Predictors of survival in infants with congenital diaphragmatic hernia - systemic oxygenation status versus dynamic compliance of the respiratory system

Abstract

Objective. To compare whether early measurement of blood gases and/or dynamic compliance of the respiratory system (CRS_dyn) predicts outcome in high-risk infants with unilateral congenital diaphragmatic hernia (CDH).
Patients and methods. A retrospective study was performed at Tygerberg Children’s Hospital between January 1992 and August 2001. High-risk infants with unilateral CDH, who presented with respiratory distress within 6 hours of birth, were included. Patients with other lethal congenital abnormalities were excluded. The first arterial blood gas value after endotracheal intubation was documented and the arterial-alveolar oxygen tension (a:A) ratio was calculated. CRS_dyn was measured within 24 hours of birth. The ability of these measurements to predict outcome (survival or death during the newborn period) was determined.
Results. Seventeen of 40 infants with CDH were categorised as high risk and included in the study. Eight of them (47%) survived the neonatal period. The best single predictors of outcome were, in order, partial pressure of oxygen in arterial blood (PaO₂), a:A ratio and dynamic compliance of the respiratory system standardised for body weight (CRS_dyn/kg). The specificity and sensitivity at a PaO₂ cut-off of 19.3 kPa were 7/8 (95% confidence interval (CI): 0.473 - 0.997) and 9/9 (95% CI: 0.634 - 1.000) respectively. Results for a:A ratio were cut-off 0.321, specificity 6/8 (95% CI: 0.349 - 0.968), and sensitivity 9/9 (95% CI: 0.634 - 1.000). Results for CRS_dyn/kg were cut-off 0.259, specificity 6/8 (95% CI: 0.349 - 0.968), and sensitivity 9/9 (95% CI: 0.634 - 1.000). A linear discriminant function based on the 3 best single predictors was found to be no more effective than the first PaO₂.
Conclusions. Early oxygenation status predicts outcome better than the CRS_dyn/kg in infants with unilateral CDH. However, both measurements predict outcome with high accuracy.