Marijka Batterham1, Linda C Tapsell2, Karen E Charlton2. 1. Statistical Consulting Centre, National Institute for Applied Statistics Research Australia, School of Mathematics and Applied Statistics, University of Wollongong, NSW, Australia. Electronic address: marijka@uow.edu.au. 2. School of Medicine, Faculty of Science, Medicine and Health, University of Wollongong, NSW, Australia.
Abstract
UNLABELLED: Attrition causes analytical and efficacy issues in weight loss trials. Consistent predictors of attrition in weight loss trials have not been identified. Trial design could be improved if factors predicting attrition are accounted for. The aim of this study is to quantify the effect of easily measured pre study and early study variables to determine their relationship with attrition in dietary weight loss trials. METHODS: Data was pooled from four previous dietary weight loss trials. Mixed effects logistic regression, Receiver Operator Curves and decision trees (classification and regression trees) were used to determine which of the variables (percent weight loss at 1 month, age, gender and baseline BMI) predicted dropout and to determine cutoffs useful for future trial design. RESULTS: The sample included 289 subjects, 73% female, with a mean age of 46.68±9.27years and average dropout of 25%. Percent weight loss at 1 month was the strongest predictor of dropout, those with a weight loss ≤2% were 4.99 times (95% CI 2.71, 9.18) more likely to drop out than those with a weight loss >2% in the first month (P<0.001). When considering only data available at the beginning of a trial those ≤50 years old were 2.07 times (95% CI 1.2, 3.5) more likely to drop out than those >50 (P=0.006). DISCUSSION: Early weight loss and age were identified as significant variables for predicting attrition in weight loss trials. Trial designs maybe improved by incorporating these variables and developing interventions targeting these factors may improve participant retention.
UNLABELLED: Attrition causes analytical and efficacy issues in weight loss trials. Consistent predictors of attrition in weight loss trials have not been identified. Trial design could be improved if factors predicting attrition are accounted for. The aim of this study is to quantify the effect of easily measured pre study and early study variables to determine their relationship with attrition in dietary weight loss trials. METHODS: Data was pooled from four previous dietary weight loss trials. Mixed effects logistic regression, Receiver Operator Curves and decision trees (classification and regression trees) were used to determine which of the variables (percent weight loss at 1 month, age, gender and baseline BMI) predicted dropout and to determine cutoffs useful for future trial design. RESULTS: The sample included 289 subjects, 73% female, with a mean age of 46.68±9.27years and average dropout of 25%. Percent weight loss at 1 month was the strongest predictor of dropout, those with a weight loss ≤2% were 4.99 times (95% CI 2.71, 9.18) more likely to drop out than those with a weight loss >2% in the first month (P<0.001). When considering only data available at the beginning of a trial those ≤50 years old were 2.07 times (95% CI 1.2, 3.5) more likely to drop out than those >50 (P=0.006). DISCUSSION: Early weight loss and age were identified as significant variables for predicting attrition in weight loss trials. Trial designs maybe improved by incorporating these variables and developing interventions targeting these factors may improve participant retention.
Authors: Linda C Tapsell; Maureen Lonergan; Marijka J Batterham; Elizabeth P Neale; Allison Martin; Rebecca Thorne; Frank Deane; Gregory Peoples Journal: BMJ Open Date: 2017-07-13 Impact factor: 2.692