Childhood Famine Exposure and Late-Life Edentulism: Evidence From Natural Experiments

To examine whether childhood famine exposure affects late-life edentulism using natural experiments from Asia, Europe and America.

We utilised historical famines as exogenous shocks. Data were pooled from 10 population-based studies across eight countries (165,586 participants). Because the famine-exposed cohort is older at survey than the comparison cohort, we separated famine exposure from the birth-cohort age gradient using a regression discontinuity in time, an interrupted time series, a region-of-birth famine-intensity difference-in-differences, a pre-famine comparison cohort and applied negative controls.

The crude positive association (risk ratio [RR] 1.69) was eliminated after adjustment for current age (RR 1.07, 95% confidence interval (CI) 0.92-1.23), and the famine cohort did not differ from the older pre-famine cohort (RR 0.97). The regression discontinuity (+0.13% points, 95% CI: -1.50 to +1.77), interrupted time series (-4.01% points, p = 0.15) and the region-of-birth difference-in-differences in the two Chinese cohorts (unadjusted, both p ≥ 0.07) were null, as were all negative controls.

Once the birth-cohort age gradient is accounted for, childhood famine exposure shows no detectable effect on late-life edentulism; the crude positive association is largely explained by age confounding, which cross-sectional famine-cohort comparisons do not address.