Non-invasive respiratory support in preterm infants as primary mode: a network meta-analysis

Numerous innovations in non-invasive respiratory support have been introduced, resulting in a variety of available modes. Given the many options, understanding the relative effectiveness of these strategies is important.

To evaluate the benefits and harms of various non-invasive respiratory support modes when used as primary support in preterm infants.

We searched CENTRAL, MEDLINE, Embase, CINAHL, Web of Science, and trial registries (to 7 January 2024).

Randomised, quasi-randomised, and cluster-randomised controlled trials comparing two or more non-invasive respiratory support modes used as primary support for preterm infants within the first 24 hours.

Critical outcomes included treatment failure, endotracheal ventilation, and moderate-severe chronic lung disease (CLD). Important outcomes included any CLD, death, death or moderate-severe CLD, pulmonary air leak syndrome, intestinal perforation, and moderate-severe neurodevelopmental impairment.

We assessed risk of bias using the Cochrane RoB 1 tool.

For direct treatment comparisons, we conducted standard pairwise meta-analyses of each treatment pair of non-invasive respiratory support modes using a random-effects model. The seven eligible non-invasive respiratory support modes evaluated included nasal continuous positive airway pressure (CPAP), nasal intermittent positive pressure ventilation (NIPPV), biphasic positive airway pressure (BiPAP), high-flow nasal cannula (HFNC), non-invasive high-frequency oscillatory ventilation (NIHFV), non-invasive neurally adjusted ventilatory assist (NIV-NAVA), and high nasal continuous positive airway pressure (H-CPAP). For indirect and mixed treatment comparisons, we used a random-effects model with the Bayesian approach to estimate relative treatment effects to generate a network risk ratio (nRR) and 95% credible interval (95% CrI) for each outcome. We assessed the certainty of evidence using GRADE, specifically adapted for network meta-analyses.

We included 61 studies (7554 preterm neonates); 44 studies had a high risk of bias. No studies reported on primary use of non-invasive respiratory support with H-CPAP.

Treatment failure (47 studies, 6045 infants): NIHFV may decrease the risk of treatment failure compared to CPAP (nRR 0.41, 95% CrI 0.23 to 0.69; low-certainty evidence); compared to HFNC it may decrease the risk, but the evidence is very uncertain (nRR 0.35, 95% CrI 0.19 to 0.62; very low-certainty evidence). NIPPV may decrease the risk of treatment failure compared to CPAP (nRR 0.63, 95% CrI 0.48 to 0.82; very low-certainty evidence) and HFNC (nRR 0.54, 95% CrI 0.39 to 0.74; very low-certainty evidence), but the evidence for both is very uncertain. Endotracheal ventilation (44 studies, 5819 infants): NIHFV may decrease the risk of endotracheal ventilation compared to CPAP (nRR 0.48, 95% CrI 0.30 to 0.74; very low-certainty evidence) and HFNC (nRR 0.56, 95% CrI 0.33 to 0.91; very low-certainty evidence). NIPPV may decrease the risk of endotracheal ventilation compared to CPAP (nRR 0.62, 95% CrI 0.50 to 0.76; very low-certainty evidence) and HFNC (nRR 0.72, 95% CrI 0.55 to 0.94; very low-certainty evidence). The evidence for all these comparisons is very uncertain. Moderate-severe CLD (20 studies, 3026 infants): the pairwise comparisons with available data suggest that there is little to no effect based on choice of non-invasive respiratory support mode, but the evidence is generally very uncertain. Sensitivity analyses (restricted to studies with a low risk of bias): NIPPV may decrease the risk of treatment failure compared to HFNC (nRR 0.24, 95% CrI 0.09 to 0.69; very low-certainty evidence), and HFNC may increase the risk of treatment failure compared to CPAP (nRR 1.84, 95% CrI 1.02 to 2.90; very low-certainty evidence), based on very uncertain evidence. NIPPV may decrease the risk of endotracheal ventilation compared to CPAP (nRR 0.45, 95% CrI 0.23 to 0.84; low-certainty evidence); while it may decrease the risk compared to HFNC (nRR 0.44, 95% CrI 0.21 to 0.98; very low-certainty evidence), the latter is based on very uncertain evidence. The available pairwise comparisons on moderate-severe CLD suggest that the choice of non-invasive respiratory support mode may have little to no effect on outcome, but the evidence is very uncertain. Stratified analyses: findings of analyses in preterm infants 28 weeks' gestational age or greater were consistent with the main analyses. NIPPV (nRR 0.70, 95% CrI 0.49 to 0.99; very low-certainty evidence) and NIHFV (nRR 0.42, 95% CrI 0.18 to 0.96; very low-certainty evidence) may reduce the risk of treatment failure compared to CPAP based on very uncertain evidence. Similarly, NIPPV (nRR 0.66, 95% CrI 0.5 to 0.86; very low-certainty evidence) and NIHFV (nRR 0.43, 95% CrI 0.22 to 0.80; very low-certainty evidence) may reduce the risk of endotracheal ventilation compared to CPAP, both based on very uncertain evidence. The available pairwise comparisons on moderate-severe CLD suggest that the choice of non-invasive respiratory support mode may have little to no effect on outcome, but the evidence is generally very uncertain. Among preterm infants less than 28 weeks' gestational age, only one trial (75 infants) provided data on treatment failure and endotracheal ventilation, whereas no studies provided data on moderate-severe CLD. There were no differences between pairwise comparisons for either of these outcomes. Additional notes: the most common reasons for downgrading the certainty of evidence were due to within-study bias, imprecision, and incoherence. Additionally, most studies did not compare the non-invasive respiratory support modes at equivalent mean airway pressures. Results of important outcomes are described in the main text of this review.

NIPPV and NIHFV may reduce the risk of treatment failure or endotracheal ventilation compared to CPAP or HFNC, but may not reduce the risk of moderate-severe CLD. However, the certainty of evidence is low to very low, precluding firm conclusions and recommendations. More data are needed for infants less than 28 weeks' gestational age, as they are currently under-represented in studies. Future research on non-invasive respiratory support modes should be conducted with equivalent mean airway pressures between different modes to demonstrate to what extent benefits are related to the unique gas flow characteristics of each mode.

No funding specific to this review.

Protocol available via DOI: 10.1002/14651858.CD014895.