DOI: 10.1097/01.ccm.0000424680.82477.15
,
Issn Print: 0090-3493
Publication Date: 2012/12/01
Facebook
Twitter
LinkedIn
Email
 |
Checking for direct PDF access through Ovid | |
Abstract
Introduction:To identify clinical and physiologic predictors of intubation among subjects with respiratory failure treated with non-invasive ventilation (NIV).
Hypothesis:Predictors of intubation on patients treated with NIV vary according to the cause of respiratory failure.
Methods:This is a retrospective study of patients admitted with acute respiratory failure and treated with NIV over a period of 6 months. Patients were classified according to the cause of acute respiratory failure in: a) pneumonia (PNA); b) COPD; c) cardiogenic pulmonary edema (CPE); d) post-operative, and e) PNA plus COPD. Clinical and physiologic parameters, such as vital signs, SOFA score, and blood gas analysis pre and post-NIV were collected. Receiver Operative Curves (ROCs) were constructed for each parameter to assess predictability of NIV failure. ROCs with areas under the curve (AUC) higher than 0.75 were considered.
Results:Out of 91 patients treated with NIV, only 60 (66%) patients presented an identified diagnosis and included in the study. Thirteen (22%) subjects had PNA, 17 (28%) had COPD, 11 (18%) had CPE, 10 (17%) were post-operative, and 9 (15%) had both COPD plus PNA. Intubation rate was 61%, 23%, 9%, 20%, and 67% in each group, respectively. The best predictors for NIV failure and intubation in the PNA group were admission pCO2 (AUC of 0.81) and SOFA (AUC of 0.96). Within the COPD group, admission pCO2 (AUC of 0.75), respiratory rate (RR) variability (AUC of 0.81), and SOFA (AUC of 0.83) presented the best predicting ability. Heart rate (HR) variability (AUC of 0.80) and SOFA (AUC of 0.80) were the best predictors in patients with CPE. Among subjects post-operative, admission HR (AUC of 0.97) and SOFA (AUC 0.88) performed better than other parameters. In patients with COPD and PNA, HR variability (AUC of 0.94), RR variability (AUC of 0.94), and SOFA (AUC of 0.94) were the best predictors of NIV failure.
Conclusions:Predictors of NIV failure vary according to the etiology of acute respiratory failure. SOFA predicts NIV failure across a group of subjects with a variety of diagnoses. Nevertheless, other parameters may need to be utilized depending of the cause of respiratory failure.
