The Uppsala birth cohort study is a lifelong follow up study of people born in 1915-29.3,4
We used official registers to follow their marital status and mortality from 1970 through to 1995. Complete data were available for 5577 men and 5227 women. We collected school results at age 10 and body height (for men) at conscription (age 18) for a random sample of 405 boys and 411 girls (8%).
To indicate fetal growth we calculated birth weight for gestational age, since this predicts cardiovascular disease better than birth weight alone.3
Birth weight for gestational age, divided into fifths, showed an association with later marital status for men but not women. In the lowest fifth nearly 12% of men remained unmarried, a figure that is 3.0-4.5% higher than in the upper three fifths; the odds ratio (adjusted for year of birth and mother's age) was 1.51 (95% confidence interval 1.14 to 2.01) when the lowest fifth was compared with the highest. Further adjustment for adult socioeconomic factors (known from 1960 and 1970 censuses), parents' social class, and marital status at childbirth changed this odds ratio by less than 1%.
Men born to unmarried women tended more often than other men to be unmarried themselves (odds ratio 1.55, 1.21-1.99; adjusted for year of birth and mother's age); additional adjustment for birth weight by gestational age hardly affected this estimate.
Phillips et al suggested that this link between early growth and later marital status provided some explanation of the differences in mortality associated with adult marital status. We compared the mortality ratios (total, ischaemic heart disease, and stroke) of never married and ever married people before and after adjusting for early biological and social factors (birth weight by gestational age, marital status of the mother, and social class at birth), and three adult socioeconomic factors (occupation, education, and income). Never married women were not significantly different from ever married women; risk estimates were always near unity. The table shows the corresponding results for men. Ratios are large yet changed little on adjustment for birth weight by gestational age; they did not change when social factors at birth were adjusted for. Controlling for adult factors, in contrast, had a substantial effect.