Latent Profiles of Physical Activity and Sedentary Behavior in Elementary School-Age Youth:Associations With Health-Related Quality of Life

Latent Profiles of Physical Activity and Sedentary Behavior in Elementary School-Age... Abstract Objective The objectives were to identify and describe profiles of elementary school-age youth based on objective measurements of total time spent in moderate-to-vigorous physical activity (MVPA) and sedentary behavior (SB) and in bouts of the activities, to examine predictors of profiles, and to examine whether profiles were differentially associated with physical and psychosocial health-related quality of life (HRQOL). Methods Participants included 204 youth (aged 8–11 years) who wore accelerometers to gather objective activity data. The average proportion of time in MVPA and SB and average proportion of time in MVPA and SB bouts were used for analyses. Participants completed a self-report measure of HRQOL. Latent variable mixture modeling was conducted. Results Overall, participants did not meet the MVPA guideline (M = 50.7 min) and spent 47% of waking hours in SB, indicating that elementary school is a critical developmental period to study activity levels. Three profiles emerged: “Active,” “Inactive,” and “Moderate.” Boys were significantly more likely to be in the Active profile than the other profiles, and older youth were significantly more likely to be in the Inactive profile. After controlling for child sex and age, participants in the Active profile reported significantly higher psychosocial HRQOL than the participants in the other profiles; however, no significant differences were found in physical HRQOL. Conclusions Identification of these naturally occurring profiles suggests need for interventions early in development focused on increasing the intensity of physical activity from light to moderate-to-vigorous for at least 60 min per day as way to improve psychosocial HRQOL. children, health-related quality of life, mixture model, physical activity, sedentary behavior Physical activity and sedentary behavior (SB) are both modifiable lifestyle behaviors that impact physical and psychosocial health outcomes in youth. For example, research has indicated that more time in moderate-to-vigorous physical activity (MVPA) is associated with better bone health, body composition, and self-concept, and with lower risk for depression, anxiety, heart disease, and diabetes (Booth, Roberts, & Laye, 2012; US Department of Health and Human Services (USDHHS), 2008). In contrast, more time in SB is associated with higher levels of adiposity and blood pressure, as well as lower levels of fitness and academic achievement (Salmon, Tremblay, Marshall, & Hume, 2011). MVPA and SB are understood to be distinct constructs with different determinants that can explain unique variance in outcomes (Fakhouri et al., 2013); therefore, the literature suggests that research should focus on both MVPA and SB. In addition to total time in MVPA and SB, there is evidence to suggest that the way in which youth accumulate time in these behaviors may be important. Specifically, emerging evidence suggests that MVPA bouts of at least 5 min are related to greater physical health outcomes, including lower weight status and waist circumference (Rowlands, Pilgrim, & Eston, 2008; Willis et al., 2015). Similarly, research has shown that SB bouts of <10 min are associated with better physical health outcomes, including lower cardiometabolic risk, fasting glucose, and weight status, independent of total time in SB and MVPA (Carson, Stone, & Faulkner, 2014; Saunders et al., 2013). Further understanding of how bouts affect physical and psychosocial outcomes in youth is needed (Altenburg & Chinapaw, 2015). As reviewed above, a number of studies have shown independent associations between total time and bouts of MVPA and SB and health outcomes in elementary school-age youth. However, predictors of poor health outcomes, including low levels of MVPA and excessive SB, often co-occur and can amplify negative physical and psychosocial outcomes (Leech, McNaughton, & Timperio, 2014). Therefore, rather than examining these constructs independently, Leech et al. (2014) recommended further research to identify patterns of behaviors. One way to accomplish this goal is to use latent variable mixture modeling (LVMM), a person-centered, model-based approach in which individuals are probabilistically assigned to unobserved groups based on similar patterns of observed variables (Berlin, Williams, & Parra, 2014). Identifying smaller homogenous profiles within a larger heterogonous population allows for the identification of characteristics and needs of the profiles to tailor interventions for these specific populations (Berlin et al., 2014). Several studies over the past decade have used LVMM to examine profiles of activity patterns in youth (Berlin et al., 2017; Heitzler et al., 2011; Huh et al., 2011; Iannotti & Wang, 2013; Trilk et al., 2012). These studies identified between three and six profiles, but most commonly, three profiles emerged: high MVPA/low SB, high SB/low MVPA, and low–moderate MVPA/SB. Common predictors of activity profiles were child age, sex, and race/ethnicity, which coincide with research showing differences in MVPA and SB based these variables (Belcher et al., 2010; Fakhouri et al., 2013). Child sex, age, and race/ethnicity were included as predictors of profiles in the current study. The primary outcome examined in relation to activity profiles in the available literature is weight status (Heitzler et al., 2011; Huh et al., 2011; Trilk et al., 2012), an objective measure of physical health. These studies found that youth who engaged in high SB/low MVPA were more likely to present with overweight/obesity. Because of the frequent examination of weight status in relation to activity profiles in the literature, weight status was not a primary variable of interest in the current study. Instead, current analyses focused on whether profiles were related to subjective reports of physical and psychosocial outcomes. Using self-reported activity, Iannotti and Wang (2013) found that adolescents in the high MVPA/low SB profile had fewer depressive symptoms, better life satisfaction, and better overall physical health. Also using self-reported activity, Berlin and colleagues (Berlin et al., 2017) found that adolescents in the high MVPA/low SB/healthy diet profile had higher levels of socioemotional protective factors (e.g., locus of control) and lower levels of socioemotional risk factors (e.g., internalizing symptoms). These studies provide preliminary evidence that youth who have high MVPA/low SB may have higher subjective reports of physical and psychosocial health. To expand on these initial examinations, an important next step is to examine health-related quality of life (HRQOL) as a measure of subjective physical and psychosocial functioning in relation to activity profiles in youth. HRQOL is a multidimensional construct of a child’s perception of his/her functioning across physical and psychosocial domains (Spieth & Harris, 1996). There is support for the independent associations of MVPA and SB with HRQOL in youth. For example, Gopinath et al. (2012) found that higher levels of MVPA were associated with higher total, physical, and psychosocial HRQOL; the opposite was true for screen-based SB. However, as noted above, health risk factors can interact to amplify negative physical and psychosocial outcomes (Leech et al., 2014). Assessing the associations among activity profiles and HRQOL can help illuminate the cumulative effects of MVPA and SB on youth’s physical and psychosocial HRQOL. The findings of the available studies that have used LVMM to examine activity profiles in youth must be considered in light of a number of limitations. First, most studies included samples of adolescents (Berlin et al., 2017; Heitzler et al., 2011); only Huh and colleagues (Huh et al., 2011) examined profiles of activity patterns in elementary school-age youth. Understanding the activity profiles in elementary school-age youth can help inform targeted intervention efforts during this critical developmental period in which activity levels dramatically change (Fakhouri et al., 2013). There are also limitations of the measurement of MVPA and SB. For example, the majority of studies included only self-report measures of MVPA and SB to inform profiles (Berlin et al., 2017; Huh et al., 2011). Because youth may be limited in their memory and recall, self-report measures may not be reliable or valid. Additionally, the two studies that used objective measures of MVPA and SB focused on the total time spent in the behaviors (Heitzler et al., 2011; Trilk et al., 2012), instead of examining bouts of MVPA and SB. The current study aimed to address gaps in the literature by using LVMM to identify and describe profiles of elementary school-age youth based on their patterns of objective MVPA and SB (total time and bouts), to examine predictors of profiles, and to examine whether profiles were differentially associated with physical and psychosocial HRQOL. First, it was hypothesized that LVMM would identify three profiles. Second, it was hypothesized that child age, sex, and race/ethnicity would predict profile membership. Third, it was hypothesized that, after controlling for significant predictors, the profile characterized by high total MVPA/more MVPA bouts and low total SB/Fewer SB bouts would be associated with the highest psychosocial HRQOL and that the profile characterized by low total MVPA/fewer MVPA bouts and high total SB/More SB bouts would be associated with the lowest psychosocial HRQOL; similar findings were hypothesized for physical HRQOL. Further, the study also examined the minimally clinically important difference (MCID) to determine whether statistically significant differences between profiles would translate into clinically significant benefits (Copay, Subach, Glassman, Polly, & Schuler, 2007). Method Participants Participants included 204 students from two Midwestern elementary schools. Eligibility criteria included children who: (a) were enrolled in mainstream 3rd through 5th grade classes, (b) spoke/read English, (c) had parental consent, and (d) wore the accelerometer for at least 1 valid weekend day and 2 valid weekdays. Census data indicated that the schools were situated in a suburban community with a median household of $46,406 (United States Census Bureau, 2015). The current sample was a part of a larger longitudinal study. At Time 1, 149 students were recruited. Data were excluded for children who were absent, moved, or declined assent (n = 4), as well as for children without adequate accelerometer data (n = 27), resulting in 118 students from Time 1. At Time 2 and Time 3, 126 additional students were recruited. Children who were absent, moved, or declined assent (n = 19) were excluded from these analyses, as well as children without sufficient accelerometer data (n = 21). Recruitment at Times 2 and 3 resulted in 86 additional participants. The final sample consisted of 204 students, which represents 55% of the total eligible population (n = 504). Only 15% of parents actively denied consent, so it is impossible to confirm that all parents received study information. Procedure Children were recruited from two schools that were chosen for diversity across demographic characteristics. Information letters and consent forms were distributed to parents during school events and in packets that were sent home with children. Assent was collected, and accelerometers were distributed. Data collection for self-report measures occurred in group format during the school day; research assistants read measures aloud to ensure that participation was not limited by reading comprehension. Objective height and weight were collected in isolated locations to ensure privacy. Procedures were approved by the authors’ institutional review board, school district’s office of research and assessment, and schools’ principals. Measures Objective Measures of MVPA and SB ActiGraph accelerometers (Model GT3X, ActiGraph, LLC) were used to obtain objective measurements of MVPA and SB. Participants were instructed to wear the devices on their nondominant hip for 3 weekdays and 2 weekend days. Raw accelerometer data were downloaded and analyzed using the ActiLife Data Analysis Software (version 6.10.4). Raw data were first binned into 10 s epochs, and sleep and nonwear periods were flagged using the Sadeh algorithm (Sadeh, Sharkey, & Carskadon, 1994) and the Troiano algorithm (Troiano et al., 2008). Waking wear time data were processed using Evenson cut points (Evenson, Catellier, Gill, Ondrak, & McMurray, 2008). Following Cain, Sallis, Conway, Van Dyck, and Calhoon (2013), valid days were those with ≥10 h of waking wear time. Participants who had at least 1 valid weekend day and 2 valid weekdays were included (Cain et al., 2013). The 3 days with the most wear time were used to calculate the average proportions of time in MVPA and SB. These proportions were calculated by dividing the average minutes by the average device wear time. Based on the best available evidence, MVPA bouts of 5 min and SB bouts of 10 min were chosen for analyses. To determine the number of MVPA bouts per day, the software calculated how many times each participant spent 5 min above 2,296 counts per minute (Evenson et al., 2008). The number of 5-min MVPA bouts was averaged, and the total number of minutes spent in 5-min bouts was divided by the average wear time. To determine the number of bouts of SB per day, the software calculated how many times each participant spent exactly 10 min under 100 counts per minute (Evenson et al., 2008). The number of 10-min SB bouts was averaged, and the total number of minutes spent in 10-min SB bouts was divided by the average wear time. Health-Related Quality of Life Participants completed the Pediatric Quality of Life Inventory (PedsQL™ 4.0; Varni, Seid, & Kurtin, 2001) as a measure of HRQOL. The PedsQL is a 23-item measure with instructions to report how much of a problem each statement has been in the past 1 month on a five-point Likert scale. The two summary scores (i.e., physical and psychosocial HRQOL) were used in current analyses. Evidence for validity indicates that the summary scores distinguished between healthy and chronically ill youth and were related to illness burden and morbidity (Varni et al., 2001). The internal consistencies for the physical and psychosocial summary scores in the current study were 0.76 and 0.86, respectively. Weight Status Staff obtained child height to the nearest 0.1 centimeter using a stadiometer. Child weight was measured to the nearest 0.1 kg using a calibrated portable digital scale. Triple measurements were averaged, and body mass index (BMI) z-scores were calculated using the SAS script provided by the Centers for Disease Control and Prevention (CDC, 2011). Weight status was used to describe children in the various activity profiles. Data Analytic Plan Preliminary Analyses Descriptive statistics and Pearson correlations were conducted for the overall sample, and logistic regression analyses were conducted to determine whether there were differences in participants and those excluded because of insufficient accelerometer data. Model Specification LVMMs were conducted using Mplus statistical software (Version 7.31; Muthén & Muthén, 1998–2012). The following indictor variables were chosen to identify the profiles: proportion of time in MVPA, SB, MVPA bouts, and SB bouts. Univariate entropy values for each variable were examined to quantify the variance that each indicator contributed to profile separation/classification. Because having a completely unrestricted model results in the estimation of a large number of parameters, minimal restrictions were applied to streamline estimation and increase parsimony (Bauer & Steinley, 2016). A “proportionality restriction” was used, which assumes that the covariance matrices are related by a constant multiplier; the correlations are restricted to be equal across profiles, but the variances are allowed to be proportionally different (Banfield & Raftery, 1993). Model Estimation First, a parsimonious one-profile model was fit to the data, followed by consecutive models with increasing numbers of profiles (up to four; Berlin et al., 2014). Maximum likelihood estimation was used to determine the parameter estimates with the highest likelihood of having produced the observed data. Owing to inclusion criteria related to accelerometer data, no missing data were present at this step. Multiple start values were used (1,000 initial-stage, 100 second-stage), and the replication of best log-likelihood was confirmed for each model (Berlin et al., 2014). Model Selection and Interpretation The Akaike information criterion (AIC; Akaike, 1987), Bayesian information criterion (BIC; Schwarz, 1978), and the sample-size adjusted BIC (ssBIC; Sclove, 1987) were examined to evaluate model fit, with the smaller values indicating better fit.1 Additionally, entropy was examined to determine the accuracy of classification; scores range from 0 to 1, with higher scores indicating better accuracy of classification (Celeux & Soromenho, 1996). Finally, the sample sizes of the profiles were considered to determine the benefit of including classes with small numbers of individuals (Lubke & Neale, 2006). Once the best fitting model was determined, descriptive statistics for the profiles were conducted. Predictors of Profiles Vermunt’s three-step approach was used to examine whether profiles could be predicted by age, sex, and race/ethnicity (Bakk, Tekle, & Vermunt, 2013). Individuals were assigned to profiles based on most likely profile membership. Next, the assigned profile was treated as an indicator variable for a new latent profile variable; demographic variables were then used as predictors. Listwise deletion was used by default; only 5% of data were missing, so no additional missing data strategies were implemented. Maximum likelihood estimation with robust standard error was used, and odds ratios (OR) were reported. Outcomes of Profiles The manual BCH method (Muthén & Muthén, 1998–2012) was used to examine the effect of the latent profile variable on the outcomes while controlling for significant covariates. In the first step, the three-profile model was estimated and the BCH weights were saved. In the second step, predictors were specified to predict profiles and outcomes, using the BCH weights as training data. Maximum likelihood estimation with robust standard error was used. Total 3% of participants were excluded from analyses because of missing data for predictors. Wald tests of equal means were examined to determine if profiles differed in means for physical and psychosocial HRQOL, after controlling for significant predictors in profiles. Minimal Clinically Important Difference For the profiles that had statistically significant differences in HRQOL, MCID was used to determine whether the differences between profiles were likely to be associated with clinically significant benefits (Copay et al., 2007). MCID signifies the smallest difference in HRQOL that the participants would identify as important, elucidating whether profile differences in HRQOL scores were greater than the variability attributed to measurement error. A common approach for measuring MCID is to compare the differences in scores with the standard error of measurement. A difference greater than or equal to the MCID is considered to be clinically meaningful (Copay et al., 2007). Results Preliminary Analyses The current study included 204 participants between the ages of 8 and 11 years (M = 9.5, SD = 0.09). The self-reported racial/ethnic makeup of the sample (presented in Table I, first column) was similar to the overall composition of the school district as indicated by district records. Most participants (71.3%) were of normal weight. Preliminary analyses revealed no differences in demographic characteristics or physical HRQOL in these participants and those who were dropped from analyses because of insufficient accelerometer data (all p > .05); study participants had significantly lower psychosocial HRQOL than those excluded (β = −.04, p = .003). Table I. Demographic Characteristics and Study Variables for Total Sample and Activity Profiles   Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)    Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)  Note. BMI = body mass; HRQOL = health-related quality of life; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior. Table I. Demographic Characteristics and Study Variables for Total Sample and Activity Profiles   Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)    Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)  Note. BMI = body mass; HRQOL = health-related quality of life; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior. Participants in the current sample accumulated similar activity levels as previous samples (Saunders et al., 2013; Willis et al., 2015), and mean physical and psychosocial HRQOL scores were also comparable (Varni et al., 2001; see Table I). Age was positively associated with total SB/SB bouts and psychosocial HRQOL (see Table II). Boys had higher total MVPA/MVPA bouts and lower total SB. Total MVPA was strongly positively associated with MVPA bouts and negatively associated with total SB/SB bouts. Similarly, total SB was strongly positively associated with SB bouts and negatively associated with MVPA bouts. MVPA bouts and SB bouts were negatively associated. Total MVPA/MVPA bouts were positively correlated with both HRQOL measures. Psychosocial and physical HRQOL were positively associated. Table II. Correlations Among Study Variables   1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50    1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50  Notes. *p < .05; **p < .01; Child sex (0 = female, 1 = male); race/ethnicity (0 = Caucasian; 1 = Minority); BMI = body mass index; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior; HRQOL = health-related quality of life. Table II. Correlations Among Study Variables   1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50    1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50  Notes. *p < .05; **p < .01; Child sex (0 = female, 1 = male); race/ethnicity (0 = Caucasian; 1 = Minority); BMI = body mass index; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior; HRQOL = health-related quality of life. Primary Analyses Results from the one-, two-, and three-profile models are presented in Table III. The four-profile model was considered, but ultimately rejected for two reasons. First, the four-profile model produced significant estimation errors, requiring many parameters to be fixed to avoid singularity of the information matrix. Second, the four-profile model resulted in one profile that included only seven participants. Results indicated that models with two or three latent profiles fit the data better than a unitary model without profiles. Further, AIC, BIC, and ssBIC favored the three-profile solution, and the overall entropy statistic favored the two-profile model. However, as noted previously, entropy is typically used to determine classification accuracy, not to select optimal number of profiles. Therefore, it was concluded that the three-profile model best fit the data. Univariate entropy values for indicators ranged from 0.46 to 0.57, supporting the retention of all indicators because of their nonnegligible contribution to profile separation/classification. Table III. Model Fit of the LVMMs Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Note. The four-profile model produced estimation errors and was omitted from comparisons; optimal models according to criteria are bolded; other fit indices are reported for completeness; AIC = Akaike’s information criterion; BIC = Bayesian information criterion; LVMM = latent variable mixture modeling; ssBIC = sample-size adjusted BIC. Table III. Model Fit of the LVMMs Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Note. The four-profile model produced estimation errors and was omitted from comparisons; optimal models according to criteria are bolded; other fit indices are reported for completeness; AIC = Akaike’s information criterion; BIC = Bayesian information criterion; LVMM = latent variable mixture modeling; ssBIC = sample-size adjusted BIC. The descriptive statistics of the profiles are presented in Table I (Columns 2–4; for graphical depiction, see Online Supplementary Material). Participants in Profile 1 (“Active”; n = 29) were characterized by the highest amounts of MVPA; further, 93% averaged over 60 min of MVPA per day, thus meeting the US Department of Health and Human Services (USDHHS, 2009) guideline. Girls were significantly less likely to be in the Active profile than Profiles 2 and 3 (OR = 4.97 and 6.51, respectively). Similarly, boys were more likely to be in the Active profile than Profiles 2 and 3 (OR = 0.20 and 0.15, respectively). The largest profile, Profile 2 (“Moderate”; n = 148), was characterized by moderate amounts of MVPA and SB. Only 25% met the 60-min guideline. The Active and Moderate profiles had similar SB (M = 396.0 vs. 382.9 min), but the Moderate profile had less MVPA (M = 46.1 vs. 94.9 min) and more light activity (M = 432.5 vs. 404.2 min). Profile 3 (“Inactive”; n = 27) was characterized by the highest levels of SB. Total 0% met the USDHHS (2008) guideline. Older youth were significantly more likely to be in the Inactive profile than the Active or Moderate profiles (OR = 0.30 and 0.34, respectively). The next step was to determine the associations between the activity profiles and psychosocial and physical HRQOL, after controlling for child sex and age, which were significant predictors of profile membership. The results indicated that participants in the Active profile reported significantly higher psychosocial HRQOL than the participants in the Moderate and Inactive profile (χ2 = 5.74, p = .02 and χ2 = 6.44, p = .01, respectively). Moderate and Inactive profiles did not significantly differ in terms of psychosocial HRQOL. MVPA profile membership was not significantly associated with physical HRQOL (p > .05). Finally, MCID was calculated to determine whether the statistically significant differences in psychosocial HRQOL between the Active profile, and the other two profiles were consistent with clinically significant benefits. MCID for psychosocial HRQOL was 5.56. The mean differences (HRQOL) between the Active profile and the Moderate and Inactive profiles were 6.4 and 6.0, respectively, indicating clinically meaningful differences for both comparisons. Discussion The first hypothesis, that three profiles would be identified, was supported. Specifically, the following profiles were identified: Active, Inactive, and Moderate. Consistent with Iannotti and Wang (2013), the majority of children in the current study fell into the Moderate profile. Consistent with the review by Leech et al. (2014), older youth, particularly females, tended to fall in the profile characterized by low MVPA and high SB and younger youth, particularly males, tended to fall in the profile characterized by high levels of MVPA. The consistency in findings across studies (Berlin et al., 2017; Heitzler et al., 2011, Iannotti & Wang, 2013) provides support for three profiles being reliable subgroups within the heterogeneous population. An anticipated advantage of the current study was that it was the first to identify profiles based on total time in MVPA and SB and time in bouts. However, time in bouts was strongly associated with total MVPA time, limiting the impact of bouts on the identification of profiles. That is, children who accumulated more time in MVPA/SB did so in bouts of activity, rather than in shorter “bursts.” An important direction for further study is whether different bout lengths might result in different outcomes. Further, the inclusion of breaks in bouts (i.e., when participant transitions from sitting to standing or from standing to sitting) might further differentiate profiles. Although the inclusion of breaks was considered in the present study, model complexity and convergence issues prevent our use of this variable in analyses. Researchers frequently expect that study samples reflect a homogenous population and that individual variability is because of random causes (Bauer & Steinley, 2016). With this mindset, researchers often examine the skewness and kurtosis of variables to confirm that they can be considered normal in subsequent analyses (e.g., ±3 for skewness; ±10 for kurtosis; Kline, 2011). If the current study had taken these steps, the study variables would have been considered “normally distributed” (i.e., skewness and kurtosis values were within normal limits). However, that decision would make the variables the focus of analyses, attempting to make them fit within a normal framework, instead of focusing on the naturally occurring groups of individuals within the population. Our findings provide support for using LVMM to examine MVPA and SB in youth, and may recommend this approach in other samples of children and youth. The second hypothesis, that child sex, age, and race/ethnicity would predict profile membership, was partially supported. Consistent with previous studies (Berlin et al., 2017; Huh et al., 2011), age and sex were found to predict membership. Contrary to previous studies (Berlin et al., 2017; Huh et al., 2011), race/ethnicity did not predict membership. A possible explanation is that elementary school-age youth may have been unable to accurately self-report on race/ethnicity. Additionally, previous studies found differences in profiles for specific minority groups (e.g., Hispanic, Black). In the current sample, only 2.1 and 5.2% of the participants self-reported as Hispanic and Black, respectively. Significant findings may be found with a larger, nationally representative sample, and with parent-reported race/ethnicity. The third hypothesis, that the Active profile would have the highest psychosocial HRQOL and the Inactive profile would have the lowest psychosocial HRQOL, was partially supported. Consistent with previous literature (Iannotti & Wang, 2013; Omorou et al., 2016), the results indicated that the Active profile had statistically and clinically significantly higher psychosocial HRQOL than the Moderate and Inactive profiles. This finding indicates that higher levels of total MVPA and MVPA bouts—specifically levels that exceed the 60-min guideline—contribute to higher levels of psychosocial HRQOL. However, in contrast to previous literature (Iannotti & Wang, 2013; Omorou et al., 2016), the Moderate and Inactive profiles had similar levels of psychosocial HRQOL, despite the Inactive profile having the dramatically higher SB. These results provide evidence that SB may not strongly contribute psychosocial HRQOL. Instead, it appears that MVPA, or more specifically, exceeding the USDHHS (2008) MVPA guideline, contributes to the significant differences. Taking into account the statistically and clinically significant findings regarding psychosocial HRQOL, several potential policy and intervention implications can be drawn. First, findings suggest that the guideline of 60 min of MVPA per day should be maintained, considering that participants who exceeded this recommendation had the highest psychosocial HRQOL. Further, findings suggest that interventions for elementary school-age youth should focus on substituting light activity with MVPA. Because the Active profile had significantly higher HRQOL than the Moderate profile, it can be argued that accumulating activity of higher intensity is driving the higher self-reported psychosocial HRQOL among the Active profile. Therefore, interventions focused on increasing children’s intensity of physical activity from light to moderate-to-vigorous would be expected to improve their psychosocial HRQOL by a statistically and clinically significant amount. Finally, there is evidence to suggest that interventions should be implemented in early elementary school, before adolescence. The current study suggests that these trajectories begin in elementary school, as evidenced by a higher proportion of older elementary school youth falling in the profile that had the lowest MVPA and highest SB. Therefore, interventions should be implemented early in development, when, theoretically, the changes required to see improvements would be smaller. In contrast to previous literature and our a priori hypothesis (Iannotti & Wang, 2013; Omorou et al., 2016), the activity profiles did not have statistically different levels of physical HRQOL. It is possible that, with a larger sample size, statically significant results would have emerged. However, our MCID analyses indicated that the Active profile did not differ meaningfully from the other two profiles on mean levels of physical HRQOL. Therefore, even if statistically significant results were found, there is evidence to suggest that physical HRQOL, as measured by the PedsQL, might not result in meaningful differences for the activity profiles. Although the preponderance of evidence in the literature led to the current hypotheses, there are also some investigations that found that activity profiles were not differentially associated with physical HRQOL. For example, Goldfield et al. (2015) found that screen time duration was associated with lower overall HRQOL and psychosocial HRQOL but not physical HRQOL. Additionally, other studies have found that another health behavior, disordered eating, was associated with psychosocial but not physical HRQOL (Jalali-Farahani et al, 2015). Therefore, the literature is mixed regarding the association of health behaviors and physical HRQOL as measured by the PedsQL. Further research should examine the association between profiles and other objective physical health outcomes, such as fitness, bone health, cholesterol, and blood pressure, or other subjective measures of physical functioning. Although some results from the current study failed to confirm the third and fourth hypotheses, the strengths may partially explain why these results differed from those in the existing literature. First, as mentioned previously, the current study used a person-centered, model-based approach to segment the population into smaller groups, taking into account total time spent in activity and bouts of activity, instead of only total time (Heitzler et al., 2011; Trilk et al., 2012). Additionally, the current study used objective measurements of activity, instead of relying on self-report measures like most previous studies (Berlin et al., 2017; Huh et al., 2011). The sample of the current study was unique, targeting elementary school-age youth during the time when activity levels dramatically change, in contrast to studies of adolescents (Berlin et al., 2017; Iannotti & Wang, 2013). Despite notable strengths, limitations must be considered when interpreting the findings of the current study. First, generalizability may be limited because of the sample being primarily Caucasian and middle-class. Future studies should examine profiles in a nationally representative sample of elementary school-age youth. Additionally, the current study is cross-sectional, which precludes statements of causality; future studies should examine longitudinal relationships between activity profiles and outcomes. Further, listwise deletion was implemented to exclude participants who did not have at least 2 valid weekdays and 1 weekend day of data. This approach limited sample size, thus reducing power and potentially producing biased estimates. Multiple imputation was considered, but this approach was not used because it could mask the identification of profiles in later analyses (Enders, 2010). Finally, although the inclusion of objective activity measurements is a strength, the current study did not include any self-report measures of MVPA or SB, limiting the ability to determine what types of activities the participants were engaging in or where they were accumulating the activity. Future studies should use a combination of self-report and objective measures. The current study was the first to use LVMM to identify and characterize profiles of elementary school-age youth based on objective measurements of total and bouts of activity. This study replicated profiles found in other samples of youth, suggesting that the profiles identified may represent reliable subgroups. Further, the current study extended the knowledge about these profiles, suggesting that interventions early in development focused on increasing intensity of physical activity from light to moderate-to-vigorous for at least 60 min per day may result in clinically meaningful differences in psychosocial HRQOL. Future studies should replicate the findings in a nationally representative sample, using objective and subjective measurements of MVPA and SB. Future studies are also needed to determine whether the profiles are associated with other outcomes and whether profiles differ based on other bout lengths or break patterns. Overall, the current study advances the literature on MVPA and SB in elementary school-age youth and has broader impacts for health promotion for youth. Supplementary Data Supplementary data can be found at: http://www.jpepsy.oxfordjournals.org/. Funding This research was supported, in part, by the University of Kansas General Research Fund (# 2301617), awarded to the second author. Conflicts of interest: None declared. Footnotes 1 Bootstrapped likelihood ratio test was considered, but ultimately not used because it tends to perform better with larger sample sizes (N > 1000: Nylund, Asparouhov, & Muthen, 2007). When this test was attempted in the current study, BLRT was not able to be computed. References Akaike H. ( 1987). Factor analysis and AIC. Psychometrika , 52( 3), 317– 332. Google Scholar CrossRef Search ADS   Altenburg T. M., Chinapaw M. J. M. ( 2015). Bouts and breaks in children's sedentary time: Currently used operational definitions and recommendations for future research. Preventive Medicine , 77, 1– 3. doi: 10.1016/j.ypmed.2015.04.019 Google Scholar CrossRef Search ADS PubMed  Bakk Z., Tekle F. B., Vermunt J. K. ( 2013). Estimating the association between latent class membership and external variables using bias-adjusted three-step approaches. Sociological Methodology , 43, 272– 311. doi: 10.1177/0081175012470644 Google Scholar CrossRef Search ADS   Banfield J. D., Raftery A. E. ( 1993). Model-based Gaussian and non-Gaussian clustering. Biometrics , 49, 803– 821. http://dx.doi.org/10.2307/2532201 Google Scholar CrossRef Search ADS   Bauer D. J., Steinley D. (Eds.) ( 2016). Cluster Analysis and Mixture Modeling Course Notes . Chapel Hill, NC: Curran-Bauer Analystics. Belcher B. R., Berrigan D., Dodd K. W., Emken B. A., Chou C. P., Spruijt-Metz D. ( 2010). Physical activity in US youth: Impact of race/ethnicity, age, gender, & weight status. Medicine and Science in Sports and Exercise , 42, 2211– 2221. Google Scholar CrossRef Search ADS PubMed  Berlin K. S., Kamody R. C., Thurston I. B., Banks G. G., Rybak T. M., Ferry R. J.Jr. ( 2017). Physical activity, sedentary behaviors, and nutritional risk profiles and relations to body mass index, obesity, and overweight in eighth grade. Behavioral Medicine , 43, 31– 39. doi: 10.1080/08964289.2015.1039956 Google Scholar CrossRef Search ADS PubMed  Berlin K. S., Williams N. A., Parra G. R. ( 2014). An introduction to latent variable mixture modeling (part 1): Overview and cross-sectional latent class and latent profile analyses. Journal of Pediatric Psychology , 39, 174– 187. doi: 10.1093/jpepsy/jst084 Google Scholar CrossRef Search ADS PubMed  Booth F. W., Roberts C. K., Laye M. J. ( 2012). Lack of exercise is a major cause of chronic diseases. Comprehensive Physiology , 2, 1143– 1211. doi: 10.1002/cphy.c110025 Google Scholar PubMed  Cain K. L., Sallis J. F., Conway T. L., Van Dyck D., Calhoon L. ( 2013). Using accelerometers in youth physical activity studies: A review of methods. Journal of Physical Activity and Health , 10, 437– 450. doi: 10.1123/jpah.10.3.437 Google Scholar CrossRef Search ADS PubMed  Carson V., Stone M., Faulkner G. ( 2014). Patterns of sedentary behavior and weight status among children. Pediatric Exercise Science , 26, 95– 102. doi: 10.1123/pes.2013-0061 Google Scholar CrossRef Search ADS PubMed  Celeux G., Soromenho G. ( 1996). An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification , 13, 195– 212. http://dx.doi.org/10.1007/BF01246098 Google Scholar CrossRef Search ADS   Centers for Disease Control and Prevention (CDC). ( 2011). About BMI for children and teens. Retrieved from http://www.cdc.gov/healthyweight/assessing/bmi/childrens_bmi Copay A. G., Subach B. R., Glassman S. D., Polly D. W., Schuler T. C. ( 2007). Understanding the minimum clinically important difference: A review of concepts and methods. The Spine Journal , 7, 541– 546. doi: 10.1016/j.spinee.2007.01.008 Google Scholar CrossRef Search ADS PubMed  Enders C. K. ( 2010). Applied Missing Data Analysis . New York, NY: Guilford. Evenson K. R., Catellier D. J., Gill K., Ondrak K. S., McMurray R. G. ( 2008). Calibration of two objective measures of physical activity for children. Journal of Sports Sciences , 26, 1557– 1565. doi: 10.1080/02640410802334196 Google Scholar CrossRef Search ADS PubMed  Fakhouri T. H., Hughes J. P., Brody D. J., Kit B. K., Ogden C. L. ( 2013). Physical activity and screen-time viewing among elementary school–aged children in the United States from 2009 to 2010. JAMA Pediatrics , 167, 223– 229. doi: 10.1001/2013.jamapediatrics.122 Google Scholar CrossRef Search ADS PubMed  Goldfield G. S., Cameron J. D., Murray M., Maras D., Wilson A. L., Phillips P., Kenny G. P., Hadjiyannakis S., Alberga A. S., Tulloch H., Doucette S., Sigal R. J. ( 2015). Screen time is independently associated with health-related quality of life in overweight and obese adolescents. Acta Paediatrica , 104, e448– e454. doi: 10.1111/apa.13073 Google Scholar CrossRef Search ADS PubMed  Gopinath B., Hardy L. L., Baur L. A., Burlutsky G., Mitchell P. ( 2012). Physical activity and sedentary behaviors and health-related quality of life in adolescents. Pediatrics , 130, e167– e174. doi: 10.1542/peds.2011-3637 Google Scholar CrossRef Search ADS PubMed  Heitzler C., Lytle L., Erickson D., Sirard J., Barr-Anderson D., Story M. ( 2011). Physical activity and sedentary activity patterns among children and adolescents: A latent class analysis approach. Journal of Physical Activity and Health , 8, 457– 467. Google Scholar CrossRef Search ADS PubMed  Huh J., Riggs N. R., Spruijt Metz D., Chou C. P., Huang Z., Pentz M. ( 2011). Identifying patterns of eating and physical activity in children: A latent class analysis of obesity risk. Obesity , 19, 652– 658. doi: 10.1038/oby.2010.228 Google Scholar CrossRef Search ADS PubMed  Iannotti R. J., Wang J. ( 2013). Patterns of physical activity, sedentary behavior, and diet in US adolescents. Journal of Adolescent Health , 53, 280– 286. http://dx.doi.org/10.1016/j.jadohealth.2013.03.007 Google Scholar CrossRef Search ADS PubMed  Jalali-Farahani S., Chin Y. S., Nasir M. T. M., Amiri P. ( 2015). Disordered eating and its association with overweight and health-related quality of life among adolescents in selected high schools of Tehran. Child Psychiatry and Human Development , 46, 485– 492. doi: 10.1007/s10578-014-0489-8 Google Scholar CrossRef Search ADS PubMed  Kline R. B. ( 2011). Principles and Practice of Structural Equation Modeling  ( 3rd ed.). New York, NY: Guilford Press. Leech R. M., McNaughton S. A., Timperio A. ( 2014). The clustering of diet, physical activity and sedentary behavior in children and adolescents: A review. International Journal of Behavioral Nutrition and Physical Activity , 11, 4. Google Scholar CrossRef Search ADS PubMed  Lubke G. H., Neale M. C. ( 2006). Distinguishing between latent classes and continuous factors: Resolution by maximum likelihood. Multivariate Behavioral Research , 41, 499– 532. doi: 10.1207/s15327906mbr4104_4 Google Scholar CrossRef Search ADS PubMed  Muthén L. K., Muthén B. O. ( 1998-2012). Mplus User's Guide  ( 7th ed.). Los Angeles, CA: Muthén & Muthén. Nylund K. L., Asparouhov T., Muthén B. O. ( 2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling , 14, 535– 569. Google Scholar CrossRef Search ADS   Omorou A. Y., Langlois J., Lecomte E., Briançon S., Vuillemin A. ( 2016). Cumulative and bidirectional association of physical activity and sedentary behaviour with health-related quality of life in adolescents. Quality of Life Research , 25, 1169– 1178. Google Scholar CrossRef Search ADS PubMed  Rowlands A. V., Pilgrim E. L., Eston R. G. ( 2008). Patterns of habitual activity across weekdays and weekend days in 9–11-year-old children. Preventive Medicine , 46, 317– 324. doi: 10.1016/j.ypmed.2007.11.004 Google Scholar CrossRef Search ADS PubMed  Sadeh A., Sharkey K. M., Carskadon M. A. ( 1994). Activity-based sleep-wake identification: An empirical test of methodological issues. Sleep , 17, 201– 207. Google Scholar CrossRef Search ADS PubMed  Salmon J., Tremblay M. S., Marshall S. J., Hume C. ( 2011). Health risks, correlates, and interventions to reduce sedentary behavior in young people. American Journal of Preventive Medicine , 41, 197– 206. doi: 10.1016/j.amepre.2011.05.001 Google Scholar CrossRef Search ADS PubMed  Saunders T. J., Tremblay M. S., Mathieu M. È., Henderson M., O’loughlin J., Tremblay A., Chaput J. P.; QUALITY Cohort Research Group. ( 2013). Associations of sedentary behavior, sedentary bouts and breaks in sedentary time with cardiometabolic risk in children with a family history of obesity. PLoS One , 8, e79143. doi: 10.1371/journal.pone.0079143 Google Scholar CrossRef Search ADS PubMed  Schwarz G. ( 1978). Estimating the dimension of a model. The Annals of Statistics , 6, 461– 464. doi: 10.1214/aos/1176344136 Google Scholar CrossRef Search ADS   Sclove S. L. ( 1987). Application of model-selection criteria to some problems in multivariate analysis. Psychometrika , 52, 333– 343. http://dx.doi.org/10.1007/BF02294360 Google Scholar CrossRef Search ADS   Spieth L. E., Harris C. V. ( 1996). Assessment of health-related quality of life in children and adolescents: An integrative review. Journal of Pediatric Psychology , 21, 175– 193. http://dx.doi.org/10.1093/jpepsy/21.2.175 Google Scholar CrossRef Search ADS PubMed  Trilk J. L., Pate R. R., Pfeiffer K. A., Dowda M., Addy C. L., Ribisl K. M., Neumark-Sztainer D., Lytle L. A. ( 2012). A cluster analysis of physical activity and sedentary behavior patterns in middle school girls. Journal of Adolescent Health , 51, 292– 298. Google Scholar CrossRef Search ADS PubMed  Troiano R. P., Berrigan D., Dodd K. W., Masse L. C., Tilert T., McDowell M. ( 2008). Physical activity in the United States measured by accelerometer. Medicine and Science in Sports and Exercise , 40, 181– 188. doi: 10.1249/mss.0b013e31815a51b3 Google Scholar CrossRef Search ADS PubMed  United States Census Bureau ( 2015). Quick facts: Lawrence, Kansas. Retrieved from https://www.census.gov/quickfacts/table/PST045215/2038900, 00 US Department of Health and Human Services (USDHHS). ( 2008). 2008 physical activity guidelines for Americans. Retrieved from http://health.gov/paguidelines/pdf/paguide.pdf Varni J. W., Seid M., Kurtin P. S. ( 2001). PedsQL (TM) 4.0: Reliability and validity of the Pediatric Quality of Life Inventory (TM) version 4.0 generic core scales in healthy and patient populations. Medical Care , 39, 800– 812. http://dx.doi.org/10.1097/00005650-200108000-00006 Google Scholar CrossRef Search ADS PubMed  Willis E. A., Ptomey L. T., Szabo-Reed A. N., Honas J. J., Lee J., Washburn R. A., Donnelly J. E. ( 2015). Length of moderate-to-vigorous physical activity bouts and cardio-metabolic risk factors in elementary school children. Preventive Medicine , 73, 76– 80. doi: 10.1016/j.ypmed.2015.01.022 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Pediatric Psychology Oxford University Press

Latent Profiles of Physical Activity and Sedentary Behavior in Elementary School-Age Youth:Associations With Health-Related Quality of Life

Loading next page...
 
/lp/ou_press/latent-profiles-of-physical-activity-and-sedentary-behavior-in-s4wRWI6anr
Publisher
Oxford University Press
Copyright
© The Author(s) 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0146-8693
eISSN
1465-735X
D.O.I.
10.1093/jpepsy/jsx149
Publisher site
See Article on Publisher Site

Abstract

Abstract Objective The objectives were to identify and describe profiles of elementary school-age youth based on objective measurements of total time spent in moderate-to-vigorous physical activity (MVPA) and sedentary behavior (SB) and in bouts of the activities, to examine predictors of profiles, and to examine whether profiles were differentially associated with physical and psychosocial health-related quality of life (HRQOL). Methods Participants included 204 youth (aged 8–11 years) who wore accelerometers to gather objective activity data. The average proportion of time in MVPA and SB and average proportion of time in MVPA and SB bouts were used for analyses. Participants completed a self-report measure of HRQOL. Latent variable mixture modeling was conducted. Results Overall, participants did not meet the MVPA guideline (M = 50.7 min) and spent 47% of waking hours in SB, indicating that elementary school is a critical developmental period to study activity levels. Three profiles emerged: “Active,” “Inactive,” and “Moderate.” Boys were significantly more likely to be in the Active profile than the other profiles, and older youth were significantly more likely to be in the Inactive profile. After controlling for child sex and age, participants in the Active profile reported significantly higher psychosocial HRQOL than the participants in the other profiles; however, no significant differences were found in physical HRQOL. Conclusions Identification of these naturally occurring profiles suggests need for interventions early in development focused on increasing the intensity of physical activity from light to moderate-to-vigorous for at least 60 min per day as way to improve psychosocial HRQOL. children, health-related quality of life, mixture model, physical activity, sedentary behavior Physical activity and sedentary behavior (SB) are both modifiable lifestyle behaviors that impact physical and psychosocial health outcomes in youth. For example, research has indicated that more time in moderate-to-vigorous physical activity (MVPA) is associated with better bone health, body composition, and self-concept, and with lower risk for depression, anxiety, heart disease, and diabetes (Booth, Roberts, & Laye, 2012; US Department of Health and Human Services (USDHHS), 2008). In contrast, more time in SB is associated with higher levels of adiposity and blood pressure, as well as lower levels of fitness and academic achievement (Salmon, Tremblay, Marshall, & Hume, 2011). MVPA and SB are understood to be distinct constructs with different determinants that can explain unique variance in outcomes (Fakhouri et al., 2013); therefore, the literature suggests that research should focus on both MVPA and SB. In addition to total time in MVPA and SB, there is evidence to suggest that the way in which youth accumulate time in these behaviors may be important. Specifically, emerging evidence suggests that MVPA bouts of at least 5 min are related to greater physical health outcomes, including lower weight status and waist circumference (Rowlands, Pilgrim, & Eston, 2008; Willis et al., 2015). Similarly, research has shown that SB bouts of <10 min are associated with better physical health outcomes, including lower cardiometabolic risk, fasting glucose, and weight status, independent of total time in SB and MVPA (Carson, Stone, & Faulkner, 2014; Saunders et al., 2013). Further understanding of how bouts affect physical and psychosocial outcomes in youth is needed (Altenburg & Chinapaw, 2015). As reviewed above, a number of studies have shown independent associations between total time and bouts of MVPA and SB and health outcomes in elementary school-age youth. However, predictors of poor health outcomes, including low levels of MVPA and excessive SB, often co-occur and can amplify negative physical and psychosocial outcomes (Leech, McNaughton, & Timperio, 2014). Therefore, rather than examining these constructs independently, Leech et al. (2014) recommended further research to identify patterns of behaviors. One way to accomplish this goal is to use latent variable mixture modeling (LVMM), a person-centered, model-based approach in which individuals are probabilistically assigned to unobserved groups based on similar patterns of observed variables (Berlin, Williams, & Parra, 2014). Identifying smaller homogenous profiles within a larger heterogonous population allows for the identification of characteristics and needs of the profiles to tailor interventions for these specific populations (Berlin et al., 2014). Several studies over the past decade have used LVMM to examine profiles of activity patterns in youth (Berlin et al., 2017; Heitzler et al., 2011; Huh et al., 2011; Iannotti & Wang, 2013; Trilk et al., 2012). These studies identified between three and six profiles, but most commonly, three profiles emerged: high MVPA/low SB, high SB/low MVPA, and low–moderate MVPA/SB. Common predictors of activity profiles were child age, sex, and race/ethnicity, which coincide with research showing differences in MVPA and SB based these variables (Belcher et al., 2010; Fakhouri et al., 2013). Child sex, age, and race/ethnicity were included as predictors of profiles in the current study. The primary outcome examined in relation to activity profiles in the available literature is weight status (Heitzler et al., 2011; Huh et al., 2011; Trilk et al., 2012), an objective measure of physical health. These studies found that youth who engaged in high SB/low MVPA were more likely to present with overweight/obesity. Because of the frequent examination of weight status in relation to activity profiles in the literature, weight status was not a primary variable of interest in the current study. Instead, current analyses focused on whether profiles were related to subjective reports of physical and psychosocial outcomes. Using self-reported activity, Iannotti and Wang (2013) found that adolescents in the high MVPA/low SB profile had fewer depressive symptoms, better life satisfaction, and better overall physical health. Also using self-reported activity, Berlin and colleagues (Berlin et al., 2017) found that adolescents in the high MVPA/low SB/healthy diet profile had higher levels of socioemotional protective factors (e.g., locus of control) and lower levels of socioemotional risk factors (e.g., internalizing symptoms). These studies provide preliminary evidence that youth who have high MVPA/low SB may have higher subjective reports of physical and psychosocial health. To expand on these initial examinations, an important next step is to examine health-related quality of life (HRQOL) as a measure of subjective physical and psychosocial functioning in relation to activity profiles in youth. HRQOL is a multidimensional construct of a child’s perception of his/her functioning across physical and psychosocial domains (Spieth & Harris, 1996). There is support for the independent associations of MVPA and SB with HRQOL in youth. For example, Gopinath et al. (2012) found that higher levels of MVPA were associated with higher total, physical, and psychosocial HRQOL; the opposite was true for screen-based SB. However, as noted above, health risk factors can interact to amplify negative physical and psychosocial outcomes (Leech et al., 2014). Assessing the associations among activity profiles and HRQOL can help illuminate the cumulative effects of MVPA and SB on youth’s physical and psychosocial HRQOL. The findings of the available studies that have used LVMM to examine activity profiles in youth must be considered in light of a number of limitations. First, most studies included samples of adolescents (Berlin et al., 2017; Heitzler et al., 2011); only Huh and colleagues (Huh et al., 2011) examined profiles of activity patterns in elementary school-age youth. Understanding the activity profiles in elementary school-age youth can help inform targeted intervention efforts during this critical developmental period in which activity levels dramatically change (Fakhouri et al., 2013). There are also limitations of the measurement of MVPA and SB. For example, the majority of studies included only self-report measures of MVPA and SB to inform profiles (Berlin et al., 2017; Huh et al., 2011). Because youth may be limited in their memory and recall, self-report measures may not be reliable or valid. Additionally, the two studies that used objective measures of MVPA and SB focused on the total time spent in the behaviors (Heitzler et al., 2011; Trilk et al., 2012), instead of examining bouts of MVPA and SB. The current study aimed to address gaps in the literature by using LVMM to identify and describe profiles of elementary school-age youth based on their patterns of objective MVPA and SB (total time and bouts), to examine predictors of profiles, and to examine whether profiles were differentially associated with physical and psychosocial HRQOL. First, it was hypothesized that LVMM would identify three profiles. Second, it was hypothesized that child age, sex, and race/ethnicity would predict profile membership. Third, it was hypothesized that, after controlling for significant predictors, the profile characterized by high total MVPA/more MVPA bouts and low total SB/Fewer SB bouts would be associated with the highest psychosocial HRQOL and that the profile characterized by low total MVPA/fewer MVPA bouts and high total SB/More SB bouts would be associated with the lowest psychosocial HRQOL; similar findings were hypothesized for physical HRQOL. Further, the study also examined the minimally clinically important difference (MCID) to determine whether statistically significant differences between profiles would translate into clinically significant benefits (Copay, Subach, Glassman, Polly, & Schuler, 2007). Method Participants Participants included 204 students from two Midwestern elementary schools. Eligibility criteria included children who: (a) were enrolled in mainstream 3rd through 5th grade classes, (b) spoke/read English, (c) had parental consent, and (d) wore the accelerometer for at least 1 valid weekend day and 2 valid weekdays. Census data indicated that the schools were situated in a suburban community with a median household of $46,406 (United States Census Bureau, 2015). The current sample was a part of a larger longitudinal study. At Time 1, 149 students were recruited. Data were excluded for children who were absent, moved, or declined assent (n = 4), as well as for children without adequate accelerometer data (n = 27), resulting in 118 students from Time 1. At Time 2 and Time 3, 126 additional students were recruited. Children who were absent, moved, or declined assent (n = 19) were excluded from these analyses, as well as children without sufficient accelerometer data (n = 21). Recruitment at Times 2 and 3 resulted in 86 additional participants. The final sample consisted of 204 students, which represents 55% of the total eligible population (n = 504). Only 15% of parents actively denied consent, so it is impossible to confirm that all parents received study information. Procedure Children were recruited from two schools that were chosen for diversity across demographic characteristics. Information letters and consent forms were distributed to parents during school events and in packets that were sent home with children. Assent was collected, and accelerometers were distributed. Data collection for self-report measures occurred in group format during the school day; research assistants read measures aloud to ensure that participation was not limited by reading comprehension. Objective height and weight were collected in isolated locations to ensure privacy. Procedures were approved by the authors’ institutional review board, school district’s office of research and assessment, and schools’ principals. Measures Objective Measures of MVPA and SB ActiGraph accelerometers (Model GT3X, ActiGraph, LLC) were used to obtain objective measurements of MVPA and SB. Participants were instructed to wear the devices on their nondominant hip for 3 weekdays and 2 weekend days. Raw accelerometer data were downloaded and analyzed using the ActiLife Data Analysis Software (version 6.10.4). Raw data were first binned into 10 s epochs, and sleep and nonwear periods were flagged using the Sadeh algorithm (Sadeh, Sharkey, & Carskadon, 1994) and the Troiano algorithm (Troiano et al., 2008). Waking wear time data were processed using Evenson cut points (Evenson, Catellier, Gill, Ondrak, & McMurray, 2008). Following Cain, Sallis, Conway, Van Dyck, and Calhoon (2013), valid days were those with ≥10 h of waking wear time. Participants who had at least 1 valid weekend day and 2 valid weekdays were included (Cain et al., 2013). The 3 days with the most wear time were used to calculate the average proportions of time in MVPA and SB. These proportions were calculated by dividing the average minutes by the average device wear time. Based on the best available evidence, MVPA bouts of 5 min and SB bouts of 10 min were chosen for analyses. To determine the number of MVPA bouts per day, the software calculated how many times each participant spent 5 min above 2,296 counts per minute (Evenson et al., 2008). The number of 5-min MVPA bouts was averaged, and the total number of minutes spent in 5-min bouts was divided by the average wear time. To determine the number of bouts of SB per day, the software calculated how many times each participant spent exactly 10 min under 100 counts per minute (Evenson et al., 2008). The number of 10-min SB bouts was averaged, and the total number of minutes spent in 10-min SB bouts was divided by the average wear time. Health-Related Quality of Life Participants completed the Pediatric Quality of Life Inventory (PedsQL™ 4.0; Varni, Seid, & Kurtin, 2001) as a measure of HRQOL. The PedsQL is a 23-item measure with instructions to report how much of a problem each statement has been in the past 1 month on a five-point Likert scale. The two summary scores (i.e., physical and psychosocial HRQOL) were used in current analyses. Evidence for validity indicates that the summary scores distinguished between healthy and chronically ill youth and were related to illness burden and morbidity (Varni et al., 2001). The internal consistencies for the physical and psychosocial summary scores in the current study were 0.76 and 0.86, respectively. Weight Status Staff obtained child height to the nearest 0.1 centimeter using a stadiometer. Child weight was measured to the nearest 0.1 kg using a calibrated portable digital scale. Triple measurements were averaged, and body mass index (BMI) z-scores were calculated using the SAS script provided by the Centers for Disease Control and Prevention (CDC, 2011). Weight status was used to describe children in the various activity profiles. Data Analytic Plan Preliminary Analyses Descriptive statistics and Pearson correlations were conducted for the overall sample, and logistic regression analyses were conducted to determine whether there were differences in participants and those excluded because of insufficient accelerometer data. Model Specification LVMMs were conducted using Mplus statistical software (Version 7.31; Muthén & Muthén, 1998–2012). The following indictor variables were chosen to identify the profiles: proportion of time in MVPA, SB, MVPA bouts, and SB bouts. Univariate entropy values for each variable were examined to quantify the variance that each indicator contributed to profile separation/classification. Because having a completely unrestricted model results in the estimation of a large number of parameters, minimal restrictions were applied to streamline estimation and increase parsimony (Bauer & Steinley, 2016). A “proportionality restriction” was used, which assumes that the covariance matrices are related by a constant multiplier; the correlations are restricted to be equal across profiles, but the variances are allowed to be proportionally different (Banfield & Raftery, 1993). Model Estimation First, a parsimonious one-profile model was fit to the data, followed by consecutive models with increasing numbers of profiles (up to four; Berlin et al., 2014). Maximum likelihood estimation was used to determine the parameter estimates with the highest likelihood of having produced the observed data. Owing to inclusion criteria related to accelerometer data, no missing data were present at this step. Multiple start values were used (1,000 initial-stage, 100 second-stage), and the replication of best log-likelihood was confirmed for each model (Berlin et al., 2014). Model Selection and Interpretation The Akaike information criterion (AIC; Akaike, 1987), Bayesian information criterion (BIC; Schwarz, 1978), and the sample-size adjusted BIC (ssBIC; Sclove, 1987) were examined to evaluate model fit, with the smaller values indicating better fit.1 Additionally, entropy was examined to determine the accuracy of classification; scores range from 0 to 1, with higher scores indicating better accuracy of classification (Celeux & Soromenho, 1996). Finally, the sample sizes of the profiles were considered to determine the benefit of including classes with small numbers of individuals (Lubke & Neale, 2006). Once the best fitting model was determined, descriptive statistics for the profiles were conducted. Predictors of Profiles Vermunt’s three-step approach was used to examine whether profiles could be predicted by age, sex, and race/ethnicity (Bakk, Tekle, & Vermunt, 2013). Individuals were assigned to profiles based on most likely profile membership. Next, the assigned profile was treated as an indicator variable for a new latent profile variable; demographic variables were then used as predictors. Listwise deletion was used by default; only 5% of data were missing, so no additional missing data strategies were implemented. Maximum likelihood estimation with robust standard error was used, and odds ratios (OR) were reported. Outcomes of Profiles The manual BCH method (Muthén & Muthén, 1998–2012) was used to examine the effect of the latent profile variable on the outcomes while controlling for significant covariates. In the first step, the three-profile model was estimated and the BCH weights were saved. In the second step, predictors were specified to predict profiles and outcomes, using the BCH weights as training data. Maximum likelihood estimation with robust standard error was used. Total 3% of participants were excluded from analyses because of missing data for predictors. Wald tests of equal means were examined to determine if profiles differed in means for physical and psychosocial HRQOL, after controlling for significant predictors in profiles. Minimal Clinically Important Difference For the profiles that had statistically significant differences in HRQOL, MCID was used to determine whether the differences between profiles were likely to be associated with clinically significant benefits (Copay et al., 2007). MCID signifies the smallest difference in HRQOL that the participants would identify as important, elucidating whether profile differences in HRQOL scores were greater than the variability attributed to measurement error. A common approach for measuring MCID is to compare the differences in scores with the standard error of measurement. A difference greater than or equal to the MCID is considered to be clinically meaningful (Copay et al., 2007). Results Preliminary Analyses The current study included 204 participants between the ages of 8 and 11 years (M = 9.5, SD = 0.09). The self-reported racial/ethnic makeup of the sample (presented in Table I, first column) was similar to the overall composition of the school district as indicated by district records. Most participants (71.3%) were of normal weight. Preliminary analyses revealed no differences in demographic characteristics or physical HRQOL in these participants and those who were dropped from analyses because of insufficient accelerometer data (all p > .05); study participants had significantly lower psychosocial HRQOL than those excluded (β = −.04, p = .003). Table I. Demographic Characteristics and Study Variables for Total Sample and Activity Profiles   Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)    Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)  Note. BMI = body mass; HRQOL = health-related quality of life; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior. Table I. Demographic Characteristics and Study Variables for Total Sample and Activity Profiles   Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)    Full sample (n = 204)  Profile 1 (n = 29)  Profile 2 (n = 148)  Profile 3 (n = 27)  Age (years)  9.5 (0.9)  9.4 (0.7)  9.4 (0.8)  10.1 (1.0)   8  9.5%  7.1%  10.4%  7.4%   9  47.2%  53.6%  50.7%  22.2%   10  27.6%  32.1%  27.1%  25.9%   11  15.6%  7.1%  11.8%  44.4%  Sex           Male  47.3%  72.4%  55.1%  33.3%   Female  52.7%  27.6%  44.9%  66.7%  Race/ethnicity           White/Non-Hispanic  67.4%  78.6%  64.5%  70.4%   Hispanic  2.1%  0%  2.9%  0%   American Indian  4.7%  7.1%  3.6%  7.4%   Black/Non-Hispanic  5.2%  3.6%  5.8%  3.7%   Asian  1.6%  0%  1.4%  3.7%   Biracial/Other  19.2%  10.7%  21.7%  14.8%  Grade in school           3rd  48.0%  55.2%  50.7%  25.9%   4th  28.0%  31.0%  30.4%  11.1%   5th  24.0%  13.8%  18.9%  63.0%  BMI z-score  0.30 (1.1)  0.2 (1.0)  0.3 (1.1)  0.2 (1.1)   Underweight  2.1%  0%  2.8%  0%   Normal weight  71.3%  72.4%  68.8%  76.0%   Overweight  13.3%  13.8%  13.9%  8.0%   Obese  13.3%  3.4%  14.6%  16.0%  MVPA           Minutes  50.7 (28.0)  94.9 (28.8)  46.1 (19.8)  28.8 (15.2)   Proportion of day  5.8% (3.1)  10.7% (3.0)  5.3% (2.3)  3.3% (1.8)  SB           Minutes  409.2 (72.7)  382.9 (57.8)  396.0 (64.1)  509.9 (46.9)   Proportion of day  46.8% (7.7)  43.5% (6.3)  45.3% (6.4)  58.6% (3.5)  MVPA bouts           Number  5.7 (4.5)  13.8 (4.3)  4.7 (2.7)  2.5 (1.9)   Proportion of Day  3.2% (2.5)  7.8% (2.3)  2.7% (1.6)  1.4% (1.1)  SB bouts           Number  12.8 (5.0)  11.0 (3.8)  11.5 (3.7)  21.7 (3.0)   Proportion of Day  12.6% (5.7)  12.5% (4.2)  13.1% (4.1)  24.9% (2.9)  Light activity           Minutes  415.1 (64.0)  404.2 (50.7)  432.5 (57.6)  331.4 (36.4)   Proportion of day  47.4% (6.5)  45.8% (5.0)  49.5% (5.6)  38.1% (2.7)  Wear time (minutes)  875.0 (61.4)  882.0 (54.7)  874.6 (62.8)  870.0 (62.3)  HRQOL           Physical  82.1 (14.0)  84.9 (14.7)  81.7 (13.2)  81.2 (16.9)   Psychosocial  75.6 (14.9)  81.1 (14.2)  74.7 (14.9)  75.1 (15.0)  Note. BMI = body mass; HRQOL = health-related quality of life; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior. Participants in the current sample accumulated similar activity levels as previous samples (Saunders et al., 2013; Willis et al., 2015), and mean physical and psychosocial HRQOL scores were also comparable (Varni et al., 2001; see Table I). Age was positively associated with total SB/SB bouts and psychosocial HRQOL (see Table II). Boys had higher total MVPA/MVPA bouts and lower total SB. Total MVPA was strongly positively associated with MVPA bouts and negatively associated with total SB/SB bouts. Similarly, total SB was strongly positively associated with SB bouts and negatively associated with MVPA bouts. MVPA bouts and SB bouts were negatively associated. Total MVPA/MVPA bouts were positively correlated with both HRQOL measures. Psychosocial and physical HRQOL were positively associated. Table II. Correlations Among Study Variables   1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50    1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50  Notes. *p < .05; **p < .01; Child sex (0 = female, 1 = male); race/ethnicity (0 = Caucasian; 1 = Minority); BMI = body mass index; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior; HRQOL = health-related quality of life. Table II. Correlations Among Study Variables   1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50    1  2  3  4  5  6  7  8  9  10  1. Child age  −                    2. Child sex  −.07  −                  3. Race/ethnicity  −.08  −.10  −                4. BMI z-score  .01  .15*  −.18*  −              5. MVPA  −.12  .32**  −.01  −.07  −            6. SB  .30**  −.16*  −.10  −.08  −.54**  −          7. MVPA bouts  −.11  .30**  −.04  −.06  .96**  −.44**  −        8. SB bouts  .29**  −.09  −.09  −.04  −.39**  .91**  −.31**  −      9. Physical HRQOL  .13  .02  .00  −.02  .18**  −.07  .21**  .01  −    10. Psychosocial HRQOL  .20**  −.01  −.05  −.09  .15*  −.01  .18*  .03  .65**  −   Minimum  8  0  0  −2.51  .52%  28.21%  .17%  4.00%  15.00  20.71   Maximum  11  1  1  2.98  16.89%  67.21%  12.97%  32.67%  100.00  100.00   Skewness  —  —  —  .11  .98  .16  1.34  .64  −1.34  −.79   Kurtosis  —  —  —  −4.32  1.02  −.50  1.98  .12  2.83  .50  Notes. *p < .05; **p < .01; Child sex (0 = female, 1 = male); race/ethnicity (0 = Caucasian; 1 = Minority); BMI = body mass index; MVPA = moderate-to-vigorous physical activity; SB = sedentary behavior; HRQOL = health-related quality of life. Primary Analyses Results from the one-, two-, and three-profile models are presented in Table III. The four-profile model was considered, but ultimately rejected for two reasons. First, the four-profile model produced significant estimation errors, requiring many parameters to be fixed to avoid singularity of the information matrix. Second, the four-profile model resulted in one profile that included only seven participants. Results indicated that models with two or three latent profiles fit the data better than a unitary model without profiles. Further, AIC, BIC, and ssBIC favored the three-profile solution, and the overall entropy statistic favored the two-profile model. However, as noted previously, entropy is typically used to determine classification accuracy, not to select optimal number of profiles. Therefore, it was concluded that the three-profile model best fit the data. Univariate entropy values for indicators ranged from 0.46 to 0.57, supporting the retention of all indicators because of their nonnegligible contribution to profile separation/classification. Table III. Model Fit of the LVMMs Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Note. The four-profile model produced estimation errors and was omitted from comparisons; optimal models according to criteria are bolded; other fit indices are reported for completeness; AIC = Akaike’s information criterion; BIC = Bayesian information criterion; LVMM = latent variable mixture modeling; ssBIC = sample-size adjusted BIC. Table III. Model Fit of the LVMMs Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Number of profiles  Log-likelihood  Free parameters  AIC  BIC  ssBIC  Entropy  1  1,923.12  14  −3,818.24  −3,771.79  −3,816.14  N/A  2  1,954.92  20  −3,869.83  −3,803.47  −3,866.84  0.90  3  1,973.58  26  −3,895.16  −3,808.89  −3,891.26  0.86  Note. The four-profile model produced estimation errors and was omitted from comparisons; optimal models according to criteria are bolded; other fit indices are reported for completeness; AIC = Akaike’s information criterion; BIC = Bayesian information criterion; LVMM = latent variable mixture modeling; ssBIC = sample-size adjusted BIC. The descriptive statistics of the profiles are presented in Table I (Columns 2–4; for graphical depiction, see Online Supplementary Material). Participants in Profile 1 (“Active”; n = 29) were characterized by the highest amounts of MVPA; further, 93% averaged over 60 min of MVPA per day, thus meeting the US Department of Health and Human Services (USDHHS, 2009) guideline. Girls were significantly less likely to be in the Active profile than Profiles 2 and 3 (OR = 4.97 and 6.51, respectively). Similarly, boys were more likely to be in the Active profile than Profiles 2 and 3 (OR = 0.20 and 0.15, respectively). The largest profile, Profile 2 (“Moderate”; n = 148), was characterized by moderate amounts of MVPA and SB. Only 25% met the 60-min guideline. The Active and Moderate profiles had similar SB (M = 396.0 vs. 382.9 min), but the Moderate profile had less MVPA (M = 46.1 vs. 94.9 min) and more light activity (M = 432.5 vs. 404.2 min). Profile 3 (“Inactive”; n = 27) was characterized by the highest levels of SB. Total 0% met the USDHHS (2008) guideline. Older youth were significantly more likely to be in the Inactive profile than the Active or Moderate profiles (OR = 0.30 and 0.34, respectively). The next step was to determine the associations between the activity profiles and psychosocial and physical HRQOL, after controlling for child sex and age, which were significant predictors of profile membership. The results indicated that participants in the Active profile reported significantly higher psychosocial HRQOL than the participants in the Moderate and Inactive profile (χ2 = 5.74, p = .02 and χ2 = 6.44, p = .01, respectively). Moderate and Inactive profiles did not significantly differ in terms of psychosocial HRQOL. MVPA profile membership was not significantly associated with physical HRQOL (p > .05). Finally, MCID was calculated to determine whether the statistically significant differences in psychosocial HRQOL between the Active profile, and the other two profiles were consistent with clinically significant benefits. MCID for psychosocial HRQOL was 5.56. The mean differences (HRQOL) between the Active profile and the Moderate and Inactive profiles were 6.4 and 6.0, respectively, indicating clinically meaningful differences for both comparisons. Discussion The first hypothesis, that three profiles would be identified, was supported. Specifically, the following profiles were identified: Active, Inactive, and Moderate. Consistent with Iannotti and Wang (2013), the majority of children in the current study fell into the Moderate profile. Consistent with the review by Leech et al. (2014), older youth, particularly females, tended to fall in the profile characterized by low MVPA and high SB and younger youth, particularly males, tended to fall in the profile characterized by high levels of MVPA. The consistency in findings across studies (Berlin et al., 2017; Heitzler et al., 2011, Iannotti & Wang, 2013) provides support for three profiles being reliable subgroups within the heterogeneous population. An anticipated advantage of the current study was that it was the first to identify profiles based on total time in MVPA and SB and time in bouts. However, time in bouts was strongly associated with total MVPA time, limiting the impact of bouts on the identification of profiles. That is, children who accumulated more time in MVPA/SB did so in bouts of activity, rather than in shorter “bursts.” An important direction for further study is whether different bout lengths might result in different outcomes. Further, the inclusion of breaks in bouts (i.e., when participant transitions from sitting to standing or from standing to sitting) might further differentiate profiles. Although the inclusion of breaks was considered in the present study, model complexity and convergence issues prevent our use of this variable in analyses. Researchers frequently expect that study samples reflect a homogenous population and that individual variability is because of random causes (Bauer & Steinley, 2016). With this mindset, researchers often examine the skewness and kurtosis of variables to confirm that they can be considered normal in subsequent analyses (e.g., ±3 for skewness; ±10 for kurtosis; Kline, 2011). If the current study had taken these steps, the study variables would have been considered “normally distributed” (i.e., skewness and kurtosis values were within normal limits). However, that decision would make the variables the focus of analyses, attempting to make them fit within a normal framework, instead of focusing on the naturally occurring groups of individuals within the population. Our findings provide support for using LVMM to examine MVPA and SB in youth, and may recommend this approach in other samples of children and youth. The second hypothesis, that child sex, age, and race/ethnicity would predict profile membership, was partially supported. Consistent with previous studies (Berlin et al., 2017; Huh et al., 2011), age and sex were found to predict membership. Contrary to previous studies (Berlin et al., 2017; Huh et al., 2011), race/ethnicity did not predict membership. A possible explanation is that elementary school-age youth may have been unable to accurately self-report on race/ethnicity. Additionally, previous studies found differences in profiles for specific minority groups (e.g., Hispanic, Black). In the current sample, only 2.1 and 5.2% of the participants self-reported as Hispanic and Black, respectively. Significant findings may be found with a larger, nationally representative sample, and with parent-reported race/ethnicity. The third hypothesis, that the Active profile would have the highest psychosocial HRQOL and the Inactive profile would have the lowest psychosocial HRQOL, was partially supported. Consistent with previous literature (Iannotti & Wang, 2013; Omorou et al., 2016), the results indicated that the Active profile had statistically and clinically significantly higher psychosocial HRQOL than the Moderate and Inactive profiles. This finding indicates that higher levels of total MVPA and MVPA bouts—specifically levels that exceed the 60-min guideline—contribute to higher levels of psychosocial HRQOL. However, in contrast to previous literature (Iannotti & Wang, 2013; Omorou et al., 2016), the Moderate and Inactive profiles had similar levels of psychosocial HRQOL, despite the Inactive profile having the dramatically higher SB. These results provide evidence that SB may not strongly contribute psychosocial HRQOL. Instead, it appears that MVPA, or more specifically, exceeding the USDHHS (2008) MVPA guideline, contributes to the significant differences. Taking into account the statistically and clinically significant findings regarding psychosocial HRQOL, several potential policy and intervention implications can be drawn. First, findings suggest that the guideline of 60 min of MVPA per day should be maintained, considering that participants who exceeded this recommendation had the highest psychosocial HRQOL. Further, findings suggest that interventions for elementary school-age youth should focus on substituting light activity with MVPA. Because the Active profile had significantly higher HRQOL than the Moderate profile, it can be argued that accumulating activity of higher intensity is driving the higher self-reported psychosocial HRQOL among the Active profile. Therefore, interventions focused on increasing children’s intensity of physical activity from light to moderate-to-vigorous would be expected to improve their psychosocial HRQOL by a statistically and clinically significant amount. Finally, there is evidence to suggest that interventions should be implemented in early elementary school, before adolescence. The current study suggests that these trajectories begin in elementary school, as evidenced by a higher proportion of older elementary school youth falling in the profile that had the lowest MVPA and highest SB. Therefore, interventions should be implemented early in development, when, theoretically, the changes required to see improvements would be smaller. In contrast to previous literature and our a priori hypothesis (Iannotti & Wang, 2013; Omorou et al., 2016), the activity profiles did not have statistically different levels of physical HRQOL. It is possible that, with a larger sample size, statically significant results would have emerged. However, our MCID analyses indicated that the Active profile did not differ meaningfully from the other two profiles on mean levels of physical HRQOL. Therefore, even if statistically significant results were found, there is evidence to suggest that physical HRQOL, as measured by the PedsQL, might not result in meaningful differences for the activity profiles. Although the preponderance of evidence in the literature led to the current hypotheses, there are also some investigations that found that activity profiles were not differentially associated with physical HRQOL. For example, Goldfield et al. (2015) found that screen time duration was associated with lower overall HRQOL and psychosocial HRQOL but not physical HRQOL. Additionally, other studies have found that another health behavior, disordered eating, was associated with psychosocial but not physical HRQOL (Jalali-Farahani et al, 2015). Therefore, the literature is mixed regarding the association of health behaviors and physical HRQOL as measured by the PedsQL. Further research should examine the association between profiles and other objective physical health outcomes, such as fitness, bone health, cholesterol, and blood pressure, or other subjective measures of physical functioning. Although some results from the current study failed to confirm the third and fourth hypotheses, the strengths may partially explain why these results differed from those in the existing literature. First, as mentioned previously, the current study used a person-centered, model-based approach to segment the population into smaller groups, taking into account total time spent in activity and bouts of activity, instead of only total time (Heitzler et al., 2011; Trilk et al., 2012). Additionally, the current study used objective measurements of activity, instead of relying on self-report measures like most previous studies (Berlin et al., 2017; Huh et al., 2011). The sample of the current study was unique, targeting elementary school-age youth during the time when activity levels dramatically change, in contrast to studies of adolescents (Berlin et al., 2017; Iannotti & Wang, 2013). Despite notable strengths, limitations must be considered when interpreting the findings of the current study. First, generalizability may be limited because of the sample being primarily Caucasian and middle-class. Future studies should examine profiles in a nationally representative sample of elementary school-age youth. Additionally, the current study is cross-sectional, which precludes statements of causality; future studies should examine longitudinal relationships between activity profiles and outcomes. Further, listwise deletion was implemented to exclude participants who did not have at least 2 valid weekdays and 1 weekend day of data. This approach limited sample size, thus reducing power and potentially producing biased estimates. Multiple imputation was considered, but this approach was not used because it could mask the identification of profiles in later analyses (Enders, 2010). Finally, although the inclusion of objective activity measurements is a strength, the current study did not include any self-report measures of MVPA or SB, limiting the ability to determine what types of activities the participants were engaging in or where they were accumulating the activity. Future studies should use a combination of self-report and objective measures. The current study was the first to use LVMM to identify and characterize profiles of elementary school-age youth based on objective measurements of total and bouts of activity. This study replicated profiles found in other samples of youth, suggesting that the profiles identified may represent reliable subgroups. Further, the current study extended the knowledge about these profiles, suggesting that interventions early in development focused on increasing intensity of physical activity from light to moderate-to-vigorous for at least 60 min per day may result in clinically meaningful differences in psychosocial HRQOL. Future studies should replicate the findings in a nationally representative sample, using objective and subjective measurements of MVPA and SB. Future studies are also needed to determine whether the profiles are associated with other outcomes and whether profiles differ based on other bout lengths or break patterns. Overall, the current study advances the literature on MVPA and SB in elementary school-age youth and has broader impacts for health promotion for youth. Supplementary Data Supplementary data can be found at: http://www.jpepsy.oxfordjournals.org/. Funding This research was supported, in part, by the University of Kansas General Research Fund (# 2301617), awarded to the second author. Conflicts of interest: None declared. Footnotes 1 Bootstrapped likelihood ratio test was considered, but ultimately not used because it tends to perform better with larger sample sizes (N > 1000: Nylund, Asparouhov, & Muthen, 2007). When this test was attempted in the current study, BLRT was not able to be computed. References Akaike H. ( 1987). Factor analysis and AIC. Psychometrika , 52( 3), 317– 332. Google Scholar CrossRef Search ADS   Altenburg T. M., Chinapaw M. J. M. ( 2015). Bouts and breaks in children's sedentary time: Currently used operational definitions and recommendations for future research. Preventive Medicine , 77, 1– 3. doi: 10.1016/j.ypmed.2015.04.019 Google Scholar CrossRef Search ADS PubMed  Bakk Z., Tekle F. B., Vermunt J. K. ( 2013). Estimating the association between latent class membership and external variables using bias-adjusted three-step approaches. Sociological Methodology , 43, 272– 311. doi: 10.1177/0081175012470644 Google Scholar CrossRef Search ADS   Banfield J. D., Raftery A. E. ( 1993). Model-based Gaussian and non-Gaussian clustering. Biometrics , 49, 803– 821. http://dx.doi.org/10.2307/2532201 Google Scholar CrossRef Search ADS   Bauer D. J., Steinley D. (Eds.) ( 2016). Cluster Analysis and Mixture Modeling Course Notes . Chapel Hill, NC: Curran-Bauer Analystics. Belcher B. R., Berrigan D., Dodd K. W., Emken B. A., Chou C. P., Spruijt-Metz D. ( 2010). Physical activity in US youth: Impact of race/ethnicity, age, gender, & weight status. Medicine and Science in Sports and Exercise , 42, 2211– 2221. Google Scholar CrossRef Search ADS PubMed  Berlin K. S., Kamody R. C., Thurston I. B., Banks G. G., Rybak T. M., Ferry R. J.Jr. ( 2017). Physical activity, sedentary behaviors, and nutritional risk profiles and relations to body mass index, obesity, and overweight in eighth grade. Behavioral Medicine , 43, 31– 39. doi: 10.1080/08964289.2015.1039956 Google Scholar CrossRef Search ADS PubMed  Berlin K. S., Williams N. A., Parra G. R. ( 2014). An introduction to latent variable mixture modeling (part 1): Overview and cross-sectional latent class and latent profile analyses. Journal of Pediatric Psychology , 39, 174– 187. doi: 10.1093/jpepsy/jst084 Google Scholar CrossRef Search ADS PubMed  Booth F. W., Roberts C. K., Laye M. J. ( 2012). Lack of exercise is a major cause of chronic diseases. Comprehensive Physiology , 2, 1143– 1211. doi: 10.1002/cphy.c110025 Google Scholar PubMed  Cain K. L., Sallis J. F., Conway T. L., Van Dyck D., Calhoon L. ( 2013). Using accelerometers in youth physical activity studies: A review of methods. Journal of Physical Activity and Health , 10, 437– 450. doi: 10.1123/jpah.10.3.437 Google Scholar CrossRef Search ADS PubMed  Carson V., Stone M., Faulkner G. ( 2014). Patterns of sedentary behavior and weight status among children. Pediatric Exercise Science , 26, 95– 102. doi: 10.1123/pes.2013-0061 Google Scholar CrossRef Search ADS PubMed  Celeux G., Soromenho G. ( 1996). An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification , 13, 195– 212. http://dx.doi.org/10.1007/BF01246098 Google Scholar CrossRef Search ADS   Centers for Disease Control and Prevention (CDC). ( 2011). About BMI for children and teens. Retrieved from http://www.cdc.gov/healthyweight/assessing/bmi/childrens_bmi Copay A. G., Subach B. R., Glassman S. D., Polly D. W., Schuler T. C. ( 2007). Understanding the minimum clinically important difference: A review of concepts and methods. The Spine Journal , 7, 541– 546. doi: 10.1016/j.spinee.2007.01.008 Google Scholar CrossRef Search ADS PubMed  Enders C. K. ( 2010). Applied Missing Data Analysis . New York, NY: Guilford. Evenson K. R., Catellier D. J., Gill K., Ondrak K. S., McMurray R. G. ( 2008). Calibration of two objective measures of physical activity for children. Journal of Sports Sciences , 26, 1557– 1565. doi: 10.1080/02640410802334196 Google Scholar CrossRef Search ADS PubMed  Fakhouri T. H., Hughes J. P., Brody D. J., Kit B. K., Ogden C. L. ( 2013). Physical activity and screen-time viewing among elementary school–aged children in the United States from 2009 to 2010. JAMA Pediatrics , 167, 223– 229. doi: 10.1001/2013.jamapediatrics.122 Google Scholar CrossRef Search ADS PubMed  Goldfield G. S., Cameron J. D., Murray M., Maras D., Wilson A. L., Phillips P., Kenny G. P., Hadjiyannakis S., Alberga A. S., Tulloch H., Doucette S., Sigal R. J. ( 2015). Screen time is independently associated with health-related quality of life in overweight and obese adolescents. Acta Paediatrica , 104, e448– e454. doi: 10.1111/apa.13073 Google Scholar CrossRef Search ADS PubMed  Gopinath B., Hardy L. L., Baur L. A., Burlutsky G., Mitchell P. ( 2012). Physical activity and sedentary behaviors and health-related quality of life in adolescents. Pediatrics , 130, e167– e174. doi: 10.1542/peds.2011-3637 Google Scholar CrossRef Search ADS PubMed  Heitzler C., Lytle L., Erickson D., Sirard J., Barr-Anderson D., Story M. ( 2011). Physical activity and sedentary activity patterns among children and adolescents: A latent class analysis approach. Journal of Physical Activity and Health , 8, 457– 467. Google Scholar CrossRef Search ADS PubMed  Huh J., Riggs N. R., Spruijt Metz D., Chou C. P., Huang Z., Pentz M. ( 2011). Identifying patterns of eating and physical activity in children: A latent class analysis of obesity risk. Obesity , 19, 652– 658. doi: 10.1038/oby.2010.228 Google Scholar CrossRef Search ADS PubMed  Iannotti R. J., Wang J. ( 2013). Patterns of physical activity, sedentary behavior, and diet in US adolescents. Journal of Adolescent Health , 53, 280– 286. http://dx.doi.org/10.1016/j.jadohealth.2013.03.007 Google Scholar CrossRef Search ADS PubMed  Jalali-Farahani S., Chin Y. S., Nasir M. T. M., Amiri P. ( 2015). Disordered eating and its association with overweight and health-related quality of life among adolescents in selected high schools of Tehran. Child Psychiatry and Human Development , 46, 485– 492. doi: 10.1007/s10578-014-0489-8 Google Scholar CrossRef Search ADS PubMed  Kline R. B. ( 2011). Principles and Practice of Structural Equation Modeling  ( 3rd ed.). New York, NY: Guilford Press. Leech R. M., McNaughton S. A., Timperio A. ( 2014). The clustering of diet, physical activity and sedentary behavior in children and adolescents: A review. International Journal of Behavioral Nutrition and Physical Activity , 11, 4. Google Scholar CrossRef Search ADS PubMed  Lubke G. H., Neale M. C. ( 2006). Distinguishing between latent classes and continuous factors: Resolution by maximum likelihood. Multivariate Behavioral Research , 41, 499– 532. doi: 10.1207/s15327906mbr4104_4 Google Scholar CrossRef Search ADS PubMed  Muthén L. K., Muthén B. O. ( 1998-2012). Mplus User's Guide  ( 7th ed.). Los Angeles, CA: Muthén & Muthén. Nylund K. L., Asparouhov T., Muthén B. O. ( 2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling , 14, 535– 569. Google Scholar CrossRef Search ADS   Omorou A. Y., Langlois J., Lecomte E., Briançon S., Vuillemin A. ( 2016). Cumulative and bidirectional association of physical activity and sedentary behaviour with health-related quality of life in adolescents. Quality of Life Research , 25, 1169– 1178. Google Scholar CrossRef Search ADS PubMed  Rowlands A. V., Pilgrim E. L., Eston R. G. ( 2008). Patterns of habitual activity across weekdays and weekend days in 9–11-year-old children. Preventive Medicine , 46, 317– 324. doi: 10.1016/j.ypmed.2007.11.004 Google Scholar CrossRef Search ADS PubMed  Sadeh A., Sharkey K. M., Carskadon M. A. ( 1994). Activity-based sleep-wake identification: An empirical test of methodological issues. Sleep , 17, 201– 207. Google Scholar CrossRef Search ADS PubMed  Salmon J., Tremblay M. S., Marshall S. J., Hume C. ( 2011). Health risks, correlates, and interventions to reduce sedentary behavior in young people. American Journal of Preventive Medicine , 41, 197– 206. doi: 10.1016/j.amepre.2011.05.001 Google Scholar CrossRef Search ADS PubMed  Saunders T. J., Tremblay M. S., Mathieu M. È., Henderson M., O’loughlin J., Tremblay A., Chaput J. P.; QUALITY Cohort Research Group. ( 2013). Associations of sedentary behavior, sedentary bouts and breaks in sedentary time with cardiometabolic risk in children with a family history of obesity. PLoS One , 8, e79143. doi: 10.1371/journal.pone.0079143 Google Scholar CrossRef Search ADS PubMed  Schwarz G. ( 1978). Estimating the dimension of a model. The Annals of Statistics , 6, 461– 464. doi: 10.1214/aos/1176344136 Google Scholar CrossRef Search ADS   Sclove S. L. ( 1987). Application of model-selection criteria to some problems in multivariate analysis. Psychometrika , 52, 333– 343. http://dx.doi.org/10.1007/BF02294360 Google Scholar CrossRef Search ADS   Spieth L. E., Harris C. V. ( 1996). Assessment of health-related quality of life in children and adolescents: An integrative review. Journal of Pediatric Psychology , 21, 175– 193. http://dx.doi.org/10.1093/jpepsy/21.2.175 Google Scholar CrossRef Search ADS PubMed  Trilk J. L., Pate R. R., Pfeiffer K. A., Dowda M., Addy C. L., Ribisl K. M., Neumark-Sztainer D., Lytle L. A. ( 2012). A cluster analysis of physical activity and sedentary behavior patterns in middle school girls. Journal of Adolescent Health , 51, 292– 298. Google Scholar CrossRef Search ADS PubMed  Troiano R. P., Berrigan D., Dodd K. W., Masse L. C., Tilert T., McDowell M. ( 2008). Physical activity in the United States measured by accelerometer. Medicine and Science in Sports and Exercise , 40, 181– 188. doi: 10.1249/mss.0b013e31815a51b3 Google Scholar CrossRef Search ADS PubMed  United States Census Bureau ( 2015). Quick facts: Lawrence, Kansas. Retrieved from https://www.census.gov/quickfacts/table/PST045215/2038900, 00 US Department of Health and Human Services (USDHHS). ( 2008). 2008 physical activity guidelines for Americans. Retrieved from http://health.gov/paguidelines/pdf/paguide.pdf Varni J. W., Seid M., Kurtin P. S. ( 2001). PedsQL (TM) 4.0: Reliability and validity of the Pediatric Quality of Life Inventory (TM) version 4.0 generic core scales in healthy and patient populations. Medical Care , 39, 800– 812. http://dx.doi.org/10.1097/00005650-200108000-00006 Google Scholar CrossRef Search ADS PubMed  Willis E. A., Ptomey L. T., Szabo-Reed A. N., Honas J. J., Lee J., Washburn R. A., Donnelly J. E. ( 2015). Length of moderate-to-vigorous physical activity bouts and cardio-metabolic risk factors in elementary school children. Preventive Medicine , 73, 76– 80. doi: 10.1016/j.ypmed.2015.01.022 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2017. Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Journal of Pediatric PsychologyOxford University Press

Published: Dec 11, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off