Commentary: History in the search of policy.

The work of Thomas McKeown has spawned a decades-long controversy about the sources of the long-term reduction in mortality, beginning sometime in the 18th century, that, in his view, was undeniably the impetus for the modern rise of population. On one side of the debate, McKeown argues that a changing economic environment, specifically increases in food production resulting in better nutrition, was the most important factor in the downward shift in mortality levels, while medicine, focused on providing individual immunizations and treatments, had little on no impact on the decline1,2 McKeown’s argument is part of a long tradition linking economic conditions to the growth of population. It incorporates the Malthusian notion that an increase in food supplies increased the population dramatically through lowered mortality levels while birth rates remained high. However, implicit in McKeown’s model is the necessity of a decline in fertility to avoid a Malthusian trap, a growth in population outstripping the growth of food supplies and producing a rise in mortality to old higher levels. Although essential to his argument, birth control and the declining birth rates are background to the falling mortality rates; McKeown offers little in the way of explanation for the widespread adoption of deliberate fertility limitation. One can only assume that, like mortality rates, birth rates were affected by economic growth and the impinging of economic realities on family behaviour.1 McKeown’s work has become canonical—although controversial—among scholars examining the history of mortality. No self-respecting scholar would attempt an explanation of the mortality transition without at least a passing reference to McKeown, though not all agree with his conclusions. The opposing view links improvements in life expectancy to retrospectively rational increases in knowledge and accompanying changes in activities to lower mortality levels both through specific medical intervention and public health programs. The debate is being carried on largely within and between two groups of scholars: historical demographers and economic historians, both of which focus not on the earliest period of mortality decline in the 18th century when little data exist but on the second half of the 19th century when mortality rates dropped precipitously and the registration of deaths was relatively more advanced in many parts of the world. In the article reprinted here, McKeown outlines the ‘historical methods and data’ that underlie his analysis of the mortality transition in England and Wales. The decline in mortality rates that had certainly begun by the middle of the 18th century in England and Wales was a break from prior demographic conditions and led to unprecedented increases in the population. However the data needed to examine the sources of the mortality change are not available until cause of death is recorded in 1838. Therefore, McKeown draws conclusions from 19th-century data and extrapolates backward to understand the 18th century. If, as he concludes, nutritional changes were the leading source of the mortality decline and medicine had little effect on mortality during the 19th century, then surely medicine could not have been more important and improvements in nutrition less important in the 18th century.3 McKeown treats the fall in cause-specific death rates from a number of infectious diseases as the proximate determinants of the overall mortality decline and links each cause of death to one of four potential broad underlying sources of its decline—improved diet as a result of rises in the standard of living, hygienic changes as reflected in improvements in water supplies and sewage disposal, a favourable change in organisms causing disease and/or in the relationship between the host and the pathogen, and improvement in the specific treatment and prevention of diseases in individuals. The list is in order of each factor’s contribution to decline in mortality in England and Wales during the second half of the 19th century. Specific prevention and treatment of disease in the individual made a very small contribution to the decline in the death rate through the effects of vaccination on the deaths from smallpox. The weight given to improved nutrition in McKeown’s analysis rests on his assertion, based on the elimination of other influences, that this single variable explains the decline in the tuberculosis death rate, which accounts for almost half the change in overall mortality.2 Uncertainties about the history of tuberculosis, however, have made the disease’s history a Rorschach test for scholars, who have presented plausible but often conflicting narratives regarding both the timing and causes of its decline.4,5 Researchers focusing on different time periods and locations have attributed the decline in tuberculosis deaths variously to rising standards of living,6 lower levels of early childhood morbidity,7 reductions in the size of susceptible populations,8 public health interventions,4,9 medical advances in its treatment and prevention,9,10 and the decline of other diseases that left their survivors more susceptible to tuberculosis.11 While McKeown’s explanation is parsimonious, economic historians and historical demographers are far from accepting it on the basis of his data and methods. The debate continues also because, in statistical terms, the model is over-determined, with explicatory variables that move together and point in the same direction with regard to mortality levels. These independent variables may in fact have been causally related to each other, if, for example, rises in the standard of living increased the revenues available for large municipal sanitation projects. They may also have interacted with each other to reduce the levels of mortality. For example, improvements in sanitation and medical practices might well have had different effects in a well-nourished population than in one with poor nutrition. Disentangling these potential sources of mortality decline, ordering them in importance and indeed, a la McKeown, attributing proportions of the mortality decline to each are tasks fraught with problems. Department of Sociology, Temple University, Philadelphia, Pennsylvania, USA 19122. E-mail: [email protected]