A proposed standardized approach to studying attrition in pediatric weight management

Pediatric obesity is a major public health issue. Lifestyle and behavioral interventions are the foundation of pediatric weight management; however, intervention effectiveness is compromised when families (children, youth, and/or parent[s]) discontinue care prematurely. Intervention attrition minimizes the potential health benefits derived from interventions, results in inefficient use of health services resources, and can magnify health disparities. Most attrition research in pediatric weight management has been descriptive, highlighting the need to advance the field, both academically and clinically. Herein, we propose a standard approach to studying attrition in pediatric weight management interventions to enhance our understanding, elevate the quality of research, enable study-to-study comparisons, and inform strategies designed to mitigate its impact. We focus on three issues. First, "Conceptualization and operationalization," whereby the processes underlying attrition from interventions should be decomposed into clinically important phases that are defined based on intervention characteristics. Relatedly, theoretically relevant variables should be identified with different mechanisms driving attrition in each phase. We propose a matrix of attrition, a tool designed to delineate the relevant stages of attrition and associated variables of analytical value. Second, "Pre-study" underscores the value of developing a plan to study attrition a priori rather than post hoc, including variable and sample size considerations, which broadens the range and quality of analysis. Finally, "Post-study" emphasizes comprehensive reporting of attrition, outlines typical comparisons of analytical interest, and statistical techniques used to handle missing data. Implications for clinical practice in pediatric weight management are discussed. (C) 2019 Asia Oceania Association for the Study of Obesity. Published by Elsevier Ltd. All rights reserved.