Level 1: Likely reliable evidence
Preoperative inspiratory muscle training prevented pneumonia in a randomized trial of 279 patients undergoing coronary artery bypass graft (CABG) surgery who were at high risk for pulmonary complications (JAMA. 2006;296:1851-1857). High risk was defined as forced expiratory volume in one second (FEV1) <80% predicted and ratio of FEV1 to forced vital capacity (FVC) <70% predicted, or any two of the following: age older than 70 years, cough and expectoration, diabetes mellitus, smoker, FEV1 27.
Patients were randomized to preoperative inspiratory muscle training or usual care. Inspiratory muscle training was performed for 20 minutes daily for at least two weeks before surgery. It included incentive spirometry with gradually increased resistance, and forced expiration and was supervised by a physical therapist once weekly. Mean time awaiting surgery was 8.2 weeks. Both groups had similar postoperative care with incentive spirometry, chest physical therapy, and mobilization. The interim analysis included all but three patients who died before surgery.
Continue Reading
Pneumonia was defined by CDC criteria, i.e., onset of pneumonia >72 hours after hospital admission; physical exam showing rales or dullness to percussion, or infiltrate on chest x-ray; and at least one of the following: purulent sputum, isolation of pathogen, diagnostic antibody titers, or histopathologic evidence of pneumonia (Am Rev Respir Dis. 1989;139:1058-1059).
Comparing preoperative inspiratory muscle training vs. usual care, outcomes that favored the intervention included rate of postoperative pulmonary complications (18% vs. 35%, P =.02, NNT 6), rate of selected complications (thoracentesis for pleural effusion, radiologic evidence of pneumonia, pneumothorax, reintubation) (7.2% vs. 17.5%, P = .01, NNT 10), rate of pneumonia (6.5% vs. 16.1%, P = .01, NNT 11), and median duration of postoperative hospitalization (7 vs. 8 days, P = .02). Ventilator dependence for >48 hours was 0.7% vs. 4.4% and showed a trend toward statistical significance (P = .09).
