K M Flegal1. 1. National Center for Health Statistics, Center for Disease Control and Prevention, Hyattsville, MD 20782, USA. kmf2@cdc.gov
Abstract
OBJECTIVE: The objective of this study is to investigate the effects of methods of age grouping on estimates of overweight prevalence for children and adolescents based on reference body mass index (BMI) percentiles. METHODS: Data for children aged 6-17 y came from three nationally representative US surveys, the Third National Health and Nutrition Examination Survey (1988-1994) and Cycles II and III of the National Health Examination Survey (1963-1965, 1966-1970). Month-specific BMI percentile values were taken from the revised US growth charts. Ages were grouped into categories 3, 6 or 12 months in length. Cut-off points were selected as the low or the mean percentile value within the category. Overweight prevalences for these groupings were compared with prevalences calculated using the month-specific values. RESULTS: The effects of grouping and cut-off point selection on overweight prevalence estimates were generally small; however, the combination of 12 month groupings and the low value led to an overestimation by up to 3 percentage points. Within the 12 month groupings, the first 6 months differed systematically from the second 6 months. CONCLUSIONS: Although age categorization may often have little effect on prevalence estimates, prevalence may sometimes be overestimated by as much as 3 percentage points. Use of narrower age categorizations than those used to construct the reference values may result in systematic biases. It is important to understand how age was handled in the construction of the reference population and to select age categories consistent with those used for the reference population.
OBJECTIVE: The objective of this study is to investigate the effects of methods of age grouping on estimates of overweight prevalence for children and adolescents based on reference body mass index (BMI) percentiles. METHODS: Data for children aged 6-17 y came from three nationally representative US surveys, the Third National Health and Nutrition Examination Survey (1988-1994) and Cycles II and III of the National Health Examination Survey (1963-1965, 1966-1970). Month-specific BMI percentile values were taken from the revised US growth charts. Ages were grouped into categories 3, 6 or 12 months in length. Cut-off points were selected as the low or the mean percentile value within the category. Overweight prevalences for these groupings were compared with prevalences calculated using the month-specific values. RESULTS: The effects of grouping and cut-off point selection on overweight prevalence estimates were generally small; however, the combination of 12 month groupings and the low value led to an overestimation by up to 3 percentage points. Within the 12 month groupings, the first 6 months differed systematically from the second 6 months. CONCLUSIONS: Although age categorization may often have little effect on prevalence estimates, prevalence may sometimes be overestimated by as much as 3 percentage points. Use of narrower age categorizations than those used to construct the reference values may result in systematic biases. It is important to understand how age was handled in the construction of the reference population and to select age categories consistent with those used for the reference population.
Authors: Deanna M Hoelscher; R Sue Day; Eun Sul Lee; Ralph F Frankowski; Steven H Kelder; Jerri L Ward; Michael E Scheurer Journal: Am J Public Health Date: 2004-06 Impact factor: 9.308