Abstract 1334 Poster Session I, Saturday, 5/1 (poster 62)

Initial severity of pulmonary disease, defined as an oxygenation index (OI) greater than 10, has been reported predictive of the development of bronchopulmonary dysplasia (BPD). However, some newborn infants with apparent mild to moderate lung disease also develop BPD. Thus, predicting BPD on the first day of life may be difficult.

Methods: We retrospectively analyzed the data from all mechanically ventilated premature infants at our institution born between January, 1996 and June, 1997. Oxygenation indices (FiO2 × mean airway pressure [Paw] × 100 / PaO2) were calculated based on the first arterial blood gas values and simultaneous ventilator parameters. We examined the predictive power of OI for subsequent development of BPD (supplemental oxygen requirement with an abnormal chest radiograph after 36 weeks post-conceptual age) by estimating sensitivity, specificity, positive and negative predictive values using an OI threshold of ≥10. We also estimated the proportion of infants with OI < 4 (e.g. FiO2 0.4; Paw 6cmH2O; PaO2 60torr) who later developed BPD. In addition, we estimated a logistic regression model with the corresponding area under the receiver operating characteristic curve (ROC) as a measure of the predictive power of OI.

Results: Data were available on 107 premature newborn infants. Twenty-five infants died (BW: 727 ± 40g [mean±sem]; GA: 25.0 ± 0.4wks; OI:11.6 ± 1.7). Surviving infants were divided into two groups: non-BPD (n=42; BW:1122 ± 26g; GA:28.6 ± 0.3wks) and BPD (n=40; BW: 902 ± 28g; GA: 26.9 ± 0.3wks). As expected, non-BPD infants had a significantly lower OI than BPD infants (6.5 ± 0.8 vs 10.4 ± 1.4, p=0.02). An OI ≥10 was associated with a near 3-fold increase in the risk of BPD (odds ratio 2.83 with a 95% C.I. 1.1-7.5). Using an OI ≥10 as the criterion for a positive test, the sensitivity of predicting BPD was 40%; specificity 81%; positive predictive value 67%; negative predictive value 59%. In particular, 60% (24/40) of all cases of BPD occurred in infants with OI <10 (i.e., false negative rate = 60% with a 95% C.I. 43 - 75%). The area under the ROC curve for the logistic model of OI ≥10 had a modest value of only 61%. Furthermore, 30% (9/30) of infants with an OI <4 subsequently developed BPD (95% C.I. 15-49%).

Conclusions: Newborn infants with early severe lung disease are more likely to develop BPD. However, whereas an OI ≥10 predicts a 3-fold higher likelihood of developing BPD, even an OI <4 is associated with a significant risk of BPD. Treatments designed to prevent the development of BPD should consider including newborn infants who initially appear to have mild to moderate lung disease.