Objectives Previous research has suggested that the economic turmoil during the Great Depression led to significant improvements in public health. However, these studies have relied on highly aggregated national data (using less than 25 data points), and employed intermediary measures of economic change, such as employment and Gross Domestic Product. We use a new historical data set of US mortality rates and bank suspensions to analyse both the immediate and underlying causes of mortality change during Great Depression.

Design Cause-specific mortality rates covering 114 US cities in 36 states were taken from the US Bureau of the Census. Bank suspensions data were taken from the Federal Deposit Insurance Corporation. Epidemiologic analysis was performed of the immediate causes of fluctuations in urban mortality rates weighted by population size. Dynamic fixed effects models were used to assess the immediate and delayed effects of bank suspensions on mortality.

Setting 114 US cities and 36 US states, 1929–1937.

Participants NA.

Main outcome measure Age-standardised all-cause and cause-specific mortality rates

Results Reductions in all-cause mortality rates (about 10% between 1929 and 1932) were attributable to declines in death rates due to pneumonia (26.4% of total), influenza (13.1% of total), and respiratory tuberculosis (11.2% of total), while death rates increased from heart disease (19.4% of total), cancer (8.1% of total) and diabetes (2.9%). Of these main causes of mortality changes, only heart disease plausibly relates to contemporary economic shocks. A higher rate of bank suspensions was associated with contemporary higher suicide rates (β=0.32, 95% CI 0.24 to 0.41) but lower death rates from motor vehicle accidents (β=−0.18, 95% CI −0.29 to −0.07); no effect was observed for other causes of death studied. There was no evidence of substantially differing delayed effects.

Conclusion In contrast with existing research, we find that the majority of rises and falls in deaths during the Great Depression was unrelated to economic shocks. Spurious correlations can occur when immediate effects are not decoupled from long-term trends, especially problematic with trending variables, such as GDP. Consistent with existing work, we observed that bank suspensions led to rapid rises in suicides and falls in road traffic fatalities. Further research should investigate alternative explanations for the reductions in infectious diseases and their marked variations across cities and states, such as nutrition, sanitation, the New Deal, Prohibition and other public health measures at the time.