Demography

Random effects model

Percentile

Statistics

Effect modification

Distributed lag

Computer science

Environmental health

Computer network

Mathematics

Geography

Population

Meta-analysis

Internal medicine

Sociology

Confidence interval

Medicine

Poisson regression

The Science of The Total Environment

Few data are available on the health impacts of heatwaves in China, and in particular, the heatwave definition and vulnerable populations remain to be identified. We collected data on daily maximum temperature and mortality from 31 Chinese capital cities during 2007-2013. A Poisson regression model allowing for over-dispersion was applied to estimate the short-term effects of heatwaves on mortality in hot season (May-September). 15 heatwave definitions combining five heat thresholds (90.0th, 92.5th, 95th, 97.5th and 99th percentiles of daily maximum temperature) and three durations (≥2, ≥3 and ≥4 days) were compared. The pooled effects were then computed using random effect meta-analysis based on the residual maximum likelihood estimation. Effect modification of heatwave-mortality association by individual-level characteristics was tested using a stratified analysis. Potential effect modification by city-level characteristics was examined by meta-regression analysis. Totally, 259 million permanent residents were covered and 4,481,090 non-accidental deaths occurred during the study period. Generally, the magnitude of heatwave impacts increased by intensities and durations of the heatwaves. Heatwave definition using daily maximum temperature ≥ 92.5th percentile with duration ≥3 days produced the best model fit. The pooled relative risks of heatwaves on non-accidental mortality at lag 0, lag 0-2 and lag 0-10 days were 1.06 (95%CI: 1.03-1.09), 1.09 (1.05-1.13) and 1.10 (1.05-1.15), respectively. Compared with non-accidental mortality, higher effect estimates of heatwaves were observed among deaths from ischemic heart diseases, stroke and respiratory diseases, although the differences were not statistically significant. Females, those ≥75 years old and the illiterates were more vulnerable to heatwaves. Cities with higher concentrations of PM2.5, higher latitudes, and lower numbers of hospital beds per 10,000 populations had higher mortality risks during heatwaves. These findings may have important implications for developing heat alert systems and early response actions on protecting the vulnerable populations from adverse health effects of heatwave in China.

Heatwave and mortality in 31 major Chinese cities: Definition, vulnerability and implications