numbers. Every form of interpretation of these numbers is pure speculation and has no scientific ground whatsoever.

Epidemiology is useful, of course. It can be used very well to search for possible cause-and-effect relationships. Proving cause and effect is usually more difficult. It can be done by taking away the supposed cause and see if the effect ceases to occur. Or it can be done by actually observing the effect and analyzing the process that triggered this effect. This depending on the branch of science involved.

Epidemiology is primarily used for analyzing health effects of environmental pollution. In these types of research, epidemiology enters the field of toxicology. Proving that a certain pollutant causes a certain health effect is very difficult. Toxicologists often have to resort to experiments with animals. These are usually conducted with very high exposures to the skin (not to the respiratory system). This appears to be necessary because incubation times would otherwise be much too long for practical research (to actually observe the effect).
Carrying out dose-response research is even more difficult, especially since it is very difficult to translate results from animal testing to estimates of toxicity for humans. Due to these difficulties “maximum safe levels” of exposure are usually set lower than strictly needed, due to precaution.

For some, this form of precaution has become a matter of principle. The precautionary principle, so to speak. This attitude has severe drawbacks, if not dangers, to it. First of all, the precautionary principle is based on the assumption that a risk-free world is feasible, which it isn’t. The second drawback is that the alleged pollutant is usually a product of some form of human activity, which would have to be banned altogether for the pollution to stop. In a normal political process, there would be a trade-off between the level of risk that the toxic substance would cause, and the amount of disturbance the banning of a human activity would cause in society. This trade-off becomes even more complicated if the substance concerned is also produced by natural processes (which means there is some form of background exposure anyway). The precautionary principle seems to avoid this trade-off by claiming that all risk should be avoided, even when the risk level is very uncertain.

The precautionary principle is primarily used for reversing the political process.
A normal political process would aim for a certain effect, e.g. to reduce the prevalence of a disease. It would then search for the most relevant risk factors in order to find the causes. Only then would it prepare a policy to reduce or remove the cause.
The reversed process starts with a negative attitude towards a certain human behavior. It then starts to look for products of this behavior, and researches possible adverse effects of these products. Which can always be found, using epidemiologic research. These adverse effects can then be used to ban or control the disliked human behavior. Reducing the adverse effect is usually given as an argument, but whether this reduction actually occurs is hardly ever monitored.

The problem is that epidemiologic research can be manipulated to show (small) risks for any substance, if you look for a relevant health effect long enough. Moreover, epidemiologic results are never more than just a number (e.g. a relative risk value). The interpretation of this (is there a cause-effect relationship?) is always subjective.
However, epidemiologic results are often used to conclude that a cause-effect relationship exists, which then gives a basis for new policy, again with reference to the precautionary principle. This even happens if toxicologic research fails to confirm the supposed cause-effect relationship, or still worse, before any toxicologic research has been carried out in the first place.

Of course, epidemiologists have devised some norms as to when to suspect a cause-effect relationship. A general rule is that a relative risk (RR) below 2.0 is not to be seen as an indication of a cause and effect relationship, because there is a great chance this relative risk could be caused by (unknown) confounding factors. This means that it is more likely that there is another cause with a much higher risk factor, which happens to have a statistical correlation with the researched cause.
Known and confirmed cause and effect relationships usually boast relative risks which are much higher (at least 10 and often over 100).
Most epidemiologists agree that a relative risk of below 3 is hardly worth looking into as far as establishing a cause-effect relationship is concerned.

It is, however, astonishing to see how quickly these guidelines are sidelined as soon as the reverse political process is followed. Because in this case it is not the effect that is the object of scrutiny, but the cause itself. It doesn’t matter what it causes, as long as it causes something bad so that the cause can be banished. In these cases, relative risks of 1.2 are large enough to “prove” that the substance has enough negative side effects to be banned. If one would have searched for relevant causes to combat the effect, a substance with a relative risk of 1.2 would never be considered a relevant threat, because there are so many substances with higher RR’s for that particular disease. The argument is usually made by rephrasing the relative risk; a RR of 1.2 would be rephrased to “20% more risk” or, even worse, a number of “excess or premature deaths”. These death tolls have no connection to reality, but are created to frighten politicians into applying the precautionary principle.

The reverse political process, combined with (often shoddy) epidemiologic research and the precautionary principle, shows that politics is moving towards a new type of policy: regulating human behavior. This seems to have become an end in itself, instead of a means to other goals (such as eliminating certain diseases).

Michiel van de Sande, March 2007