In an early released article in Pediatrics entitled “State and Local Government Expenditures and Infant Mortality,” Shivani J. Sowmyan (Kennedy Krieger Institute, Baltimore, MD), Ashley H. Hirai (Health Resources and Services Administration), and Jay S. Kaufman (McGill University, Montreal, CA) conducted an updated analysis of data from a prior study in which they linked data about government expenditures from 41 states with infant mortality rate (IMR) in those states 2 years later (10.1542/peds.2023-063571).
In short, the authors found an even stronger association between increased government spending and reductions in infant mortality than the prior work had shown.
This thoughtful work has some technical complexity, but as a non-statistician myself, I hope to aid you in navigating this well-written article, and the informative commentary (10.1542/peds.2024-068252) by Ciaran S Phibbs, PhD (Stanford University, CA) is very helpful and could be read first.
Why re-analyze data from the original article? The authors of the newer article aimed to build on the original analysis by adding recent data and by revising the analysis to address methodological limitations. Phibbs’ commentary explains these issues in a nutshell:
- Adding 2015–2019 data to the 2000–2014 information makes intuitive sense for an update.
- The original article reported the “interaction effect” rather than the “main effect” of the analysis, and this was revised.
This “interaction effect” was a lot smaller than the “main effect,” so the original analysis underestimated the effect of government spending on IMR. The “main effect” is the direct influence of an independent variable (e.g., government spending) on the dependent variable (IMR), while an “interaction effect” measures any interaction between the independent variables (government spending, year, expenditure type) that then affects the dependent variable (i.e., if the level of one independent variable influences the effect of another independent variable).
- The new analysis has now adjusted for spending differences across states while including longitudinal (over time) within-state changes, and
- The new analysis adds adjustment for inflation, regional differences in prices, and annual changes in state poverty and in state population.
Sowmyan et al. include additional details regarding the types of state expenditures (social, environment, educational) that made a difference, and the effects of spending on differing racial and ethnic groups.
In summary, this new and updated analysis found that a one standard deviation (SD) increase in per capita total government spending was associated with a reduction of 0.35 infant deaths per 1,000 live births, compared to the reduction of 0.02 deaths per 1,000 live births reported in the original analysis.
This well-written and accessible research provides needed evidence that supports meaningful government policy and spending that can invest in and positively impact infant health outcomes.
