Predicting pulmonary complications after nonthoracic surgery: a systematic review of blinded studies

Purpose To determine the performance of variables commonly used in the prediction of postoperative pulmonary complications in patients undergoing nonthoracic surgery. Methods We conducted a systematic review of the literature in English, using MEDLINE (1966–2001), manual searches of identified articles, and contact with content experts. All studies reporting independent and blinded comparisons of preoperative or operative factors with postoperative pulmonary complications were included. Two reviewers independently abstracted inclusion and exclusion criteria, study designs, patient characteristics, predictors of interest, and the nature and occurrence of postoperative pulmonary complications. Results Seven studies fulfilled the inclusion criteria. The definition of postoperative pulmonary complications differed among studies, and the incidence of postoperative pulmonary complications varied from 2% to 19%. Of the 28 preoperative or operative predictors that were evaluated in the 7 studies, 16 were associated significantly with postoperative pulmonary complications, although only 2 (duration of anesthesia and postoperative nasogastric tube placement) were significant in more than one study. The positive (2.2 to 5.1) and negative (0.2 to 0.8) likelihood ratios for these 16 variables suggest that they have only modest predictive value. Neither hypercarbia nor reduced spirometry values were independently associated with an increased risk of postoperative pulmonary complications. Conclusion Few studies have rigorously evaluated the performance of the preoperative or operative variables in the prediction of postoperative pulmonary complications. Prospective studies with independent and blinded comparisons of these variables with postoperative outcomes are needed.