Within-Summer Adaptation To Heat: A Multi-Country Analysis

Abstract

Introduction: A number of studies have examined the variation in mortality risk associated to heat within the summer period. Here we investigate the issue in a large multi-country data set, applying flexible statistical techniques based on time-varying distributed lag non-linear models (DLNMs). Methods: We collected daily time series data of temperature and all-cause mortality for 272 locations in 7 countries, including 21,572,892 deaths occurring in summer months between 1985-2012. The analysis is based on two-stage design. The heat-mortality relationship was estimated in each location with time-varying DLNMs, based on a bivariate spline to model the exposure-lag-response over lag 0-10. The within-summer variation was expressed through an interaction between the bivariate spline variables and day of the season. Estimates were pooled by country through multivariate meta-analysis. Results: Comparison of predicted overall cumulative exposure-response curves for early and late summer indicates a strong evidence of a reduction in risk along the season. The excess risk for the 99th percentile vs the minimum mortality temperature decreases from 31% (95%CI: 7;60%) to 25% (6;47%) in Australia, from 17% (6;30%) to 2% (-8;13%) in Canada, from 21% (14;29%) to 3% (0;7%) in Japan, from 35% (12;62%) to -9% (-17;7%) in South Korea, from 64% (49;80%) to 26% (16;36%) in Spain, from 13% (4;24) to 11% (5;17%) in the UK, and from 16% (12;21%) to 3% (0;7%) in the USA. The analysis of the lag-response curves suggests that the attenuation in late summer is due to a reduction of risk at longer lags. Conclusions: This multi-country analysis consistently indicates that the mortality risk due to heat decreases along the summer. This phenomenon can be explained by several factors, such as within-season adaptation and acclimatization, or harvesting effects that deplete the pool of susceptible in early summer. This issue has important implications on public health policies and climate change projections.