Physical Activity Self-Efficacy and Fitness: Family Environment Relationship Correlates and Self-Esteem as a Mediator among Adolescents Who Are Overweight or Obese

Abstract

Background:

Little is known regarding how dimensions of the family social environment relate to fitness levels and physical activity self-efficacy (PASE) among adolescents who are overweight or obese and whether these relationships are mediated by self-esteem.

Methods:

Potential associations were evaluated between relationship subdomains (cohesion, conflict, expressivity) of the Family Environment Scale (FES), self-esteem (Rosenberg Self-Esteem Scale, RSES), and PASE and fitness, using recovery heart rate [RHR, beats per minute (bpm)] from a 3-minute submaximal step test at baseline. Participants were 108 adolescents who were overweight or obese and were seeking weight-loss treatment as part of the Healthy Kids, Healthy Weight 12-week multidisciplinary pediatric weight management program. Structural equation modeling (SEM) was used to simultaneously evaluate paths between these variables and test for mediation.

Results:

In multivariable models, higher FES cohesion (β = −2.18, s.e. = 0.98; p = 0.02), expressivity (β = −1.97, s.e. = 0.99; p < 0.05), and PASE (β = −0.64, s.e. = 0.33; p < 0.05) scores were associated with lower RHR, which represents higher fitness. Furthermore, higher FES conflict scores were associated with lower RSES scores (β = −0.83, s.e. = 0.29; p < 0.01), and FES conflict (β = −0.63, s.e. = 0.22; p < 0.01) and RSES (β = 0.33, s.e. = 0.07; p < 0.01) were associated with PASE scores. In a good-fitting multivariate SEM [Comparative Fit Index (CFI) = 1.00; Standardized Root Mean Square Residual (SRMR) = 0.02; Tucker–Lewis index (TLI) = 1.22; Root Mean Squared Error of Approximation (RMSEA) <0.01], RSES mediated the relationship between FES conflict and PASE (sum of indirect paths: β = −0.30, s.e. = 0.11; p < 0.01) scores.

Conclusions:

Our results highlight the importance of the relationship domain of the family environment on self-esteem, PASE, and physical fitness in adolescents who are overweight or obese.

Introduction

Rates of overweight and obesity in children and adolescents have been rising over the past several decades. Although recent data suggest a plateauing of this trend,¹ rates of type 2 diabetes in adolescents, which are largely attributed to obesity, continue to rise² and may decrease life span and quality of life. Physical inactivity contributes to overweight and obesity, and the World Health Organization (WHO) estimates that ∼70% of premature deaths in adulthood are attributed to poor health behaviors adopted during adolescence.³ Furthermore, positive physical activity patterns established in childhood have been shown to predict higher physical activity and fitness levels in adulthood.^4–6

The family environment plays a key role in establishing adolescent health behaviors, particularly physical activity. However, most of the studies conducted have focused on parenting style, parental monitoring, parental support, and parental (modeling) behaviors, as well as the physical environment.^7,8 Only a few studies have evaluated associations between relationship domains of the family environment and physical activity in adolescents. One study used an overall health measure (a health “cluster” based on physical activity, BMI, sleep, diet, smoking, and alcohol use), rather than a single measure of physical activity, and found that adolescents in the “most healthy clusters” reported having the highest levels of family cohesion as measured using the Family Adaptability and Cohesion scale.⁹

Subsequently, Ornelas et al.¹⁰ found that greater family cohesion was associated with meeting self-reported moderate-to-vigorous physical activity (MVPA) guidelines, which was defined as 5 or more weekly bouts of at least 30 minutes of MVPA (five to eight metabolic equivalents, METs). In this study, family cohesion was measured by summing responses to three items (how much people in their family understand them, how much they and their family have fun together, and how much their family pays attention to them). Using the relationship dimensions of the Family Environment Scale (FES), Kalavana et al.¹¹ reported that family cohesion was positively associated with self-reported physical activity, but no statistically significant association between family conflict and physical activity was found. Recently, Bigman et al.¹² reported that family cohesion but not family conflict, as measured using subscales of the Family Life Questionnaire, was associated with self-reported MVPA, defined as engaging in 60 minutes or more of physical activity on 5 or more days of the week in the past week.

Even less is known about how the family environment affects fitness levels, which may provide a more objective and accurate measure of typical, habitual physical activity patterns. In particular, recovery heart rate (RHR) following the 3-minute step test has been shown to be an effective measure for evaluating fitness in children and adolescents.^13–15

Furthermore, very little is known regarding whether the associations observed between the relationship dimensions of the family environment and physical activity (and, putatively, physical fitness) might be mediated by self-efficacy and/or self-esteem. It is well known that physical activity leads to higher self-esteem in adolescents, especially those who are overweight.¹⁶ Furthermore, self-esteem is an important determinant of psychological well-being as well as increased self-efficacy relative to physical activity and other health behaviors.^17,18 Interventions promoting activity in adolescents,¹⁹ including those conducted outside of the school system,²⁰ show the effects of the intervention on objectively measured physical activity are mediated by self-efficacy. In addition, intervention programs that aim to improve physical activity and self-esteem have shown improvement in physical activity and self-esteem, as well as self-efficacy²¹; however, potential relations between the family environment, self-esteem, self-efficacy, and physical fitness have not been well explored.

Therefore, we evaluated potential associations between the relationship subdomains (cohesion, conflict, expressivity) of the FES, self-esteem, physical activity self-efficacy (PASE), and fitness using RHR from a 3-minute submaximal step test at baseline. Participants were 108 adolescents who were overweight or obese seeking weight-loss treatment as part of the Healthy Kids, Healthy Weight (HKHW) 12-week multidisciplinary pediatric weight management program. We hypothesized that higher family cohesion and expressivity and lower conflict would be associated with higher levels of fitness. We further hypothesized that these associations between the family environment and fitness would be mediated by self-esteem and PASE.

Methods

Study Population

HKHW is an ongoing family-based behavioral weight management program at University Hospitals Rainbow Babies & Children's Hospital, details of which have been described elsewhere.²² Briefly, children and adolescents between 7 and 18 years of age with a BMI at or above the 85th percentile are eligible. Child and adolescent participants and their adult caregivers (legal guardian) are recruited from urban, suburban, and rural areas of northeast Ohio to represent diverse ethnic and socioeconomic backgrounds. The program has been approved by the Institutional Review Board at University Hospitals, and all participants and their caregivers provide informed assent and consent. The baseline evaluations include a medical evaluation, psychological screening, and physical fitness and body composition assessments.

For the current study, we include a subsample of the overall study population of participants and their primary adult caregiver who completed the fitness and body composition tests as well as all of the questionnaires listed below at baseline (n = 108). In this article, the term “participant” is used when describing the adolescent participating in the baseline testing (for fitness and body composition) and completing questionnaires (for self-efficacy, self-esteem, and physical activity: please see “Behavioral Measures” below for details) at baseline, and the term “caregiver” is used to describe the primary adult caregiver (legal guardian) of the adolescent participant who completed questionnaires for demographics and family environment (please see “Behavioral Measures” below for details). The primary adult caregiver was, in most cases, the mother.

Fitness Measure

Fitness was measured in the adolescent participant at baseline using the YMCA 3-minute step test performed on a 30 cm bench at a rate of 24 steps per minute to the beat of a metronome set at 96 beats per minute (bpm).¹⁵ Immediately after the 3-minute step test, the heart rate of subjects was measured for 60 seconds at the radial artery to determine RHR, expressed in bpm. RHR following the 3-minute step test has been shown to be an effective measure for evaluating fitness in children and adolescents.^13–15

Body Composition Measures

In the adolescent participant, height was measured using a stadiometer and weight was measured with a digital scale. BMI percentiles and Z-scores were calculated using CDC growth charts.²³

Demographic and Behavioral Questionnaire Measures

Demographics were obtained from a questionnaire that included questions regarding child age, gender, race, parental education, and annual family income that was completed by the primary adult caregiver.

Physical activity was measured by a single question regarding how much time was spent performing MVPA per day with possible scores ranging from 1 to 7 (1 = <20 minutes; 2 = 20–30 minutes; 3 = 30–60 minutes; 4 = 1–2 hours; 5 = 2–3 hours; 6 = 3–4 hours; 7 = >4 hours). This question was completed by the adolescent participant.

Family environment was assessed using the relationship dimension of three subscales (conflict, cohesion, expressivity) from the FES.^24,25 Each subscale consists of nine statements evaluated on a 5-point scale. Cohesion measures the degree of commitment, help, and support within family members; conflict evaluates discord among family members; and expressivity assesses the extent to which family members are encouraged to express their feelings.^24,25 Each item was coded so that higher scores represent a higher level of the attribute. Based on data from our participants, the FES conflict subscale (Cronbach's α = 0.75) and FES cohesion subscale (Cronbach's α = 0.70) demonstrated good to acceptable internal consistency reliability, but the expressivity subscale had poor internal reliability (Cronbach's α = 0.54). This questionnaire was completed by the primary adult caregiver.

Self-esteem was measured with the Rosenberg Self-Esteem Scale (RSES).²⁶ The 10 items, each evaluated on a 4-point scale (strongly agree to strongly disagree), were coded so that higher scores represent higher self-esteem. Based on data from our participants, the RSES (Cronbach's α = 0.91) demonstrated very good internal consistency reliability. This questionnaire was completed by the adolescent participant.

Self-efficacy was assessed with the PASE questionnaire.²⁷ The eight items are coded so that higher scores represent higher self-efficacy of being active despite encountering barriers, which are summed to provide a total score and can be averaged for a mean score. Based on data from our participants, the total PASE scores ranged from 11 to 32 and the 8-item average PASE score ranged from 1.37 to 4. Furthermore, the PASE scale (Cronbach's α = 0.88) demonstrated very good internal consistency reliability. This questionnaire was completed by the adolescent participant.

Statistical Analysis

Descriptive statistics (mean, standard deviation, counts, frequencies) were first calculated for the study population, which is a subsample of the total study population (comprising 108 families: adolescent participant and primary adult caregiver) with complete data on all aforementioned baseline tests and questionnaires. Internal consistency reliability of the behavioral measure scales was estimated using Cronbach's α. Then, we evaluated univariate/crude and multivariable general linear models adjusting for age, gender, race, BMI Z-score, parent education, and income using SAS v9.4 (SAS Institute, Inc., Cary, NC).

Structural equation modeling (SEM) (“path analysis”) was used to evaluate overall model fit and hypothesized paths (relations) between observed (not latent) variables using maximum likelihood estimation in MPlus v7.11 (Muthen & Muthen). Overall model fit was assessed using the Chi-square test (χ²; degrees of freedom, df), the Comparative Fit Index (CFI), the Root Mean Squared Error of Approximation (RMSEA), and the Standardized Root Mean Square Residual (SRMR).²⁸ Because of the sensitivity of the Chi-square test to sample size, the Tucker–Lewis index (TLI; also known as the non-normed fit index (NNFI)) was also evaluated, which is based on the χ²/df ratio.²⁸ Models were classified as having a good model fit based on the following criteria for the fit indices: CFI ≥0.95; RMSEA ≤0.06; TLI >0.90; and SRMR ≤0.08.^28,29

Bootstrapping with 10,000 bootstrap samples was used to estimate the standard errors of the direct and indirect effects when testing for mediation. Initially, a full model was evaluated (Fig. 1) that included all hypothesized direct and indirect effects between FES subscales (conflict, cohesion), self-esteem (RSES), self-efficacy (PASE), and fitness (RHR) based upon results from the multiple regression models. We note that because the internal consistency reliability of the FES expressivity subscale in our population was so low (Cronbach's α = 0.54), we did not include the FES expressivity subdomain in the initial model. We removed paths from the full model that were not even marginally statistically significant (p > 0.20) and, then, pruned the model further to include only paths with p < 0.10 in the final model (Fig. 2).

FIG. 1.

SEM of paths between FES cohesion (COH) and conflict (CON), self-esteem (RSES), PASE, and RHR after a 3-minute step test. Betas and standard errors of betas from 10,000 bootstrap samples. Paths adjusted for age, race, gender, BMI z-score, education, and income (not shown for clarity). Overall model fit indices: CFI = 1.000; TLI = 1.084; SRMR = 0.056; RMSEA <0.01 (90% CI: <0.01–0.083). CFI, Comparative Fit Index; FES, Family Environment Scale; PASE, physical activity self-efficacy; RHR, recovery heart rate; RMSEA, Root Mean Squared Error of Approximation; RSES, Rosenberg Self-Esteem Scale; SEM, structural equation(s) model; SRMR, Standardized Root Mean Square Residual; TLI, Tucker–Lewis index.

FIG. 2.

Reduced SEM of paths between FES conflict (CON), self-esteem (RSES), PASE, and RHR after the 3-minute step test. Betas and standard errors of betas from 10,000 bootstrap samples. Paths adjusted for age, race, gender, BMI z-score, education, and income (not shown for clarity). Overall model fit indices: CFI = 1.000; TLI = 1.218; SRMR = 0.018; RMSEA <0.01 (90% CI: <0.01 to 0.086).

Results

Characteristics of the study population are provided in Table 1. Adolescent participants were, on average, ∼14 years of age, and there were slightly more females than males (58.5% vs. 41.5%) and more Caucasians than African Americans (56.8% vs. 40.0%). The overall study population of participants (n = 394) has similar gender (61% females) and ethnic (53% Caucasians, 35% African Americans) distributions, and the mean age of the overall study population is slightly younger (∼11 years of age) than the subsample evaluated.

Table 1.

Characteristics of the Healthy Kids Healthy Weight Study Population at Baseline

Characteristic	Mean (SD) or Number (%)
Age (years)	13.9 (1.8)
Gender
Males	49 (41.5%)
Females	69 (58.5%)
Race
Caucasian	67 (56.8%)
African American	46 (40.0%)
Other	5 (4.2%)
BMI percentile	98.3 (1.2)
BMI Z-score	2.36 (0.35)
Total family income
Less than $5000	13 (11.2%)
$5000–$9999	7 (6.0%)
$10,000–$14,999	5 (4.3%)
$15,000–$19,999	1 (0.9%)
$20,000–$24,999	10 (8.6%)
$25,000–$29,999	7 (6.0%)
$30,000–$34,999	4 (3.4%)
$35,000–$39,999	6 (5.2%)
$40,000–$54,999	10 (8.6%)
$55,000–$74,999	13 (11.2%)
$75,000–$100,000	25 (21.6%)
Over $100,000	15 (12.9%)
Parent education
Junior high school	0 (0%)
Some high school	6 (5.1%)
High school graduate or GED	20 (16.9%)
Partial college	41 (34.7%)
College/University Bachelor's Degree	25 (21.2%)
Graduate degree	26 (22.0%)
Resting heart rate (bpm)	105.1 (18.1)
Moderate-to-vigorous physical activity	2.11 (1.47)
FES: cohesion	16.2 (1.8)
FES: conflict	12.0 (2.3)
FES: expressivity	15.1 (1.9)
Rosenberg Self-Esteem Scale	21.2 (6.1)
Physical activity self-efficacy	2.8 (0.6)

bpm, beats per minute; FES, Family Environment Scale.

First, potential associations were evaluated between family environment subscales, self-esteem (RSES), self-efficacy (PASE), and fitness, as measured using RHR after a 3-minute step test (Table 2). In multivariable models adjusted for age, gender, race, BMI z-score, parent education, and income, higher FES cohesion (β = −2.18, s.e. = 0.98; p = 0.02), higher FES expressivity (β = −1.97, s.e. = 0.99; p < 0.05), and higher PASE (β = −0.64, s.e. = 0.33; p < 0.05) scores were associated with lower RHR, which represents better fitness (Table 2). While potential associations between family environment subscales, self-esteem (RSES), self-efficacy (PASE), and self-reported MVPA were evaluated, only the association between PASE scores and MVPA was statistically significant at the 0.05 level (cohesion: β = 0.14, s.e. = 0.08, p = 0.08; conflict: β = 0.03, s.e. = 0.07, p = 0.68; expressivity: β = 0.05, s.e. = 0.08, p = 0.50; RSES: β = 0.01, s.e. = 0.02, p = 0.83; PASE: β = 0.10, s.e. = 0.03, p < 0.01) (data not shown).

Table 2.

Multiple Regression Between Family Environment Relationship Domains, Self-Esteem, and Physical Activity Self-Efficacy and Recovery Heart Rate After a 3-Minute Step Test

Predictors	β	s.e. of β	p
FES: cohesion	−2.1803	0.9755	0.02
FES: conflict	0.0686	0.8751	0.94
FES: expressivity	−1.9694	0.9922	0.05^a
Rosenberg Self-Esteem Scale	−0.4489	0.2688	0.09
Physical activity self-efficacy	−0.6442	0.3266	0.05^a

Adjusted for age, gender, race, BMI z-scores, parent education, income.

p < 0.05, rounded to 0.05.

Next, potential associations between the family environment subscales and self-esteem (RSES) were evaluated. Higher FES conflict scores were associated with lower RSES scores (β = −0.83, s.e. = 0.29; p < 0.01) (Table 3). Then, potential associations between family environment subscales, self-esteem (RSES), and PASE were evaluated. Higher FES conflict scores were associated with lower PASE scores (β = −0.63, s.e. = 0.22; p < 0.01), and higher RSES scores were associated with higher PASE scores (β = 0.33, s.e. = 0.07; p < 0.01) (Table 4).

Table 3.

Multiple Regression Between Family Environment Relationship Domains and Self-Esteem

Predictors	β	s.e. of β	p
FES: cohesion	0.4109	0.3455	0.23
FES: conflict	−0.8263	0.2931	0.01^a
FES: expressivity	0.3842	0.3321	0.24

Adjusted for age, gender, race, BMI z-scores, parent education, income.

p = 0.005 rounded to p < 0.01.

Table 4.

Multiple Regression Between Family Environment Relationship Domains, Self-Esteem, and Physical Activity Self-Efficacy

Predictors	β	s.e. of β	p
FES: cohesion	0.4362	0.2585	0.09
FES: conflict	−0.6325	0.2167	<0.01^a
FES: expressivity	0.3553	0.2397	0.13
Rosenberg Self-Esteem Scale	0.3312	0.0695	<0.01^b

Adjusted for age, gender, race, BMI z-scores, parent education, income.

p = 0.003 rounded to p < 0.01.

p < 0.001 rounded to p < 0.01.

A full model that included all hypothesized direct and indirect effects was tested between FES subscales with adequate to good reliability in our study population (conflict, cohesion), self-esteem (RSES), self-efficacy (PASE), and fitness (RHR) (Fig. 1). Although the overall model fit was good (CFI = 1.000; SRMR = 0.056; TLI = 1.08; RMSEA <0.01), several paths were not statistically significant, including paths from FES cohesion to RSES and FES cohesion to PASE. Therefore, the model was pruned to remove all paths that were not at least marginally statistically significant (p < 0.10), including the path from FES conflict to fitness (RHR) (p = 0.18) (Fig. 2). In a good-fitting multivariate (SEM) model (CFI = 1.00; SRMR = 0.02; TLI = 1.22; RMSEA <0.01), self-esteem (RSES) mediated the relationship between FES conflict and self-efficacy (PASE) scores (sum of indirect paths: β = −0.30, s.e. = 0.11; p < 0.01; β_std = −0.15, s.e. = 0.05; p < 0.01) and higher PASE scores were marginally associated with lower RHR, which represents higher fitness (β = −0.76, s.e. = 0.44; p = 0.08; β_std = −0.18, s.e. = 0.10; p = 0.08).

Discussion

In multivariable models, higher family cohesion and expressivity scores were associated with lower RHR, which represents better physical fitness levels, and higher PASE was associated with better fitness as well. In addition, higher family conflict and lower self-esteem were related to lower PASE (i.e., lower perceived ability to overcome barriers to being physically active). In a good-fitting multivariate SEM, self-esteem partially mediated the relationship between family conflict and PASE. Our findings, therefore, indicate that relationship elements of the family environment may be important targets for intervention to improve the health of adolescents who are obese, in particular targeting how family members work as a team, communicate about their feelings, and manage disagreements.

As an example of such an intervention, Robin and Foster's Behavioral Family Systems Therapy (BFST) seeks to reduce parent–child conflict and improve communication and has been a part of interventions to improve adherence to treatment for adolescents with health conditions such as type 1 diabetes and cystic fibrosis.^30–33 Using these strategies for improving the adherence of children who are obese to physical activity recommendations may be an avenue worth pursuing further. In addition, more general child clinical techniques, such cognitive behavioral therapy strategies for improving self-esteem, may also hold promise for achieving health behavior goals.³⁴

Our multivariable results are generally consistent with prior studies that evaluated family conflict and cohesion on self-reported physical activity.^10–12 The association between family cohesion and self-reported MVPA in our study population was only marginally significant (p = 0.08); however, our measure of MVPA used was based on a single question with seven ordinal response categories, which may have limited our ability to fully observe the association. Furthermore, our use of an objective measure of fitness, RHR after a 3-minute step test, provides for a much better method of capturing chronic MVPA than the self-report of fitness used in previous research.^13–15 Although the normative data for RHR from the 3-minute step test are very limited, our study population (mean RHR: 108 ± 18 bpm) would be classified, on average, as having “good” to “very good” cardiorespiratory fitness and less than 10% of our population would be classified as having “poor” or “very poor” fitness.¹³ Our study adds to the research knowledge in this area by including a novel construct, the FES expressivity subdomain of the family environment, which was significantly associated with our objective measure of fitness. However, we note that the internal consistency reliability of the FES expressivity subscale in our population was low, which limits the interpretation of these findings. As a result, the FES expressivity subdomain was not retained in our multivariate SEM analyses.

Furthermore, no prior studies have examined the potential mediation of the associations between family cohesion and family conflict by self-esteem or self-efficacy. In our full model with all potentially relevant paths included (Fig. 1), we observed that paths from family cohesion to self-esteem and family cohesion and PASE were not statistically significant in the presence of family conflict and the other variables shown. In a reduced model that pruned all paths without statistically significant (p ≤ 0.05) or marginally significant associations (p < 0.10), self-esteem mediated the relationship between family conflict and PASE. Higher PASE was only marginally associated with lower RHR, which represents higher fitness, in a good-fitting model (Fig. 2).

Although it is well established that self-efficacy is a key determinant of physical activity and physical fitness,^35–37 less is known about the relationships between family conflict, self-esteem, and self-efficacy in the context of physical activity. Self-esteem has been associated with physical activity in observational studies,³⁸ and a recent meta-analysis of randomized controlled trials found that self-esteem is improved by physical activity interventions.³⁹ Furthermore, self-esteem affects sports participation, and this association has been shown to be mediated by perceived sports competence in adolescents.⁴⁰

In addition, family adversity (domestic violence, family substance use, broken parental marriage) has been shown to be associated with self-esteem (using the RSES), and self-esteem has been shown to mediate the relationship between parent–adolescent conflict and suicide attempts⁴¹ in adolescents. However, whether or not family conflict is associated with self-esteem and PASE, and, in turn, physical fitness, is largely unknown. Our findings suggest that family conflict (measured using the FES) is associated with self-esteem (measured using the RSES) and self-efficacy (measured using the PASE), which, in turn, is marginally predictive of fitness (measured using RHR from the 3-minute step test). However, additional studies are needed to confirm these findings. It would be important to determine longitudinally whether interventions to reduce family conflict would relate to improved PASE and eventually to better fitness levels, if adolescents acted on their improved confidence in overcoming barriers to engaging in regular physical activity.

Strengths of our study include our diverse population of adolescents who are primarily obese and comprised about 40% of African Americans. In addition, as mentioned earlier, an objective measure of physical fitness (RHR from the 3-minute step test) was used versus relying on self-reported physical activity levels. The use of SEM enabled the illumination of which subscales of the family environment remained statistically significant in the presence of each other and in the presence of other putative mediators.

Limitations of our study include a relatively small sample size, which limited our statistical power to conduct multivariable subgroup and multivariate multiple group SEM analyses by gender and race. Furthermore, although our sample size was close to the lower bound of the general minimum recommended sample size (of ∼100) for SEM,^28,42,43 our model was relatively simple with only a small number of parameters estimated. In addition, we used bootstrapping to estimate the standard errors of the parameters,⁴⁴ which is the recommended procedure for testing for mediation.^42,45 We evaluated overweight and obese adolescents within a limited range of BMI percentiles (range: 92%–99%; mean: 98.3 ± 1.2) and BMI z-scores (range: 1.47–3.24; mean: 2.36 ± 0.35), which limits the generalizability of our findings.

In addition, the cross-sectional nature of our data is an inherent limitation that prohibits any inference of causality. We did not collect information on the number of siblings or the number of people living in the household, which could potentially affect the family environment. Furthermore, in our study, the FES was completed by the primary caregiver and not the adolescent participant, as we believe that the primary caregiver in this situation is better suited to provide information about the family environment. We do not believe that having the caregiver (proxy) report of the family environment would change the interpretation of the results presented; however, future studies should evaluate whether or not there are substantive differences in responses to the FES between adolescent participants and primary caregivers. Future studies should also evaluate how the family environment may affect fitness differently with respect to race and gender and should evaluate how lifestyle programs may potentially modify associations between the relationship dimensions of the family environment and fitness through modification of mediators (e.g., self-efficacy) as mentioned previously.

In summary, this is the first report to evaluate the relationship domain of the family environment on physical fitness. Our results suggest the relationship domain of the family environment is an important determinant of PASE and, potentially, physical fitness, and that the relationship between family conflict and PASE is mediated by self-esteem. Our results also highlight the importance of using multivariate analyses (SEM) to better understand the key determinants of fitness in adolescents seeking treatment for weight management.

Footnotes

Acknowledgments

The authors would like to acknowledge all of the children and their families who have been evaluated and taken part in the Healthy Kids, Healthy Weight program; the many staff members who have assisted with this clinical research program; and Kathy Sanniti, who has been a consistent supporter and advocate for our program. This work was supported by the Rainbow Babies and Children's Foundation, which is a supporting organization of University Hospitals Rainbow Babies and Children's Hospital.

Authors Disclosure Statement

No competing financial interests exist.

References

Ogden

, Carroll

, Kit

, Flegal

. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA, 2014; 311:806–814.

Dabelea

, Mayer-Davis

, Saydah

, et al. Prevalence of type 1 and type 2 diabetes among children and adolescents from 2001 to 2009. JAMA, 2014; 311:1778–1786.

World Health Organization (WHO). 2001. www.who.int/maternal_child_adolescent/documents/frh_adh_98_18/en/ (last accessed May 23, 2016).

Huotari

, Nupponen

, Mikkelsson

, et al. Adolescent physical fitness and activity as predictors of adulthood activity. J Sports Sci, 2011; 29:1135–1141.

Huotari

, Mikkelsson

, Kujala

, et al. Physical activity and fitness in adolescence as predictors of self-estimated fitness in adulthood. J Sports Med Phys Fitness, 2013; 53:177–184.

Kwon

, Janz

, Letuchy

, et al. Developmental trajectories of physical activity, sports, and television viewing during childhood to young adulthood: Iowa bone development study. JAMA Pediatr, 2015; 169:666–672.

Hsu

, Chou

, Nguyen-Rodriguez

, et al. Influences of social support, perceived barriers, and negative meanings of physical activity on physical activity in middle school students. J Phys Act Health, 2011; 8:210–219.

Lawman

, Wilson

. A review of family and environmental correlates of health behaviors in high-risk youth. Obesity (Silver Spring), 2012; 20:1142–1157.

De Bourdeaudhuij

, Van Oost

. Family characteristics and health behaviors of adolescents and families. Psychol Health, 1998; 13:785–803.

10.

Ornelas

, Perreira

, Ayala

. Parental influences on adolescent physical activity: A longitudinal study. Int J Behav Nutr Phys Act, 2007; 4:3.

11.

Kalavana

, Lazarou

, Christodoulou

. Family environment in relation to eating and health risk behaviors in adolescents. 2011; 7:15–25.

12.

Bigman

, Rajesh

, Koehly

, et al. Family cohesion and moderate-to-vigorous physical activity among Mexican origin adolescents: A longitudinal perspective. J Phys Act Health, 2015; 12:1023–1030.

13.

Jankowski

, Niedzielska

, Brzezinski

, Drabik

. Cardiorespiratory fitness in children: A simple screening test for population studies. Pediatr Cardiol, 2015; 36:27–32.

14.

Laguna

, Aznar

, Lara

, et al. Heart rate recovery is associated with obesity traits and related cardiometabolic risk factors in children and adolescents. Nutr Metab Cardiovasc Dis, 2013; 23:995–1001.

15.

Simhaee

, Corriveau

, Gurm

, et al. Recovery heart rate: An indicator of cardiovascular risk among middle school children. Pediatr Cardiol, 2013; 34:1431–1437.

16.

Schmalz

, Deane

, Birch

, Davison

. A longitudinal assessment of the links between physical activity and self-esteem in early adolescent non-Hispanic females. J Adolesc Health, 2007; 41:559–565.

17.

Strauss

. Childhood obesity and self-esteem. Pediatrics, 2000; 105:e15.

18.

Strong

, Malina

, Blimkie

, et al. Evidence based physical activity for school-age youth. J Pediatr, 2005; 146:732–737.

19.

Lewis

, Marcus

, Pate

, Dunn

. Psychosocial mediators of physical activity behavior among adults and children. Am J Prev Med, 2002; 23(2 Suppl):26–35.

20.

Lytle

, Murray

, Evenson

, et al. Mediators affecting girls' levels of physical activity outside of school: Findings from the trial of activity in adolescent girls. Ann Behav Med, 2009; 38:124–136.

21.

Tirlea

, Truby

, Haines

. Pragmatic, randomized controlled trials of the girls on the go! Program to improve self-esteem in girls. Am J Health Promot, 2015. [Epub ahead of print.]

22.

Chuang

, Fehr

, Ievers-Landis

, et al. Associations of sleep duration and regularity with level of obesity among youth in a weight loss program. Trans Issues Psych Sci, 2015; 1:45–56.

23.

Kuczmarski

, Flegal

. Criteria for definition of overweight in transition: Background and recommendations for the United States. Am J Clin Nutr, 2000; 72:1074–1081.

24.

Moos

, Insel

, Humphrey

. Family Environment Scale. Consulting Psychologist Press: Palo Alto, CA, 1974.

25.

Moos

, Moos

. Life Stressors and Social Resources Inventory: Adult Form Manual. Psychological Assessment Resources: Odessa, FL, 1994.

26.

Rosenberg

Rosenberg self-esteem scale (RSE). Acceptance Commitment Ther Measures Package, 61.

27.

Motl

, Dishman

, Trost

, et al. Factorial validity and invariance of questionnaires measuring social-cognitive determinants of physical activity among adolescent girls. Prev Med, 2000; 31:584–594.

28.

Kline

. Principles and Practice of Structural Equation Modeling, pp. 133–145. Guilford Press: New York, 2005.

29.

, Bentler

. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Eq Model, 6:1–55.

30.

DeLambo

, Ievers-Landis

, Drotar

, Quittner

. Association of observed family relationship quality and problem-solving skills with treatment adherence in older children and adolescents with cystic fibrosis. J Pediatr Psychol, 2004; 29:343–353.

31.

Harris

, Freeman

, Beers

. Family therapy for adolescents with poorly controlled diabetes: Initial test of clinical significance. J Pediatr Psychol, 2009; 34:1097–1107.

32.

Wysocki

, Harris

, Buckloh

, et al. Effects of behavioral family systems therapy for diabetes on adolescents' family relationships, treatment adherence, and metabolic control. J Pediatr Psychol, 2006; 31:928–938.

33.

Wysocki

, Harris

, Buckloh

, et al. Randomized, controlled trial of Behavioral Family Systems Therapy for Diabetes: Maintenance and generalization of effects on parent-adolescent communication. Behav Ther, 2008; 39:33–46.

34.

Vancampfort

, Probst

, Adriaens

, et al. Changes in physical activity, physical fitness, self-perception and quality of life following a 6-month physical activity counseling and cognitive behavioral therapy program in outpatients with binge eating disorder. Psychiatry Res, 2014; 219:361–366.

35.

Bauman

, Reis

, Sallis

, et al. Correlates of physical activity: Why are some people physically active and others not?. Lancet, 2012; 380:258–271.

36.

Babic

, Morgan

, Plotnikoff

, et al. Physical activity and physical self-concept in youth: Systematic review and meta-analysis. Sports Med, 2014; 44:1589–1601.

37.

Young

, Plotnikoff

, Collins

, et al. Social cognitive theory and physical activity: A systematic review and meta-analysis. Obes Rev, 2014; 15:983–995.

38.

Ahn

, Fedewa

. A meta-analysis of the relationship between children's physical activity and mental health. J Pediatr Psychol, 2011; 36:385–397.

39.

Liu

, Wu

, Ming

. How does physical activity intervention improve self-esteem and self-concept in children and adolescents? Evidence from a meta-analysis. PLoS One, 2015; 10:e0134804.

40.

Wagnsson

, Lindwall

, Gustafsson

. Participation in organized sport and self-esteem across adolescence: The mediating role of perceived sport competence. J Sport Exerc Psychol, 2014; 36:584–594.

41.

Kuhlberg

, Pena

, Zayas

. Familism, parent-adolescent conflict, self-esteem, internalizing behaviors and suicide attempts among adolescent Latinas. Child Psychiatry Hum Dev, 2010; 41:425–440.

42.

Hancock

, Mueller

. Structural Equation Modeling: A Second Course. Information Age Publishing: Greenwich, CT, 2006.

43.

Hoyle

, Gottfredson

. Sample size considerations in prevention research applications of multilevel modeling and structural equation modeling. Prev Sci, 2015; 16:987–996.

44.

Ievers-Landis

, Burant

, Hazen

. The concept of bootstrapping of structural equation models with smaller samples: An illustration using mealtime rituals in diabetes management. J Dev Behav Pediatr, 2011; 32:619–626.

45.

Preacher

, Hayes

. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods, 2008; 40:879–891.