Objective: The aim of this study was to test the following hypotheses: the application of early CPAP in the treatment of RDS decreases both the intubation rate and the incidence of bronchopulmonary dysplasia.
Design: This retrospective study compares the ventilatory management of premature infants between two periods of 12 months. The first period (July 1996-June 1997) precedes the application of early CPAP, the second (July 1998-June 1999) follows the systematic introduction of early CPAP.
Population: All inborn infants, < 34 weeks gestational age and admitted in the Division of Neonatology, CHUV, Lausanne, are included.
Results: The intubation rate was reduced significantly from 63.6% in the first period to 23.8% in the second and, if we only consider infants < or = 1500 g, from 72.1 to 30.8% (p < 0.01). The incidence of bronchopulmonary dysplasia, defined as the persistence of oxygen needs at 28 days of life, decreases from 10% in the first period to 4.8% in the second period and, if we only consider infants < or = 1500 g, from 14.7 to 7.7%. These differences are not significant. However we noticed a significant decrease in the duration of O2 therapy from 8.8 days in the first period to 4.6 days in the second (p = 0.04). If we only consider infants < or = 1500 g, the O2 therapy lasted 12.1 days for the first period and 6.8 days for the second. This decrease is not significant. Mortality is identical for both periods: survival rate is 90% in the first period and 90.5% in the second, and 85% and 86.2% respectively if we only consider infants < or = 1500 g. We noticed no significant differences in the morbidity rate (intraventricular haemorrhage, leucomalacia, retinopathy of prematurity and patent ductus arteriosus), except that for infants < or = 1500 g the incidence of pneumothorax was higher in the second period.
Conclusion: The application of early CPAP significantly reduces the intubation rate without increasing mortality. A decrease in the incidence of bronchopulmonary dysplasia was observed but this finding was not statistically significant.