INTRODUCTION
Most US children do not meet the recommended 60 min of moderate-to-vigorous physical activity (MVPA) per day. Given that schools serve more than 95% of US children aged 5–17 yr (1), the school environment is a valuable setting to engage students in physical activity (PA). School administrators and teachers may be more likely to promote and implement school-based PA if there is a clear positive relationship between PA and academic achievement (AA), but current literature is conflicting. Several systematic reviews have identified a small positive association between PA and AA, whereas others have identified no consistent relationship (2,3). Moreover, analyses from the present study found no meaningful association between school-day MVPA and AA (4).
Beyond a direct relationship with AA, PA might have an indirect relationship through one of its many health benefits for children. Two clear health benefits of regular PA are a healthy body mass index (BMI) and cardiorespiratory fitness (CRF). Both healthy BMI and CRF have positive health impacts for children—a healthy BMI avoids the health complications of obesity, and higher CRF is associated with improved cardiometabolic health (5), lowered risk of premature cardiovascular disease (6), and improved mental health (7). Both healthy BMI and higher CRF have possible associations with AA through a range of mechanisms. Higher BMI may hinder academic performance through brain development and cognition (8), school absenteeism due to poorer health (9), social isolation (10), weight-based stigmatization by teachers (11), or poor sleep due to breathing issues (12). CRF could be linked to enhanced AA through positive changes in brain structure and function that affect attention, memory, and overall cognition (13).
There is limited previous research on BMI and CRF as mediators of the PA–AA relationship, with more evidence of mediation by CRF than by BMI. A 2022 systematic review examining mediation of the PA–AA relationship found that, out of nine prior studies examining mediation by physical fitness (eight of which specifically examined CRF), eight reported significant mediation by fitness and one reported null findings (14). Among five studies analyzing adiposity variables (including BMI) as mediators, one longitudinal study found significant mediation among both female and male students, whereas one cross-sectional study found that BMI mediated the PA–AA association in female but not male students (14). Some hypothesize that BMI could more substantially mediate the PA–AA relationship among female than male students because higher BMI is more stigmatized for female students (14).
The existing literature on mediation by BMI and CRF has several limitations. First, there are relatively few existing studies on mediation of the PA–AA relationship. CRF is the most widely studied mediator but has only been assessed as a mediator in fewer than a dozen studies, the vast majority of which are cross-sectional (14). Many studies also rely on self-reported questionnaires instead of objectively measured PA and do not account for PA intensity (15).
The current analysis extends a previous study examining the relationship between MVPA and AA by exploring potential mediation by BMI and CRF in a diverse population of 4936 elementary-aged students enrolled in grade 4 and followed through grade 5 (4). Analyses evaluate whether school-day MVPA is associated cross-sectionally and longitudinally with BMI and CRF, and whether BMI and CRF mediate the relationship between school-day MVPA and academic outcomes. Analyses also examine this mediation separately by sex. We hypothesize that BMI and CRF both positively mediate the MVPA–AA relationship and that there will be more evidence of mediation by BMI among female students. This study addresses the literature’s limitations because of its large sample, longitudinal design, and use of objectively-measured PA.
METHODS
Study Design
Students from 40 public elementary schools in suburban Georgia were prospectively followed from grade 4 to grade 5, including grade 4 fall (fall 2018; T1), grade 4 spring (spring 2019; T2), grade 5 fall (fall 2019; T3), and grade 5 spring (spring 2020; T4), although study activities ended midway through T4 in March 2020 because of the coronavirus disease 2019 (COVID-19) pandemic. The study utilized the evidence-based Health Empowers You! Intervention through a cluster-randomized controlled trial design (20 intervention schools, 20 control schools) with the goal of sustainably elevating student school-day MVPA (16,17). The intervention also ensured some students experienced higher school-day MVPA levels, allowing for more rigorous assessment of the relationship between school-day MVPA and AA. Additional details about school recruitment and the intervention have been previously reported (4,18).
Study Participants and Recruitment
The school district administration, district institutional review board (IRB), and Emory University IRB (IRB00095600) approved this study. This study was retrospectively registered on ClinicalTrials.gov (NCT03765047). Informed consent agreements (available in English, Spanish, Vietnamese, and several other languages) were required. Enrollment in the study included providing parental consent and student assent for participation in PA measurement via accelerometry and authorizing the school district to share education data with the research team. All grade 4 students not enrolled in a full-time special education classroom at participating schools at the beginning of the 2018–2019 school year were eligible for enrollment in the study. Of 6525 eligible grade 4 students across the 40 schools, 4936 (76%) were enrolled in the study. Additional details about the study’s protocol have been previously reported (18).
Data Sources
Two sources of data were used for this study: 1) school district records of student academic and demographic data and 2) accelerometers.
School district records of student academic, demographic, and physical fitness data
School district data included student standardized test scores, course grades, demographics (sex and race/ethnicity), free/reduced-price lunch (FRL) eligibility status, English language learner (ELL) status, student with disabilities (SWD) status (including physical and learning disabilities), special education participation, attendance (including days enrolled, absences, and tardies), and data from the FITNESSGRAM assessment. To maintain participant confidentiality, the district provided data linked to each student by a unique subject identification number.
Accelerometers
The study used ActiGraph wGT3X-BT 3-axis accelerometers (ActiGraph LLC, Pensacola, FL) worn on the waist. Students put on their assigned accelerometer on a waist belt at the beginning of the school day and removed it before leaving school. ActiLife software (ActiGraph LLC) was used to download and score the data and filter such that only school-day minutes were used in scoring. Nonwear time was defined as 60 consecutive minutes of zero counts, allowing for up to 2 min of counts between 0 and 100 (19). Data were collected in 15-s epochs and scored using Evenson cut points for activity thresholds (20).
Data Measures
Exposure
Accelerometer-measured school-day PA was the primary exposure. Criteria for a valid day required wearing the accelerometer for at least 80% of the school day. Students needed at least 3 valid days of wear time during the 5-d measurement period each semester to be included in that semester’s analyses. A single measure of mean MVPA minutes was calculated in each semester for students who met the 3-d criteria.
Primary outcomes
Analyses evaluated two AA variables as the primary outcomes: teacher-assigned course grades and standardized test scores. Teacher-assigned course grades were reported each semester (T1, T2, and T3) for math, reading, spelling, and writing on a 100-point scale. Standardized test scores were from the Georgia Milestone standardized test in English language arts (ELA) and math, which is administered each spring for students in grades 3 through 8 (21). Participating students’ results from the spring 2019 grade 4 Georgia Milestones tests were used. Because of COVID-19 disrupting the spring 2020 semester and canceling standardized tests, grades and standardized test scores were not collected for grade-5 spring.
Mediators
Student CRF and BMI were collected from the school district FITNESSGRAM assessment (22). The district’s physical education (PE) instructors are routinely trained on FITNESSGRAM data collection. PE instructors measured student height and weight to calculate student BMI using standardized equipment. Results from the FITNESSGRAM Progressive Aerobic Cardiovascular Endurance Run Test (a 20-m shuttle run) were used to estimate CRF. Complete FITNESSGRAM data were collected in T1, T2, and T3 but were not collected in T4 because of COVID-19-related disruptions.
Covariates
School district data were used for student-level, teacher-level, and school-level covariates. Student sex, race and ethnicity, SWD status, participation in special education courses, ELL status, FRL eligibility status, prior academic grades, prior absenteeism, and prior tardiness were controlled for in all models. Student sex was documented as either male or female, and student race and ethnicity was classified as Asian, Black, Hispanic, multiracial and/or multiethnic, or White. SWD status (including physical and learning disabilities), ELL status, and FRL status were dichotomized as yes or no. Student FRL status was used as a proxy for socioeconomic status. Students were eligible for FRL if their family household income was at or below 185% of the federal poverty level (23). Students’ enrollment in special education courses was incorporated as a variable ranging from zero to four based on the number of special education courses students were enrolled in across math, reading, spelling, and writing. Student prior achievement was measured as the previous year’s course grade or standardized test score. A student’s prior absenteeism and tardiness were measured by percent days absent and tardy in grade 3.
Some teachers were departmentalized at the teacher level. Departmentalized teachers teach some but not all core school subjects, meaning students rotate between teachers to take their classes. The teacher level was not included in the multilevel analyses because students with departmentalized teachers did not remain with the same teacher across core subjects.
At the school level, the percentage of students who were female, Black, Hispanic, and receiving FRL were calculated from school data. Schools’ intervention or control status was also controlled for in all analyses.
Statistical Analysis
Variables were missing data either because students were not enrolled in the participating schools for the entirety of the study or because their observation did not meet the inclusion criteria (e.g., did not meet the 3-valid-day criteria for accelerometer data in a given semester). Multiple imputation addressed missing data. Twenty imputed data sets were created using the multilevel multiple imputation program Blimp (24). Implausible imputed values were set to variables’ upper or lower bounds. Descriptive statistics were run on the nonimputed data.
Two-level random-intercepts models controlled for student demographics, school demographics, school intervention or control status, and grade 3 AA. Two-level models (accounting for students nested in schools) were used instead of three-level models because some analyzed schools used a departmentalized-teacher model where students rotated between teachers for core school subjects.
Analyses first assessed whether school-day MVPA was associated cross-sectionally with CRF and BMI and whether CRF and BMI were associated cross-sectionally with AA in T1, T2, and T3. Indirect effects of MVPA on academic outcomes through CRF and BMI were calculated cross-sectionally and longitudinally. Cross-sectional and longitudinal analyses were conducted to test different time durations over which there could be an MVPA–AA relationship through CRF and BMI. The indirect effect was determined by the product method, which multiplies the coefficients from 1) a model where the focal predictor is predicting the mediator and 2) a model in which the mediator is predicting the outcome. The indirect effect was estimated regardless of the statistical significance of the total effect, deviating from initial recommendations of Baron and Kenny, but now considered best practice (25,26). The 95% confidence interval (CI) for the indirect effect was found using the Monte Carlo method (27). A Bonferroni-corrected α of 0.00038 was used to address multiple testing. The data set was then stratified by sex, and cross-sectional and longitudinal analyses were performed again for female students and male students to test the hypothesis that mediation could differ between the sexes.
RESULTS
Descriptive Analysis of the Study Population
Analyses include data from 4936 grade 4 students from 40 public schools. The sample was racially and ethnically diverse, with 12.2% of the sample identifying as Asian, 25.2% as Black, 33.2% as Hispanic, 4.3% as multiracial/multiethnic, and 24.8% as White (Table 1). Approximately half (53.1%) were FRL eligible during the study period, and 50.0% were female. About a third of participants were current or former ELL (35.1%), and 12.9% had learning or physical disabilities.
TABLE 1 -
Student and School Demographics, AA, and PA Data Grades 4 to 5.
|
Student-Level Data (n = 4936)
a
|
| Variable |
Average or Percent |
SD or Percent |
% Missing Data |
| Sex |
|
|
|
| Female |
2466 |
(50.0) |
0 |
| Male |
2465 |
(49.9) |
| Race/Ethnicity |
| Asian |
601 |
(12.2) |
0.1 |
| Black |
1243 |
(25.2) |
| Hispanic |
1640 |
(33.2) |
| Multiracial and/or multiethnic |
213 |
(4.3) |
| White |
1226 |
(24.8) |
| FRL recipient |
| Yes |
2622 |
(53.1) |
0.1 |
| No |
2309 |
(46.8) |
| ELL |
| Yes |
1735 |
(35.1) |
0.1 |
| No |
3196 |
(64.7) |
| SWD |
|
|
|
| Yes |
637 |
(12.9) |
0.1 |
| No |
4294 |
(87.0) |
| Grade 3 absences (%) |
3.1 |
(3.0) |
0.1 |
| Grade 3 tardies (%) |
1.9 |
(3.6) |
0.1 |
| Special education participant |
| Yes |
242 |
(4.9) |
0 |
| No |
4694 |
(95.1) |
| Grade 3 average course grades |
| Math |
83.5 |
(9.4) |
8.4 |
| Reading |
82.9 |
(9.0) |
8.4 |
| Spelling |
87.9 |
(9.0) |
12.2 |
| Writing |
84.4 |
(7.8) |
9.1 |
| Grade 4 average course grades |
| Math |
82.4 |
(10.9) |
3.5 |
| Reading |
81.4 |
(9.6) |
3.6 |
| Spelling |
87.1 |
(9.9) |
7.2 |
| Writing |
83.5 |
(8.3) |
3.8 |
| Grade 5 fall course grades
b
|
| Math |
82.0 |
(11.5) |
15.1 |
| Reading |
82.4 |
(9.4) |
14.4 |
| Spelling |
87.6 |
(9.6) |
16.5 |
| Writing |
84.9 |
(8.2) |
15.0 |
| Change in course grade, grade 4 fall to grade 5 fall |
| Math |
−0.11 |
(8.9) |
15.3 |
| Reading |
0.89 |
(7.6) |
14.7 |
| Spelling |
0.65 |
(10.0) |
18.1 |
| Writing |
1.35 |
(7.6) |
15.4 |
| Grade 3 standardized test scores
c
|
| Math |
541.4 |
(49.9) |
6.0 |
| ELA |
527.1 |
(58.2) |
6.0 |
| Lexile |
728.6 |
(219.6) |
6.0 |
| Grade 4 standardized test scores
c
|
| Math |
549.2 |
(53.7) |
3.3 |
| ELA |
535.2 |
(55.5) |
3.3 |
| Lexile |
897.2 |
(221.1) |
3.3 |
| Average daily MVPA minutes |
| Grade 4 fall |
21.1 |
(9.2) |
12.5 |
| Grade 4 spring |
21.9 |
(10.0) |
23.0 |
| Grade 5 fall |
18.9 |
(8.9) |
27.3 |
| BMI (kg·m−2) |
|
|
|
| Grade 4 fall |
19.4 |
(4.1) |
6.3 |
| Grade 4 spring |
19.7 |
(4.4) |
12.1 |
| Grade 5 fall |
20.2 |
(4.7) |
20.0 |
| CRF (V˙O2max) |
|
|
|
| Grade 4 fall |
42.2 |
(4.5) |
7.5 |
| Grade 4 spring |
42.9 |
(4.6) |
14.9 |
| Grade 5 fall |
42.6 |
(5.1) |
18.4 |
|
School-Level Data (n = 40)
|
| School % female |
48.6% |
(4.2) |
0 |
| School % Black |
27.3% |
(12.1) |
0 |
| School % Hispanic |
32.2% |
(19.8) |
0 |
| School % FRL |
56.1% |
(26.7) |
0 |
| Teacher departmentalization |
| Yes |
133 |
(46.5%) |
0 |
| No |
153 |
(53.5%) |
0 |
a Not all tabulations add to 4936 because of missing data.
b Because of COVID-19-related disruptions, grade 5 spring course grades are not incorporated in analyses.
c Because of COVID-19-related disruptions, no standardized tests were conducted in grade 5.
V˙O2max, maximal oxygen consumption.
Mean course grades were similar for the sample across semesters from grade 3 fall to grade 5 fall. Standardized test descriptive statistics are also detailed in Table 1. Average daily school-day MVPA declined slightly over the study period. Mean CRF was consistent across semesters. The intervention increased average school-day MVPA by 3–5 min in intervention schools compared with control schools, as described elsewhere (4). BMI increased across the study period.
Mediation Analyses
A cross-sectional analysis showed that average school-day MVPA was significantly associated with lower BMI in the three semesters (β = −0.03, P < 0.001 in T1; β = −0.05, P < 0.001 in T2; and β = −0.03, P < 0.001 in T3; Table 2). Average school-day MVPA and CRF were positively and significantly associated in T1 (β = 0.06, P < 0.001), T2 (β = 0.09, P < 0.001), and T3 (β = 0.10, P < 0.001). BMI was not a significant predictor of semester course grades except in T3 (βMath = −0.09, P < 0.001; βWriting = −0.08, P < 0.001). CRF was positively and significantly associated with most semester course grades and standardized test scores during the study period.
TABLE 2 -
Cross-Sectional Associations of Physical Fitness with Student MVPA and AA (
n = 4936).
| Predictor |
Outcome |
Association |
P
|
| Grade 4 fall MVPA |
Grade 4 fall BMI |
−0.03 |
<0.001
|
| Grade 4 fall BMI |
Grade 4 fall course grades |
|
|
|
Math |
−0.03 |
0.26 |
|
Reading |
0.00 |
0.99 |
|
Writing |
−0.05 |
0.03 |
|
Spelling |
−0.02 |
0.42 |
| Grade 4 spring MVPA |
Grade 4 spring BMI |
−0.05 |
<0.001
|
| Grade 4 spring BMI |
Grade 4 spring course grades |
|
|
|
Math |
−0.05 |
0.05 |
|
Reading |
−0.02 |
0.48 |
|
Writing |
−0.04 |
0.98 |
|
Spelling |
−0.05 |
0.03 |
|
Grade 4 standardized test scores |
|
Math |
−0.12 |
0.24 |
|
ELA |
−0.14 |
0.20 |
|
Lexile |
−0.13 |
0.79 |
| Grade 5 fall MVPA |
Grade 5 fall BMI |
−0.03 |
<0.001
|
| Grade 5 fall BMI |
Grade 5 fall course grades |
|
|
|
Math |
−0.09 |
<0.001
|
|
Reading |
−0.05 |
0.03 |
|
Writing |
−0.08 |
<0.001
|
|
Spelling |
−0.06 |
0.02 |
| Grade 4 fall MVPA |
Grade 4 fall CRF |
0.06 |
<0.001
|
| Grade 4 fall CRF |
Grade 4 fall course grades |
|
|
|
Math |
0.14 |
<0.001
|
|
Reading |
0.06 |
0.03 |
|
Writing |
0.09 |
<0.001
|
|
Spelling |
0.09 |
0.006 |
| Grade 4 spring MVPA |
Grade 4 spring CRF |
0.09 |
<0.001
|
| Grade 4 spring CRF |
Grade 4 spring course grades |
|
|
|
Math |
0.12 |
<0.001
|
|
Reading |
0.08 |
<0.001
|
|
Writing |
0.10 |
<0.001
|
|
Spelling |
0.09 |
0.002 |
|
Grade 4 standardized test scores |
|
Math |
0.38 |
<0.001
|
|
ELA |
0.51 |
<0.001
|
|
Lexile |
1.79 |
<0.001
|
| Grade 5 fall MVPA |
Grade 5 fall CRF |
0.10 |
<0.001
|
| Grade 5 fall CRF |
Grade 5 fall course grades |
|
|
|
Math |
0.14 |
<0.001
|
|
Reading |
0.08 |
<0.001
|
|
Writing |
0.11 |
<0.001
|
|
Spelling |
0.08 |
0.001 |
Models adjusted for student sex, race/ethnicity, FRL status, ELL status, SWD status, grade 3 absences, grade 3 tardies, special education course enrollment, school percentage female, school percentage Black, school percentage Hispanic, school percentage FRL, school cohort (intervention or control), departmentalization, and prior achievement for the specific academic outcome (e.g., when assessing math grade as outcome, used grade 3 average math grade) (
28).
Cross-sectional mediation analyses produced consistently negative total effects and generally positive indirect effects (Table 3). There were no significant cross-sectional indirect effects through BMI. There was a significant positive indirect effect from school-day MVPA through CRF for T1 math course grades (0.008; 95% CI, 0.002–0.017); T2 math (0.010; 95% CI, 0.001–0.020) and writing (0.009; 95% CI, 0.001–0.018) course grades; grade 4 standardized ELA test scores (0.051; 95% CI, 0.007–0.106); and T3 math (0.013; 95% CI, 0.004–0.025), reading (0.008; 95% CI, 0.001–0.017), and writing (0.011; 95% CI, 0.004–0.020) course grades.
TABLE 3 -
Cross-Sectional Associations of Student School-Day MVPA and AA, Mediated by Physical Fitness (
n = 4936).
| Predictor, Mediator |
Academic Outcome |
Total Effect |
a
|
b
|
Indirect Effect |
Indirect Effect ES |
95% CI of Indirect Effect |
| Grade 4 fall MVPA, BMI |
Grade 4 fall course grades |
|
Math |
−0.055 (0.016) |
−0.029 (0.008) |
−0.030 (0.030) |
0.0009 |
0.0007 |
−0.003 to 0.005 |
|
Reading |
−0.050 (0.013) |
−0.029 (0.008) |
−0.005 (0.026) |
−0.0002 |
−0.0002 |
−0.004 to 0.004 |
|
Writing |
−0.056 (0.013) |
−0.029 (0.008) |
−0.051 (0.025) |
0.0015 |
0.0016 |
−0.001 to 0.006 |
|
Spelling |
−0.069 (0.017) |
−0.030 (0.008) |
−0.022 (0.033) |
0.0006 |
0.0005 |
−0.003 to 0.005 |
| Grade 4 spring MVPA, BMI |
Grade 4 spring course grades |
|
Math |
−0.052 (0.014) |
−0.053 (0.008) |
−0.045 (0.027) |
0.0024 |
0.0023 |
−0.003 to 0.008 |
|
Reading |
−0.031 (0.013) |
−0.053 (0.008) |
−0.025 (0.025) |
0.0013 |
0.0013 |
−0.004 to 0.006 |
|
Writing |
−0.031 (0.012) |
−0.053 (0.008) |
−0.048 (0.025) |
0.0025 |
0.0029 |
−0.002 to 0.009 |
|
Spelling |
−0.030 (0.016) |
−0.053 (0.008) |
−0.057 (0.030) |
0.0030 |
0.0029 |
−0.003 to 0.009 |
| Grade 4 MVPA, BMI |
Grade 4 standardized test scores |
|
Math |
−0.270 (0.075) |
−0.060 (0.010) |
−0.153 (0.104) |
0.0092 |
0.0016 |
−0.014 to 0.035 |
|
ELA |
−0.199 (0.077) |
−0.059 (0.010) |
−0.142 (0.116) |
0.0084 |
0.0015 |
−0.016 to 0.039 |
|
Lexile |
−0.494 (0.345) |
−0.059 (0.010) |
0.039 (0.534) |
0.0023 |
0.0001 |
−0.122 to 0.131 |
| Grade 5 fall MVPA, BMI |
Grade 5 fall course grades |
|
Math |
−0.002 (0.017) |
−0.036 (0.011) |
−0.093 (0.026) |
0.0034 |
0.0026 |
−0.0003 to 0.0098 |
|
Reading |
−0.018 (0.014) |
−0.036 (0.011) |
−0.057 (0.022) |
0.0020 |
0.0019 |
−0.0012 to 0.0068 |
|
Writing |
−0.027 (0.014) |
−0.037 (0.011) |
−0.088 (0.021) |
0.0032 |
0.0035 |
−0.0002 to 0.0088 |
|
Spelling |
−0.022 (0.018) |
−0.036 (0.011) |
−0.077 (0.028) |
0.0028 |
0.0026 |
−0.0005 to 0.0092 |
| Grade 4 fall MVPA, CRF |
Grade 4 fall course grades |
|
Math |
|
0.058 (0.009) |
0.138 (0.031) |
0.0080
|
0.0061 |
0.002 to 0.017 |
|
Reading |
|
0.057 (0.008) |
0.075 (0.026) |
0.0043 |
0.0038 |
−0.001 to 0.011 |
|
Writing |
|
0.057 (0.008) |
0.078 (0.026) |
0.0044 |
0.0045 |
−0.001 to 0.011 |
|
Spelling |
|
0.057 (0.008) |
0.078 (0.035) |
0.0044 |
0.0036 |
−0.003 to 0.013 |
| Grade 4 spring MVPA, CRF |
Grade 4 spring course grades |
|
Math |
|
0.087 (0.009) |
0.112 (0.028) |
0.0098
|
0.0092 |
0.001 to 0.020 |
|
Reading |
|
0.085 (0.009) |
0.072 (0.026) |
0.0061 |
0.0062 |
−0.001 to 0.014 |
|
Writing |
|
0.085 (0.009) |
0.103 (0.026) |
0.0088
|
0.0101 |
0.001 to 0.018 |
|
Spelling |
|
0.085 (0.009) |
0.091 (0.034) |
0.0077 |
0.0075 |
−0.002 to 0.020 |
| Grade 4 MVPA, CRF |
Grade 4 standardized test scores |
|
Math |
|
0.102 (0.009) |
0.328 (0.114) |
0.0333 |
0.0060 |
−0.007 to 0.083 |
|
ELA |
|
0.099 (0.009) |
0.517 (0.124) |
0.0511
|
0.0088 |
0.007 to 0.106 |
|
Lexile |
|
0.098 (0.009) |
1.631 (0.573) |
0.1604 |
0.0070 |
−0.035 to 0.382 |
| Grade 5 fall MVPA, CRF |
Grade 5 fall course grades |
|
Math |
|
0.099 (0.011) |
0.135 (0.026) |
0.0133
|
0.0103 |
0.0042 to 0.0254 |
|
Reading |
|
0.098 (0.011) |
0.083 (0.022) |
0.0082
|
0.0078 |
0.0006 to 0.0170 |
|
Writing |
|
0.099 (0.011) |
0.113 (0.022) |
0.0113
|
0.0123 |
0.0039 to 0.0202 |
|
Spelling |
|
0.097 (0.011) |
0.089 (0.027) |
0.0087 |
0.0081 |
−0.0010 to 0.0195 |
Models adjusted for student sex, race/ethnicity, FRL status, ELL status, SWD status, grade 3 absences, grade 3 tardies, special education course enrollment, school percentage female, school percentage Black, school percentage Hispanic, school percentage FRL, school cohort (intervention or control), departmentalization, and prior achievement for the specific academic outcome (e.g., when assessing math grade as outcome, used grade 3 average math grade). The 95% CI for the indirect effect was found using the Monte Carlo method. In bold are the indirect effects found to be statistically significant using a Bonferroni corrected
α of 0.00038. The
a path is from a model in which the focal predictor is predicting the mediator, and the
b path is from a model in which the mediator is predicting the outcome. The effect size (ES) for the indirect effect is found by multiplying the product of
a and
b by the SD of the focal predictor and dividing by the SD of the outcome (referred to as the index of mediation (
28)).
Longitudinal mediation analyses produced consistently negative total effects and positive indirect effects (Table 4). There were significant indirect effects through BMI from grade 4 averaged school-day MVPA to T3 math (0.006; 95% CI, 0.001–0.013) and writing (0.005; 95% CI, 0.000–0.011) course grades. There were significant positive indirect effects from averaged grade 4 school-day MVPA through CRF to T3 math (0.016; 95% CI, 0.005–0.029), reading (0.009; 95% CI, 0.001–0.020), and writing (0.014; 95% CI, 0.005–0.024) course grades.
TABLE 4 -
Longitudinal Associations of Change in Student School-Day MVPA and AA, Mediated by Physical Fitness (
n = 4936).
| Predictor, Mediator |
Academic Outcome |
Total |
a
|
b
|
Indirect Effect |
Indirect Effect ES |
95% CI of Indirect Effect |
| Grade 4 fall MVPA, BMI |
Grade 4 spring course grades |
|
Math |
−0.034 (0.012) |
−0.030 (0.008) |
−0.024 (0.019) |
0.0007 |
0.0006 |
−0.002 to 0.004 |
|
Reading |
−0.038 (0.011) |
−0.030 (0.008) |
−0.002 (0.019) |
0.0001 |
0.0001 |
−0.002 to 0.003 |
|
Writing |
−0.022 (0.010) |
−0.031 (0.008) |
−0.008 (0.019) |
0.0003 |
0.0003 |
−0.002 to 0.003 |
|
Spelling |
−0.039 (0.013) |
−0.030 (0.008) |
−0.038 (0.023) |
0.0011 |
0.0010 |
−0.002 to 0.004 |
| Grade 4 MVPA, BMI |
Grade 5 fall course grades |
|
Math |
−0.017 (0.018) |
−0.063 (0.011) |
−0.095 (0.025) |
0.006
|
0.0050 |
0.001 to 0.013 |
|
Reading |
−0.028 (0.014) |
−0.062 (0.011) |
−0.044 (0.021) |
0.003 |
0.0031 |
−0.001 to 0.008 |
|
Writing |
−0.040 (0.014) |
−0.063 (0.011) |
−0.073 (0.020) |
0.005
|
0.0059 |
0.0003 to 0.011 |
|
Spelling |
−0.009 (0.018) |
−0.063 (0.011) |
−0.063 (0.026) |
0.004 |
0.0040 |
−0.001 to 0.011 |
| Grade 4 fall MVPA, CRF |
Grade 4 spring course grades |
|
Math |
|
0.078 (0.009) |
0.049 (0.020) |
0.004 |
0.0035 |
−0.001 to 0.011 |
|
Reading |
|
0.077 (0.009) |
0.046 (0.020) |
0.004 |
0.0037 |
−0.002 to 0.010 |
|
Writing |
|
0.078 (0.009) |
0.058 (0.019) |
0.005 |
0.0053 |
−0.001 to 0.011 |
|
Spelling |
|
0.077 (0.009) |
0.052 (0.024) |
0.004 |
0.0036 |
−0.002 to 0.011 |
| Grade 4 MVPA, CRF |
Grade 5 fall course grades |
|
Math |
|
0.126 (0.011) |
0.127 (0.025) |
0.016
|
0.0134 |
0.005 to 0.029 |
|
Reading |
|
0.124 (0.011) |
0.076 (0.021) |
0.009
|
0.0092 |
0.001 to 0.020 |
|
Writing |
|
0.125 (0.011) |
0.110 (0.021) |
0.014
|
0.0164 |
0.005 to 0.024 |
|
Spelling |
|
0.122 (0.011) |
0.090 (0.026) |
0.011 |
0.0110 |
−0.0001 to 0.025 |
Models adjusted for student sex, race/ethnicity, FRL status, ELL status, SWD status, grade 3 absences, grade 3 tardies, special education course enrollment, school percentage female, school percentage Black, school percentage Hispanic, school percentage FRL, school cohort (intervention or control), departmentalization, and prior achievement for the specific academic outcome (e.g., when assessing math grade as outcome, used grade 3 average math grade). The 95% CI for the indirect effect was found using the Monte Carlo method. In bold are the indirect effects found to be statistically significant using a Bonferroni corrected
α of 0.00038. The
a path is from a model in which the focal predictor is predicting the mediator, and the
b path is from a model in which the mediator is predicting the outcome. The effect size (ES) for the indirect effect is found by multiplying the product of
a and
b by the SD of the focal predictor and dividing by the SD of the outcome (referred to as the index of mediation (
28)).
For female students, cross-sectional analyses showed no significant indirect effects through BMI. Female students did show significant indirect effects through CRF in cross-sectional analyses for T1 math (0.009; 95% CI, 0.001–0.023), reading (0.006; 95% CI, 0.000–0.016), and writing (0.007; 95% CI, 0.000–0.018) grades; T2 writing grades (0.007; 95% CI, 0.000–0.019); and T3 math grades (0.011; 95% CI, 0.001–0.028) (Supplemental Content 1, table, https://links.lww.com/TJACSM/A213). The only significant indirect effects in longitudinal analyses among female students were found through CRF from grade 4 school-day averaged MVPA to T3 math (0.015; 95% CI, 0.002–0.034) and reading (0.010; 95% CI, 0.000–0.023) grades (Supplemental Content 2, table, https://links.lww.com/TJACSM/A214).
Similar to females, cross-sectional analyses among male students found no significant indirect effects through BMI (Supplemental Content 3, table, https://links.lww.com/TJACSM/A215). However, unlike female students, cross-sectional analyses among male students found almost no significant effects through CRF, except for T3 writing grades (0.014; 95% CI, 0.002–0.030). In longitudinal analyses among male students, the only significant indirect effect was through CRF from grade 4 averaged school-day MVPA to T3 writing grades (0.018; 95% CI, 0.004–0.035) (Supplemental Content 4 table, https://links.lww.com/TJACSM/A216).
DISCUSSION
Results from this study suggest that school-day MVPA could contribute to improved teacher-assigned grades by increasing CRF and (to a lesser extent) by decreasing BMI. However, the magnitude of the indirect effects is negligible. Based on longitudinal analyses from grade 4 to grade 5 fall, the increased CRF associated with a 10-min increase in daily school-day MVPA would be expected to boost math, reading, and writing grades by 0.16, 0.09, and 0.14 points, respectively, on a 100-point scale. The reduced BMI associated with a 10-min increase in daily school-day MVPA would be expected to boost math and writing grades by 0.06 and 0.05 points, respectively. Results also showed more evidence of mediation among female students compared with male students specifically for CRF, but again, the indirect effects were negligible in practical terms.
The small magnitude of mediation by CRF aligns with previous literature that objectively measured PA using accelerometers. A 2021 study of 186 schoolchildren aged 9 to 11 yr who wore accelerometers for 7 d found significant mediation from accelerometer-measured MVPA to AA through CRF, but the standardized effect size for the indirect effect was never greater than 0.100 (29). A 2018 study of 970 children aged 9 to 15 yr in Finland only found a significant indirect positive association from PA to AA through CRF when relying on self-reported PA—accelerometer-based PA was not associated with AA (30). A 2013 study in the United States of 401 students aged 6 to 9 yr who wore accelerometers for 3 or more days found significant mediation from PA through CRF for mathematics achievement but not reading or spelling; however, the indirect effect through CRF for mathematics was small (b = 0.003, P < 0.01) (31).
The total effect of school-day MVPA on AA is consistently negative. As outlined in a previous article, any statistically significant associations between school-day MVPA and AA were functionally negligible given their very small magnitude (32). Although the total effect was negative, the significant indirect effects through CRF and BMI were positive. This inconsistent mediation suggests that there are other significant unmeasured mediators between school-day MVPA and AA beyond CRF and BMI (33). Future research should include other potential mediators such as time on task, motivation, or mental well-being.
Our results indicating a smaller indirect effect for BMI compared with CRF align with previous research. In a 2022 systematic review of mediators between PA and AA, CRF was the lone variable for which a majority of studies identified significant mediation (14). Of nine studies examining physical fitness as a mediator, six reported a positive indirect effect through fitness (14). The same systematic review found mixed results for mediation by adiposity measures, including BMI (14). Of five studies looking at mediation by adiposity variables, two found an indirect association (14). Beyond mediation, the literature on BMI and CRF as predictors of AA also points to a stronger role for CRF (34).
Previous literature suggests that weight status is a stronger mediator for female than male individuals given greater weight stigma for female individuals (14); however, this study found stronger evidence for mediation among female students through CRF. One possible explanation may be higher levels of attention-deficit/hyperactivity disorder-related hyperactivity among male students, suggesting that male students with higher CRF are more likely to experience academically hindering hyperactivity than female students with higher CRF (35). However, as previously mentioned, the magnitude of significant mediation among females in this study is very small.
Regardless of associations with AA, school-based PA’s health benefits, like higher CRF and lower BMI, are even more critical in the wake of the COVID-19 pandemic. In a cohort of 432,302 US infants, children, and adolescents aged 2–19 yr, the rate of BMI increase doubled during the COVID-19 pandemic compared with a prepandemic period, and accelerated weight gain was especially pronounced among younger school-aged children (36). The reduction in structured PA opportunities, including school-based PA, lost because of pandemic-related school closures, likely contributed to this accelerated pediatric weight gain (36). These patterns during the COVID-19 pandemic highlight the importance of protecting time for school-based PA to promote children’s physical health.
Strengths
This study has several strengths. First, this is the largest study of the association between objectively measured PA and AA in the United States to date. Second, the use of an objective measure of PA by accelerometers increases the validity of PA measurement. Third, BMI and CRF were measured objectively by trained staff in PE classes instead of relying on self-report. Fourth, PA measurement was repeated over 15 d across three semesters to strengthen validity; most prior studies have been limited by their cross-sectional design (14). Fifth, the study used both teacher-assigned course grades and standardized test scores. Finally, the study’s consent rate was high at 76% across all grade 4 students at baseline.
Limitations
Despite these strengths, our study has some limitations. First, this study did not include possible academic behaviors that may mediate the effects of PA and AA (e.g., time on task, motivation). Second, follow-up time decreased by one semester because of COVID-19 disruptions. Third, the extrapolation of 3–5 d of accelerometer data to a full semester of activity levels may have been less accurate than a longer accelerometer measurement period. Fourth, teacher-assigned grades might be less objective than standardized test scores when measuring AA. This potential bias was addressed by evaluating both teacher-assigned grades and standardized test scores as academic outcomes, and by the fact that the studied school district uses a district-wide procedure for assigning grades to increase consistency across schools. It is also worth noting that research suggests student grade-point averages based on teacher-assigned grades may be more predictive of longer-term achievements (e.g., college graduation) than standardized test scores (37).
Conclusions
The study findings indicate that higher CRF associated with school-based MVPA might slightly boost AA, although the magnitude of this indirect effect is negligible. Evidence for mediation through BMI was even more limited. The negative total effects and positive indirect effects observed in this population suggest that there are other unmeasured mediators between school-day MVPA and AA.
Future research should continue relying on objective measures of MVPA, BMI, and CRF. It is also important to examine longer periods to better understand how children’s MVPA, BMI, and CRF dynamics change in adolescence, and how those changes impact AA. Finally, future studies should examine other potential mediators (e.g., time on task, student engagement) because the findings suggest that there are other significant unmeasured mediators in this population.
Regardless of the lack of meaningful mediation identified in this study, school administrators and teachers should note the importance of MVPA to children’s health and development through higher CRF and lower BMI and ensure time dedicated to MVPA in schools is protected. This is particularly important given the rise in pediatric obesity in the United States during the COVID-19 pandemic. Schools can also incorporate MVPA in ways that do not compete with other academic subjects, such as through physical activities that incorporate academic content. Promoting higher-intensity PA, which more readily improves CRF, is especially important.
This study was retrospectively registered on ClinicalTrials.gov (NCT03765047). Additional thanks go to Blessing Falade and Chuck Truett, the physical activity specialists from HealthMPowers, members of the study’s advisory board, and participating school district administrators. The results of the present study do not constitute endorsement by the American College of Sports Medicine.
The authors declare that they have no competing interests. This study was funded by the Robert Wood Johnson Foundation (ID: 74281). The study also received supplementary grant funding from the Ardmore Institute of Health. These study sponsors did not have a role in study design, collection, analysis, and interpretation of data; writing the report; or the decision to submit the report for publication.
REFERENCES
1. Centers for Disease Control and Prevention.
CDC Healthy Schools [Internet]. Washington (DC): U.S. Department of Health and Human Services; 2019 [cited 2020 Aug 24]. Available from:
https://www.cdc.gov/healthyschools/about.htm.
2. Barbosa A, Whiting S, Simmonds P, et al. Physical activity and academic achievement: an umbrella review.
Int J Environ Res Public Health. 2020;17(16):5972.
3. Wassenaar TM, Williamson W, Johansen-Berg H, et al. A critical evaluation of systematic reviews assessing the effect of chronic physical activity on academic achievement, cognition and the brain in children and adolescents: a systematic review.
Int J Behav Nutr Phys Act. 2020;17(1):79.
4. Elish PN, Bryan CS, Boedeker PJ, et al. The longitudinal association between objectively-measured school-day physical activity and academic achievement in US elementary school students.
Int J Behav Nutr Phys Act. 2022;19(1):90.
5. Lang JJ, Belanger K, Poitras V, et al. Systematic review of the relationship between 20 m shuttle run performance and health indicators among children and youth.
J Sci Med Sport. 2018;21(4):383–97.
6. Högström G, Nordström A, Nordström P. High aerobic fitness in late adolescence is associated with a reduced risk of myocardial infarction later in life: a nationwide cohort study in men.
Eur Heart J. 2014;35(44):3133–40.
7. Lubans D, Richards J, Hillman C, et al. Physical activity for cognitive and mental health in youth: a systematic review of mechanisms.
Pediatrics. 2016;138(3):e20161642.
8. Liang J, Matheson BE, Kaye WH, et al. Neurocognitive correlates of obesity and obesity-related behaviors in children and adolescents.
Int J Obes (Lond). 2014;38(4):494–506.
9. An R, Yan H, Shi X, Yang Y. Childhood obesity and school absenteeism: a systematic review and meta-analysis.
Obes Rev. 2017;18(12):1412–24.
10. Gunnarsdottir T, Njardvik U, Olafsdottir AS, et al. Teasing and social rejection among obese children enrolling in family-based behavioural treatment: effects on psychological adjustment and academic competencies.
Int J Obes (Lond). 2012;36(1):35–44.
11. MacCann C, Roberts RD. Just as smart but not as successful: obese students obtain lower school grades but equivalent test scores to nonobese students.
Int J Obes (Lond). 2013;37(1):40–6.
12. Tan E, Healey D, Schaughency E, et al. Neurobehavioural correlates in older children and adolescents with obesity and obstructive sleep apnoea.
J Paediatr Child Health. 2014;50(1):16–23.
13. Chaddock L, Hillman CH, Pontifex MB, et al. Childhood aerobic fitness predicts cognitive performance one year later.
J Sports Sci. 2012;30(5):421–30.
14. Visier-Alfonso ME, Sánchez-López M, Álvarez-Bueno C, et al. Mediators between physical activity and academic achievement: a systematic review.
Scand J Med Sci Sports. 2022;32(3):452–64.
15. Watson A, Timperio A, Brown H, et al. Effect of classroom-based physical activity interventions on academic and physical activity outcomes: a systematic review and meta-analysis.
Int J Behav Nutr Phys Act. 2017;14(1):114.
16. SNAP-Ed Connection.
Health EmPowers You! [Internet]. Washington (DC): U.S. Department of Agriculture; 2015 [cited 2021 Jul 24]. Available from:
https://snaped.fns.usda.gov/library/materials/health-empowers-you.
17. HealthMPowers.
Who We Are [Internet]. Norcross (GA): HealthMPowers; [cited 2021 Jul 24]. Available from:
https://healthmpowers.org/about-us/.
18. Boedeker P, Turner L, Calvert H, et al. Study protocol for testing the association between physical activity and academic outcomes utilizing a cluster-randomized trial.
Contemp Clin Trials Commun. 2021;21:100747.
19. Troiano RP, Berrigan D, Dodd KW, et al. Physical activity in the United States measured by accelerometer.
Med Sci Sports Exerc. 2008;40(1):181–8.
20. Evenson KR, Catellier DJ, Gill K, et al. Calibration of two objective measures of physical activity for children.
J Sports Sci. 2008;26(14):1557–65.
21. Georgia Department of Education.
Georgia Milestones Assessment System [Internet]. Atlanta: Georgia Department of Education; [cited 2021 Jul 24]. Available from:
https://www.gadoe.org/Curriculum-Instruction-and-Assessment/Assessment/Pages/Georgia-Milestones-Assessment-System.aspx.
22. Meredith M.
Parental Overview of FITNESSGRAM Assessment. Dallas (TX): The Cooper Institute; 2008. 32p.
23. Food and Nutrition Service.
Child Nutrition Programs: Income Eligibility Guidelines (2019–2020). Washington (DC): U.S. Department of Agriculture; 2019.
24. Enders CK, Keller BT, Levy R. A fully conditional specification approach to multilevel imputation of categorical and continuous variables.
Psychol Methods. 2018;23(2):298–317.
25. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations.
J Pers Soc Psychol. 1986;51(6):1173–82.
26. Zhao X, Lynch JG Jr., Chen Q. Reconsidering Baron and Kenny: myths and truths about mediation analysis.
J Consum Res. 2010;37(2):197–206.
27. Preacher KJ, Selig JP. Advantages of Monte Carlo confidence intervals for indirect effects.
Commun Methods Meas. 2012;6(2):77–98.
28. Preacher KJ, Hayes AF. Contemporary approaches to assessing mediation in communication research. In: Hayes AF, Slater MD, Snyder LB, editors.
The SAGE Sourcebook of Advanced Data Analysis Methods for Communication Research. Thousand Oaks (CA): Sage Publications, Inc.; 2008. pp. 13–54.
29. Visier-Alfonso ME, Álvarez-Bueno C, Sánchez-López M, et al. Fitness and executive function as mediators between physical activity and academic achievement.
J Sports Sci. 2021;39(14):1576–84.
30. Syväoja HJ, Kankaanpää A, Kallio J, et al. The relation of physical activity, sedentary behaviors, and academic achievement is mediated by fitness and bedtime.
J Phys Act Health. 2018;15(2):135–43.
31. Lambourne K, Hansen DM, Szabo AN, et al. Indirect and direct relations between aerobic fitness, physical activity, and academic achievement in elementary school students.
Ment Health Phys Act. 2013;6(3):165–71.
32. Elish PN, Bryan CS, Boedeker P, et al. The longitudinal association between objectively-measured physical activity and academic achievement in US elementary school students.
Int J Behav Nutr Phys Act. 2022;19:90.
33. MacKinnon DP, Fairchild AJ, Fritz MS. Mediation analysis.
Annu Rev Psychol. 2007;58:593–614.
34. Álvarez-Bueno C, Hillman CH, Cavero-Redondo I. Aerobic fitness and academic achievement: a systematic review and meta-analysis.
J Sports Sci. 2020;38(5):582–9.
35. Quinn PO, Madhoo M. A review of attention-deficit/hyperactivity disorder in women and girls: uncovering this hidden diagnosis.
Prim Care Companion CNS Disord. 2014;16(3):PCC.13r01596.
36. Lange SJ, Kompaniyets L, Freedman DS, et al. Longitudinal trends in body mass index before and during the COVID-19 pandemic among persons aged 2–19 years—United States, 2018–2020.
MMWR Morb Mortal Wkly Rep. 2021;70(37):1278–83.
37. Allensworth EM, Clark K. High school GPAs and ACT scores as predictors of college completion: examining assumptions about consistency across high schools.
Educ Res. 2020;49(3):198–211.
