Short-term health impacts of PM2.5 exposure on pediatric ambulance dispatches in India using air quality data developed by machine learning

Poor ambient air quality poses a significant global health concern. However, accurate measurement remains challenging, particularly in countries like India, where ground monitors are scarce despite high expected exposure and health burdens. This lack of precise measurements impedes understanding of changes in pollution exposure over time and across populations, limiting effective public health responses. India faces severe air pollution issues, with fine particulate matter (PM_2.5) levels consistently exceeding the World Health Organization (WHO) guidelines, leading to various health problems, including respiratory and cardiovascular diseases, injuries, and deaths. Existing health impact research on PM_2.5 in India is limited, particularly for pediatric populations in diverse and socioeconomically varied regions.

In this study, we developed an open-source daily PM_2.5 dataset at a 10 km resolution for India from 2005 to 2023 using a two-stage machine learning model. This model integrates data from satellite sensors, meteorological variables, and land-use information, validated against held-out monitor data to generate accurate daily PM_2.5 estimates. We then linked this dataset with over one million pediatric ambulance dispatch records across 11 states in India from 2013 to 2015 to investigate the short-term effects of PM_2.5 exposure on pediatric emergency health outcomes. We employed a fixed-effects Poisson regression model combined with an instrumental variable (IV) approach to address potential endogeneity issues, such as reverse causality and omitted variable bias. The primary instrument used is thermal inversion, a meteorological phenomenon associated with elevated PM_2.5 levels. Our outcome measure is the number of ambulance dispatches per 100,000 people per day, categorized by cause (illness or injury) to reduce misclassification bias. Our fixed-effects model controls for time-invariant differences and temporal confounders, isolating effects of PM_2.5. Using thermal inversion as an instrument further confirms the robustness of the causal link between short-term exposure to PM_2.5 and increased ambulance dispatches.

Our analysis reveals significant associations between short-term PM_2.5 exposure and increased pediatric ambulance dispatches. For all-cause and illness-related calls, we observed more than a 2% increase in ambulance dispatches per 10 μg/m³ increase in PM_2.5 exposure, with cumulative lagged effects up to 7 days. Furthermore, for injury-related dispatches, there was more than a 5% increase associated with a 10 μg/m³ increase in PM_2.5 exposure, with cumulative effects observed within just 0 to 1 day of exposure. These findings emphasize the severe public health implications of PM_2.5 exposure on vulnerable populations, particularly children, underscoring the necessity for stringent air quality regulations and public health interventions across India.