Accuracy of Parent-Measured and Parent-Estimated Heights and Weights in Determining Child Weight Status

Abstract

This cohort study assesses whether parent measurements of children’s height and weight are more accurate than their respective estimates for determining weight status and how both compare with professionally measured data.

Accurate measurements of height and weight are necessary for assessing weight status of children in population-based studies. Professionally measured heights and weights represent the criterion standard for calculating body mass index (BMI), the most feasible and frequently used measure of body weight status. However, collecting professionally measured data on sufficient samples of children in population studies is extremely resource intensive.¹ As a result, these studies often rely on child- or parent-estimated heights and weights. Although these estimated values are easy to collect, they are prone to misreporting, with systematic differences across sociodemographic characteristics.²

A viable alternative, explored in only a few studies so far,^3,4 is parent measurement of height and weight—a method that is less expensive than professionally measured data and arguably more accurate than parent estimates. However, the accuracy of parent measurements has not been empirically tested in the United States. In this cohort study, using a subsample of children derived from the New Jersey Child Health Study, we evaluated whether parent-measured heights and weights were more accurate than parent-estimated heights and weights for classifying children’s weight status. We compared each method with the criterion standard, professionally measured heights and weights.

Methods

The New Jersey Child Health Study was a longitudinal study that followed up on 2 panels of children between 2009 and 2017 to examine the association of changes in food and physical activity environments with changes in body weight status. In a subsample, 3 approaches were used in the following order to measure the height and weight of participants aged 3 to 18 years (n = 226): (1) parent estimates, obtained through a telephone interview; (2) parent measurements, whereby parents weighed and measured their children according to an established protocol⁵ that included using instructions and a tape measure that were mailed to the child’s home, recording the anthropometric data on a supplied worksheet, and mailing the worksheet back to the research team; and (3) professional measurements conducted by trained nurses in children’s homes using the standard Centers for Disease Control and Prevention protocol¹ and standardized equipment. The institutional review board of Arizona State University approved this study. Oral consent was obtained from parents for collecting the estimated and measured heights and weights, and written consent (from parents and adolescents) and oral assent (from younger children) were obtained for collecting professionally measured data.

The 3 types of measurements were used to calculate age- and sex-specific BMI percentiles for participants based on US Centers for Disease Control and Prevention growth reference charts.⁶ Professionally measured data were used to obtain the true weight status classification: overweight or obese (≥85th BMI percentile) vs not overweight or obese (<85th BMI percentile). The sensitivity and specificity of parent-estimated and parent-measured values for accurately classifying children’s weight status relative to the criterion standard were also calculated. Analyses were run using Stata 14.0 (StataCorp).

Results

The analytical sample included 226 children from low-income families from 4 New Jersey cities: Camden, New Brunswick, Newark, and Trenton (Table). Among those in the sample, 123 children (54.4%) were female and 103 (45.6%) were male, 118 (52.5%) were aged 3 to 11 years and 108 (47.8%) aged 12 to 18 years, and 9 (4.0%) were non-Hispanic white, 144 (63.7%) non-Hispanic black, 62 (27.4%) Hispanic, and 11 (4.9%) another race/ethnicity. The parent-estimated, parent-measured, and professionally measured methods classified similar proportions of children as overweight or obese: 104 of 207 children (50.2%), 105 of 222 children (47.3%), and 114 of 226 children (50.4%), respectively. The sensitivity and specificity (Figure) of parent-measured values (77.7% and 83.6%, respectively) were similar to those of parent-estimated values (78.1% and 78.4%, respectively). The 95% CIs for the relative sensitivity and relative specificity of the 2 approaches suggested no statistically significant difference in the accuracy of children’s weight status classification using the 2 methods. Subgroup analyses using child’s age, sex, and race; parent educational level and BMI (obese vs not obese) calculated from self-reported heights and weights; and household annual income yielded similar results. Classifying children’s weight status as obese vs nonobese also did not alter our findings.

Table. Demographic Description of the Analytical Sample.

Characteristic	Participants, No. (%) (n = 226)
Sex
Male	103 (45.6)
Female	123 (54.4)
Age, y
3-11	118 (52.2)
12-18	108 (47.8)
Race/ethnicity
Non-Hispanic white	9 (4.0)
Non-Hispanic black	144 (63.7)
Hispanic	62 (27.4)
Other	11 (4.9)
Mother’s educational level
Less than high school	29 (12.9)
High school	99 (44.2)
College or higher	96 (42.9)
Parent’s BMIa
Not obese	120 (53.1)
Obese	106 (46.9)
Household poverty level, FPLb
<200%	163 (72.1)
≥200%	63 (27.9)

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); FPL, federal poverty level.

^aParent’s BMI: not obese was defined as a BMI less than 25; obese was defined as a BMI of 25 or greater.

^bHousehold income is presented as a percentage of the FPL in the year in which the parent was interviewed.

Figure. Sensitivity and Specificity of Parent-Measured and Parent-Estimated Values for Classifying Children’s Weight Status.

Professional measurements were used as true values for calculating sensitivity and specificity. Subgroup analyses were conducted using child age, sex, and race/ethnicity; parent body mass index and educational level; and household income.

Discussion

Collection of parent-measured heights and weights based on standardized instructions is feasible and less expensive than collection of professionally measured data. However, in our study parent measurements did not improve the accuracy of children’s weight status classification over parent estimates, the least expensive of the 3 methods. A possible explanation for this finding is that parent estimates yielded relatively high sensitivity and specificity, leaving limited room for improvement. A European study of preschool children reported lower sensitivity (47%) and similar specificity (93%) of parent estimates relative to our findings and reached the same conclusion: parent measurements did not improve overweight/obese status classification over parent estimates.⁴

Our study has some limitations. The children included in this sample were predominantly non-Hispanic black and Hispanic and were from low-income households, limiting the generalizability of our findings to other populations. In addition, the subsample from the New Jersey Child Health Study that we used depended on parents’ willingness to participate in additional measurements. However, the demographic characteristics of the subsample were similar to those of the overall New Jersey Child Health Study.

Our study indicates that parent estimates and parent measurements of children’s height and weight are both effective in classifying weight status. Therefore, when professionally measured data cannot be collected, parent estimates represent the most efficient alternative, as they are less resource intensive than parent measurements but without any trade-off in accuracy.