Logistic Regression Analysis of Risk Factors for Invasive Candidiasis Among Infants in 12 NICUs, 1998–2001

VariableOdds Ratio, Point Estimate95% Confidence Interval
Center (same as Table 1)
Noncenter variables with P < .15
    Exposure to BSA2.1571.422–3.271
    Gestational age0.7520.672–0.841
    Maternal hypertension0.5320.323–0.877
    Gravida status1.0730.999–1.153
    ROM >24 h1.5301.077–2.174
    Age at first enteral feed1.0321.015–1.049
     ≥2 doses surfactant1.4381.007–2.052
    PDA no surgery vs no PDA1.9351.340–2.793
    PDA surgery vs no PDA1.5861.003–2.507
    SGA status1.6660.925–2.999
  • Variables in multivariable logistic regression model for the primary outcome, candidiasis. Noncenter variables tested in the regression that were removed by backward elimination using a filter of P < .15 were: prenatal steroids, maternal marital status, maternal antibiotics, tocolytics, duration of rupture of membranes (ROM), cesarean section delivery, race (4 levels: black, white, Hispanic, other), Apgar scores at 1 and 5 minutes, duration of umbilical artery catheter, any surfactant versus no surfactant, indomethacin prophylaxis (any indomethacin doses in the first 24 hours of life), grades III or IV intraventricular hemorrhage, changes from birth weight (birth weight regained, birth weight not lost, and birth weight lost and not regained), initial antibiotic course >5 days, or initial antibiotic course <4 days. Both antibiotic variables required a negative postnatal day-1 blood culture. Max-rescaled R2 was 0.2073 for the final model. PDA indicates patent ductus arteriosus; SGA, small for gestational age.

  • a The reference center had lowest incidence of candidiasis among postnatal day-3 survivors.