IT鈥橲 a startling fact that the Black Death killed about one-third of Europeans in the three years between 1347 and 1350. Had epidemiological surveillance been what it is today, that episode of bubonic plague would have generated some fabulous statistics. And had we known what we know now about plague and how it spreads, it would have been quite clear what we should have done about it.
But this is an exceptional case. In public health, it is rare for absolute numbers to suggest a defensible course of action. This is spectacularly illustrated by the current debate in the US over smallpox inoculations. The risk of terrorists carrying out a smallpox attack seems small. Yet there is more to it than that when it comes to deciding what action to take. The mere possibility of an attack, however unlikely, provokes great fear. The public response to an attack or the imminent threat of one is likely to be hysterical, for the consequences for those infected would be horrific. So rather than take the minuscule risk at face value, the US government has chosen to deploy a national system of bioterrorism monitoring equipment and inoculate some 10 million medical workers to guard against it.
Most public health problems, and the statistics that go with them, lie in the area somewhere between rampant plague and unquantifiable menace. Governments and funding bodies face all sorts of pitfalls when trying to interpret statistics, for statistics mask what is often a complex situation. Death has the mathematical virtue of being an unambiguous state. But governments and funding bodies need to be careful, and should resist making decisions on the basis of simple 鈥渉eadline鈥� figures.
Advertisement
For example, the easiest way to compare the potency of deadly diseases is to rank them in a league table according to how many people they kill. Yet this raises all sorts of questions. Choosing only those diseases caused by single infectious agents gives the greatest prominence to AIDS, TB and malaria. Advertising and lobbying by health bodies and non-governmental organisations raised global awareness of the threat posed by these three infections and led to the establishment last year of the Global Fund to Fight AIDS, TB and Malaria.
The fund now has more than $2 billion at its disposal. While this is far less than the target of at least $10 billion in the first year, it is still a marvellous opportunity to combat three major diseases of poverty. However, by concentrating on 鈥渢he big three鈥� caused by single agents, it inevitably excludes many other worthy causes. There are, for instance, at least as many deaths each year from childhood pneumonia or diarrhoea, caused by an assortment of bacteria and viruses, as there are from malaria or TB.
What鈥檚 more, any doctor filling out a death certificate knows that the immediate cause of death is often not the only cause. Yet for each death recorded by the World Health Organization鈥檚 International Statistical Classification of Diseases (ICD), only one cause is listed. This is not simply a statistical problem. The interactions between different kinds of illness have profound implications for the way we set priorities for disease control, for how we intervene to save lives, and for what we expect to gain from prevention and treatment.
For example, it is now conventional to classify all deaths of HIV-infected people as deaths from AIDS. So the quarter of a million HIV-positive people who died with TB last year were lost not only to TB statistics but also, perhaps 鈥� given that figures lead to funding 鈥� to the global budget for TB control. There are many reasons why these patients died, but some of them could have been saved by the standard treatment for TB.
The argument over how to invest in malaria prevention and treatment is more subtle. We now know from extensive research in Africa that bed nets treated with insecticide prevent more child deaths than have traditionally been attributed to malaria. This means that malaria contributes to deaths from other primary causes, such as pneumonia, and preventing mosquito bites carries a bonus that is not captured by ICD statistics. Even more striking is the implication that some child deaths could be averted either by preventing malaria, or by treating pneumonia 鈥� it doesn鈥檛 matter which.
The perfect statistic will remain elusive because there is no universally agreed goal for public health. The WHO鈥檚 constitution sees health as 鈥渁 state of complete physical, mental and social well-being鈥�. Few would disagree with the sentiment, but it cannot easily be captured in a single number. And there will always be some subjectivity because individuals will evaluate health threats according to their own perceptions of risk.
鈥淎 single death is a tragedy, a million deaths is a statistic.鈥� Joseph Stalin鈥檚 oft-quoted remark puts in a nutshell every epidemiologist鈥檚 dilemma: how to quantify tragedy so as to rouse the world, not merely to action, but to the best course of action.