Factors Associated With Heat-Related In-Hospital Mortality: A Multicity Case–Crossover Analysis

MAA1-O-07 Introduction: Several studies have identified strong effects of high temperatures on mortality. Only a few, however, investigated individual medical susceptibility to these effects. In our previous report (Epidemiology, in press) we noted that in-hospital mortality was related to increases in temperature. The finding implicates potential public health measures because people in the hospital are supposed to be under medical attention, avoiding the consequences of temperature stress. We applied a multicity case-crossover analysis to evaluate the factors associated with heat-related in-hospital mortality. Methods: We studied 267,611 residents ages 35 years or older of 4 Italian cities who died during 1997–2004 from all natural causes. The mortality datasets were linked with the hospital discharge files to retrieve information on place of death (out-of-hospital 34%, discharged in the last 28 days 8%, in-hospital 54%, nursing home 4%). Hospital wards and primary diagnoses were considered for people who died in the hospital. A city-specific case-crossover analysis was performed to evaluate the association between mean apparent temperature (lag 0-1) and mortality. Control days were chosen according to the time-stratified approach. The exposure variable was modeled as a linear piece-wise spline with 2 inner knots. Pooled odds ratios (ORs) and 95% confidence intervals (95% CIs) of dying at 30°C (mean apparent temperature) compared to 20°C were estimated in a random-effect meta-analysis. Results: We estimated an overall OR of 1.35 (95% CI = 1.25–1.44). The OR was similar regardless of place of death (out-of-hospital OR = 1.37, in-hospital OR = 1.33). Most of the effect on in-hospital deaths was related to admissions immediately after the temperature increase (lag 0-1; OR = 1.41), but it remained high and statistically significant also in subjects already in the hospital (OR = 1.29, 95% CI = 1.23–1.35). We restricted the analysis to the latter category (113,443 subjects) and identified the hospital ward as an important risk indicator: those staying in general medicine were at higher risk compared with those in intensive care units (1.39 vs 1.10). Psychoses (OR = 2.09; 95% CI = 1.27–3.42), hypertension (2.15; 1.09–4.24), heart failure (1.70; 1.37–2.11), cerebrovascular diseases (1.59; 1.39–1.83), and chronic pulmonary diseases (2.42; 1.30–4.47) registered the higher effect estimates. Discussion and Conclusions: The study indicates in-hospital mortality as being strongly associated to high temperatures. The effect is negligible for patients staying in intensive care units, whereas it is mostly concentrated in those in general medicine wards and in subjects suffering from cardiac and cerebrovascular diseases. These findings suggest that maintaining a comfortable temperature in hospitals, especially in general medicine wards, can be an immediate and necessary measure against the health effects of heat.