Change in the number of PM2.5-attributed deaths in China from 2000 to 2010: Comparison between estimations from census-based epidemiology and pre-established exposure-response functions

Long-term exposure to ambient fine particulate matter (PM2.5) has been identified as a major contributor to disease burden in many countries, including China. The effects of long-term PM2.5 exposure have been evaluated by risk assessment studies, which are based on an exposure-response function (ERF) derived from a specific epidemiological study or multiple studies. To explore whether generalization from the pre-established ERFs (including the integrated exposure-response [IER] model and the global exposure mortality model [GEMM]) introduces biases into risk assessment of a specific local population, we conducted the first census-based epidemiological study of PM2.5, aimed at the entire population of mainland China. Using a difference-in-difference analysis at the county level, we associated mortality changes from 2000 to 2010 to corresponding PM2.5 changes, with adjustments made for multiple socioeconomic factors. Based on the epidemiological linkage between PM2.5 and total mortality, we calculated the change in PM2.5-attributed deaths and compared this value with the results derived from the pre-established models (IER and GEMM). According to the epidemiological model, a 10-μg/m3 increment in PM2.5 was associated with a 4.3% (95% confidence interval [CI]: 1.9%–6.7%) increment in total mortality, and the change in PM2.5-attributed deaths from 2000 to 2010 was estimated as 561,000 (95% CI: 539,000–581,000). The census-based estimation was in better agreement with the GEMM results (545,000–612,000) than was the IER result (354,000 [95% CI: 286,000–421,000]). In sensitivity analyses of the epidemiological model, the association between PM2.5 exposure and mortality was stronger among younger adults, consistent with the pre-established models. Due to the potential limitations of our findings, we cannot conclusively state that GEMM is more reliable than IER in China. Future studies are warranted to confirm or refute our findings. •Pre-established PM2.5-mortality function (IER or GEMM) has been widely used.•Generalizing a pre-established function to a local population may cause bias.•We estimate a census-based function for the general population of China.•We compare the census-based function with pre-established ones in China.•The Census-based result is in better agreement with GEMM's result than IER's.