Mortality Associated with Ambient Exposure in India: Results from the Million Death Study

Studies on the extent to which long-term exposure to ambient particulate matter (PM) with aerodynamic diameter $$ ($$) contributes to adult mortality in India are few, despite over 99% of Indians being exposed to levels that the World Health Organization (WHO) considers unsafe.

We conducted a retrospective cohort study within the Million Death Study (MDS) to provide the first-ever quantification of national mortality from exposure to $$ in India from 1999 to 2014.

We calculated relative risks (RRs) by linking a total of ten 3-y intervals of satellite-based estimated $$ exposure to deaths 3 to 5 y later in over 7,400 small villages or urban blocks covering a total population of $$. We applied using a model-based geostatistical model, adjusted for individual age, sex, and year of death; smoking prevalence, rural/urban residency, area-level female illiteracy, languages, and spatial clustering and unit-level variation.

$$ exposure levels increased from 1999 to 2014, particularly in central and eastern India. Among 212,573 deaths at ages 15-69 y, after spatial adjustment, we found a significant RR of 1.09 [95% credible interval (CI): 1.04, 1.14] for stroke deaths per $$ increase in $$ exposure, but no significant excess for deaths from chronic respiratory disease and ischemic heart disease (IHD), all nonaccidental causes, and total mortality (after excluding stroke). Spatial adjustment attenuated the RRs for chronic respiratory disease and IHD but raised those for stroke. The RRs were consistent in various sensitivity analyses with spatial adjustment, including stratifying by levels of solid fuel exposure, by sex, and by age group, addition of climatic variables, and in supplementary case-control analyses using injury deaths as controls.

Direct epidemiological measurements, despite inherent limitations, yielded associations between mortality and long-term $$ inconsistent with those reported in earlier models used by the WHO to derive estimates of $$ mortality in India. The modest RRs in our study are consistent with near or null mortality effects. They suggest suitable caution in estimating deaths from $$ exposure based on MDS results and even more caution in extrapolating model-based associations of risk derived mostly from high-income countries to India. https://doi.org/10.1289/EHP9538.