Number Needed to Treat in Fluconazole Prophylaxis in the Neonatal Intensive Care Unit

Abstract

To the Editors: In a recent publication in this journal, Rueda et al1 conducted a study in a developing country and demonstrated that routine prophylactic use of fluconazole at birth versus placebo in 523 neonates less than 1250 g is associated with significant impact on the frequency of documented systemic Candida infection (CI). The control group developed CI in 21 (7.7%), whereas the treatment group had only 3 (1.1%) CI. The benefit of prophylactic fluconazole (PF) to prevent a systemic CI was represented by the number needed to treat (NNT) with a nominal value of 7 according to Rueda et al.1 This measure is calculated on the inverse of the absolute risk reduction.2 NNT of 7 means that it is necessary to treat 7 neonates with PF to prevent 1 CI. This NNT is calculated incorrectly. Considering the absolute risk reduction of 6.6% (7.7% − 1.1%), the NNT should be 15.2 (calculated by 1/6.6%) instead of 7. Although this error does not change the statistical significance of the benefit of fluconazole in the study of Rueda et al (P = 0.007),1 for pharmacoeconomic analysis studies, this error has an important impact. There is scant information from developing countries about PF in neonates,1 and the estimated NNT of 15.2 is acceptable. A broader view on the updated systematic review results of the Cochrane meta-analysis of 5 of 8 eligible trials on “systemic antifungal agents prophylatic in low birth weight infants,” issue 12, 2010,3 revealed that NNT for PF is in line with the confidence intervals of PF, with significant reduction in the risk of invasive fungal infection in the infants who received prophylaxis (NNT = 11; 95% confidence interval, 7–33).3 Although it requires confirmation from future studies, the prophylactic use of fluconazole in premature infants of lower weight, with higher natural risk of systemic candidiasis (extremely low birth weight <1000 g),4 suggests a greater benefit: the calculated NNT is 2.68 for fungal colonization (22% PF × 60% placebo) and NNT of 5 for systemic fungal infection (0% PF × 20% placebo).4 Moreover, the systematic reviews still show a nonsignificant decrease in the risk of death before hospital discharge (typical relative risk, 0.74 [95% confidence interval, 0.51–1.09]).3 To date, the only untoward effects reported with fluconazole use in preterm infants are reversible elevations in hepatic enzymes or bilirubin, without sequelae.5 ACKNOWLEDGMENT The authors thank Dr. Débora Cristina Raulik Shieh of Universidade Federal do Paraná for her assistance in the preparation of the manuscript. Huei Hsin Shieh, MD Pediatric Intensive Care Unit University Hospital University of Sao Paulo Silvia Maria Ibidi, MD, MSc Neonatal Intensive Care Unit University Hospital University of Sao Paulo Alfredo Elias Gilio, MD, PhD Division of Pediatrics University Hospital University of Sao Paulo Sao Paulo, Brazil