Effect of School-Based Body Mass Index Reporting in California Public Schools

This cluster randomized clinical trial assesses the effect of school-based body mass index reporting on weight status and adverse outcomes (weight stigmatization and weight-related perceptions and behaviors) among a diverse population of students in California public schools.

Key Points

Question

Does school-based body mass index (BMI) screening and reporting improve weight status or have unintended consequences among diverse students in grades 3 to 8?

Findings

In this cluster randomized clinical trial among 79 California schools (28 641 students participating up to 3 years), BMI reports sent to parents had no effect on BMI z scores at 1 year or 2 years of follow-up, with mixed results found on the effect of these reports on adverse outcomes.

Meaning

Although thousands of schools currently send BMI reports to parents, the practice alone has no impact on pediatric obesity and may decrease student weight satisfaction; schools should consider dedicating resources to evidence-based approaches to improve student health.

Abstract

Importance

Annually, US schools screen millions of students’ body mass index (BMI) and report the results to parents, with little experimental evidence on potential benefits and harms.

Objective

To determine the impact of school-based BMI reporting on weight status and adverse outcomes (weight stigmatization and weight-related perceptions and behaviors) among a diverse student population.

Design, Setting, and Participants

Cluster randomized clinical trial. The Fit Study (2014-2017) randomized 79 California schools to BMI screening and reporting (group 1), BMI screening only (group 2), or control (no BMI screening or reporting [group 3]) in grades 3 to 8. The setting was California elementary and middle schools. Students in grades 3 to 7 at baseline participated for up to 3 years. A modified intent-to-treat protocol was used. Data analysis was conducted from April 13, 2017, to March 26, 2020.

Interventions

School staff assessed BMI each spring among students in groups 1 and 2. Parents of students in group 1 were sent a BMI report each fall for up to 2 years.

Main Outcomes and Measures

Changes in BMI z score and in adverse outcomes (based on surveys conducted each fall among students in grades 4 to 8) from baseline to 1 and 2 years of follow-up.

Results

A total of 28 641 students (14 645 [51.1%] male) in grades 3 to 7 at baseline participated in the study for up to 3 years. Among 6534 of 16 622 students with a baseline BMI in the 85th percentile or higher (39.3%), BMI reporting had no effect on BMI z score change (−0.003; 95% CI, −0.02 to 0.01 at 1 year and 0.01; 95% CI, −0.02 to 0.03 at 2 years). Weight dissatisfaction increased more among students having BMI screened at school (8694 students in groups 1 and 2) than among control participants (5674 students in group 3). Results of the effect of BMI reporting on other adverse outcomes were mixed: compared with the control (group 3), among students weighed at school (groups 1 and 2), weight satisfaction declined more after 2 years (−0.11; 95% CI, −0.18 to −0.05), and peer weight talk increased more after 1 year (0.05; 95% CI, 0.01-0.09); however, concerning weight control behaviors declined more after 1 year (−0.06; 95% CI, −0.10 to −0.02).

Conclusions and Relevance

Body mass index reports alone do not improve children’s weight status and may decrease weight satisfaction. To improve student health, schools should consider investing resources in evidence-based interventions.

Trial Registration

ClinicalTrials.gov Identifier: NCT02088086

Introduction

In the United States, almost 1 in 5 youths have obesity, with prevalence substantially higher among Hispanic (26%) and African American (22%) youths than among White youths (14%).¹ As of 2013, a total of 25 states required schools to screen students’ body mass index (BMI), and 11 states required that schools report BMI results to parents in an effort to reduce pediatric obesity.²

Although widely implemented, BMI reporting has not been shown to reduce childhood obesity.³ However, existing studies have had important limitations, including the use of BMI reports with higher than recommended literacy levels or potentially insensitive language.^4,5 In addition, despite evidence suggesting that parents of different racial/ethnic groups respond differently to BMI reports,^6,7 to the best of our knowledge, no studies have examined the effect of BMI reporting on weight status by race/ethnicity, nor have studies compared effects between younger students, whose parents may have more control over weight-related behaviors,⁸ and older students with greater autonomy.

Furthermore, case studies suggest that school-based BMI screening may increase weight stigmatization, including weight-based teasing^9,10 and other forms of weight-related talk,¹¹ both of which have been found to be factors in disordered eating behaviors in adolescents.^12,13 Weight stigmatization is particularly prevalent and harmful among students at higher BMI levels.^14,15 In addition, BMI reporting has been shown to increase the likelihood that parents will put their children on diets,^10,16 which has been found to predict weight gain in adolescents.^17,18 However, to our knowledge, no experimental studies have examined the potential adverse outcomes of BMI screening and reporting. It is important that efforts to curb obesity leave youth empowered to make healthy changes, not increasingly stigmatized, which may foster the development of eating disorders.¹⁹

The Fit Study sought to determine the effect of a BMI report, developed with input by parents from diverse racial/ethnic backgrounds,²⁰ on weight status among at-risk students (BMI ≥85th percentile) in grades 3 to 8 and to identify potential differential impacts by race/ethnicity and grade level. We also aimed to identify potential adverse outcomes of BMI screening and reporting and to elucidate differential impacts by weight status.

Methods

Study Design

This cluster randomized clinical trial was approved by participating school districts and the University of California, Berkeley’s Committee for the Protection of Human Subjects. The Committee for the Protection of Human Subjects waived the requirement for informed consent; parents were able to opt out instead. Schools sent parents a letter describing the study that asked them to return an enclosed form if they did not want their child to participate.

California Education Code requires annual height and weight assessments in grades 5, 7, and 9, but BMI reporting to parents is optional.²¹ The Fit Study (2014-2017) randomized 79 elementary and middle schools in 5 California school districts that did not send BMI reports to parents to 1 of the following 3 groups (the CONSORT flow diagram²² is shown in the Figure): group 1 (BMI screening and reporting, with BMI assessed and reported to parents [27 schools]); group 2 (BMI screening only, with BMI assessed but not reported to parents [27 schools]); or group 3 (control, with no BMI screening or reporting [25 schools]). The 3-group design (eFigure 1 in Supplement 1) allowed us to compare changes from baseline to 1 and 2 years of follow-up in the following: (1) weight status between students whose families received a report stating they were overweight or at risk for overweight (BMI ≥85th percentile) in group 1 and similar students who had BMI assessed but whose families did not receive a report (group 2), (2) child- and peer-related adverse outcomes between students who had their BMI assessed at school (groups 1 and 2) and students who did not (group 3), and (3) weight stigmatization by families between students who received a BMI report (group 1) and those who did not (groups 2 and 3). Schools were the unit of randomization and intervention; recruitment and study design have been described previously.²³

Figure. The Fit Study: CONSORT Flow Diagram²².

The trial protocol and statistical analysis plan are available in Supplement 2. The study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.

Participants

Students in grades 3 to 7 in fall 2014 and fall 2015 (eFigure 2 in Supplement 1) were eligible for participation in the study (N = 30 542). Surveys were only administered in grades 4 and higher (younger students are less reliable respondents²⁴); therefore, in control schools, where data were solely collected via student survey (BMI was not assessed), eligibility was limited to grades 4 to 7 at baseline. Opt-out rates were similar across groups (6.2%, P = .57). Enrolled students in grades 3 to 7 at baseline (N = 28 641 [30 542 eligible students across groups minus 1901 who opted out]) participated for up to 3 years (eFigure 3 in the Supplement).

Intervention and Fidelity

Body mass index screening by school staff was conducted in groups 1 (BMI reporting) and 2 (BMI screening) in the spring of 2015, 2016, and 2017. Parents of students in group 1 were sent a BMI report (eFigure 4 in Supplement 1) each fall for up to 2 years (October of 2015 and 2016, approximately 6 months after BMI assessments). Reports, which were developed based on focus groups with diverse parents,²⁰ classified children as overweight (BMI≥95th percentile for sex and age), at risk for overweight (BMI ≥85th percentile and <95th percentile), healthy weight (BMI≥5th percentile and <85th percentile), or underweight (BMI<5th percentile).²⁵ Reports included an infographic visually presenting family-oriented health recommendations.^26,27 To ensure intervention fidelity, a third-party vendor mailed BMI reports directly to parents using each school’s name in the return address; the 1.7% of reports returned to sender were delivered to the appropriate school to be given to students to bring home.

Measures

Change in BMI z Score

School staff assessed BMI among participating group 1 and 2 students in grades 3 to 8 using research-grade equipment²³; height and weight measurements were equivalent to those of trained researchers.²⁸ The California Department of Education excused control schools (group 3) from assessing BMI during study years; however, all study schools conducted fitness assessments per usual practice.

Adverse Outcomes

Researchers (nonauthors) administered surveys each fall to students in grades 4 to 8. In group 1 and 2 schools, follow-up surveys occurred 6 to 9 months after BMI assessments (and 1-2 months after BMI reports were mailed). Survey items were adapted from Project EAT (Eating and Activity in Teens [formerly Eating Among Teens]) ^12,13,19 and the Family Experiences Related to Food Questionnaire.²⁹

A peer weight teasing index (range, 1-5) was used to assess the frequency of the student or “other kids” being teased or made fun of at school because of weight. Responses on 5-point scales ranged from “never” to “almost every day.”

To evaluate weight-related perceptions and behaviors, weight satisfaction was assessed on a 5-point scale ranging from “very unhappy” to “very happy.” An index for concerning weight control behaviors (range, 0-3) summed 3 binary indicators for dieting, skipping meals, or eating very little food in the last year to lose weight.

To assess family behaviors, a weight talk index (range, 1-5) averaged responses to 2 questions concerning family talk about the student’s weight or size and frequency of weight teasing by family (higher scores reflect greater stigmatization); family encouraging the student to diet was assessed on a 4-point scale ranging from not at all to very much. Students were also asked if they perceived themselves to be very underweight, somewhat underweight, about the right weight, somewhat overweight, or very overweight (very and somewhat underweight responses were collapsed in analyses).

Schools provided parent-reported sex, race/ethnicity, and grade for participating students. School-level free and reduced-price meal eligibility (a proxy for socioeconomic status) and school enrollment were obtained from the California Department of Education website.³⁰

Statistical Analysis

Primary analyses were limited to complete cases, namely, students with a valid BMI (0.8% of BMI values were biologically implausible [eAppendix in Supplement 1]) or complete survey data at baseline and at least 1 of the 2 follow-up assessments (eFigure 3 in Supplement 1). Among students enrolled in groups 1 and 2 (n = 20 482 [21 826 eligible students minus 1344 who opted out]), 5.3% had no (or invalid) baseline BMI data, and 13.5% had no follow-up data, with no differences between groups (eFigure 3 in Supplement 1), yielding 16 622 (of 19 390 students with a valid baseline BMI, 16 543 had a valid BMI at year 1; an additional 79 students without a BMI at year 1 had a valid BMI at year 2) complete cases. Among 20 937 enrolled students in grades 4 to 7 at baseline, 14.9% across groups 1 to 3 had no (or incomplete) baseline survey data, and an additional 16.5% had no follow-up data, with no differences between groups (eFigure 3 in Supplement 1), yielding 14 368 (of 17 827 students with a complete baseline survey, 14 033 had a complete survey at year 1; an additional 335 students without a complete survey at year 1 had a complete survey at year 2) complete cases. For all outcomes, we used linear mixed-effects models with a group by time interaction term (including main effects) and random intercepts for students nested within schools to account for clustering. Models were adjusted for student sex, race/ethnicity, grade, and school district, as well as school-level free and reduced-price meal eligibility and calendar year. Stata/SE, version 15.1 (StataCorp) was used for statistical analyses. Two-sided P < .05 was considered statistically significant for interaction and main effects.

We compared change in BMI z score between group 1 and 2 students with a baseline BMI in the 85th percentile or higher (primary outcome), for whom BMI reports are likely to have the greatest effect. Participant flow among overweight students was similar to that of all group 1 and 2 students (eFigure 5 in Supplement 1). We explored race/ethnicity (Hispanic vs non-Hispanic) and elementary grade status (grades 3-5 at baseline vs grades 6-7) as effect modifiers. We used a modified intent-to-treat protocol, in which students who left their school during the study were not followed up, but we conducted sensitivity analyses using multiple imputation for all missing and biologically implausible follow-up BMI data. Accounting for students moving out of schools,²³ the sample size provided 80% power to detect a difference between groups 1 and 2 in 1-year change in BMI z score of 0.02.

To identify adverse outcomes of assessing BMI in schools (secondary outcomes), models compared all students in groups 1 and 2 with students in group 3 (controls). Models included all students because disordered eating and body dissatisfaction are not limited to students with overweight or obesity.^31,32 To identify adverse outcomes of sending BMI reports to parents, models compared group 1 with groups 2 and 3. Models also adjusted for baseline weight status (BMI was only assessed in groups 1 and 2; therefore, perceived weight status from surveys was used) and explored perceived weight status as an effect modifier.

Results

A total of 28 641 students (14 645 [51.1%] male and 13 996 [48.9%] female) in grades 3 to 7 at baseline participated in the study for up to 3 years. Baseline characteristics are listed in Table 1; there were no differences in sex, race/ethnicity, or grade between groups. Among students in groups 1 and 2, BMI z scores were similar, as was the prevalence of a BMI in the 85th percentile or higher. Among group 1 and 2 students, 39.6% had a baseline BMI in the 85th percentile or higher, among whom baseline BMI z scores (mean [SE], 1.74 [0.43]) were similar between groups (P = .14), as were the proportion of female students (46.0%, P = .85), elementary school students (66.4%, P = .95), and Hispanic students (70.2%, P = .13). Among the 14 368 students with complete survey data (49.5% female), baseline measures of adverse outcomes and perceived weight status were similar between groups 1 and 2.

Table 1. Baseline Characteristics Among 28 641 Students in Grades 3 to 7 From 79 California Schools.

Variable	Enrolled students				Complete cases
	Group 1: BMI screening and reporting (n = 10 041)	Group 2: BMI screening only (n = 10 441)	Group 3: Control (n = 8159)	Total (N = 28 641)	Weight status analyses, groups 1 and 2, BMI ≥85th percentile (n = 6534)a	Weight stigmatization analyses, groups 1, 2, and 3; grades 4-7 (n = 14 318)b
Sex, %
Male	51.9	50.7	50.8	51.1	54.3	50.5
Female	48.1	49.3	49.2	48.9	45.7	49.5
Race/ethnicity, %
Hispanic	56.9	62.9	56.3	58.9	70.2	58.0
Asian	17.8	10.7	17.5	15.1	9.7	17.6
African American	5.2	9.6	7.8	7.5	5.2	6.2
White	16.9	14.9	14.6	15.5	12.8	15.1
Other	3.2	2.0	3.8	2.9	2.1	3.0
Grade, %
3	40.0	35.3	NA	26.9	36.6	NA
4	22.1	19.6	43.4	27.3	20.7	35.7
5	9.2	8.7	9.2	9.0	9.1	12.2
6	16.3	22.0	22.3	20.1	20.2	28.6
7	12.4	14.5	25.0	16.8	13.5	23.6
BMI z score, mean (SD)c	0.57 (1.16)	0.65 (1.14)	NA	0.61 (1.15)	1.17 (0.43)	0.61 (1.15)
BMI ≥85th percentile, %c	38.5	40.6	NA	39.6	100	40.1
Perceived weight status, %d
Underweight	21.7	21.5	22.5	22.0	12.2	21.8
About the right weight	52.9	51.4	50.6	51.5	39.9	51.7
Somewhat overweight	20.1	22.2	21.0	21.1	38.3	21.4
Very overweight	5.4	4.9	5.9	5.4	9.6	5.1
School district, %
1 Northern California	25.4	9.8	17.9	17.5	20.0	21.4
2 Central California	17.7	20.8	29.3	22.2	20.9	21.4
3 Southern California	13.2	24.1	2.7	14.2	18.1	11.6
4 Southern California	18.3	29.9	30.1	25.9	23.8	22.2
5 Southern California	25.4	15.4	20.0	20.2	17.2	23.5
School-level FRPM eligibility, %	64.4	72.5	68.3	68.4	69.5	67.6

Abbreviations: BMI, body mass index; FRPM, free and reduced-price meal; NA, not applicable.

^aComplete cases for weight status: students in groups 1 and 2 with valid baseline BMI in the 85th percentile or higher and 1 valid follow-up BMI.

^bComplete cases for weight stigmatization: students in grades 4 to 7 at baseline with complete survey data at baseline and at 1 follow-up period.

^cValues restricted to students with a valid baseline BMI (n = 8458); BMI was not assessed in group 3.

^dValues restricted to students in grades 4 to 7 who completed surveys; grade 3 students did not complete surveys.

BMI z Score

Among 6534 of 16 622 complete case students with a baseline BMI in the 85th percentile or higher (39.3%), there was no difference in change in BMI z score between group 1 and group 2 students after 1 year (−0.003; 95% CI, −0.02 to 0.01; P = .71) or 2 years (0.01; 95% CI, −0.02 to 0.03; P = .62) of BMI reporting. Grade level did not modify the effect of the intervention, nor did race/ethnicity, although there was a suggestion that BMI z score increased more among Hispanic youth receiving a BMI report than among non-Hispanic youth (Table 2). Multiple imputation models and models exploring additional categories of race/ethnicity yielded similar results (eTable 2 in Supplement 1). In exploratory models stratified on baseline weight status (eTable 1 in Supplement 1), among students with a baseline BMI less than the 5th percentile, BMI z scores increased more among students whose parents received a BMI report (group 1) than among those in group 2 after 1 year (0.08; 95% CI, 0.01-0.15) and 2 years (0.18; 95% CI, 0.08-0.27). Among students with a BMI at or above the 5th percentile but less than the 85th percentile, BMI z scores decreased more in group 1 students than group 2 students after 1 year (−0.02; 95% CI, −0.04 to −0.01) but increased more after 2 years (0.03; 95% CI, 0.00-0.05).

Table 2. Adjusted BMI z Scores Among Students With a Baseline BMI in the 85th Percentile or Higher (n = 6534), by Group^a.

Variable	Mean (SE) z score			z Score 1-y change (95% CI)			z Score 2-y change (95% CI)
	Baseline	1 y	2 y	Within-group	Between-groupb	P valuec	Within-group	Between-groupb	P valuec
All students
BMI reporting	1.74 (0.01)	1.69 (0.01)	1.69 (0.02)	−0.05 (−0.08 to −0.02)	−0.003 (−0.02 to 0.01)	.71	−0.05 (−0.11 to 0.00)	0.01 (−0.02 to 0.03)	.62
BMI screening	1.75 (0.01)	1.70 (0.01)	1.69 (0.02)	−0.05 (−0.07 to −0.02)	NA	NA	−0.06 (−0.12 to −0.00)	NA	NA
Hispanic students
BMI reporting	1.77 (0.01)	1.73 (0.01)	1.73 (0.02)	−0.04 (−0.07 to −0.01)	0.003 (−0.01 to 0.02)	.70	−0.04 (−0.09 to 0.02)	0.02 (−0.00 to 0.05)	.07
BMI screening	1.78 (0.01)	1.73 (0.01)	1.72 (0.02)	−0.05 (−0.07 to −0.02)	NA	NA	−0.06 (−0.12 to −0.00)	NA	NA
Non-Hispanic students
BMI reporting	1.67 (0.02)	1.61 (0.02)	1.58 (0.02)	−0.06 (−0.09 to −0.03)	−0.01 (−0.04 to 0.01)	.25	−0.09 (−0.15 to −0.03)	−0.03 (−0.07 to 0.01)	.12
BMI screening	1.69 (0.02)	1.64 (0.02)	1.62 (0.03)	−0.05 (−0.08 to −0.01)	NA	NA	−0.06 (−0.12 to −0.00)	NA	NA
Elementary school studentsd
BMI reporting	1.76 (0.01)	1.71 (0.01)	1.69 (0.02)	−0.05 (−0.08 to −0.03)	−0.01 (−0.03 to 0.01)	.32	−0.07 (−0.12 to −0.02)	−0.01 (−0.03 to 0.02)	.70
BMI screening	1.77 (0.01)	1.72 (0.01)	1.70 (0.02)	−0.05 (−0.07 to −0.02)	NA	NA	−0.07 (−0.12 to −0.02)	NA	NA
Non–elementary school students
BMI reporting	1.73 (0.02)	1.66 (0.02)	1.61 (0.03)	−0.08 (−0.11 to −0.06)	0.01 (−0.01 to 0.03)	.43	−0.12 (−0.18 to −0.07)	0.02 (−0.02 to 0.06)	.24
BMI screening	1.74 (0.02)	1.65 (0.02)	1.60 (0.02)	−0.09 (−0.12 to −0.06)	NA	NA	−0.15 (−0.20 to −0.09)	NA	NA

Abbreviations: BMI, body mass index; NA, not applicable.

^aAnalyses adjusted for sex, race/ethnicity (except for models stratified by Hispanic students), grade (except for models stratified by grade level), school district, school-level percentage of students eligible for free and reduced-price meal, and calendar year.

^bBetween-group difference: BMI reporting group minus BMI screening group.

^cP value for between-group difference.

^dElementary: in grades 3 to 5 at baseline.

Adverse Outcomes of Assessing BMI in Schools

Weight satisfaction declined more in students weighed at school (groups 1 and 2) than in controls after 2 years of BMI screening (−0.11; 95% CI, −0.18 to −0.05), and the frequency of peer weight talk increased more after 1 year (0.05; 95% CI, 0.01-0.09); however, concerning weight control behaviors declined more after 1 year (−0.06; 95% CI, −0.10 to −0.02) (Table 3). Students’ perceived weight status did not modify the effect of BMI screening on peer-related outcomes.

Table 3. Adjusted Weight Stigmatization Outcomes for 14 318 Students in Grades 4 to 7 With Complete Surveys, by Group^a.

Variable	Mean (SE)			Value (95% CI)
	Baseline	1 y	2 y	1-y Change			2-y Change
				Within-group	Between-group	P valueb	Within-group	Between-group	P valueb
Child and peer outcomesc
Peer weight teasing index, range 1-5
BMI reporting and BMI screening	1.73 (0.03)	1.68 (0.02)	1.65 (0.03)	−0.06 (−0.10 to −0.02)	0.01 (−0.02 to 0.04)c	.67	−0.08 (−0.15 to −0.01)	−0.02 (−0.07 to 0.03)c	.35
Control	1.76 (0.03)	1.70 (0.03)	1.70 (0.04)	−0.06 (−0.11 to −0.02)	NA	NA	−0.06 (−0.13 to 0.01)	NA	NA
Peer weight talk, range 1-5
BMI reporting and BMI screening	1.67 (0.02)	1.70 (0.01)	1.67 (0.03)	0.03 (−0.01 to 0.07)	0.05 (0.01 to 0.09)c	.007	−0.00 (−0.08 to 0.07)	−0.00 (−0.07 to 0.06)c	.99
Control	1.69 (0.02)	1.67 (0.02)	1.69 (0.03)	−0.03 (−0.07 to 0.02)	NA	NA	−0.00 (−0.08 to 0.08)	NA	NA
Teacher weight talk, range 1-5
BMI reporting and BMI screening	1.14 (0.01)	1.10 (0.01)	1.06 (0.02)	−0.04 (−0.06 to −0.01)	−0.01 (−0.03 to 0.02)c	.53	−0.07 (−0.11 to −0.03)	−0.00 (−0.04 to 0.03)c	.82
Control	1.13 (0.01)	1.11 (0.01)	1.07 (0.02)	−0.03 (−0.06 to −0.00)	NA	NA	−0.07 (−0.11 to −0.02)	NA	NA
Weight satisfaction, range 1-5
BMI reporting and BMI screening	3.43 (0.02)	3.45 (0.02)	3.37 (0.03)	0.02 (−0.03 to 0.06)	−0.03 (−0.07 to 0.01)c	.13	−0.06 (−0.15 to 0.02)	−0.11 (−0.18 to −0.05)c	.001
Control	3.41 (0.02)	3.46 (0.02)	3.46 (0.04)	0.05 (−0.00 to 0.10)	NA	NA	0.05 (−0.04 to 0.14)	NA	NA
Weight control behaviors index, range 0-3
BMI reporting and BMI screening	1.14 (0.02)	1.13 (0.01)	1.12 (0.03)	−0.01 (−0.05 to 0.03)	−0.06 (−0.10 to −0.02)c	.001	−0.02 (−0.10 to 0.06)	−0.04 (−0.10 to 0.02)c	.19
Control	1.13 (0.02)	1.18 (0.02)	1.15 (0.04)	0.05 (0.00 to 0.10)	NA	NA	0.02 (−0.06 to 0.11)	NA	NA
Family outcomesc
Family weight teasing, range 1-5
BMI reporting	1.32 (0.02)	1.28 (0.02)	1.28 (0.03)	−0.03 (−0.07 to 0.01)	−0.01 (−0.04 to 0.03)d	.73	−0.04 (−0.11 to 0.03)	−0.01 (−0.06 to 0.04)d	.59
BMI screening and control	1.33 (0.01)	1.30 (0.01)	1.30 (0.02)	−0.03 (−0.06 to 0.01)	NA	NA	−0.03 (−0.09 to 0.03)	NA	NA

Abbreviations: BMI, body mass index; NA, not applicable.

^aAnalyses adjusted for baseline perceived weight status, sex, race/ethnicity, grade, school district, school-level percentage of students eligible for free and reduced-price meal, and calendar year. For all outcomes except weight satisfaction, higher scores reflect poorer outcomes (increased teasing, talk, or concerning weight control behaviors).

^bP value for between-group difference.

^cBetween-group difference: BMI reporting group and BMI screening group combined minus control group.

^dBetween-group difference: BMI reporting group minus BMI screening group and control group combined.

Adverse Outcomes of Sending BMI Reports

Students’ perceived weight status modified the effect of BMI reporting on all family-related outcomes except teasing (Table 3). Among students who considered themselves very overweight, family encouraging the child to diet increased more after 2 years in the BMI reporting group (group 1) than in the screening-only group (group 2) (0.44; 95% CI, 0.06-0.82), although family talk decreased more (−0.24; 95% CI, −0.47 to −0.00) (eTable 3 in Supplement 1).

Discussion

Forty percent of US children³³ live in states where schools are required or known to send BMI reports to parents.^2,34 In this first cluster randomized clinical trial to date of BMI screening and reporting in the United States, we find that BMI reports do not improve student weight status, nor are there impacts among subgroups based on race/ethnicity or grade. With respect to the effect of BMI reports on adverse outcomes, the results were mixed. For example, weight satisfaction decreased and peer weight talk increased among students who were weighed at school, but unhealthy weight control behaviors also declined. Overall, the findings of the present study suggest that the use of BMI reports alone does not improve children’s weight status and may decrease weight satisfaction and increase peer weight talk.

With respect to null effects on weight status, the findings from this cluster randomized clinical trial are consistent with those of previous research. The only previous randomized trial of BMI reporting, conducted among more than 2700 elementary school children in Mexico, found no decrease in weight or BMI over a follow-up period of 5 months.³⁵ We found that BMI reporting had no impact after up to 2 years of follow-up, supporting the findings from the other trial. A study³⁶ using a regression discontinuity design found that BMI reporting had no effect among students in grades kindergarten to 11 in New York City. However, those BMI reports used the term obese to describe children with a BMI in the 95th percentile or higher, which some parents find perjorative.³⁷ To maximize parents’ receptivity to BMI reporting, the Fit Study BMI report was crafted with attention to format and language based on feedback from diverse parents.²⁰ In addition, the report was sent in English and either Spanish or Chinese (depending on each school’s routine practices) to ensure cultural sensitivity and comprehension, and it included color-coded BMI results and an infographic for easier interpretation. Therefore, further enhancements to BMI reports would be unlikely to improve their effect on weight status.

Although BMI reports have widespread reach, the magnitude of the intervention is small and may not be sufficiently salient for parents. Most Fit Study BMI reports were delivered as intended over 2 years (<2% of reports were returned to sender); however, among a random sample of parents, recall of the report was 54% after receiving 1 report and 70% after receiving 2 reports,³⁸ consistent with previous studies.^6,7,39,40 In addition, parents of children in the Fit Study with a BMI in the 85th percentile or higher had poorer recall of report results than other parents, and only 1 in 5 parents were surprised by the results, suggesting that BMI reports alone are not sufficient to change parents’ behaviors.³⁸ Increasing evidence suggests that interventions to reduce obesity should focus on changing social, socioeconomic, and built environment factors, all of which play major roles in the development of obesity.^41,42

Measuring student BMI at school is controversial,^43,44 and numerous experts have raised concerns that school-based BMI screening may have unintended negative consequences.^16,45,46 Massachusetts, which instituted BMI reporting in 2009, stopped sending BMI reports in 2014 because of concerns about stigmatization.⁴⁷ The weight stigmatization results in the present study, the first to date based on a cluster randomized clinical trial design, were inconclusive. Weight satisfaction decreased and peer weight talk increased among students weighed at school, both of which are associated with the development of disordered eating behaviors.¹⁹ However, there were also protective effects because concerning weight control behaviors declined among youth exposed to BMI screening compared with controls. Furthermore, no weight stigmatization results were consistent across both 1 and 2 years of reporting, making it difficult to draw definitive conclusions about adverse outcomes. The findings of the present study are consistent with those of a study⁴⁸ conducted in Arkansas that found no difference in the prevalence of disordered eating behaviors among high school students exposed to BMI screening and reporting and historical controls. Few studies^49,50 have asked students directly about their experience with BMI screening: although most students did not express concerns, a small minority were uncomfortable with the process. Similarly, adverse outcomes of sending BMI reports to parents were mixed in the present study, with families encouraging their overweight children to diet after receiving 2 consecutive reports, a problematic response because dieting in adolescents has been associated with weight gain.^17,18 Conversely, family weight talk, which is associated with body dissatisfaction and unhealthy weight control behaviors,¹³ declined after 2 years.

Health screenings, in which every individual is assessed, are recommended when an effective intervention exists for those identified as being at risk.⁵¹ The findings of the present study demonstrate that sending BMI reports to families of children in grades 3 to 8 does not improve weight status; therefore, we recommend that BMI screening should not be done for the purpose of sending reports to families unless effective interventions can be identified and made available. With respect to screening student BMI in schools without sending BMI reports, many researchers and advocates rely on BMI data to study the impact and cost-effectiveness of obesity prevention efforts.^52,53 However, given that we documented an increase in peer weight talk and a decrease in weight satisfaction as a result of school-based BMI screening, we believe that every attempt should be made to identify other sources for BMI data. There is rapid movement toward centralizing electronic health record (EHR) data in California. For example, the University of California has pooled EHRs from 5 medical centers, making deidentified data on more than 15 million patients available to University of California researchers,⁵⁴ and Kaiser Permanente has EHRs for 9 million Californians. With such coverage, it is worth exploring the use of EHR data for BMI surveillance in California; however, we recognize that other states may not be situated to move as quickly in this direction.

Strengths and Limitations

To our knowledge, the well-powered Fit Study was the first investigation to test for differences in the effect of BMI reporting by race/ethnicity and age group. An additional strength of the Fit Study was the large number of youths enrolled for all 3 years of the investigation, which permitted assessment of the cumulative effect of 2 annual BMI reports.

This trial has limitations. It included only California public schools (serving 1 in 8 of the country’s youths); therefore, the results may not generalize to other states. The small proportion of African American students could also limit generalizability. Among students in grades 5 to 7 with a BMI in the 85th percentile or higher, more control students than intervention students were missing follow-up data, which could have biased the results in either direction, although the results from multiple imputation models showed null effects. Student surveys were administered approximately 6 months after BMI assessments, which may not have captured immediate harmful effects; however, we were interested in persistent adverse effects. Finally, the Fit Study did not explore the effects of BMI reporting among children in grade 2 or younger or over periods longer than 2 years. It is possible that BMI reporting could have had a different effect with those additional factors.

Conclusions

Body mass index reporting is widely used in an effort to reduce pediatric obesity. It is important to convey to national stakeholders who recognize the importance of schools in addressing population health—including researchers, practitioners, school districts, and education and health departments—that BMI reports alone do not improve student weight status and that resources should be directed toward comprehensive evidence-based interventions.

Supplement 1.

eAppendix. Supplemental Appendix

eTable 1. Adjusted BMI z-Scores, by Students’ Baseline Weight Category and Group

eTable 2. Adjusted BMI z-Scores by Study Group, Among Students With a Baseline BMI ≥85th Percentile for Sex and Age in Grades 3-7 Based on Estimation Using Multiply Imputed Datasets (n = 7672)

eTable 3. Adjusted Weight Stigmatization Outcomes in Grades 4-8, Stratified by Baseline Perceived Weight Status (n = 14318)

eFigure 1. Study Groups Constituting the Exposed/Intervention and Control Groups for Each Study Outcome

eFigure 2. Longitudinal Data Collection

eFigure 3. Participant Flow Among Eligible Students (N = 30542) in Grades 3-7 in 79 California Schools

eFigure 4. BMI Report

eFigure 5. Participant Flow Among Students With a Baseline BMI ≥85th Percentile for Age and Sex

eReferences.

Supplement 2.

Trial Protocol

Supplement 3.

Data Sharing Statement