Methods: We used nonlinear distributed lag time-series models to evaluate the association between heat index and daily emergency department admissions in 7 cities in NH, 7 cities in ME, and in the state of RI between May and September from 2000 to 2010. Hourly weather station data from the NOAA National Climatic Data Center (NCDC) were used to calculate daily maximum heat index. For NH and ME, the geographic unit of analysis was defined as all towns in a 10-mile radius surrounding each weather station. All-cause daily ED visits were obtained from hospital discharge data in each state. ED admissions with heat or dehydration (ICD-9 267.5; 992; E900) as a primary or secondary cause, or renal (ICD-9 580-589) as a primary cause were classified as heat-specific. We applied overdispersed Poisson constrained distributed lag models controlling for long-term time trends, day of week, and federal holidays to each study site. All models considered heat index over the previous 0-7 days and allowed for non-linear exposure-response functions. City-specific risk estimates were then pooled in a meta-analysis to provide a single regional estimate for all-cause and heat-related risk.
Results: This analysis focuses on New England and covers 100%, 60%, and 66% of the populations of Rhode Island, Maine, and New Hampshire, respectively. The total study population represents 74% of the population across the 3 states and 19% of the population in New England. Maximum daily heat indices from May-September ranged from 39°F to 110°F, with a median of 74°F. We observed a higher risk of all-cause ED visits associated with higher heat index. In a pooled analysis of NH and RI, rates of all-cause ED visits were 1.0% (95% CI: 0.6%, 1.5%) higher on days with a heat index 90°F versus 80°F, while the cumulative risk over a 1-week lag period was 3.3% (95% CI: 2.2%, 4.4%) higher. Rates of heat-specific ED visits, on days with a heat index of 90°F versus 80°F, were 27.3% (95% CI: 19.7%, 35.4%) higher on the same day and 39.9% (95% CI: 30.0%, 50.6%) higher over the 1-week lag period.
Conclusions: This project is a unique regional collaboration across multiple state health departments and universities. Our preliminary results suggest that there may be adverse health impacts associated with heat index below the current NWS excessive heat advisory or warning thresholds. These locally-relevant results will be communicated to NWS partners in an attempt to revise the current threshold and will inform future local public health actions.