During the 19th century, advances in the health of the population mostly came from improvements in nutrition and the environments in which people lived and worked. In the 20th century, the focus shifted from the environment to the behaviour of individuals. According to this view, the great achievement of epidemiology was to nail down smoking and, to a lesser extent, cholesterol. Having done that, the epidemiologists’ job was largely complete. In the future, advances in human health will come from the revolution in molecular biology and genetic approaches to combating disease, and the role of epidemiology will, in its evidence-based, clinical form, be to support the assessment of individual risk, diagnosis, and treatment. Concern with individuals and parts of individuals has been the dominant trend in epidemiology. Ecological analysis, the correlation between the characteristics of groups, has been seen as a second-rate way to approach individual risks. Although poor sanitation and malnutrition have been solved in developed countries, the circumstances in which people live and work are still crucial determinants of disease rates and, therefore, potentially provide the place for effective interventions to improve the public’s health. As Rose1 states,“The primary determinants of disease are mainly economic and social, and therefore its remedies must also be economic and social”(p 129). If the environment is important, the appropriate analysis should be at the environmental level. Thus, ecological analyses are not second rate but are the most useful way to examine the effect of social environment on health.

Death rates provide a good example of the social determinants of disease. Why are death rates so fixed? In the UK there are 152000 deaths per year from coronary heart disease. Next year, there may be fewer deaths because, happily, there is a long-term secular decline in coronary heart disease and because the ageing of the population makes little year-on-year impact. However, the death rate will not be as few as 100 000 or even 130 000; it will turn out to be close to 186 per 100000 individuals. Similarly, next year the mortality rate in Hungary will be higher than that in the UK, about 250 per 100 000 people. Cause-specific death rates are characteristic of societies, so there must be causes of these patterns. Those of us trained in medicine, or in social psychology, tend to start with the characteristics of individuals—were the individuals who died smokers or did they have high cholesterol? But the individuals who died this year will not contribute to next year’s death rate. Thus, the characteristics of societies, over and above the characteristics of individuals, determine the death rate:“No description, however good, of particular cases will ever tell us which ones have a sociological character... The social rate must be taken directly as the object of analysis: progress must be from the whole to the parts”(p 148). 2 Durkheim, in the late 19th century, criticised individualistic accounts of suicide and emphasised the “social element”. 2 In the late 20th century, Geoffrey Rose questioned individualistic accounts of coronary heart and other diseases and pointed to the importance of population determinants. My two great teachers in epidemiology were Leonard Syme and Geoffrey Rose. Syme started out as a Durkheimian and Rose ended as one. Rose’s The strategy of preventive medicine1 largely developed from the study of cardiovascular disease and its risk factors, but also has