• Mark E Cohen, Yaoming Liu, Jason B Liu, Clifford Y Ko, and Bruce L Hall.
• Division of Research and Optimal Patient Care, American College of Surgeons, Chicago, IL. Electronic address: markcohen@facs.org.
• J. Am. Coll. Surg. 2018 Mar 1; 226 (3): 309-316.

BackgroundAmerican College of Surgeons NSQIP risk-adjustment models rely on the designated "principal" CPT code to account for procedure-related risk. However, if hospitals differ in their propensity to undertake multiple major operations under the same anesthetic, then risk adjustment using only a single code could bias hospital quality estimates. This study investigated this possibility for bias.Study DesignWe examined hospital odds ratios (ORs) when either the principal CPT code was used for risk adjustment (along with other standard NSQIP predictor variables) or when this code was used in addition to the remaining reported CPT code with the highest associated risk. We examined models for general surgery mortality and morbidity and morbidity in datasets that included mastectomy and/or breast reconstruction, or hysterectomy and/or gynecologic reconstruction as areas known to likely involve more than 1 procedure.ResultsHospital ORs based on 1 vs 2 CPT codes for risk adjustment were essentially the same for mortality and morbidity general surgery models and for the mastectomy and/or breast reconstruction model. For hysterectomy and/or gynecologic reconstruction, the 1 CPT code model tended to slightly overestimate ORs compared with the 2 CPT codes model, when the hospital's OR and the proportion of combined operations were large.ConclusionsConditions under which practice-pattern-associated modeling bias might exist appear to be uncommon and to have a small impact on quality assessments for the areas examined. The evidence suggests that, within the American College of Surgeons NSQIP modeling paradigm, the principal CPT code adequately risk adjusts for operative procedure in performance assessments.Copyright © 2018 American College of Surgeons. Published by Elsevier Inc. All rights reserved.

26 keywords...

hide…