BACKGROUND: With climate change, the number of extreme heat (EH) days are expected to increase, along with heat-related morbidity and mortality. The Michigan Climate and Health Program needed county-level estimates of the burden of current and projected heat-related morbidity. As Michigan lacks a statewide database of emergency department (ED) visits, EPA’s BenMAP was used to estimate the fraction of non-traumatic ‘all-cause’ ED visits which could be attributed to heat.
METHODS: BenMAP is designed to estimate morbidity and mortality attributable to changes in air pollutants. We adapted its inputs to include the distribution of historic and projected temperature and age categories across Michigan, along with heat-attributable morbidity risk ratios from the literature. ‘All-cause’ ED visits (all natural causes, ICD-9 CM <800, and heat-related external causes, E900.0, E900.2) were estimated by adapting a method used by BenMAP for estimating asthma ED visits for states lacking actual data. Age- and cause-specific ED visits were derived from the Healthcare Cost and Utilization Project (HCUP) State Emergency Department Databases. All-cause ED visit rates were estimated using age-specific HCUP data from the Midwest region applied to all Michigan counties, using county age distributions derived from 2007 Census. Present (years 1971-2000) and future (years 2041-2070) ED visits attributable to EH were estimated as follows. For each county, time period, temperature, and age group, we calculated the daily attributable fraction which was then multiplied by the number of EH days. This result was imported into BenMAP as the quantity in the burden-of-disease calculation which varies spatially and temporally. The burden-of-disease function was then defined in BenMAP as BODcap, the burden of disease due to heat in county c in age group a in time period p (present or future). BODcap estimates were summed across age groups to generate county and time-period-specific estimates. County estimates were summed to generate statewide estimates.
RESULTS: Statewide heat-associated ED rates were estimated as 12 per 100,000 persons (1,218 visits) in the present and 68 per 100,000 persons (7,845 visits) in the projected period. There was significant heterogeneity between counties, with a 16-fold variation in ED rates in the present and a 6-fold variation in the projected period. All counties saw an increase in ED visits between the two time periods.
CONCLUSIONS: This method produced estimates of current and future morbidity that could occur due to extreme heat. The results can be used to prioritize county level measures to protect public health during hot summer weather.