Commentary: William Ogilvy Kermack and the childhood origins of adult health and disease

The 1934 Lancet paper by Kermack, McKendrick and McKinlay was a landmark in the discussion of birth cohort influences on adult disease risk.1 It pre-dated Wade Hampton Frost's paper on cohort influences on tuberculosis mortality,2 and, whilst it was pre-dated by several demographic analyses highlighting birth cohort influences,3,4 was prescient in using these analyses to inform hypotheses regarding early-life exposures and their influence on later disease. As Kermack, McKendrick and McKinlay concluded, the data behaved as if ‘the expectation of life was determined by the conditions which existed during the child's early years’, and concluded, ‘the health of the child is determined by the environmental conditions existing during the years 0–15, and … the health of the man is determined preponderantly by the physical constitution which the child has built up’.1 The idea that adult disease is determined in early-life is currently most influentially reflected in the ‘fetal origins’ work of Professor David Barker and his colleagues. In this formulation the critical period of early-life development which is influenced by environmental exposures and has long-term effects on later health is the intra-uterine period.5 Interestingly Kermack et al. noted that one exception to the general pattern they observed—of mortality rates initially declining in the young, and only declining at older ages when the cohorts experiencing the initial mortality declines had entered these older age groups —was mortality under one year of age, which started to fall only after mortality at later ages had done so. They suggested that infant mortality was dependent upon the health of the mother, and thus improvement in infant mortality followed the generational improvement in the vitality of women of childbearing age. Kermack et al. suggested that the fall in the death rate at ages 1–4 was also somewhat delayed, but to a lesser extent than infant mortality. They did not provide data to support this assertion, however, and as Harris shows, at least in the case of Scotland, the fall in 1–4 year-old mortality fitted in well with the cohort-specific falls, occurring 10 years before the fall in 5–14 year-old mortality.6 The data suggest that very early-life development—before or within the first year of life—could not determine the mortality risk of birth cohorts as they aged. Other recent research—taking a lead from the seminal work of Anders Forsdahl7,8—has investigated the possibility that exposures acting right across the growth and developmental phase—from pre-natal growth through to the end of adolescence —may importantly influence later health.9,10 Some epidemiologists maintain that there is nothing special about the early-life period in terms of embodied (i.e. irreversible) outcomes of early-life exposure, with early-life conditions only influencing adult health through the social trajectories into which they place people.11 Clearly the issues raised by Kermack et al. nearly 70 years ago still resonate within current epidemiology.