Effects of Interventions and Quality of the Evidence
Figure B show forest plots of the results of the meta-analysis of RCTs. The GRADE evidence profile table summarizes the assessment of the certainty of evidence for each outcome (Table 2). Mortality at discharge (or at a postmenstrual age of 36 weeks) was not significantly different between the two respiratory strategies (65/345 vs. 52/348; RR 1.26 [95% CI 0.91–1.75]; ARD 39 more per 1000 [95% CI 13 fewer to 112 more]; low certainty of evidence). The rates of BPD were not significantly different between the two respiratory strategies (94/342 vs. 101/345, RR 0.94 [95% CI 0.74–1.18]; ARD 11 more per 1000 [95% CI 50 fewer to 86 more per 1000]; very low certainty of evidence). Notably, the largest RCT (Thome 2015) reported a significant increase in the rate of NEC with the use of permissive hypercapnia. In our systematic review, the difference in NEC rates between the groups was not significant; however, the estimated risk of NEC was higher in the permissive hypercapnia group than in the normocapnia group, with a lower boundary of 95% CI close to and just below 1.0 (32/345 vs. 19/348, RR 1.69 [95% CI 0.98–2.91]; ARD 38 more per 1000 [95% CI 1 fewer to 104 more per 1000, very low certainty of evidence]). An observational study (Hogan, 2008) reported no significant differences in IVH rates before NICU discharge and early childhood behavior or function scores at 2–3 years of age between the permissive hypercapnia and normocapnia groups.