Of all the policy areas that invoke scientific evidence as their foundation, public health appears to have the strongest claim. Whether it’s ‘plain packs’ for cigarettes or enlisting transport policy to get us more physically active, lifestyle interventions have the studies and the stats to support them, prompting headlines like ‘Minimum alcohol pricing would save 860 lives a year, study finds’. But a closer look may find the science less convincing. Epidemiological studies are notoriously susceptible to confounding factors: does this lifestyle choice cause the poor health outcomes, or is it mere correlation? And the impressively specific number of lives that minimum alcohol pricing would save each year is the output of a computer model: however rigorous the methods, the output depends on a matrix of assumptions about how real people would behave.  Moreover, some policies seem to have little basis in medical research. Are e-cigarettes really a dangerous alternative form of nicotine delivery? Calls to ban them seem to focus more on what they look like rather than any effects on human health.

Does this matter? If public-health campaigns achieve their goals of improving our health by changing our behaviour, many would argue that some selective presentation of the medical science is justified. Isn’t it more effective to say alcohol is bad for you, than to muddy the message by discussing the health benefits of moderate drinking? Simple campaigns like ‘five a day’ may not be strictly accurate, but they are memorable and thus, so the argument goes, more likely to increase the population’s intake of healthy fruit and vegetables.  Indeed, one branch of public-health research lies closer to social science or even to market research than biomedical science. What is the projected outcome of this or that campaign? Which tactics are more effective in nudging the population away from their cigarettes and on to their bicycles? This research claims its share of the unquestioning reverence that science, when wearing the white coat of moral purpose, can still command.

But if evidence-based policy gives way to ‘policy-based evidence’, can public-health policy truly claim to be based on science? A strategy designed to change our habits by editing the evidence owes more to politics than to the transparent sharing of research to which science aspires. However robust the biomedical science, there are public-health questions that it cannot answer. It can tell us that smoking significantly increases the smoker’s risk of a range of life-shortening diseases, but not what we should do about it. Who should decide what is in the public’s best interest when it comes to health? Governments? Individuals? Doctors? If public health is truly scientific rather than political, should it also be more humble in its claims and recommendations?