Retrospective validation and comparison of prediction calculators for weight loss and comorbidity resolution after bariatric surgery
Jacqueline Paolino MD, Dmitry Nepomnayshy MD
Background: There are a variety of prediction tools available to clinicians in bariatric surgery to assist in patient and procedure selection. The Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MBSAQIP®) and the Michigan Bariatric Surgery Collaborative (MBSC) have created online calculators to predict one year weight loss outcomes, comorbidity remission and 30 day morbidity and mortality after bariatric operations. Methods: We compared the actual and predicted outcomes of 306 patients who underwent primary laparoscopic Roux-en-Y gastric bypass (RYGB) or sleeve gastrectomy (SG) between 8/2013 -8/2018 at a single center.
Results: Our study population was 78.8% female. 40.5% underwent RYGB. Mean preoperative body mass index was 43.6 ± 6.4. At 1 year, mean total body weight loss (TBWL) was 25.7 ± 9.0%. Both calculators overpredicted TBWL with low goodness-of-fit (MBSAQIP +4.8 ± 8.5%, R2= 0.11; MBSC +4.9 ± 8.7%, R2= 0.08). For 30 day morbidity and 1 year comorbidity resolution, only diabetes and hypertension resolution predictions achieved a c-statistic greater than 0.7 for either MBSAQIP or MBSC.
Conclusion: Both the MBSAQIP and MBSC calculators poorly predicted the weight loss achieved by our patients, as there was great variability among actual patient outcomes. The predictions for morbidity and for resolution of hyperlipidemia, obstructive sleep apnea, and gastroesophageal reflux disease showed poor discrimination, but the predictions for hypertension and diabetes resolution showed reasonable discrimination. These tools should be used with caution when counseling patients.
Back to 2020 Posters