Three-year changes in social correlates of physical activity, physical activity and health -related fitness among adolescents: the PAHL study

Abstract

Background: Physical inactivity has been ranked the fourth leading risk factor for global mortality. Despite the health risks, physical inactivity is common. Studies to identify the correlates of physical activity to inform the design of interventions and to reduce the disease burden associated with physical inactivity, have become a public health imperative. Research evidence has been consistent in substantiating social support as significant indicators of children and adolescents’ participation in physical activity (PA). The understanding of the associated social correlates of physical activity can be significantly enhanced by being examined in a longitudinal study, because physical activity behaviour is assumed to track over time. The objective of the study therefore, was to investigate the changes in social correlates of physical activity, physical activity and health-related fitness in a three-year follow-up study among learners in high schools in the Tlokwe local municipality, South Africa. Methods: Data from a total of 206 (where boys: 73 and girls: 133) in 2012, 160 (where boys: 62 and girls: 98) in 2013 and 138 (where boys: 87 and girls: 51) in 2014 at the three measurements of 2012 to 2014 in the Physical Activity and Health Longitudinal Study (PAHLS) were used. The participants who were aged 14 years and in grade 8 were purposefully selected from class lists provided so that they could be successfully followed for the duration of the 5 year PAHL study before completing high school at the age of 18 years. The International Physical Activity Questionnaire-Short Form (IPAQ-SF) was used to determine the levels of physical activity. The cardiorespiratory endurance, muscle strength and endurance, and flexibility tests were conducted according to the standard procedures of the EUROFIT. Anthropometric measurements of height, weight, skinfold thickness and waist circumferences were determined using the standard procedures described by the International Society for the Advancement of Kinanthropometry (ISAK). Waist-to-height ratio (WHtR), body mass index (BMI) and percentage body fat (%BF) were calculated. A standardised questionnaire on the ‘Social Support for Physical Activity’ was used to gather information on social correlates for physical activity. Descriptive statistics including frequency, percentage, mean and standard deviations were used to explore the data. For comparing the continuous and categorical data t-test and chi-square were used. Non-parametric repeated-measures ANOVA with the Friedman test was used to assess changes in the correlates between test measurement number one (T1), test measurement number two (T2) and test measurement number three (T3). Since the statistical significant found with Friedman test does not pinpoint which groups in particular differ from each other, post-hoc analyses with Wilcoxon signed-rank tests was conducted with a Bonferroni correction applied for multiple comparison, which makes it more likely to declare a results significant when there was no Type I error. To comply with the rules of Bonferroni test, we divided the p-value of 0.05 by the number of the tests (i.e. 0.05/3=0.017). As such, the new significant level used in this test was 0.017; and that means that if the p-value was larger than 0.017, we do not have a statistical significant result. Effect sizes (partial Eta squared (ηp²) were used to assess the magnitude of these changes. Tracking (stability) was assessed using Spearman correlation coefficient. Attrition analysis was performed using independent sample t-test and chi-square of proportions to determine the difference in baseline characteristics between participants and drop-outs (n=49; 20%). No significant difference was observed between the dropouts and the actual data used in this thesis. Kolmogorov-Smirnov tests for normality was used to check if the data was normally distributed. Chi-square was calculated to determine the differences between variables. Age-adjusted Pearson correlation analysis was performed to study the development of social correlates of physical activity, physical activity and physical activity in relation with health-related fitness. Linear regression analysis with adjusted age and maturation was performed to study the relationship of changes in social correlates of physical activity, physical activity, and physical activity in relation with health-related fitness. All analyses were performed by making use of the SPSS version 21.0 (IBM SPSS Inc., Chicago III 2013) statistical programme. Results: There were significant statistical (p<0.05) changes and a high correlation coefficient (ranged from r=0.to 90 r=0.97) as well as large practical developmental changes (d≥0.8) (partial Eta Square (ηp²))) in BMI, %BF and WHtR over a three year period. Small practical but insignificant statistical (p>0.05) changes in social correlates (encouragement, coactivity, transportation) were found. A significant change (p=0.04) for someone who watched you participate in PA or sport among girls, was revealed. There was strong significant differences (p<0.001) in mean standing broad jump (SBJ), sit-up (SUP) and sit-and-reach (SAR), stature and body mass (p=0.002) and BMI among the boys and girls. The results show an increase in stature, body mass and BMI for the entire sample. The SBJ, sit-up and sit-and-reach seemed to decrease through the three measurement points. The boys had higher body mass as compared to the girls, while the girls had higher BMI. There was a statistically significant differences in body mass (X2(df=2) = 10.354, p=0.006) and BMI (X2(df=2) = 11.400) over a period of three years. Post hoc signed-rank analyses with Wilcoxon signed-rank tests was conducted with Bonferroni correction applied, resulting in a statistical significant level set at p<0.017. There was no significant differences between BMI T1 (Z = -2.240, p=0.025) and BMI T2 or between the BMI T1 and BMI T3 (Z= -2.313, p=0.021) measurements, despite an overall decrease in the BMI measurements. However, there was a statistical decrease in BMI measured at time point 1 (T2) and BMI at measurement point 2 (T3) (Z= -3.034, p=0.002). The partial Eta square of the effect size was moderate (d=0.66) for SJB, small (d=0.36) and (d=0.29) for SUP and SAR respectively. Furthermore, the results revealed a significant high correlation between body mass and stature, and a moderate correlation between stature and body mass. There was also an insignificant correlation between stature and BMI (r = -0.04; p = 0.64) and (r = -0.07; p = 0.40). While the SUP, SBJ and SAR showed a significantly weak correlation with the body mass. The girls showed a significant moderate correlation between stature and SUP in T3 (r = 0.50; p = 0.00). The results also showed significant changes for vigorous and moderate exercise and minutes spend watching TV/Sitting (p<0.001) for the period of measurement. The practical effect size (ηp²) of the changes was medium for vigorous activity minutes per week (d=0.11), medium for moderate activity minutes per week (d=0.07) and moderate for minutes spend watching TV/Sitting (d=0.61). The practical effect size of the changes was relatively small (d≤ 0.2) for all the variables. The results also revealed a significant relation for the questions “has someone done a physical activity or played sports with you?” (p<0.001), “has someone provided transportation to a place where you can do physical activities or play sports?”(p= 0.03) and “has someone watched you participate in physical activities or sports?”(p= 0.01). There were high mean values in social correlates and physical activity for the boys as compared to the girls. There was a significantly high association between “During a typical week has someone told you that you are doing well in physical activity?” and vigorous physical activity (r = 0.61; p = 0.03) per week. There was no statistical significance between measurement point 1 vs measurement 2 (Z=-0.929, p=0.353) and measurement points 3 and 4 (Z= -1.152, p=0.249), respectively. When ANOVA for repeated measure with Friedman test, statistical significant (p=0.017) was found for the physical activity measure of vigorous (X2(df=2)=11.382, p=0.003), moderate physical activity (X2 (df=2)=13.446, p=0.001) minutes spent watching TV/sitting (X2 (df=2)=29.531, p=0.000) and total physical activity (X2 (df=2)=29.531, p=0.000). Though no statistical significant differences (p>0.017) in TPA in all the three measurement points (T 1 vs T2, Z= -2.071, p=0.038; T2 vs T3, Z= -0.088, p=0.930 & T1 vs T3, Z= -2.367, p=0.018), total physical activity declined over a period of time. When a post-hoc followed was performed, the median (IQR) for “During a typical week has someone watched you participate in physical activities of sports?”, for measurement at points 2 and 3 was Z=-2.909, p=0.004. There was no statistical significance between measurement point 1 vs measurement 2 (Z= -0.929, p=0.353) and measurement points 3 and 4 (Z= -1.152, p=0.249), respectively. The results also showed a significant association between moderate physical activity and minutes spent watching TV or sitting (r = 0.67; p = 0.01) per week. A moderate significant positive correlation coefficients were observed respectively for SBJ (r = 0.31; p = 0.01) and SUP T2 (r = 0.32; p = 0.01), and total physical activity (TPA 2013). Significant positive moderate correlation coefficient was found between SUP T2 (r = 0.49; p = 0.001), and SUP T3 (r = 0.37; p = 0.05) and TPA 2013 respectively for the boys. Conclusions: There were high correlation coefficients for the developmental changes in body mass, stature, BMI, %BF and WHtR over a period of time. The adolescents did not receive any transportation support over time. Adolescents were motivated by being watched by others for participation in physical activity. There is significant gender difference in SBJ, SUP, SAR, stature, body mass and BMI. There were some developmental changes in the health-related fitness variables and the effect size was medium for SBJ and small for SUP and SAR. The girls received less social support as compared to the boys. The girls participated less in physical activity as compared to the boys. The girls spent more minutes watching TV/Sitting in 2012 and 2014 as compared to the boys. The study also revealed that the children participated in vigorous physical activity when friends and family or someone told them that they were doing well in physical activity and sport. Explosive strength was significantly correlated with physical activity, while functional strength test was associated with physical activity in boys over a period of time