Abstract

Does Choice of Ambient Heat Metric Affect Heat-Related Epidemiologic Analyses in the US Medicare Population?

Keith Spangler¹, Quinn Adams¹, Danielle Braun², Francesca Dominici², Gregory Wellenius¹ (1)Boston University School of Public Health, (2)Harvard University T.H. Chan School of Public Health

APHA 2022 Annual Meeting and Expo

Background: Although most epidemiologic studies of heat assess ambient temperature exposures, other thermal comfort indices theoretically confer greater physiological relevance by incorporating additional meteorological variables. However, the extent to which these metrics differentially predict risks of death or hospitalizations is unknown.

Methods: We calculated daily maximum and minimum ambient temperature (Tmax/min), heat index (HImax/min), wet-bulb globe temperature (WBGTmax/min), and the universal thermal climate index (UTCImax/min) for populous conterminous US counties and linked with all-cause mortality and heat-applicable hospitalizations among Medicare beneficiaries (2006-2016). We fit distributed-lag nonlinear models for each heat metric separately to assess relationships with either mortality or hospitalizations. We extracted relative risks (RR) at the 99th percentile compared to the minimum mortality/morbidity percentile (MMP) of each heat metric for intercomparison.

Results: Days of extreme heat were associated with higher risks of morbidity and mortality across all heat metrics. Mortality RRs ranged from 1.016 (95% CI: 1.009, 1.023) for UTCImin to 1.032 (95% CI: 1.025, 1.040) for HImax, and hospitalization RRs ranged from 1.093 (95% CI: 1.080, 1.105) for WBGTmin to 1.137 (95% CI: 1.124, 1.151) for HImax. Associations were more pronounced for maximum daily heat values versus the corresponding minimum for the same metric. The starkest example was between HImax (RR: 1.137 [95% CI: 1.124, 1.151]) and HImin (RR: 1.098 [95% CI: 1.085, 1.112]) for hospitalizations. When comparing RRs across heat metrics, we found no statistically significant differences within the minimum and maximum heat values (i.e., no significant differences between Tmax/HImax/WBGTmax/UTCImax or between Tmin/HImin/WBGTmin/UTCImin).

Conclusions: Among Medicare beneficiaries in populous US counties, daily maximum values of ambient heat are associated with greater RR of morbidity and mortality versus minimum values of the same metric. Choice of heat metric (e.g., temperature versus heat index) does not appear to substantively affect the calculation of excess heat-related risk in this population.