Risk Factors for Recurrent Clostridium difficile Infection in Children: A Nested Case-Control Study

Abstract

To identify risk factors for recurrent Clostridium difficile infection (RCDI) in children. A nested case-control study was performed to identify RCDI risk factors using a pediatric cohort of inpatients and outpatients diagnosed with Clostridium difficile infection by tcdB polymerase chain reaction (PCR) at an academic children's hospital between December 9, 2012, and June 30, 2014. Strict inclusion criteria were adopted to limit selection bias related to inappropriate inclusion of patients with probable C difficile colonization. Thirty children with RCDI were compared with 94 children with non-RCDI. Statistically significant associations were identified between RCDI and malignancy (OR 2.8, 95% CI 1.0-7.4, P = .044), tracheostomy tube dependence (OR 5.2, 95% CI 1.1-24.7, P = .037), and tcdB PCR cycle threshold (OR 0.87, 95% CI 0.78-0.97, P = .01) using multivariable logistic regression modeling. The receiver operator characteristic curve for PCR cycle threshold as a predictor of RCDI demonstrated area under the curve = 0.67. The highest predictive rate (75%) for RCDI was demonstrated at cycle threshold cutpoint ≤ 20. The difference between sensitivity (64%) and specificity (68%) was minimized at cycle threshold cutpoint ≤ 23. Compared with controls with non-RCDI, children excluded because of probable C difficile colonization had a similar cycle threshold value (27.5 vs 27.2, P = .77). Malignancy and tracheostomy tube dependence were identified as RCDI risk factors. Although RCDI was associated with positivity at a lower tcdB PCR cycle threshold, the clinical utility of cycle threshold as a tool to predict recurrence was limited. Better methods to predict RCDI are needed to prioritize pediatric populations to target for RCDI prevention efforts.