Predictive value for infection of febrile morbidity after vaginal surgery

Objective: To determine the screening value of febrile morbidity for detecting infections after vaginal surgery. Methods: A cohort of 431 consecutive women had vaginal surgery at the M. S. Hershey Medical Center from September 1988 through June 1995. Outcomes of febrile morbidity and infection were analyzed. Results: Fifty-four of 431 patients (12.5%) had febrile morbidity. Thirty-five infections (8.1%) were identified, of which only 13 were accompanied by febrile morbidity. Forty-one patients (9.5%) had unexplained fevers. The sensitivity of febrile morbidity for postoperative infection was 40%, specificity was 98%, positive predictive value was 26%, and negative predictive value was 94%. Stepwise logistic regression found blood loss (odds ratio 1.001/mL; confidence interval 1.0001-1.0035), uterine weight (0.987/g; 0.976-0.999), and parity (1.570; 1.146-2.050) as significant independent variables for developing fever. Patient weight (0.984/lb; 0,971- 0.998) and type of procedure (2.16; 2.12-6.38) were confirmed as significant independent variables for postsurgical infections. Conclusion: Febrile morbidity had limited value as a screening test for postoperative infection, with poor sensitivity and positive predictive value after vaginal surgery.