Child Mortality during the Demographic Transition. A Longitudinal Analysis of a Rural Population in Southern Sweden, 1766-1894

Abstract: The great decline in mortality has had a vital impact on our societies. Life expectancy has more than doubled in the past 200 years, but no acceptable explanation has yet been given for the earlier part of the decline in mortality, even though it has attracted a very broad interdisciplinary interest. When carefully examined, many of the stated explanations, such as better hygiene, more food, and improved medical technology, are acceptable for later stages of the general mortality decline but only the lower virulence and/or improved immunity explanation is plausible for the earlier stages. Most of the decline before 1900 was in infant and child mortality, and this study concerns child mortality during the demographic transition. Since conditions in very early life have been shown to have affected later life mortality in adult ages during this period, this study considers the possibility that child mortality in this period may also have been affected by conditions in very early life. The theoretical grounds for such effects are discussed, for example the Barker hypothesis, and it is concluded that there may be several links between conditions in very early life and health in later life; both for adults and children. A theoretical child mortality model is derived and then operationalised into an empirical model for the investigation of possible determinants of child mortality. These are current and lagged nutrition, socio-economic status, family-belonging, sex, maternal age at birth, and parish. Several links from conditions in very early life to childhood are included: foetal stage and infancy disease load, foetal stage and infancy nutrition intake, and birth season. Mortality for the age group 1 to 14 in four Scanian rural populations for the period 1766 to 1894 is analysed by means of a multivariate hazard regression model with family-based frailty. A three-period subdivision of this model accounts for enclosures, land partitionings, and industrialisation, making it possible to compare how the transition from one type of society to another affected child mortality. The empirical analysis shows that there were some differences in child mortality due to socio-economic conditions in the first sub-period (1766-1814), and that differences increased over time and culminated in the last sub-period (1865-1894). It also shows that if maternal age at birth of child was 35 or higher, the risk of mortality increased by about 30%. Further, there was an effect of family-belonging which increased by about 70% from the first sub-period to the second sub-period, and then remained at the same level. Both current and lagged current nutrition intakes seem to have affected child mortality. The estimates also show that conditions very early in life influenced mortality in childhood: children with a high disease load during the foetal stage and infancy experienced increased mortality risk by up to 200%. There are few signs of an effect of constrained nutrition in early life when measured with food price, but there are nutritional effects via birth season that indicates a link between low nutrition in early life and higher mortality risk in childhood.