Amanda Doggett1, Ashok Chaurasia2, Jean-Philippe Chaput3,4, Scott T Leatherdale2. 1. School of Public Health Sciences, University of Waterloo, Waterloo, ON, Canada. adoggett@uwaterloo.ca. 2. School of Public Health Sciences, University of Waterloo, Waterloo, ON, Canada. 3. Department of Pediatrics, University of Ottawa, Ottawa, ON, Canada. 4. Healthy Active Living and Obesity Research Group, Children's Hospital of Eastern Ontario Research Institute, Ottawa, ON, Canada.
Abstract
BACKGROUND: Youth body mass index (BMI), derived from self-reported height and weight, is commonly prone to nonreporting. A considerable proportion of overweight and obesity (OWOB) research relies on such self-report data, however little literature to date has examined this nonreporting and the potential impact on research conclusions. The objective of this study was to examine the characteristics and predictors of missing data in youth BMI, height, and weight. METHODS: Using a sample of 74,501 Canadian secondary school students who participated in the COMPASS study in 2018/19, sex-stratified generalized linear mixed models were run to examine predictors of missing data while controlling for school-level clustering. RESULTS: In this sample, 31% of BMI data were missing. A variety of diet, exercise, mental health, and substance use variables were associated with BMI, height, and weight missingness. Perceptions of being overweight (females: 95% CI (1.42,1.62), males: 95% CI (1.71,2.00)) as well as intentions to lose weight (females: 95% CI (1.17,1.33), males: 95% CI (1.13,1.32)) were positively associated with BMI missingness. CONCLUSIONS: Findings from this study suggest that nonreporting in youth height and weight is likely somewhat related to the values themselves, and hint that social desirability may play a substantial role in nonreporting. The predictors of missingness identified in this study can be used to inform future studies on the potential bias stemming from missing data and identify auxiliary variables that may be used for multiple imputation approaches.
BACKGROUND: Youth body mass index (BMI), derived from self-reported height and weight, is commonly prone to nonreporting. A considerable proportion of overweight and obesity (OWOB) research relies on such self-report data, however little literature to date has examined this nonreporting and the potential impact on research conclusions. The objective of this study was to examine the characteristics and predictors of missing data in youth BMI, height, and weight. METHODS: Using a sample of 74,501 Canadian secondary school students who participated in the COMPASS study in 2018/19, sex-stratified generalized linear mixed models were run to examine predictors of missing data while controlling for school-level clustering. RESULTS: In this sample, 31% of BMI data were missing. A variety of diet, exercise, mental health, and substance use variables were associated with BMI, height, and weight missingness. Perceptions of being overweight (females: 95% CI (1.42,1.62), males: 95% CI (1.71,2.00)) as well as intentions to lose weight (females: 95% CI (1.17,1.33), males: 95% CI (1.13,1.32)) were positively associated with BMI missingness. CONCLUSIONS: Findings from this study suggest that nonreporting in youth height and weight is likely somewhat related to the values themselves, and hint that social desirability may play a substantial role in nonreporting. The predictors of missingness identified in this study can be used to inform future studies on the potential bias stemming from missing data and identify auxiliary variables that may be used for multiple imputation approaches.
Authors: Iris Eekhout; R Michiel de Boer; Jos W R Twisk; Henrica C W de Vet; Martijn W Heymans Journal: Epidemiology Date: 2012-09 Impact factor: 4.822
Authors: Amalia Karahalios; Laura Baglietto; John B Carlin; Dallas R English; Julie A Simpson Journal: BMC Med Res Methodol Date: 2012-07-11 Impact factor: 4.615
Authors: Daphne P Guh; Wei Zhang; Nick Bansback; Zubin Amarsi; C Laird Birmingham; Aslam H Anis Journal: BMC Public Health Date: 2009-03-25 Impact factor: 3.295