Relative likelihood of bronchopulmonary dysplasia based on pulmonary mechanics measured in preterm neonates during the first week of life.

We studied 143 low birth weight infants (less than or equal to 1500 gm) with respiratory distress syndrome who required mechanical ventilation, and determined the efficacy of using the alterations in pulmonary mechanics (measured at 1 to 3 days, 4 to 7 days, 2 weeks, and 4 weeks) as possible predictors for the subsequent diagnosis of bronchopulmonary dysplasia (BPD). The sensitivity and specificity of pulmonary compliance and resistance as predictors of BPD were ascertained by logistic regression correlation (p less than 0.01) and receiver operating characteristic curves. With these pulmonary mechanics data and logistic discriminant analysis techniques, we developed BPD prediction models based on pulmonary mechanics, measured between 4 and 7 days of age, to define the likelihood ratio for the subsequent diagnosis of BPD. Eight different BPD prediction models were developed by using combinations of four vectors (pulmonary compliance, total pulmonary resistance, birth weight, gestational age), and each model was validated in a subsequent low birth weight study population (n = 53). All models were deemed accurate for negative prediction of BPD. The models dependent on gestational age and dynamic pulmonary compliance had the highest positive predictive accuracy. The predictive impact of total pulmonary resistance appeared to be minimal. These prediction models may be used to calculate the likelihood ratio for a subsequent BPD diagnosis and thereby objectively categorize both the risk and the magnitude of acute lung injury by the first week of life.