Background Several theoretical life course models (critical period, sensitive period, accumulation) have been proposed, all of which may be relevant for understanding of when and how socioeconomic inequalities in health arise. Our aim was to investigate whether the effect of socio-economic position on all-cause mortality accumulates over the life course or if some periods of the life course are more important than others.

Methods We followed 3,951 men and 3,601 women born in Uppsala, Sweden, in 1915–1929 with known SEP at birth (age 0), during childhood (10 years), in adulthood (30–45 years) and in later life (50–65 years) from September 1980 until emigration, death, or until December 2010. Data on parents’, partner’s and own occupational status (a measure of SEP), marital status, deaths and emigrations were abstracted from birth records, parish records, school records, Census 1930 and routine registers. From the eligible sample who were alive and living in Sweden in September 1980 (n=11,336), 67% (n=7552) had SEP recorded at all four-time points. We compared a set of nested Cox proportional regression models, each corresponding to a specific life course model (critical, sensitive and accumulation models), to a fully saturated model, to ascertain which model best describes the relationship between SEP and mortality. An alternative analysis employed latent class trajectories of SEP across same four time points. Analyses were stratified by gender.

Results The effect of SEP across the life course on all-cause mortality was best described by the sensitive period model in both genders with social advantage in later life (ages 50–65 years) having the largest protective effect (HR 0.80, 95%CI 0.73–0.87 in men and HR 0.82, 95%CI 0.75–0.91 in women). A linear accumulation model also provided a good fit of the data for women. Only 5% and 12% of individuals experienced downward and upward social mobility during childhood respectively. The sensitive period model indicated that being advantaged at age 10 appeared to be more protective than at birth for males, while there was no difference between SEP at birth and age 10 in their effect on all-cause mortality among women. Additional adjustments for marital status did not affect the results appreciably and main results were also consistent with analyses that employed latent class trajectories of SEP.

Conclusion Our results lead to a conclusion that an individual’s socio-economic position over the life course, including during early childhood does affect their risk of all-cause mortality in later life. These findings indicate that improvements in social conditions at any stage of the life course can contribute to reducing mortality at old age.