BACKGROUND: Although currently available surgical scoring systems have good outcome predictive power, their use is often limited by complexity and their non-dynamic nature. The aim of this study was to develop and test a risk adjustment for general surgical audit which is both simple and dynamic, while preserving a high predictive power for surgical morbidity. METHODS: Twelve easily measured, well defined prognostic variables for morbidity were identified from the Otago Surgical Audit data collection form and stratified into suitable categories. Logistic regression was used to adjust for confounding between factors, identifying risk factors with the strongest prognostic value for the outcome of severe and intermediate complications. The resulting model was tested by back-validation and validation. RESULTS: The derived risk adjustment included all 12 variables. Adjusted odds ratios for all variables were markedly lower than unadjusted values. After logistic regression, the strongest predictors of postoperative morbidity were duration of operation, operation category, inpatient status and organ system in which the procedure was carried out. The area under the receiver operating characteristic curve was 0.86. CONCLUSION: A simple dynamic model for surgical morbidity has been developed which is comparable to previously published surgical scoring systems in terms of predictive power. This risk adjustment tool can be incorporated into the existing audit system, enabling comparison of surgical unit performance.
BACKGROUND: Although currently available surgical scoring systems have good outcome predictive power, their use is often limited by complexity and their non-dynamic nature. The aim of this study was to develop and test a risk adjustment for general surgical audit which is both simple and dynamic, while preserving a high predictive power for surgical morbidity. METHODS: Twelve easily measured, well defined prognostic variables for morbidity were identified from the Otago Surgical Audit data collection form and stratified into suitable categories. Logistic regression was used to adjust for confounding between factors, identifying risk factors with the strongest prognostic value for the outcome of severe and intermediate complications. The resulting model was tested by back-validation and validation. RESULTS: The derived risk adjustment included all 12 variables. Adjusted odds ratios for all variables were markedly lower than unadjusted values. After logistic regression, the strongest predictors of postoperative morbidity were duration of operation, operation category, inpatient status and organ system in which the procedure was carried out. The area under the receiver operating characteristic curve was 0.86. CONCLUSION: A simple dynamic model for surgical morbidity has been developed which is comparable to previously published surgical scoring systems in terms of predictive power. This risk adjustment tool can be incorporated into the existing audit system, enabling comparison of surgical unit performance.
Authors: Vivian E Strong; Luke V Selby; Mindy Sovel; Joseph J Disa; William Hoskins; Ronald Dematteo; Peter Scardino; David P Jaques Journal: Ann Surg Oncol Date: 2014-10-16 Impact factor: 5.344
Authors: John C Woodfield; Ross A Pettigrew; Lindsay D Plank; Michael Landmann; Andre M van Rij Journal: World J Surg Date: 2007-10 Impact factor: 3.282
Authors: James Studnicki; Christopher Craver; Christopher M Blanchette; John W Fisher; Sara Shahbazi Journal: BMC Surg Date: 2014-08-16 Impact factor: 2.102
Authors: Michala Skovlund Sørensen; Klaus Hindsø; Thea Bechmann Hovgaard; Michael Mørk Petersen Journal: Medicine (Baltimore) Date: 2016-04 Impact factor: 1.889
Authors: S M M de Castro; J T Houwert; S M Lagarde; J B Reitsma; O R C Busch; T M van Gulik; H Obertop; D J Gouma Journal: World J Surg Date: 2009-07 Impact factor: 3.352