Validation of Remote Child Weight and Height Measurements Within a Weight Management Trial: Brief Cutting Edge Report

Abstract

Objective:

The aim of this sub-study within the TEAM UP pragmatic clinical trial was to compare the validity of anthropometric measurements collected remotely vs. in-person (≤7 days apart) among youth with obesity ages 6 to 15 years.

Methods:

Child (n=37) weight and height were measured in-person by a trained data assessor (DA). These were compared with measurements taken remotely by the child’s parent with live videoconferencing observation by a study DA. In-person and remote measurements were compared using Bland-Altman plots, Pearson correlations, and two 1-sided paired t-tests. A priori bounds of acceptability were set at ±0.68 kg to allow for typical weight fluctuations within the 7-day comparison period.

Results:

Measurements were highly correlated (height: r=.991, p<.0001; weight: r=.999; p=.03). For height, two 1-sided t-tests for upper, t(36)=3.95, and lower, t(36)=−2.63, bounds (−1,1) revealed an overall p=0.006, absolute error was 3.5 cm. For weight, two 1-sided t-tests for upper, t(36)=1.93, and lower, t(36)=−7.91, bounds (−0.68, 0.68) revealed an overall p=0.03, absolute error was 1.7 kg.

Conclusions:

The present findings support the utility and interpretation of remotely assessed weight management outcomes for both research and clinical purposes. These procedures may offer greater accessibility to evidence-based measurement.

Introduction

Barriers to accessing intensive health behavior and lifestyle intervention is an obstacle, for a multitude of reasons, to many children seeking obesity treatment.¹ In light of these barriers, as well as the challenges of the COVID-19 pandemic, researchers and clinicians are embracing eHealth technologies to expand access to effective treatment and care.² eHealth refers to health interventions remotely delivered using internet and related technology components.³ These technologies demonstrate promising results for reducing child body mass index (BMI)⁴ and enhancing engagement and commitment to weight management treatment.⁵ With the increasing use of digital interventions,⁶ there is a growing need for valid remote assessment procedures. Current procedures, like remote self-report of height and weight, are perceived as non-valid.⁷ Several studies have conducted height and weight assessments outside of standard clinic and research settings, including self-/parent-report of height and weight, and employing inexperienced observers.^8–11 Few have evaluated these procedures with remote oversight by study staff and guidance provided to parents and caregivers outside of these settings, thus offering a potential solution to resolving bias inherent to self-report. The purpose of the current sub-study was to determine the validity of physical measurements collected by parents during remote visits, compared to the same measurements collected by trained data assessors (DAs) at in-person visits, among a sample of children with obesity.

Methods

Participants

Participants were youth 6–15 years old completing a screening visit for the PCORI-funded Treatment Efforts Addressing Child Weight Management by Unifying Patients, Parents, and Providers study (TEAM UP; NCT03843424). Children were recruited from two metropolitan areas of the southern U.S.

Procedures

Visit 1 took place in-person at the child’s primary care provider (PCP)’s clinic or remotely. At this visit written parent consent and child assent were obtained, child height and weight were measured, and parents completed screening questionnaires. Within one week of visit 1, participating children were asked to complete another height and weight for visit 2, which took place either in-person if visit 1 was remote, or vice versa.

For remote measurement, families were provided with a calibrated scale, carpenter’s square, measuring tape, and written and video instructions. Video instructions informed parents what materials to use and provided a demonstration of each step (i.e., procedures described in the written instructions; Figure 1). All procedures were approved by the Washington University in St. Louis Institutional Review Board.

Figure 1.

TEAM UP Remote Height and Weight Collection Instructions for Parents/Caregivers

Measures

In-person height was measured by trained DAs to the nearest 0.1 cm, using a Seca 213 portable stadiometer. Children were asked to remove hair ornaments/buns/braids as well as shoes. Children stood with their arms naturally at their sides while looking straight ahead in the Frankfort Plane. Height measurements were collected twice, with a third measurement obtained if a difference >0.3 cm was found between the first two.¹²

Remote height was collected while trained DAs observed via secure videoconferencing. Parents were instructed to take the measurements on a level, non-carpeted flat surface, in a location visible to the DA. Children were asked to remove shoes and hair ornaments/buns/braids and shoes. Parents taped a piece of paper to a wall at the approximate height of the child’s head, had the child stand in front of the paper, take a deep breath while looking straight ahead, and the parent rested the carpenter’s square against the child’s head and the wall at a 90-degree angle. Next, children stepped away while parents held the carpenter’s square in place, and then parents marked the paper at the bottom of the square with a pencil. The tape measure was then used to assess height from the floor to the mark. Two heights were completed, with a third measurement obtained if the first two differed by >0.1 cm.

In-person weight was measured by trained DAs to the nearest 0.1 kg, using a Seca 876 digital scale. Children were instructed to empty their pockets, to remove shoes/additional outerwear, and to look straight ahead while standing in the middle of the scale.¹³ This measurement was completed twice, with a third measurement taken if a difference >0.1 kg was obtained between the first two.¹²

Remote weight was measured using a study-provided scale, under the observation of a trained DA via videoconferencing. Parents were instructed to obtain measured weight on a level, non-carpeted, flat surface. Children removed shoes/heavy clothing items and were instructed to stand still, in the middle of the scale, and look straight ahead. Prior to the first measurement, the family turned on the scale and waited for it to zero out. Weight was then measured two times, and a third measurement was obtained if the first two differed by >0.1 kg.

For all four of these measures, the recorded value for statistical analyses was the average of the two measures, or (where a third measurement was taken) the three measures.

Statistical Analysis

Descriptives were obtained for in-person and remote measurements of height and weight. To measure absolute agreement, Bland-Altman plots were used to show the calculation of absolute error (|remote minus in-person|). Limits of agreement were calculated as the mean difference (difference in measurements between remote and in-person) ± 1.96 times the standard deviation of the mean difference. Two 1-sided t-tests were used to determine whether the remote and in-person measurements were equivalent. A priori bounds were set at ±0.68 kg to allow for typical weight fluctuations within the 7-day comparison period,¹⁴ where weights not significantly different than these limits were determined to be equivalent. Pearson correlations provided the degree of similarity between these measurements. Data were managed using REDCap,^15,16 and all analyses were performed using SAS 9.4.¹⁷

Results

A total of 37 children completed this sub-study. Demographics and descriptive statistics are presented in Table 1. For height, two 1-sided t-tests accounting for upper, t(36)=3.95, and lower, t(36)=−2.63, bounds (−1.00, 1.00) revealed an overall p=0.006 (90% confidence interval: −0.31, 0.71); thus, we conclude the measurements were equivalent. The absolute error value was 3.51 cm, and three participants had errors beyond the 95% limits of agreement (Figure 2A). Pearson correlations revealed high correlation between in-person and remote measurement (r=.991, p<.0001).

Table 1.

Demographics and Descriptive Statistics

	N	%	Mean	Std Dev
Child
Sex
Boys	12	32
Girls	25	68
Race
Black/African American	20	54
Multi-Race (Black and White)	2	5
Other	1	3
White	14	38
Ethnicity
Hispanic/Latino(a)	3	8
Not Hispanic/Latino(a)	34	92
Age (years)	37		11.0	2.5
Body Mass Index Percentile	37		98.9	1.0
Remote Height (cm)	37		150.9	13.7
In-Person Height (cm)	37		151.4	14.0
Remote Weight (kg)	37		78.1	26.1
In-Person Weight (kg)	37		77.7	26.0
Parent
Sex
Male	1	3
Female	36	97
Age	38	8.2
Parental education
Less than high school	1	3
High school or General Educational	18	49
Development (GED)
Associate degree or 1–3 years in college	5	14
Bachelor’s degree	7	19
Graduate/professional degree	6	16
Total Number of Household Members	37		4	1.6

Figure 2.

Height (A) and weight (B) Bland-Altman plots

For weight, two 1-sided t-tests accounting for upper, t(36)=1.93, and lower, t(36)=−7.91, bounds (−0.68, 0.68) revealed an overall p=0.03 (90% confidence interval: 0.65, −0.18); thus, we conclude the measurements were equivalent. The absolute error value was 1.7 kg, with three participants’ error beyond the 95% limits of agreement (Figure 2B). Pearson correlations revealed high correlation between in-person and remote measurement (r=.999, p<.0001).

Discussion

No significant differences in means were observed in measurements taken remotely with live videoconferencing observation compared to estimates collected using in-person procedures by trained DAs. Furthermore, measurements were highly correlated. Six datapoints were outside of the limits of agreement (LOA). In these instances, remote height was underestimated, and remote weight was overestimated. Despite our efforts at parent preparation, we speculate that other factors including in-home structures like carpeting, baseboards, and/or precision of positioning the child may have contributed to these non-significant differences. Providing families with a tripod for these appointments may provide more reliable video feed so the DA can better guide the measurement.¹⁸

Others have conducted studies to evaluate the validity of parent-reported height and weight, with varying results.^8,10,11,18 Forseth and colleagues (2022) provided similar instructions to families for parent-report height and weight and observed no significant differences between these measures compared to in-person measurements obtained by study staff within the same day. The present study is distinct in that the procedures accounted for a larger range of time and reduced potential bias via inclusion of remote observation. The aforementioned study relied on self-report, and some families were measured using in-person procedures prior to self-reporting their height and weight later that same day without video confirmation. The inclusion of remote observation increases objectivity, thus reducing risk of measurement error and social desirability bias.

We recognize limitations to the generalizability of the current study, the first being cost prohibition for researchers and clinicians to absorb the expenses of providing and shipping materials, whereas families’ own equipment may be faulty and not standardized. A second limitation is the range of LOA in the study, where a measured height ±3.5 cm or weight ±1.7 kg from the true value, as determined by in-person measurement, could potentially lead to differential and inappropriate diagnosis and treatment based on BMI percentile calculated from these values. Stricter limits of agreement should be employed for medical decision-making. The third limitation recognizes concerns about internet connectivity and digital literacy as barriers to using remote methods. Poor connectivity for videoconferencing particularly affects low-income, rural-dwelling, and racial and ethnic minority households, thus threatening to exclude these families from such interventions.¹⁹ Assessment of and alternatives for those with a digital divide must be considered.

In summary, remote measurement of height and weight by a child’s parent with live videoconferencing observation was valid and comparable to estimates collected using in-person procedures by trained DAs. These results are timely as more child weight-management interventions and treatments are proving to be feasible, acceptable, and effective when delivered via telehealth.²⁰

Study Importance.

• There is limited support for the validity of remotely collected weight and height measurements used in the evaluation of pediatric treatment outcomes.

• The current study found remote measurement of height and weight among a sample of youth with obesity to be valid and equivalent to gold standard in-person measurement.

• These findings offer greater accessibility to evidence-based measurement for researchers and clinicians.