Causes and consequences of fever complicating critical surgical illness

Fever may have malign consequences in the postoperative period. This study was performed to determine the causes and consequences of fever in critically ill surgical patients. The specific hypothesis tested is that postoperative fever is associated with adverse clinical outcomes, including increased organ dysfunction and risk of death.

Inception-cohort study of critically ill surgical patients who manifested a core temperature of >/=38.2 degrees C for the first time. The episode of fever was monitored until resolution, which was defined as a core temperature of <38.2 degrees C for at least 72 consecutive h. Demographic data collected included age, gender, admission diagnosis, admission status (elective/emergency), severity of illness (APACHE III), the systemic inflammatory response syndrome (SIRS) score, the cumulative multiple organ dysfunction score, cause of fever (infectious/non-infectious), ICU and hospital length of stay, and mortality. The day of onset of fever in the ICU, peak temperature, ICU day of peak temperature, and duration of fever episode were recorded. All diagnostic and therapeutic interventions were recorded, including the type and duration of antibiotic therapy. Univariate results of possible significance (alpha < 0.15) were tested in logistic regression models for independence of effect upon mortality after auto-correlation was excluded by matrix correlations and the Durbin-Watson statistic. Cases where both non-infectious and infectious causes of fever were present were analyzed as part of the infectious group, whereas the cumulative MOD score was dichotomized (< 5, >/=5 points) at a value known to be associated with increased mortality.

Among 2,419 screened patients, 626 patients (26%) developed fever. Febrile patients were older, sicker, more likely to have undergone emergency surgery, more likely to develop organ dysfunction, and more likely to die (all, p < 0.0001). The mean day of onset of fever was day 1 and the mean peak temperature for the episode was 39.1 +/- 0.1 degrees C. For most patients, it was their only episode of fever, with a mean of 1.4 +/- 0.1 episodes/patient. Forty-six percent of febrile patients were found to have an infectious cause of fever. Nearly all patients had SIRS, and nearly all developed organ dysfunction to some degree. By logistic regression, the presence of SIRS (as opposed to fever in isolation), emergency status, higher APACHE III score and the peak temperature were associated with increased mortality, with peak temperature being the most powerful predictor in the model (OR 2.20, 95% Cl 1.57-3.19). Gender had no bearing on outcome, and there was a trend toward a protective effect from an infectious etiology of fever.

Postoperative fever is deleterious to critically ill patients. The magnitude of fever is a determinant of mortality, whereas an infectious etiology of fever may not be. The impacts of nosocomial infection and suppression of fever on critically surgical patients deserve further study.