Nighttime temperature exposure and heat-related illness risk: a small-area distributed lag non-linear analysis in North Carolina, USA, 2011–2021

期刊论文
发表日期：

2026年6月12日

查看全文

Sophia C. Ryan,
Margaret M. Sugg,
Matthew Hefner,
Jennifer D. Runkle

摘要

High temperatures and extreme heat kill more people in the United States than any other extreme weather event, and higher nighttime temperatures and associated health risks are expected to increase under future climate change. This study applied a small-area, case time series distributed lag non-linear model (DLNM) to characterize the association between temperature exposure and heat-related illness (HRI) emergency department (ED) visits (n = 31,765) using daily ED visit data (April—September 2011–2021) and daily nighttime temperature data at the zip code tabulation area (ZCTA) level in North Carolina (n = 802). Stratified models investigate how individual and community-level factors predict HRI risks at the 80th, 90th and 95th percentiles of minimum temperature compared to HRI risks at the 50th percentile. The cumulative risk of HRI increases exponentially above the 80th percentile, with individuals more than two times (RR:2.06 [CI: 1.88–2.25]) as likely to experience HRIs at the 95th nighttime temperature percentile. Temporally, the risk of HRIs is highest on the day of exposure (RR_Lag0:1.80 [CI:1.73–1.88]), with risk remaining elevated up to two days after exposure (RR_Lag2:1.06 [CI:1.05–1.08]) and increasing again on lag day seven (RR_Lag7:1.06 [CI:1.03–1.08]). Results indicate differential HRI risk above the 80th percentile among young people (i.e., ages 24 and under) (RR: 2.14, CI:2.01–2.27), older adults (65 +) (RR:2.18, CI:2.05–2.32), and low-income individuals (RR:2.23, CI:2.09–2.38), and higher risk at the community level for rural (RR:2.09, CI:1.91–2.30), coastal (RR:2.18, CI:1.87–2.54), high income (RR:2.15, CI:1.94–2.38) and racially diverse communities (RR:2.23, CI:2.01–2.48), compared to median nighttime temperatures. Results emphasize the need for heat-health messaging and risk reduction efforts across community types that are widely accessible, regardless of individual context.