Uses of epidemiology

Abstract

Until ∼1900 death rates in middle age were high and worsening (Figure 1a), but about the turn of the century sanitary reform began to show results in this age group. Mortality rates for both men and women began to fall, and they continued to fall fairly sharply until the 1920s. Then something happened. Female mortality maintained its downward course; but the reduction of male mortality slackened and almost stopped. One result of this is that death rates for these men, which were ∼10% higher than for women a hundred years ago, and ∼33% higher after the First World War, are now 90% higher. What happened? As we now know, many strange things were happening, and are reflected, in the vital statistics of the inter-war years. The most important was the emergence from obscurity of three diseases, particularly affecting males, and very common in middle age: duodenal ulcer, cancer of the bronchus and ‘coronary thrombosis’. The first of these is mainly important as a cause of morbidity; the other two are now major causes of death, killing annually over 20 000 middle-aged men. Figure 1b shows the figures for 1929–53, and the contribution of these two diseases to the course of mortality: the trend among men is very different without them. (a) Mortality in middle age during the last 100 years in England and Wales. All causes.1 (b) Mortality in middle age in England and Wales during 1928–531,2 Figure 1a and b illustrate one use of epidemiology—in historical study. But first let me explain that what I am speaking of is the study of health and disease of populations and groups, the epidemiology of which Farr, Snow and Goldberger are the masters. In contrast with clinical medicine, the unit of study in epidemiology is the group, not the individual: deaths, or any other event, are studied only if information can be obtained, or inferred, about the group in which the events occurred. The clinician deals with cases. The epidemiologist deals with cases in their population. He may start with a population and seek out the cases in it; or start with cases and refer them back to a population, or what can be taken to represent a population. But always the epidemiologist ends up with some estimate of (cases/population). In consequence he can sometimes ask questions that the clinician may also ask, and get better or different information in reply. Sometimes he can ask questions that cannot be asked in clinical work at all. He can, for example, calculate the rates of occurrence, or frequency, of phenomena in the population—such as the deaths, from all and from particular causes, per 1000 aged 55–64, a hundred years ago and now, to make possible the kind of comparison shown in Figure 1a and b. In this article I am considering epidemiology as a procedure for finding things out, of asking questions, and of getting answers that raise further questions—that is, as a method—and I will have less time to consider the results, the information, obtained in reply. I shall confine myself to the non-infectious diseases, and try to illustrate them mostly from investigations carried out from the Social Medicine Research Unit, or with material worked up in that unit. Seven ‘uses’ of epidemiology are described—different ways of looking at epidemiological data, or applications of the method. Historical statements made in medicine are of two broad kinds. The first describes the decline of infections, for example, and of nutritional deficiencies, and the main trends are usually very obvious. The other raises problems about the possible increase of various disorders, which is quite another matter. The questions usually put (‘Have disk syndromes become commoner?’ for example) are bedevilled by uncertainty about diagnosis and nomenclature in the past, and the lack of quantitative estimates of frequency at any time: how many cases occurred annually per 1000 men, aged x, in the 1930s and in the early 1950s? In such problems as the frequency of psychoneuroses, historical questions, which are often asked, are hopeless of direct answer; but even in disorders like leukaemia, urinary cancer or cerebral tumour, sub-arachnoid haemorrhage, dissecting aneurysm and the collagen diseases, it is exceedingly difficult to estimate how much a recent apparent increase reflects a true increase of disease, and how much it is the product merely of better recognition and greater availability of diagnostic services, etc. Such questions are clearly important because the role of environmental factors in aetiology, and of recent social change which may be associated with the increase, arises. As a result of a great deal of work, the increase of duodenal ulcer, cancer of the bronchus and coronary heart disease must now be accepted as a working hypothesis and guide to environmental study. Epidemiology may further be defined as the study of health and disease of populations in relation to their environment and ways of living. In a society that is changing as rapidly as our own, epidemiol