Long-term causal effects of access to hospital maternity wards on dementia risk

We analyze the early-life determinants of later-life dementia risk, exploiting an intervention in Sweden that affected the supply of hospital maternity wards during the 1926–46 period. Using exogenous this variation in the supply of maternity wards to instrument the likelihood of institutional delivery, we find that delivery in hospital has substantial effects on late-life dementia. Exploiting the longitudinal dimension of our data, we are able to show that an improvement in cognitive abilities is visible already in childhood, and also manifest in the task contents of mid-life occupations. We argue that a decrease in child morbidity due to better treatment of complications is a likely mechanism. We also study the selection into treatment and argue that there seems to be very strong selection into treatment, which implies that the estimated LATE would seriously overstate the population-level treatment effect.

Joint work with Martin Fischer (Lund University), Martin Lövdén (University of Gothenburg), Therese Nilsson (Lund University), Nikolaos Prodromidis (University of Duisburg-Essen)