PREDICTION OF CHRONIC NEONATAL LUNG-DISEASE ON DAY-4 OF LIFE

Clinical parameters, available on day 4, were collected for 204 ventil ated neonates < 32 weeks gestation. Logistic regression was used to id entify factors significantly and independently associated with chronic neonatal lung disease (CNLD) at 36 weeks postconception, which develo ped in 29% of neonates. At 36 weeks birth weight, peak inspiratory ven tilator pressure and requirement for assisted ventilation on day 4 wer e such factors. The logistic regression equation for this association was applied to each infant resulting in a value between 0 and 1. By kn owing which neonates developed CNLD, the discriminatory ability of thi s value was assessed. A receiver-operator characteristic curve for thi s value had an area under the curve of 0.97 (SE 0.03) in an unrelated population. A logistic equation value > 0.4 had a sensitivity of 90% a nd a specificity of 88% in predicting CNLD at 36 weeks. Conclusion Use of logistic regression to identify factors independently associated w ith chronic lung disease at 36 weeks postconception, allowed accurate prediction of this disorder. This would allow reduction in size of ran domised trials of early intervention in chronic lung disease.