Uses of Epidemiology

Clues to Causes

The main function of epidemiology is to discover groups in the population with high rates of disease, and with low, so that causes of disease and of freedom from disease can be postulated. The most obvious and direct examples are the original observations on the nutritional deficiencies (scurvy, beriberi, pellagra and goitre); the geographical study of cancer (especially of the skin and liver); the industrial cancers (bladder, for instance); and industrial accidents (of coal-miners or railway workers). The biggest promise of this method lies in relating diseases to the ways of living of different groups, and by doing so to unravel 'causes' of disease about which it is possible to do something.

'Ways of living' can usually be described only in simple terms, and the kinds of causes of health and disease postulated in them tend therefore to be in rather simple terms, not of intimate biological mechanisms, but of social factors in the satisfaction of elementary human needs, of large-scale environmental features, of major aspects of behaviour. They are thus often 'general' rather than specific factors of health, causes of dis-ease, or diseases, rather than of a particular disease: as in the relations of water supply to bowel infections (not merely the cholera), living space and respiratory infections (as a class), income levels, nutrition and growth. We are only just beginning to identify such factors in ways of life, mass habits and social customs that may be related to many of the important problems of our own highly advanced society. For example, overnutrition (obesity; and ? atherosclerosis, thrombosis, dental caries, diabetes and toxaemia of pregnancy); physical inactivity (? 'coronary thrombosis' and ? how much else); tobacco and smoking (oral and lung cancer—a 'specific', rather, this—'coronary thrombosis'; and ? how much else); atmospheric pollution (lung cancer; ? chronic chest disease of many kinds, and ? acute, and ? other, non-respiratory conditions); cultural factors (genital cancer in Jews and non-Jews); eating customs (? alimentary cancer); 'social isolation' (? schizophrenia, senile psychosis and suicide). And, of course, there is still an overabundant heritage of 19th century (and earlier) poverty and crowding, insanitation and bad habits that contribute over-fully to modern ill health.

Figure 7 illustrates from some recent studies, and includes a wide range of data, from the first turning of the ground to highly advanced observations. I have included (b)^[28] the famous analysis from the 'Metropolitan Life' on the dangers of 'overweight'; an elaboration of Doll and Bradford Hill's classic data on cigarette smoking and cancer of the lung (e);^[10,29] some figures (unique so far as I know) from France on consanguineous marriages in relation to stillbirth and early neonatal mortality (g)^[27] and example (h)^[14] of the recent pioneering by the General Register Officer in psychosis, the darkest area of all; figures from Revans on the size of industrial units in relation to the frequency of accidents (d),^[30] which raise large questions of morale and the human environment, of group functioning as such, rather than the properties of the individuals who aggregate the group (compare immunity, and endemicity of infections). The distributions shown (a)^[22,31] on physical activity (they are different for gastric ulcer, incidentally, and for diabetes in younger persons, and do not show with other diseases), and on the disadvantages of town dwellers in respect of respiratory disease (c),^[1,2] merely set a stage for further inquiry; the natural experiment, or 'experiment of opportunity' provided by miners and clerks (f)^[8] provides, at least, contrast groups in which it may be profitable to seek factors that are significant in modern infant mortality.

(Enlarge Image)

(a–h) Seeking clues to causes. (a) Mortality in relation to physical activity of work. Men aged 45–64 in social class III (skilled workers). England and Wales. 1903–2. (b) Experiences of overweight men in comparison with standard policy holders. Metropolitan Life Assurance Company. Actual number of deaths per 100 of expected. Ages attained 25–74. (c) Mortality of town and country dwellers from bronchitis, pneumonia, respiratory tuberculosis, lung cancer; and from all other diseases (males). Aged 45–64. England and Wales. 1950–2. A=Conurbations. B, C, and D=Large, medium, and small urban areas. E=Rural areas. (The trend for females is very similar.) (d) Frequency of accidents and size of pits. (e) Smoking and lung cancer. Number of men per 1,000 aged 25 expected to die from cancer of lung before ages shown. (f) Infant mortality in the families of miners and clerks. England and Wales. 1949–50. (g) Stillbirths and deaths in the first four weeks of the life. Infants of consanguineous and other marriages. Two French departments. (h) Social class and admissions to mental hospitals. England and Wales. 1949

The great advantage of this kind of approach to prevention is that it may be applicable in the early stages of our knowledge of diseases, to disrupt the pattern of causation before the intimate nature of diseases is understood. Sufficient facts may be established for this by epidemiological methods alone, or in combination with others. The opportunity may thus offer to deal with one 'cause', or with various combinations of causes. Moreover, the possibility of two types of control may be opened up—environmental (as in the fluoridation of water) and personal (through alteration of diet and hygiene).

Int J Epidemiol. 2007;36(6):1165-1172. © 2007 Oxford University Press
Copyright 2007 International Epidemiological Association. Published by Oxford University Press. All rights reserved.

Read in opening a discussion at the Section of Preventive Medicine and Infectious Diseases at the Annual Meeting of the British Medical Association, Glasgow, 1954, and since expanded. First published BMJ 1955;2:395-401. Reprinted with permission.

Cite this: Uses of Epidemiology - Medscape - Dec 01, 2007.