A causal modelling framework for short-term effects of PM2.5 on hospitalisations: A nationwide time series study in Brazil

Yuming Guo, Yao Wu, Tingting Ye, Lei Zhang, Amanda Johnson, Shanshan Li

Research output: Contribution to journalArticleResearchpeer-review


Accurate estimates of the causal effect of air pollution on health outcomes, are critical when calculating attributable disease burdens. Brazil has a large population exposed to fast-growing emissions of air pollutants, however no national level studies have been conducted to examine the causal effect of PM2.5 exposure on health outcomes. This study proposes a novel approach, to accurately estimate the causal relationship between daily PM2.5 exposure and hospitalisations, across 1,814 Brazilian cities during 2000–2015. A variant of the difference-in-differences (DID) approach was applied under a counterfactual framework. Daily time series data were divided into panels. Seasonality and long-term trend were controlled using indicators for the panel. Variables which do not change within a short-period were controlled using a dummy variable for the day. Controls for variables which vary day by day, were included in the model. We found the proposed model exhibited competitive power performance in detecting causal associations between short-term PM2.5 exposure and hospitalisations in Brazil. A 10 μg/m3 increase in PM2.5 concentrations over four days (lag 0–3) was associated with a 1.06 % (95 % CI: 0.94 to 1.17) increase in all-cause hospitalisations and accounted for 1.26 % (95 % CI: 1.12–1.39) of total hospitalisations. Larger effects were found for children aged 0–4 years and the elderly aged 80+ years, suggesting policies should be developed to minimise the exposure of these age groups.

Original languageEnglish
Article number107688
Number of pages8
JournalEnvironment International
Publication statusPublished - Jan 2023


  • Causal inference
  • Fine particulate matter
  • Hospitalisation
  • Variant difference-in-differences

Cite this