Noninvasive versus invasive ventilation for acute respiratory failure in patients with hematologic malignancies: a 5-year multicenter observational survey.

BACKGROUND: Mortality is high among patients with hematologic malignancies admitted to intensive care units for acute respiratory failure. Early noninvasive mechanical ventilation seems to improve outcomes.
OBJECTIVE: To characterize noninvasive mechanical ventilation use in Italian intensive care units for acute respiratory failure patients with hematologic malignancies and its impact on outcomes vs. invasive mechanical ventilation. DESIGN, SETTING, PARTICIPANTS: Retrospective analysis of observational data prospectively collected in 2002-2006 on 1,302 patients with hematologic malignancies admitted with acute respiratory failure to 158 Italian intensive care units. MEASUREMENTS: Mortality (intensive care unit and hospital) was assessed in patients treated initially with noninvasive mechanical ventilation vs. invasive mechanical ventilation and in those treated with invasive mechanical ventilation ab initio vs. after noninvasive mechanical ventilation failure. Findings were adjusted for propensity scores reflecting the probability of initial treatment with noninvasive mechanical ventilation.
RESULTS: Few patients (21%) initially received noninvasive mechanical ventilation; 46% of these later required invasive mechanical ventilation. Better outcomes were associated with successful noninvasive mechanical ventilation (vs. invasive mechanical ventilation ab initio and vs. invasive mechanical ventilation after noninvasive mechanical ventilation failure), particularly in patients with acute lung injury/adult respiratory distress syndrome (mortality: 42% vs. 69% and 77%, respectively). Delayed vs. immediate invasive mechanical ventilation was associated with slightly but not significantly higher hospital mortality (65% vs. 58%, p=.12). After propensity-score adjustment, noninvasive mechanical ventilation was associated with significantly lower mortality than invasive mechanical ventilation. LIMITATIONS: The population could not be stratified according to specific hematologic diagnoses. Furthermore, the study was observational, and treatment groups may have included unaccounted for differences in covariates although the risk of this bias was minimized with propensity score regression adjustment.
CONCLUSIONS: In patients with hematologic malignancies, acute respiratory failure should probably be managed initially with noninvasive mechanical ventilation. Further study is needed to determine whether immediate invasive mechanical ventilation might offer some benefits for those with acute lung injury/adult respiratory distress syndrome.