The diagnostic value of fungal surveillance cultures in critically ill patients.

BACKGROUND: Heavy fungal colonization is a known risk factor for fungal infection, yet the value of fungal surveillance cultures is uncertain. METHODS: To evaluate the utility of fungal surveillance cultures in predicting fungal infections, we evaluated surveillance fungal cultures over a three month period in a prospective, cohort study conducted at a university medical center with a large tertiary referral population. We enrolled 172 patients in the Oncology Center and the medical and surgical intensive care units at Johns Hopkins Hospital. RESULTS: Surveillance cultures from five sites were obtained twice weekly and evaluated for prediction of subsequent fungal infection. Infections were prospectively defined and evaluated by a panel of clinicians. Test characteristics were assessed. Of 159 eligible patients, 14 (9%) developed invasive fungal infections. Having two or more surveillance sites positive in a single day had an odds ratio of 8.2 (1.1-358.0) (p = 0.03), a negative predictive value of 0.98, sensitivity of 0.92, and a likelihood ratio of 1.6 for a fungal infection. In a multiple logistic regression model and Kaplan-Meier analysis, fungal burden was strongly and independently associated with infection (p < 0.05). CONCLUSIONS: Surveillance cultures are helpful in determining fungal colonization but do not have a high positive predictive value for fungal infection in a broad population of intensive care unit patients. However, fungal infection is more likely in heavily colonized patients, and surveillance cultures show that fungal infection is extremely unlikely in patients without fungal colonization.