Improving Identification of Postoperative Respiratory Failure Missed by the Patient Safety Indicator Algorithm

Abstract

The Patient Safety Indicator postoperative respiratory failure (PRF) flags cases using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes for acute respiratory failure or mechanical ventilation/intubation. The authors examined how frequently PRF missed events and ways to improve event identification. A total of 125 high-risk unflagged cases were selected based on predicted probability and presence of clinically relevant codes. False-negative (FN) proportion and associated reasons were determined through chart review, and likelihood ratios (LRs) of associated codes were calculated. In all, 27% of elective cases were FNs; 55% of FNs lacked ventilation/intubation codes. “Respiratory arrest,” 799.1, had the highest LR (5.4) but occurred in only 8% of FNs. All other individual diagnostic or procedure codes had relatively low LRs (≤3.1), despite occurring in up to 36% of FNs. Among high-risk cases, the PRF algorithm often missed events. Improved ventilation/intubation coding will have more of an impact on case identification than adding ICD-9-CM codes to the algorithm.