The Use of Predictive Markers for the Development of a Model to Predict Lowest Quartile Weight Loss Following Vertical Sleeve Gastrectomy

Samuel Cottam1, Daniel Cottam, MD1, Mitchell Roslin2, Hinali Zaveri, MD1, Amit K Surve, MD1, Christina Richards1, Walter Medlin1, Austin Cottam1. 1Bariatric Medicine Institute, 2Lenox Hill Hospital

Introduction: Although average percent excess weight loss data is commonly discussed preoperatively to guide patient expectations, there is a wide range and variation following Vertical Sleeve Gastrectomy (VSG). Unfortunately, there are relatively few predictors that allow for individual guidance of weight loss.  This analysis’ purpose is to compare patients percent weight loss and determine if any statistically significant variables are predictive of weight loss and develop a model that predicts weight loss underperformance. 

Method: 124 Patients who underwent the SG had data points that were adequate for analysis. These patients underwent surgery between October 2008 and April 2014. Data was gathered. Non-linear regression was performed in order to interpolate patient weights at one year. Multivariate analysis was used to find factors that effected weight loss.  Then using that data a model was constructed to predict weight loss performance in any practice.

Results: Patients were divided in %EWL quartiles. Multiple logistic regression was used to find that Diabetes and Preoperative BMI were influencing factors in %EWL at one year.  Using this data our model had Positive and Negative predictor values of 86% and 87% respectively.

Conclusion: Patients with a BMI greater than 54 or patients with Type 2 Diabetes Mellitus (DM) and a BMI greater than 43 are more likely to be in the bottom quarter.  This information should be discussed and consideration given to performing a more aggressive procedure when a patient’s objective is to have a BMI below the obesity threshold of 30.

« Return to SAGES 2016 abstract archive