Predictors of Noninvasive Ventilation Failure in the Post-Extubation Period: A Systematic Review and Meta-Analysis*

Objectives: To identify factors associated with failure of noninvasive ventilation (NIV) in the post-extubation period. Data Sources: We searched Embase Classic +, MEDLINE, and the Cochrane Database of Systematic Reviews from inception to February 28, 2022. Study Selection: We included English language studies that provided predictors of post-extubation NIV failure necessitating reintubation. Data Extraction: Two authors conducted data abstraction and risk-of-bias assessments independently. We used a random-effects model to pool binary and continuous data and summarized estimates of effect using odds ratios (ORs) mean difference (MD), respectively. We used the Quality in Prognosis Studies tool to assess risk of bias and the Grading of Recommendations, Assessment, Development and Evaluations to assess certainty. Data Synthesis: We included 25 studies ( n = 2,327). Illness-related factors associated with increased odds of post-extubation NIV failure were higher critical illness severity (OR, 3.56; 95% CI, 1.96–6.45; high certainty) and a diagnosis of pneumonia (OR, 6.16; 95% CI, 2.59–14.66; moderate certainty). Clinical and biochemical factors associated with moderate certainty of increased risk of NIV failure post-extubation include higher respiratory rate (MD, 1.54; 95% CI, 0.61–2.47), higher heart rate (MD, 4.46; 95% CI, 1.67–7.25), lower Pa o 2 :F io 2 (MD, –30.78; 95% CI, –50.02 to –11.54) 1-hour after NIV initiation, and higher rapid shallow breathing index (MD, 15.21; 95% CI, 12.04–18.38) prior to NIV start. Elevated body mass index was the only patient-related factor that may be associated with a protective effect (OR, 0.21; 95% CI, 0.09–0.52; moderate certainty) on post-extubation NIV failure. Conclusions: We identified several prognostic factors before and 1 hour after NIV initiation associated with increased risk of NIV failure in the post-extubation period. Well-designed prospective studies are required to confirm the prognostic importance of these factors to help further guide clinical decision-making.