Predictors of outcome for patients with severe respiratory failure requiring non invasive mechanical ventilation.

Objective: Failure to identify patients in whom non-invasive ventilation (NIV) would be unsuccessful may cause inappropriate delay in intubation. Aim of this study was to determine predictors of NIV failure.

Methods: We retrospectively evaluated COPD patients, undergoing NIV for at least 12h because of hypercapnic acute respiratory failure. Univariable and multivariable analyses were performed on: age, gender, APACHE II and GCS, gas exchange at admission, during NIV and at discharge/death, length of stay. ROC analysis for pH START and APACHE II were performed.

Results: Among 201 individuals enrolled, NIV failed in 50. In the logistic regression model only APACHE II was found to have an independent effect on the outcome (p < 0.0001, OR 1.179, 95% CI 1.101-1.263 as quantitative variable; p < 0.0001, OR 3.753, 95% CI 1.798-7.835, as qualitative variable, > 20.5).

Conclusions: APACHE II may be a crucial parameter in predicting NIV failure; further multicentric studies are needed to better define NIV indications.