|| Checking for direct PDF access through Ovid
Despite increasing use, evidence is mixed as to the appropriate use of noninvasive ventilation in patients with pneumonia. We aimed to determine the relationship between receipt of noninvasive ventilation and outcomes for patients with pneumonia in a real-world setting.We performed a retrospective cohort study of Medicare beneficiaries (aged > 64 yr) admitted to 2,757 acute-care hospitals in the United States with pneumonia, who received mechanical ventilation from 2010 to 2011.Noninvasive ventilation versus invasive mechanical ventilation.The primary outcome was 30-day mortality with Medicare reimbursement as a secondary outcome. To account for unmeasured confounding associated with noninvasive ventilation use, an instrumental variable was used—the differential distance to a high noninvasive ventilation use hospital. All models were adjusted for patient and hospital characteristics to account for measured differences between groups. Among 65,747 Medicare beneficiaries with pneumonia who required mechanical ventilation, 12,480 (19%) received noninvasive ventilation. Patients receiving noninvasive ventilation were more likely to be older, male, white, rural-dwelling, have fewer comorbidities, and were less likely to be acutely ill as measured by organ failures. Results of the instrumental variable analysis suggested that, among marginal patients, receipt of noninvasive ventilation was not significantly associated with differences in 30-day mortality when compared with invasive mechanical ventilation (54% vs 55%; p = 0.92; 95% CI of absolute difference, –13.8 to 12.4) but was associated with significantly lower Medicare spending ($18,433 vs $27,051; p = 0.02).Among Medicare beneficiaries hospitalized with pneumonia who received mechanical ventilation, noninvasive ventilation use was not associated with a real-world mortality benefit. Given the wide CIs, however, substantial harm associated with noninvasive ventilation could not be excluded. The use of noninvasive ventilation for patients with pneumonia should be cautioned, but targeted enrollment of marginal patients with pneumonia could enrich future randomized trials.