Risk factors and outcomes for prolonged versus brief fever: a prospective cohort study

Introduction: Prolonged fever occurs with infectious and noninfectious diseases but is poorly studied in intensive care units. The aims of this prospective multicenter noninterventional study were to determine the incidence and etiologies of prolonged fever in critically ill patients and to compare outcomes for prolonged fever and short-lasting fever. Methods: The study involved two periods of 2 months each, with 507 patients hospitalized >= 24 hours. Fever was defined by at least one episode of temperature >= 38.3 degrees C, and prolonged fever, as lasting > 5 days. Backward stepwise logistic regression was performed to identify the independent factors associated with prolonged fever versus short-lasting fever. Results: Prolonged or short-lasting fever occurred in 87 (17%) and 278 (55%) patients, respectively. Infectious and noninfectious causes were found in 54 (62%) and 27 (31%) of 87 patients, respectively; in six patients (7%), prolonged fever remained unexplained. The two most common sites of infection were ventilator-associated pneumonia (n = 25) and intraabdominal infection (n = 13). Noninfectious fever (n = 27) was neurogenic in 19 (70%) patients and mainly associated with cerebral injury (84%). Independent risk factors for prolonged fever were cerebral injury at admission (OR = 5.03; 95% CI, 2.51 to 10.06), severe sepsis (OR = 2.79; 95% CI, 1.35 to 5.79), number of infections (OR = 2.35; 95% CI, 1.43 to 3.86), and mechanical-ventilation duration (OR = 1.05; 95% CI, 1.01 to 1.09). Older patients were less likely to develop prolonged fever. ICU mortality did not differ between the two groups. Conclusions: Prolonged fever was common, mainly due to severe infections, particularly ventilator-associated pneumonia, and mixed infectious causes were frequent, warranting systematic and careful search for multiple causes. Neurogenic fever was also especially frequent.