Background Increases in PaCO2 lead to increases in cerebral blood flow (CBF) and may therefore predispose to intraventricular hemorrhage (IVH). Periventricular leukomalacia (PVL) correlates with hypocarbia in ventilated very low birth weight (VLBW) infants possibly due to decreases in CBF. The safe upper and lower limits of PaCO2 in VLBW infants need to be determined.
Hypothesis Higher levels of PaCO2 are associated with an increased risk of IVH.
Design/Methods Medical records review of 561 VLBW infants 401 to 1250 g at birth admitted to a Level III neonatal intensive care unit between 1/1/ 2000 and 12/31/ 2002. Data collected included main prenatal and neonatal variables, including severe IVH (grades III and IV) and blood gases in the first 4 days after birth. The highest, lowest, and time-weighted average PaCO2 were collected and included in uni- and multivariable analyses with severe IVH as the dependent variable.
Results Mean birth weight (± SD) was 840 g (± 209) and median gestational age was 26 weeks. The maximum PaCO2 (median [25th-75th centiles]) 72 [62-88] vs. 59 [50-70] mm Hg, p < .001) and the time-weighted average CO2 (mean 49 vs. 47 mm Hg, p = .003) were significantly higher, and the minimum PaCO2 was lower (32 [28-37] vs. 36 [30-42], p < .001) in patients with severe IVH. Both extremes of PaCO2 predicted severe IVH but the time-weighted average PaCO2 had lower accuracy (AUC of the ROC curve for maximum PaCO2 = 0.71; minimum PaCO2 = 0.66; time-weighted average PaCO2 = 0.58 [p < .05 vs. max and min PaCO2]). A maximum PaCO2 > 60 mm Hg had 79% sensitivity and 53% specificity, and a minimum PaCO2 < 38 mm Hg had 78% sensitivity and 45% specificity for severe IVH. Multiple logistic regression analyses identified maximum PaCO2, minimum PaCO2, time-weighted average PaCO2, gestational age, 5 minute Apgar score, and use of prenatal steroids as the major variables independently associated with severe IVH.
Conclusions Extreme levels of PaCO2 are associated with severe IVH in VLBW infants, but bias cannot be excluded. Although causality cannot be inferred from this analysis, a careful evaluation of the data from randomized trials on minimal ventilation on the relationship between PaCO2 and IVH is required to determine if a “threshold effect” does exist.