ABSTRACT
Compared with mortality, the impact of weather and climate on human morbidity is less well understood, especially in the cold season. We examined the relationships between weather and emergency department (ED) visitation at hospitals in Roanoke and Charlottesville, Virginia, two locations with similar climates and population demographic profiles. Using patient-level data obtained from electronic medical records, each patient who visited the ED was linked to that day's weather from one of 8 weather stations in the region based on each patient's ZIP code of residence. The resulting 2010-2017 daily ED visit time series were examined using a distributed lag non-linear model to account for the concurrent and lagged effects of weather. Total ED visits were modeled separately for each location along with subsets based on gender, race, and age. The relationship between the relative risk of ED visitation and temperature or apparent temperature over lags of one week was positive and approximately linear at both locations. The relative risk increased about 5% on warm, humid days in both cities (lag 0 or lag 1). Cold conditions had a protective effect, with up to a 15% decline on cold days, but ED visits increased by 4% from 2 to 5 days after the cold event. The effect of thermal extremes tended to be larger for non-whites and the elderly, and there was some evidence of a greater lagged response for non-whites in Roanoke. Females in Roanoke were more impacted by winter cold conditions than males, who were more likely to show a lagged response at high temperatures. In Charlottesville, males sought ED attention at lower temperatures than did females. The similarities in the ED response patterns between these two hospitals suggest that certain aspects of the response may be generalizable to other locations that have similar climates and demographic profiles.