Evaluation of Structured Complication Scores in Operative and Non-Operative Emergency General Surgery.

Abstract

Introduction:

Emergency General Surgery (EGS) is a field growing in volume and specialty recognition. EGS Quality assessment is lagging due to a heterogenous patient group including ~70% nonoperative admissions, and a lack of validated quality assessment frameworks. Central to such assessments are working definitions and classifications of patient morbidity-none have been validated in this field. This study aims to evaluate the performance of the Clavien-Dindo (CD) classification of complications and related scores in EGS.

Methods:

A preliminary review of ten-years of EGS patients at our institution found poor recording of complications to audit, leading us to construct a smaller detailed cohort and a complication dataset. Surgeon’s interpretations of the CD complication definition were surveyed. Correlation and multivariate regression were used to analyse the relationship between the Clavien-Dindo based complication scores, length of stay and cost outcome; guided by our surgeons survey and Delphi process.

Results:

Our institution managed a >2-fold increase in the EGS workload over ten years. Complications were underreported-only 4.3% of all complications were entered. The survey revealed support for pre-op and non-operative complications despite the Clavien-Dindo definition excluding such events, the definition was inconsistently interpreted. The complication scores correlated to the LOS/Cost outcomes. The regression analyses revealed the complication scores as the predominant independent predictor of LOS and Cost variability in surviving patients.

Conclusion:

Four Key findings emerged. Emergency general surgery is expanding in our institution. The Clavien-Dindo complication definition lacks specificity and is inconsistently interpreted by surgeons. Complication scores performed well in correlative and regression analysis, emerging as the primary predictors of the LOS and Cost in surviving patients. Application choice of scores should be determined by use case.