OBES SURG DOI 10.1007/s11695-015-1699-2
ORIGINAL CONTRIBUTIONS
Risk Stratification Models: How Well do They Predict Adverse Outcomes in a Large Dutch Bariatric Cohort? Noëlle Geubbels 1 & L. Maurits de Brauw 1 & Yair I. Z. Acherman 1 & Arnold W. J. M. van de Laar 1 & Sjoerd C. Bruin 1
# Springer Science+Business Media New York 2015
Abstract Background Risk prediction models are useful tools for informing patients undergoing bariatric surgery about their risk for complications and correcting outcome reports. The aim of this study is to externally validate risk models assessing complications after laparoscopic Roux-en-Y gastric bypass (LRYGB) surgery. Methods All 740 patients who underwent a primary LRYGB between December 2007 and July 2012 were included in the validation cohort. PubMed was systematically searched for risk prediction models. Eight risk models were selected for validation. We classified our complications according to the Clavien-Dindo classification. Predefined criteria of a good model were a non-significant Hosmer and Lemeshow test, Nagelkerke R2 ≥0.10, and c-statistic ≥0.7. Results There were 85 (7.8 %) grade 1, 54 (7.3 %) grade 2, 5 (0.7 %) grade 3a, 14 (1.9 %) grade 3b, and 14 (1.9 %) grade 4a complications in our validation cohort. Only one model predicted adverse events satisfactorily. This model consisted of one patient-related factor (age) and four surgeon- or center related factors (conversion to open surgery, intraoperative
* Noëlle Geubbels [email protected] L. Maurits de Brauw [email protected] Yair I. Z. Acherman [email protected] Arnold W. J. M. van de Laar [email protected] Sjoerd C. Bruin [email protected] 1
Department of Metabolic and Bariatric Surgery, Slotervaart Hospital, Louwesweg 6, 1066 EC Amsterdam, The Netherlands
events, the need for additional procedures during LRYGB and the learning curve of the center). Conclusions The overall majority of the included risk models are unsuitable for risk prediction. Only one model with an emphasis on surgeon- and center-related factors instead of patient-related factors predicted adverse outcome correctly in our external validation cohort. These findings support the establishment of specialty centers and warn benchmark data institutions not to correct bariatric outcome data by any other patient-related factor than age. Keywords Morbid obesity . Bariatric surgery . Gastric bypass . Risk stratification . Risk factors . Complications . Adverse events . Validation . Benchmark data . Outcome reporting
Introduction Mortality and morbidity resulting from laparoscopic Roux-enY gastric bypass (LRYGB) surgery have decreased over the past decades. Early mortality came down from about 0.3 % measured over 1990–2006 [1] to
ORIGINAL CONTRIBUTIONS
Risk Stratification Models: How Well do They Predict Adverse Outcomes in a Large Dutch Bariatric Cohort? Noëlle Geubbels 1 & L. Maurits de Brauw 1 & Yair I. Z. Acherman 1 & Arnold W. J. M. van de Laar 1 & Sjoerd C. Bruin 1
# Springer Science+Business Media New York 2015
Abstract Background Risk prediction models are useful tools for informing patients undergoing bariatric surgery about their risk for complications and correcting outcome reports. The aim of this study is to externally validate risk models assessing complications after laparoscopic Roux-en-Y gastric bypass (LRYGB) surgery. Methods All 740 patients who underwent a primary LRYGB between December 2007 and July 2012 were included in the validation cohort. PubMed was systematically searched for risk prediction models. Eight risk models were selected for validation. We classified our complications according to the Clavien-Dindo classification. Predefined criteria of a good model were a non-significant Hosmer and Lemeshow test, Nagelkerke R2 ≥0.10, and c-statistic ≥0.7. Results There were 85 (7.8 %) grade 1, 54 (7.3 %) grade 2, 5 (0.7 %) grade 3a, 14 (1.9 %) grade 3b, and 14 (1.9 %) grade 4a complications in our validation cohort. Only one model predicted adverse events satisfactorily. This model consisted of one patient-related factor (age) and four surgeon- or center related factors (conversion to open surgery, intraoperative
* Noëlle Geubbels [email protected] L. Maurits de Brauw [email protected] Yair I. Z. Acherman [email protected] Arnold W. J. M. van de Laar [email protected] Sjoerd C. Bruin [email protected] 1
Department of Metabolic and Bariatric Surgery, Slotervaart Hospital, Louwesweg 6, 1066 EC Amsterdam, The Netherlands
events, the need for additional procedures during LRYGB and the learning curve of the center). Conclusions The overall majority of the included risk models are unsuitable for risk prediction. Only one model with an emphasis on surgeon- and center-related factors instead of patient-related factors predicted adverse outcome correctly in our external validation cohort. These findings support the establishment of specialty centers and warn benchmark data institutions not to correct bariatric outcome data by any other patient-related factor than age. Keywords Morbid obesity . Bariatric surgery . Gastric bypass . Risk stratification . Risk factors . Complications . Adverse events . Validation . Benchmark data . Outcome reporting
Introduction Mortality and morbidity resulting from laparoscopic Roux-enY gastric bypass (LRYGB) surgery have decreased over the past decades. Early mortality came down from about 0.3 % measured over 1990–2006 [1] to
