Comprehensive Analysis of Temperature in Hospitalized Patients

Fever and hypothermia do not affect the all-cause 30-day hospital readmission

BACKGROUND

One of the goals of the Affordable Care Act is to decrease hospital readmissions. While widely adhered to, there is no published research to support the practice of delaying discharge if patients exhibit fever or hypothermia in the preceding 24 h, which is the focus of our study.

METHODS

Retrospective analysis of the minimal (Tmin) and maximal (Tmax) body temperatures collected during the last 24 h before discharge of 19,038 inpatients. Fever was defined as Tmax >99.5F (+1SD from the mean Tmax) or >100.2F (+2SDs), and hypothermia as Tmin <97.1F (-1SD from the mean Tmin) or <96.7F (-2SDs).

RESULTS

The overall readmission rate was 10.2% (highest for General Medicine and Pediatrics). The rate of readmission was not different between normothermic patients and those with abnormal body temperature, except for higher readmission rate (12.2%) for patients with fever at 1SD from Tmax compared with normothermic patients (9.96%). Neither fever nor hypothermia was associated with shorter time to readmission, except for fever at 2 SDs from Tmax (10.6 days) compared with normothermic patients (12.6 days). Surprisingly, univariate analysis revealed that higher Tmax and older age were associated with lower readmission probability. Both uni- and multivariate analysis showed that the presence of fever is associated with lower readmission probability. Evaluating 200 individual cases, the most common explanation for body temperature abnormality was infection and 90% of the preventable readmissions were due to infection.

CONCLUSIONS

Abnormal body temperature 24 h prior to discharge was not useful for predicting the probability of readmission.

Disruption of the Body Temperature Circadian Rhythm in Hospitalized Patients

BACKGROUND

Dysregulation of the circadian rhythm is linked to immune response deficiencies. Diurnal temperature variation-a surrogate for the circadian rhythm, exists in humans, yet its preservation during illness is not well understood.

MATERIALS AND METHODS

Temperatures measured within one-half hour of 8 am, 12 pm, 4 pm, 8 pm, 12 am, and 4 am from 16,245 hospitalized patients were statistically analyzed.

RESULTS

Although we found a diurnal pattern when analyzing the ensemble of temperatures from all patients, stratified by measurement site, the trough-to-peak difference was only 0.2°F, while that for healthy volunteers had been in the 0.5°F to 1.9°F range. The peaks occurred at 8 pm for all patients, regardless of age or sex, which is similar to healthy people. However, the troughs were shifted to later times compared with the 6 am in healthy people-for young patients (age 20-30 years) the trough was at 8 am and for elderly patients (age 70-80 years), at 12 pm, again regardless of patients' sex. Analysis of individual patients showed that less than 20% of patients exhibited diurnal variation and among those showing variation, the trend was present only on the minority of hospitalization days. The presence or absence of an infectious process or fever did not influence the proportion of patients showing diurnal variation.

CONCLUSIONS

Hospitalization is associated with disruption in the circadian rhythm as reflected by patients' body temperature. Since abnormality in body temperature is known to affect patient outcomes, an understanding of the diurnal cycle during hospitalization is the first step towards devising approaches to re-establish the circadian rhythm.