Rationale: Abnormally high or low ambient temperature often causes physiological complications that warrant hospitalization. With extreme heat and cold episodes increasing in frequency and intensity, adverse health outcomes associated with anomalous fluctuations in temperature could increase.
Objective: We use information from 2006, 2007 and 2008 to derive national estimates of the health burden attributed to hyperthermia and hypothermia. These estimates represent the number of hospital admissions, associated costs, and the geographic and demographic distribution of cases.
Methods: The Health Care Utilization Project (HCUP) provides a 20% sample of hospital admissions in the United States. We use a combination of ICD9 codes and E-codes to identify admissions due to heat and cold (for heat, ICD9 = 992 and E-code = E900.0, E900.9; for cold, ICD9 = 991 and E-code = E901.0, E901.9). In order to further ensure that we identify admissions related to environmental causes, we restrict admissions to months of May to September for heat-related cases and October to March for cold-related cases. Cost-to-charge ratio files provided by HCUP were used to calculate the healthcare cost associated with each episode. All estimates are appropriately weighted to account for sampling design.
Results: The estimated total number of heat-related cases with 95% confidence intervals considering only principal diagnosis were 4610 [4149, 5072], 3654 [3267, 4042] and 3158 [ 2832, 3483] in 2006, 2007 and 2008 respectively. The corresponding numbers for cold-related cases were 2362 [2056, 2668], 3146 [2788, 3504] and 3054 [2654, 3456]. If case definitions mentioned above are applied to all available diagnoses instead of the principal diagnosis, the number of cases is twice as large for heat-related cases and six times larger for cold-related cases. The direct cost of heat-related hospital admissions with 95% confidence intervals were 25.5 [21.3, 29.7], 17.5 [14.3, 20.7] and 15.9 [13.3, 18.6] million USD in 2006, 2007 and 2008 respectively. The corresponding numbers for cold-related cases were 25.4 [19.8, 31.1], 27.9 [22.5, 33.4] and 28.1 [21.7, 34.6] million USD. When considering all diagnoses, these figures are 2.5 times higher for heat-related cases and eight times for cold-related cases. Compared to younger age groups, seniors (above 65 years) had to be most frequently hospitalized for both type of events. Hospitals in rural locations had more frequent admissions compared to those in urban areas. Admissions were most frequent from zip-codes falling in the lowest median income quartile. Heat-related admissions were more frequent in the South region, while cold-related frequency was higher in the Midwest region. While 40% of the admissions were Medicare patients, around 15% were uninsured.
Conclusions: We find a substantial health burden related to heat and cold. The demographic and geographic variation in hospital admission provides inputs for tailoring adaptation strategies across different regions. The estimates are highly sensitive to the choice between principal and secondary diagnosis that have methodological implications for case identification in future research.