Noninvasive ventilation on reintubation in patients with obesity and hypoxemic respiratory failure following abdominal surgery: a post hoc analysis of a randomized clinical trial.

Abstract

Although noninvasive ventilation (NIV) may reduce reintubation in patients with acute hypoxemic respiratory failure following abdominal surgery, this strategy has not been specifically assessed in patients with obesity. We conducted a post hoc analysis of a multicenter randomized controlled trial comparing NIV delivered via facial mask to standard oxygen therapy among patients with obesity and acute hypoxemic respiratory failure within 7 days after abdominal surgery. The primary outcome was reintubation within 7 days. Secondary outcomes were invasive ventilation-free days at day 30, intensive care unit (ICU)-acquired pneumonia and 30-day survival. Among 293 patients with hypoxemic respiratory failure following abdominal surgery, 76 (26%) patients had obesity and were included in the intention-to-treat analysis. Reintubation rate was significantly lower with NIV (13/42, 31%) than with standard oxygen therapy (19/34, 56%) within 7 days (absolute difference: - 25%, 95% confidence interval (CI) - 49 to - 1%, p = 0.03). NIV was associated with significantly more invasive ventilation-free days compared with standard oxygen therapy (27.1 ± 8.6 vs 22.7 ± 11.1days; p = 0.02), while fewer patients developed ICU-acquired pneumonia (1/42, 2% vs 6/34, 18%; p = 0.04). The 30-day survival was 98% in the NIV group (41/42) versus 85% in the standard oxygen therapy (p = 0.08). In patients with body mass index (BMI) < 30kg/m2, no significant difference was observed between NIV (36/105, 34%) and standard oxygen therapy (47/109, 43%, p = 0.03). An interaction test showed no statistically significant difference between the two subsets (BMI ≥ 30kg/m2 and BMI < 30kg/m2). Among patients with obesity and hypoxemic respiratory failure following abdominal surgery, use of NIV compared with standard oxygen therapy reduced the risk of reintubation within 7 days, contrary to patients without obesity. However, no interaction was found according to the presence of obesity or not, suggesting either a lack of power to conclude in the non-obese subgroup despite existing differences, or that the statistical difference found in the overall sample was driven by a large effect in the obese subsets.