Hypothermia is a bedside predictor of imminent death in patients with congestive heart failure

Background Most studies on predictors of mortality for patients with congestive heart failure (CHF) have described predictors that are either difficult to measure in routine practice or are only modestly sensitive and specific. Having observed 3 patients whose body temperature decreased shortly before death, we hypothesized that hypothermia may predict inhospital mortality. Methods The medical records of 291 patients with a primary discharge diagnosis of CHF were selected from 423 admissions to Memorial Hermann Hospital, Houston, Tex, 1998, after excluding patients with comorbidities that confound body temperature, deaths for causes other than progressive pump failure, and readmissions except the lost. Three groups were defined on the basis of admission body temperature (T-adm): hypothermia groups T-adm (95.5 degrees F-96.5 degrees F) and T-adm <95.5 degrees F, and reference group T-adm >= 96.6 degrees F. Several other known CHF risk factors were studied for confounding, and adjusted hazard ratios were calculated using Cox regression. Results of the 291 patients (mean age 73 years, 47% men), 17 (6%) had hypothermia on admission. Mean hospital stay was 5 days. Of the 17 (6%) patients who died of pump failure, 5 had been hypothermic on admission. Hypothermia was significantly associated with survival, and after adjusting for New York Heart Association functional class, hazard ratio for T-adm < 95.5 degrees F was 4.46 (95% confidence interval 1.38-14.3) (P trend = .0283). Conclusions Hypothermia predicted inhospital death in these patients with CHF. If confirmed by future studies, this finding could prove useful, because temperature can be measured continuously, rapidly, and inexpensively, in or out of the hospital.