Fitness Trends and Disparities Among School-Aged Children in Georgia, 2011-2014

Abstract

Objectives:

Although FitnessGram fitness data on aerobic capacity and body mass index (BMI) have been collected in public schools in Georgia since the 2011-2012 school year, the data have not been analyzed. The primary objective of our study was to use these data to assess changes in fitness among school-aged children in Georgia between 2011 and 2014. A secondary objective was to determine if student fitness differed by school size and socioeconomic characteristics.

Methods:

FitnessGram classifies fitness into the Healthy Fitness Zone (HFZ) or not within the HFZ for aerobic capacity and BMI. We used data for 3 successive school years (ie, 2011-2012 to 2013-2014) obtained from FitnessGram testing of students in >1600 schools. We calculated the percentage of students who achieved the HFZ for aerobic capacity and BMI. We used growth curve models to estimate the annual changes in these proportions, and we determined the effect of school size and socioeconomic status on these changes.

Results:

Both elementary school boys (β = 1.31%, standard error [SE] = 0.23%, P < .001) and girls (β = 1.53%, SE = 0.26%, P < .001) had significant annual increases in achievement of HFZ for aerobic capacity. Elementary school boys (β = 3.11%, SE = 0.32%, P < .001) and girls (β = 3.09%, SE = 0.32%, P < .001) also had significant increases in their BMI HFZ achievement proportions, although these increases occurred primarily between 2011-2012 and 2012-2013. Body mass index HFZ achievement proportions were mixed for middle school students, and we did not observe increases for high school students. Larger school size and higher school socioeconomic status were associated with better aerobic capacity and BMI fitness profiles.

Conclusions:

Surveillance results such as these may help inform the process of designing state and local school-based fitness promotion and public health programs and tracking the results of those programs.

Addressing the high prevalence of obesity in children and adolescents is a high priority for public health educators and policy makers. Surveillance systems are critically important to this effort because they provide up-to-date information about patterns and trends in obesity-related behaviors, and they enable planning for and evaluation of programs and policies aimed at improving health profiles, particularly in at-risk populations.¹ Since the 1970s, the National Health and Nutrition Examination Survey program has collected nationally representative data on physical activity and the prevalence of obesity among children and adolescents.² Although the data from these programs have contributed to national public health efforts, more comprehensive data from local and state sources are needed to facilitate program planning and evaluation. In particular, local or regional surveillance systems that can provide comprehensive data on fitness profiles (including obesity) across all age groups are needed.³

Although standardized physical fitness testing has been a component of some school physical education programs for years, increased emphasis has been placed on school-based fitness assessments in recent years. A promising tool for examining local- and state-level fitness patterns among school-aged children is the FitnessGram youth fitness assessment system. The FitnessGram field testing battery assesses 6 aspects of health-related fitness: (1) aerobic capacity, (2) body mass index (BMI), (3) abdominal strength and endurance, (4) trunk extensor strength and flexibility, (5) upper-body strength and endurance, and (6) flexibility. Details on the battery are provided elsewhere.⁴ FitnessGram uses established standards to evaluate fitness performance. The test results are classified into 2 or 3 general categories depending on the assessment: the categories, Healthy Fitness Zone (HFZ) and Needs Improvement Zone, are used for musculoskeletal and flexibility fitness tests, and the Needs Improvement Zone and Needs Improvement–Health Risk Zone are used for aerobic capacity and BMI. This system has already been adopted by several states, thereby creating the potential for pooling school-based data for surveillance and research.^5–7

FitnessGram can systematically capture school-based data on a large scale. This promises to open up new opportunities for evaluating the effect of environmental factors and school programming on fitness profiles of school-aged children.^8,9 We recently established standardized processing methods for school-based FitnessGram data and reported on fitness patterns observed in a large sample of schools involved in the NFL PLAY 60 FitnessGram Partnership Project, a research project that promotes physical activity and healthy eating programs among school-aged children at >1000 US schools.^10,11 Based on our experience, we aimed to determine if the application of similar methods to a larger statewide sample could help advance the understanding of patterns and trends in school-aged fitness at the state level.

The opportunity to use statewide FitnessGram data grew out of a decision by the Georgia state legislature to pass the Student Health and Physical Education (SHAPE) Act in 2009.¹² The Georgia Shape initiative built on the success of that act, bringing together governmental, philanthropic, academic, and business partners to tackle childhood obesity in Georgia.¹³ As part of this initiative, FitnessGram data have now been systematically collected since the 2011-2012 school year; however, no analysis of the data has been conducted to date.

The primary objective of our study was to use statewide fitness testing data to characterize changes in 2 school-aged fitness indicators, aerobic capacity and BMI, in Georgia between 2011 and 2014. A secondary objective was to determine if student fitness differed by school size and socioeconomic characteristics.

Methods

Study Population

We obtained de-identified data from the FitnessGram database for elementary, middle, and high school students who were part of the Georgia Shape initiative. Records from 1 194 212 students in the 2011-2012, 1 582 295 students in the 2012-2013, and 1 517 217 students in the 2013-2014 school years with FitnessGram assessments were submitted by school teachers to the FitnessGram database server and made available to us for analysis with permission from the Georgia Shape initiative. This study was considered exempt by the Iowa State University Institutional Review Board because of the de-identified nature of the data.

Outcome Measures

Although students participating in the Georgia Shape initiative took all 6 FitnessGram tests, our study used the results of only 2 tests, aerobic capacity and BMI. In Georgia, physical education teachers received training in FitnessGram testing protocols and data management, and participating schools were given equipment, software, and technical support.

Teachers measured weight and height and calculated BMI using the standard formula, BMI = weight (kg) / height (m²).⁴ For aerobic capacity assessment, schools could choose between 2 field tests: (1) a Progressive Aerobic Cardiovascular Endurance Run, in which students run 20-meter shuttles with an increasing pace each minute until the pace can no longer be maintained, after which the number of laps completed is recorded; or (2) recording the time for a standard 1-mile run. Using the field test scores, VO₂ max, which reflects aerobic capacity, was then estimated for each student from equations developed for the Progressive Aerobic Cardiovascular Endurance Run or 1-mile run.⁴

School Demographic Measures

We obtained data on student eligibility for free and/or reduced-price lunch and school enrollment for all 3 school years through a public database posted on the Georgia Department of Education website (https://oraapp.doe.k12.ga.us/ows-bin/owa/fte_pack_enrollgrade.entry_form; https://oraapp.doe.k12.ga.us/ows-bin/owa/fte_pack_frl001_public.entry_form). We identified each school’s socioeconomic status level by using the percentage of students eligible for free and/or reduced-price lunch.

Data Processing

We processed data using the FitnessGram criterion-referenced fitness standards, which are based on health risks associated with the metabolic syndrome.^14,15 Student test results for aerobic capacity and BMI were placed into 1 of 3 categories, based on how their scores compared with these HFZ standards: (1) HFZ (sufficient fitness for good health), (2) Needs Improvement Zone (potential for future health risks), or (3) Needs Improvement–Health Risk Zone (potential for future health problems).

Because the HFZ standards for BMI approximated the Centers for Disease Control and Prevention (CDC) BMI growth chart categories, the FitnessGram program formally adopted the CDC values as part of its standards.¹⁶ Thus, the HFZ for BMI is equivalent to the CDC BMI category for normal weight (BMI 5th to <85th percentile), the Needs Improvement Zone is equivalent to the CDC BMI category for overweight (BMI 85th to <95th percentile), and the Needs Improvement–Health Risk Zone is equivalent to the CDC BMI category for obese (BMI ≥95th percentile).

We determined the proportion of school-aged children achieving the HFZ for aerobic capacity and BMI. At each school and for each sex, we calculated zone achievement proportions for each grade-level observation using the following formula:

$$\text{\% HFZ achievement = Number of students in the HFZ per grade,divided by total number of students taking FitnessGram test per grade}$$

We then screened data using established methods to determine whether the grade-level sample was sufficiently representative for inclusion in the final sample.¹⁷ After screening, we excluded grade-level data with <15 students per grade-level observation or a boy-to-girl ratio of <0.5 or >2.0. We then merged the screened HFZ data with the corresponding school demographic data. We also excluded data from any schools missing ≥1 year of fitness or demographic data. Finally, to streamline the interpretation of longitudinal patterns of change, we apportioned the data into 3 distinct school levels: elementary school (first through fifth grades), middle school (sixth through eighth grades), and high school (ninth through 12th grades).

Statistical Methods

We present the descriptive data as proportion means and standard deviations (SDs). We used descriptive statistics to summarize, for each school year, the relative distributions of grade-level observations in each FitnessGram zone for aerobic capacity and BMI, stratified by sex and school level.

We estimated the average annual rate of change in HFZ achievement for aerobic capacity and BMI using growth curve models. We considered 2 key demographic indicators (socioeconomic status and school enrollment) as fixed effects because they were similar throughout the study period. To facilitate interpretation, we centered these indicators by subtracting the mean from every value of the variable before entering the result into the statistical model. We defined school as the sole random effect in the model. The final model combined fixed- and random-effects models as follows:

$$Y_{ij}={\Upsilon }_{00}+{\Upsilon }_{\text{1}0}\left(\text{Year}\right)+{\Upsilon }_{\text{2}0}{\left(\text{Year}\right)}^{\text{2}}+{\mu }_{0j}+{\mu }_{\text{1}j}\left(\text{Year}\right)+r_{ij}$$

where

• Y_ij is the expected mean aerobic capacity (or BMI) HFZ at year i for school j,

• ϒ₀₀ is the fixed intercept and represents the mean aerobic capacity (or BMI) HFZ at year 1,

• μ_0j is the random intercept for school j,

• ϒ₁₀ is the fixed slope for year, estimating the linear rate of change in mean aerobic capacity (or BMI) HFZ achievement,

• μ_1j is the random linear slope for school j,

• ϒ₂₀ is the fixed slope for quadratic term year, estimating the acceleration in the change of mean aerobic capacity (or BMI) HFZ achievement, and

• r_ij is the residual term for school j at year i.

If the quadratic term of year in the regression model was not significant in the multivariate regression model, we excluded it from the baseline model. In subsequent growth curve models, we added socioeconomic status and school enrollment, respectively, as fixed covariates predicting the intercept of the baseline HFZ achievement (ϒ₀₀) for aerobic capacity and BMI. We also added socioeconomic status and school enrollment to predict the slopes of the HFZ trajectories (ϒ₁₀) for aerobic capacity and BMI.

We used the Akaike information criterion to compare model fit, and we retained the model with the smallest Akaike information criterion. For some models in which we could not estimate the variance of year, we could not treat year as a random effect; as such, we used regular multiple regression models. We defined significance at P < .05, and we analyzed data using SAS version 9.4.¹⁸

Results

We included 1627 schools that submitted fitness data for all 3 school years in the final sample for aerobic capacity, resulting in 6318 grade-level observations (3159 for boys and 3159 for girls). Because the number of students enrolled and tested differed each year, the number of students in the aerobic data sample totaled 393 122 in 2012, 364 742 in 2013, and 356 611 in 2014. The sample for BMI was larger than the sample for aerobic capacity and consisted of 1707 schools and 11 980 grade-level observations (5990 for boys and 5990 for girls). The BMI data sample included 721 687 students in 2012, 692 928 in 2013, and 683 131 in 2014.

HFZ Achievement

For aerobic capacity, the mean HFZ achievement proportions in all grade-level observations at any point in time ranged from 45% to 63% for boys and from 32% to 51% for girls (Table 1). The mean Needs Improvement–Health Risk Zone achievement proportions ranged from 15% to 43% for boys and from 19% to 52% for girls. Aerobic capacity fitness levels generally declined with higher school levels.

Table 1.

Achievement of FitnessGram Healthy Fitness Zone (HFZ)^a for aerobic capacity among 6318 school grade-level observations,^b by sex and school level, Georgia, 2011-2014 school years

		Achievement of FitnessGram HFZ for Aerobic Capacity, % Mean (SD)
		Boys			Girls
Fitness Zonea and School Levelc	No. of Grade-Level Observationsd	2011-2012	2012-2013	2013-2014	2011-2012	2012-2013	2013-2014
HFZ
Elementary	1857	60.8 (21.9)	62.4 (21.2)	63.4 (21.4)	47.7 (24.5)	49.2 (24.3)	50.8 (24.4)
Middle	966	57.6 (19.3)	57.2 (20.0)	57.5 (20.4)	37.7 (21.6)	38.2 (22.6)	39.5 (22.8)
High	336	47.1 (20.2)	48.0 (20.0)	45.2 (20.9)	31.7 (20.4)	33.0 (21.4)	32.2 (21.4)
Needs Improvement Zone
Elementary	1857	22.1 (12.7)	21.9 (13.4)	21.4 (12.7)	31.3 (15.8)	31.0 (16.1)	30.5 (15.8)
Middle	966	14.4 (7.5)	14.8 (8.4)	14.8 (9.0)	26.0 (12.3)	26.8 (13.5)	26.5 (13.5)
High	336	12.4 (6.0)	12.3 (6.9)	11.9 (7.1)	16.6 (9.7)	16.8 (8.7)	15.9 (9.4)
Needs Improvement–Health Risk Zone
Elementary	1857	17.1 (16.6)	15.6 (15.6)	15.2 (15.4)	21.0 (19.9)	19.7 (19.7)	18.7 (19.2)
Middle	966	28.0 (17.5)	28.0 (18.1)	27.7 (18.7)	36.3 (23.8)	35.0 (24.0)	34.0 (24.5)
High	336	40.5 (19.8)	39.6 (19.9)	42.9 (22.0)	51.8 (24.0)	50.2 (23.9)	51.9 (24.1)

^aBased on FitnessGram⁴ criterion-referenced standards for health fitness: HFZ (sufficient fitness for good health), Needs Improvement Zone (potential for future health risks), Needs Improvement–Health Risk Zone (potential for future health problems).

^bFrom FitnessGram database for elementary, middle, and high school students in Georgia who were part of the Georgia Shape fitness initiative.¹³

^cSchool levels: elementary (first through fifth grade), middle (sixth through eighth grade), high (ninth through 12th grade).

^dNumber of grade-level observations within each school level applies equally to boys and girls and was the same for all 3 years.

For BMI, HFZ achievement proportions in all grade-level observations at any point in time ranged from 57% to 61% for boys and from 56% to 63% for girls (Table 2). Mean Needs Improvement–Health Risk Zone achievement proportions ranged from 20% to 25% for boys and from 18% to 24% for girls. Body mass index fitness levels did not decline with higher school levels.

Table 2.

Achievement of FitnessGram Healthy Fitness Zone (HFZ)^a for BMI among 11 980 grade-level observations,^b by sex and school level, Georgia, 2011-2014

		Achievement of FitnessGram HFZ for BMI,c % Mean (SD)
		Boys			Girls
Fitness Zonea and School Leveld	No. of Grade-Level Observationse	2011-2012	2012-2013	2013-2014	2011-2012	2012-2013	2013-2014
HFZ
Elementary	4668	56.8 (11.4)	58.9 (11.6)	58.9 (11.4)	56.7 (11.6)	58.7 (11.8)	58.6 (11.6)
Middle	973	56.6 (9.5)	57.7 (10.4)	57.6 (10.1)	56.0 (11.4)	56.2 (12.5)	56.3 (11.8)
High	349	59.8 (9.3)	60.9 (10.0)	60.7 (10.5)	63.1 (11.8)	62.3 (12.6)	61.9 (13.0)
Needs Improvement Zone
Elementary	4668	18.5 (6.4)	17.8 (6.6)	17.9 (6.5)	18.9 (6.6)	18.4 (6.6)	18.5 (6.7)
Middle	973	19.0 (5.5)	18.5 (6.2)	18.5 (6.0)	20.7 (6.3)	20.6 (7.1)	20.6 (6.5)
High	349	19.4 (6.7)	19.3 (6.8)	18.6 (7.1)	18.6 (6.8)	19.6 (7.2)	19.8 (7.4)
Needs Improvement–Health Risk Zone
Elementary	4668	24.7 (9.6)	23.3 (9.6)	23.3 (9.3)	24.4 (9.7)	22.9 (9.9)	22.9 (9.6)
Middle	973	24.4 (8.1)	23.8 (9.0)	23.9 (8.7)	23.4 (9.7)	23.2 (10.3)	23.1 (10.1)
High	349	20.8 (7.6)	19.8 (8.2)	20.7 (8.5)	18.3 (9.1)	18.2 (9.8)	18.3 (9.6)

Abbreviation: BMI, body mass index.

^aBased on FitnessGram⁴ criterion-referenced standards for health fitness: HFZ (sufficient fitness for good health), Needs Improvement Zone (potential for future health risks), and Needs Improvement–Health Risk Zone (potential for future health problems).

^bFrom FitnessGram database for elementary, middle, and high school students in Georgia who were part of the Georgia Shape fitness initiative.¹³

^cHFZ for BMI corresponds with Centers for Disease Control and Prevention (CDC) BMI normal weight category (BMI 5th to <85th percentile), Needs Improvement Zone corresponds with CDC BMI overweight category (BMI 85th to <95th percentile), and Needs Improvement–Health Risk Zone corresponds to CDC BMI obese category (BMI ≥95th percentile).

^dSchool levels: elementary (first through fifth grade), middle (sixth through eighth grade), high (ninth through 12th grade).

^eNumber of grade-level observations within each school level applies equally to boys and girls and was the same for all 3 years.

HFZ Trends

In longitudinal modeling for aerobic capacity, none of the final models had significant quadratic terms for the year variable, indicating linear annual changes in aerobic capacity HFZ achievement proportions at all school levels. Both elementary school boys (β = 1.31%, standard error [SE] = 0.23%, P < .001) and girls (β = 1.53%, SE = 0.26%, P < .001) had significant annual increases in their HFZ achievement rates for aerobic capacity. Healthy Fitness Zone achievement for aerobic capacity during the 3-year period did not change significantly in middle school boys (β = 0.08%, SE = 0.30%, P = .79) and declined slightly in high school boys (β = –0.43%, SE = 0.38%, P = .27). Conversely, the increase in HFZ achievement for aerobic capacity was significant in middle school girls (β = 0.95%, SE = 0.30%, P = .002) but not in high school girls (β = 0.50%, SE = 0.36%, P = .16) (Table 3).

Table 3.

Trends^a in achievement of aerobic capacity Healthy Fitness Zone (HFZ)^b among 6318 grade-level observations,^c by sex and school level,^d Georgia, 2011-2014

	Changes in HFZ Achievement for Aerobic Capacity
	Boys			Girls
Variable	Elementary Schoold	Middle Schoold	High Schoold	Elementary Schoold	Middle Schoold	High Schoold
Fixed effects, % (SE) [P value]
Intercept	62.08 (0.50) [<.001]e	56.86 (0.53) [<.001]e	48.86 (0.68) [<.001]e	49.10 (0.56) [<.001]e	37.00 (0.55) [<.001]e	29.58 (0.72) [<.001]e
Level 1f
Yearg	1.31 (0.23) [<.001]e	0.08 (0.30) [.793]	–0.43 (0.38) [.266]	1.53 (0.26) [<.001]e	0.95 (0.30) [.002]e	0.50 (0.36) [.157]
Level 2h
Socioeconomic status	–0.30 (0.01) [<.001]e	–0.32 (0.02) [<.001]e	–0.33 (0.03) [<.001]e	–0.39 (0.02) [<.001]e	–0.46 (0.02) [<.001]e	–0.39 (0.02) [<.001]e
School enrollment	0.49 (0.17) [.003]e	0.70 (0.15) [<.001]e		0.55 (0.19) [.003]e	0.57 (0.15 [<.001]e	0.34 (0.10) [.001]e
Year × school enrollment		–0.24 (0.08) [.004]e	0.04 (0.01) [.006]e		–0.18 (0.08) [.027]i
Year × socioeconomic status				0.02 (0.01) [.015]i
Variance components (SE) [P value]
Level 1 within personj	155.14 (5.09) [<.001]e	131.84 (6.00) [<.001]e	139.35 (7.74) [<.001]e	184.72 (6.06) [<.001]e	131.71 (5.99) [<.001]e	128.18 (7.12) [<.001]e
Level 2 interceptk	242.19 (12.19) [<.001]e	153.53 (13.00) [<.001]e	180.02 (17.67) [<.001]e	300.95 (15.76) [<.001]e	174.38 (13.84) [<.001]e	167.99 (16.37) [<.001]e
Level 2 slopel	23.45 (4.18) [<.001]e	21.32 (4.97) [<.001]e	25.16 (6.54) [<.001]e	29.38 (5.01) [<.001]e	19.37 (4.90) [<.001]e	17.91 (5.78) [.001]e

Abbreviation: SE, standard error.

^aBased on growth curve regression analysis.

^bBased on FitnessGram⁴ criterion-referenced standards for health fitness: HFZ (sufficient fitness for good health), Needs Improvement Zone (potential for future health risks), Needs Improvement–Health Risk Zone (potential for future health problems).

^cFrom FitnessGram database for elementary, middle, and high school students in Georgia who were part of the Georgia Shape fitness initiative.¹³

^dSchool levels: elementary (first through fifth grade), middle (sixth through eighth grade), high (ninth through 12th grade).

^eSignificant at P < .01.

^fVariable used to estimate the annual rate of change in aerobic capacity.

^gTime variables nested within the same observation.

^hVariables controlled for the demographic characteristics of different schools.

ⁱSignificant at P < .05.

^jResidual variance across all occasions of measurement for individuals in the population.

^kResidual variance in true intercept across all individuals in the population.

^lResidual variance in true slope across all individuals in the population.

In longitudinal modeling for BMI, models for elementary school boys and girls had significant quadratic year terms. However, none of the models for middle and high school boys and girls had significant quadratic terms for the year variable, indicating linear annual HFZ achievement changes for BMI for these 2 school levels. Among elementary school boys, the annual HFZ achievement rates for BMI increased 2.1% from 2011-2012 to 2012-2013 and stayed the same from 2012-2013 to 2013-2014. Among elementary school girls, the achievement proportion increased 2.0% from 2011-2012 to 2012-2013 and 0.2% from 2012-2013 to 2013-2014. The increase in HFZ achievement for BMI was significant in middle school boys (β = 0.50%, SE = 0.21%, P = .02) but not in middle school girls (β = 0.18%, SE = 0.23%, P = .43). The increase in achievement proportion in high school boys (β = 0.28%, SE = 0.22%, P = .21) and the decrease in achievement proportion in high school girls (β = –0.37%, SE = 0.29%, P = .19) were not significant (Table 4).

Table 4.

Trends^a in achievement of Healthy Fitness Zone (HFZ)^b for BMI among 11 980 grade-level observations,^c by sex and school level,^d Georgia, 2011-2014

	Boys			Girls
Variable	Elementary Schoold	Middle Schoold,e	High Schoold	Elementary Schoold	Middle Schoold,e	High Schoold,e
Fixed effects, % (SE) [P value]
Intercept	56.94 (0.15) [<.001]f	56.53 (0.27) [<.001]f	58.16 (0.35) [<.001]f	57.26 (0.16) [<.001]f	55.59 (0.29) [<.001]f	57.89 (0.39) [<.001]f
Level 1g
Yearh	3.11 (0.32) [<.001]f	0.50 (0.21) [.017]i	0.28 (0.22) [.210]	3.09 (0.32) [<.001]f	0.18 (0.23) [.433]	–0.37 (0.29) [.194]
Year × year	–1.04 (0.16) [<.001]f			–1.10 (0.15) [<.001]f
Level 2j
Socioeconomic status	–0.16 (0.004) [<.001]f	–0.13 (0.01) [<.001]f	–0.10 (0.01) [<.001]f	–0.20 (0.00) [<.001]f	–0.22 (0.01) [<.001]f	–0.20 (0.01) [<.001]f
School enrollment		0.33 (0.05) [<.001]f	0.24 (0.04) [<.001]f	0.34 (0.05) [<.001]f	0.46 (0.05) [<.001]f	0.49 (0.04) [<.001]f
Variance components (SE) [P value]
Level 1 within personk	76.25 (1.57) [<.001]f		64.71 (3.55) [<.001]f	75.02 (1.54) [<.001]f
Level 2 interceptl	35.15 (2.41) [<.001]f		13.30 (4.73) [.003]f	29.34 (2.28) [<.001]f
Level 2 slopem	1.45 (1.13) [.095]		0.99 (2.55) [.349]	1.39 (1.11) [.069]

Abbreviations: BMI, body mass index; SE, standard error.

^aBased on growth curve or multiple regression analyses.

^bBased on FitnessGram⁴ criterion-referenced standards for health fitness: HFZ (sufficient fitness for good health), Needs Improvement Zone (potential for future health risks), Needs Improvement–Health Risk Zone (potential for future health problems).

^cFrom FitnessGram database for elementary, middle, and high school students in Georgia who were part of the Georgia Shape fitness initiative.¹³

^dSchool levels: elementary (first through fifth grade), middle (sixth through eighth grade), high (ninth through 12th grade).

^eMultiple regression analysis used (instead of growth curve regression analysis) because year could not be treated as a random effect.

^fSignificant at P < .01.

^gTime variables nested within the same observation.

^hVariable used to estimate the annual rate of change in BMI.

ⁱSignificant at P < .05.

^jVariables controlled for the demographic characteristics of different schools.

^kResidual variance across all occasions of measurement for individuals in the population.

^lResidual variance in true intercept across all individuals in the population.

^mResidual variance in true slope across all individuals in the population.

School Socioeconomic Status and Enrollment Levels

Healthy Fitness Zone achievement proportions for aerobic capacity were associated with school socioeconomic status and enrollment (Table 3). We noted a significant association between baseline HFZ achievement proportions and school socioeconomic status at all school levels and for both boys and girls. In general, a 1% increase in the number of students eligible for free and/or reduced-price lunch was associated with a 0.30% to 0.33% lower baseline HFZ achievement proportion for aerobic capacity in boys and a 0.39% to 0.46% lower achievement proportion for aerobic capacity in girls. We also noted a significant association between HFZ achievement proportions for aerobic capacity and school enrollments at all school levels and for both boys and girls, with the exception of high school boys. An increase in school enrollment of 100 students was associated with higher HFZ achievement proportions for aerobic capacity of 0.49% in elementary school boys, 0.70% in middle school boys, 0.55% in elementary school girls, 0.57% in middle school girls, and 0.34% in high school girls.

For aerobic capacity, we also found a significant interaction between the year variable and school demographic factors, especially school enrollment, at some school levels (Table 3). For example, compared with students at schools with lower enrollment, boys at middle schools (0.08% – 0.24% = –0.16%) and boys at high schools (–0.43% + 0.04% = –0.39%) with above-average enrollment had declining HFZ achievement for aerobic capacity, whereas girls at middle schools (ie, 0.95% – 0.18% = 0.77%) with above-average enrollment had increasing achievement proportions.

Baseline HFZ achievement proportions for BMI were also significantly associated with both school socioeconomic status and enrollment at all school levels and for boys and girls (Table 4). A 1% decrease in the number of students eligible for free and/or reduced-price lunch was associated with higher baseline HFZ achievement proportions for BMI: in elementary school boys of 0.16% and girls of 0.20%, in middle school boys of 0.13% and girls of 0.22%, and in high school boys of 0.10% and girls of 0.20%. We also noted a significant association between higher baseline HFZ achievement proportions for BMI and larger school enrollment at all school levels and for both boys and girls, with the exception of elementary school boys. An increase in school enrollment of 100 students was associated with a higher baseline HFZ achievement proportion for BMI in middle school boys of 0.33%, in high school boys of 0.24%, in elementary school girls of 0.34%, in middle school girls of 0.46%, and in high school girls of 0.49%.

Discussion

Although the proportional changes in HFZ achievement for aerobic capacity and BMI among elementary school students in our study appeared small, when these changes were applied to a statewide population, the absolute number of children with favorable changes is substantial. For example, when applied to the entire population of approximately 483 000 Georgia elementary school students, an increase in the HFZ achievement proportion for BMI of 2% per year represented a shift of 9660 overweight or obese elementary school students into the HFZ for BMI. Thus, even small positive changes in the proportions of students achieving the HFZ may result in substantial improvements in the absolute number of children who attain health-related fitness. This finding suggests that public health efforts, such as statewide programs to enhance student physical education, can have a substantial effect on improving the health of a large number of school-aged children.

The reduction or plateauing in childhood obesity in our study and in the literature could be the result of prevention initiatives or a predisposition to obesity. For example, obesity prevention and health promotion programs in schools and communities through the years may have reversed the previous trend of an increasing prevalence of obesity.^19,20 Another possibility is that those children who are genetically, socially, or environmentally predisposed to obesity reach that state early on, whereas other children are more resistant to excessive weight gain. Nevertheless, whether a plateau in students under the Georgia Power Up for 30 program,²¹ which launched in elementary schools in fall 2013, occurred, why it has occurred, and whether it will continue are unclear.

Significant increases in HFZ achievement for aerobic capacity and BMI occurred primarily in elementary school students. These results are similar to those of a recent international meta-analysis of 32 studies of school-based obesity programs published between 2006 and 2012.²² Eighteen studies that focused exclusively on children aged 5-12 reported significant declines in BMI, whereas 11 studies that focused exclusively on adolescents aged 13-18 found declines in BMI that were not significant. One explanation for these findings is that younger students may be more responsive than older students to school-based physical activity promotion programs. If this finding is true, the Georgia Power Up for 30 program may be expected to yield particularly good results.²¹ Another possible explanation for our findings is that the Georgia Shape initiative and state requirements for fitness testing may have increased the awareness of and access to physical activity programs primarily in elementary schools. As of November 2015, >750 Georgia elementary schools had pledged to join this effort. However, it is unclear why there would be less awareness of, interest in, or involvement in physical activity programs among students in middle school and high school. Accordingly, additional studies examining the relative effect of physical activity programs on students in elementary, middle, and high school may be warranted.

Reports in the literature about the patterns of change in aerobic capacity among school-aged children have been less consistent than those concerning BMI. In 2007, Tomkinson and Olds reported a global secular decline in aerobic fitness since the 1970s. They reviewed 33 studies published between 1958 and 2003, which involved 25 455 527 children and adolescents aged 6-19 from 27 countries in 5 regions. They observed that aerobic performance improved from the 1950s to the 1970s but subsequently declined globally at a rate of about 5% per decade.²³ In a 2014 study of US cardiorespiratory fitness using the submaximal treadmill test, data were collected from 4 different cycles of the National Health and Nutrition Examination Survey.²⁴ This study reported substantial decreases in the level of cardiorespiratory fitness of adolescents aged 12-15 and noted that this age group was the only one that had been consistently tested in all cycles of the National Health and Nutrition Examination Survey. Of note, although the National Health and Nutrition Examination Survey data have been informative, other researchers have called attention to a quarter-century period after the mid-1980s when national youth fitness field testing was not conducted, creating a substantial gap in determining national secular aerobic fitness trends during that time.²⁵

Finally, we found disparities in the HFZ achievement proportions for both aerobic capacity and BMI between those attending large versus small schools and between those attending schools with higher versus lower socioeconomic status. Several studies have raised concerns about large disparities in fitness, including in obesity, between children of different races and socioeconomic status.^26,27 More work is needed to assess and address these disparities in fitness achievement and obesity, because the gaps seem to be widening in many age groups.²⁶

Limitations

Our study had several limitations. First, it was designed to characterize and assess changes in health-related fitness profiles among Georgia school-aged children as a result of the Georgia Shape initiative and not to determine the effect of any particular interventions. However, additional efforts to improve physical activity may have been underway at some schools during the study period, which could have affected our results. The Georgia Shape initiative established the Power Up for 30 program only for elementary school-aged children and was active only during the final year of our study, so it is not likely that this program contributed substantially to the improvements we identified. Second, the differing types of schools (magnet, charter, urban, rural), teaching formats, and faculty turnover rates could have influenced some of our results. Finally, because our study was focused on just 1 US state, and given that our sampling method differed somewhat from that used in a nationally representative survey (National Health and Nutrition Examination Survey), our results may not be generalizable to other states or the United States.

Conclusions

Additional similar standardized nationwide or state-pooled surveillance may help better characterize secular changes in fitness and provide insight into the factors contributing to disparities in fitness achievement related to socioeconomic status. Surveillance results such as ours may help inform the process of designing state and local school-based fitness and public health programs and tracking the results of those programs.