Predictive accuracy of risk factors for RSV-related hospitalizations among infants in low-income families born at 32 to 35 weeks of gestation
Abstract Objective: To examine the accuracy of risk factors available from administrative data to predict hospitalization due to severe respiratory syncytial virus (RSV) infection among moderately premature infants in low-income families. Design: Retrospective cohort study using birth certificate and Medicaid enrollment and claims data. Setting and participants: 2098 infants in Texas born at 32 to 35 weeks of gestation in 1997 who had continuous Medicaid coverage and had not received RSV prophylaxis in their first year of life. Measurements: Risk factors examined included age relative to the start of the RSV season (ie, born within 6 months of RSV season, born July–November 1997), low birth weight, presence of siblings, unmarried mother, male gender, birth stay of 7 days or more, teenaged mother, NICU stay at birth, maternal smoking during pregnancy, and ventilator assistance at birth. Results: Only 4 of the 11 risk factors were significantly associated with RSV hospitalization—age at the start of the RSV season, the presence of siblings, birth stay of 7 days or more, and NICU stay at birth. Birth from July through November was the most accurate predictor of RSV hospitalization, with a sensitivity of 0.67 and a specificity of 0.62. The addition of other risk factors to age at the start of RSV season increased specificity but decreased sensitivity. Conclusion: Multifactor prediction tools for RSV-related hospitalizations estimated from administrative data may not be valid for targeting moderately premature infants in a low-income population for receipt of RSV prophylaxis. Several promising risk factors (eg, child care attendance) remain to be investigated for this patient population.