measured and body mass index was calculated in order to to classify subjects as overweight or obese (IOTF classification). Multiple logistic regression and ...
Journal of Physical Activity and Health, 2010, 7, 633-640 © 2010 Human Kinetics, Inc.
Physical Activity Patterns and Obesity Status Among 10- to 12-Year-Old Adolescents Living in Athens, Greece George Antonogeorgos, Anastasios Papadimitriou, Demosthenes B. Panagiotakos, Kostas N. Priftis, and Polyxeni Nikolaidou Background: Childhood obesity has become a modern epidemic with escalating rates. The aim of our study was to identify physical activity patterns among Greek schoolchildren and to examine their relationship with obesity. Methods: 700 adolescents age 10 to 12 years were evaluated through a standardized questionnaire. Several demographic, socioeconomic, and physical activity characteristics were recorded. Physical activity was assessed and adolescents were characterized as active and nonactive. Body height and weight were measured and body mass index was calculated in order to to classify subjects as overweight or obese (IOTF classification). Multiple logistic regression and multivariate techniques (principal components analysis) were performed. Results: Eight physical activity patterns were identified, including increased physical activity in weekdays and weekends, sports physical activity, vigorous, moderate, and low physical activity. Increased physical activity on weekends and vigorous physical activity in boys were negatively associated with being overweight or obese (OR: 0.65; 95% CI: 0.48–0.90 and OR: 0.66; 95% CI: 0.49–0.88, correspondingly) and moderate physical activity was marginally positively associated in girls (OR: 1.28; 95% CI: 0.97–1.69), after adjusting for several confounders. Conclusions: Our findings demonstrate the important role of vigorous physical activity in the maintenance of normal weight of adolescents Keywords: vigorous, weight, maintenance, exercise, moderate
Background Childhood obesity has become a major public health concern in the last 4 decades, affecting both developed and developing countries globally. It is estimated that worldwide, approximately 1 out of 10 children are overweight (a total of 155 million) and among them 35 to 45 million children of school age and 22 million children under 5 years are obese.1 Most alarming is the fact that the prevalence of childhood obesity is rising rapidly. The prevalence of overweight and obese children in United States age 2 to 5 years and adolescents age 12 to 19 years has doubled during the past 3 decades and has tripled for the age group of 6 to 11 years.2 In Europe, about 400,000 children every year are affected by the obesity epidemic and approximately 14 million children are overweight or obese.3 Moreover, it has been suggested that childhood obesity is one of the main risk factors for adulthood obesity.4 It has been associated with increased risk of developing cardiovascular diseases,5–7 hypercholesterolemia, Antonogeorgos, Papadimitriou, and Nikolaidou are with the 3rd Dept of Pediatrics, University of Athens, Haidari, Greece. Panagiotakos is with the Dept of Nutrition Science-Dietetics, Harokopio University of Athens, Greece. Priftis is with the Allergy Dept, Penteli Children’s Hospital, P. Penteli, Greece.
hypertension,6 and increased prevalence of early onset of type 2 diabetes mellitus.8,9 Most important are the psychological consequences of childhood obesity, which includes low self-esteem, depression, and body dissatisfaction.10,11 One of the most important factors implicated in the multifactorial etiology of childhood obesity is physical activity. Obesity is the result of an imbalance of energy intake and expenditure. As a key determinant of energy expenditure, physical activity (PA) is essential to energy balance and weight control. Obese children tend to follow lifestyles with decreased levels of physical activity and tend to spend more time in sedentary activities like television viewing or video game playing than their nonobese counterparts.12,13 In Greece, the prevalence of childhood obesity is high and follows a secular trend, comparable to that reported for most European countries.14,15 Limited data exist about the association of physical activity and obesity in Greek adolescents. In a study by Lagiou et al, time spent watching television or operating the computer per day was positively associated with overweight among primary school students.16 Moreover, Krassas et al reported that overweight in 2495 Greek children age 6 to 17 years that participated in a cross-sectional study in North Greece’s region surrounding Thessaloniki was also positively influenced by hours of daily television viewing.17 633
634 Antonogeorgos et al
The purpose of this study was to identify patterns of physical activity among adolescents age 10 to 12 years old from an urban environment and to assess their relationship with obesity.
Methods Study Sample From 2005 to 2006, 700 students (323 male, 377 female), age 10 to 12 years (4th–6th grade), were selected from 18 schools located in the greater Athens area. The overall participation rate was 83.5%. Participation of subjects was on a voluntary basis. Adolescents’ guardians were fully informed on the objectives and methods of the study before acceptance and signed an informed consent. The schools were randomly selected from a list of schools provided by the regional education offices. In order for the sample to be representative, the enrolled schools were selected from various areas of the Athens region. The number of enrolled adolescents was adequate (ie, statistical power 80%) to evaluate standardized differences between various groups greater than 0.5 at probability level < 0.05. The study was designed according to the principles of the declaration of Helsinki (1989). The protocol of the study has been approved by the Research Department of the Education Institute of the Hellenic Ministry of Education, which is responsible for establishing ethical approval for any study that takes place in school settings (approval number 29712/G7/2006).
Measurements Standing height was measured with a Raven Minimeter (Raven Equipment Limited, Essex, United Kingdom) to the nearest 0.1 cm and body weight to the nearest 0.1 kg with a Seca weighing scale (Seca, Hanover, MD) after students had removed their heavy clothes (coats, sweaters, blouses, etc) and shoes. Measurements took place in the school setting. Overweight and obesity were defined using the international body mass index (BMI) cut-off points established for children and youth.18 These cutoff points are based on health-related adult definitions of overweight (25–29.9 kg/m2) and obesity (≥30 kg/ m2) and are adjusted to specific age and sex categories for children. Parents self-reported their body height and weight. According to their BMI value adolescents were classified as normal, overweight (corresponding adult BMI: 25.0–29.9 kg/m2) or obese (corresponding adult BMI: ≥30.0 kg/m2). Adolescents’ birth weight (BW) was reported by the parents who were instructed to use the standard health record booklet that accompanies each child in Greece from birth and the information was entered into the following 5 BW categories: 3500 g. Parents also completed a nonvalidated questionnaire assessing their level of education (basic referring to no education, elementary or high school education, or
higher referring to college educational level), the lifestyle characteristics of their adolescents (hours of viewing television/playing videogames, if they have their own bedroom, number of automobiles per family, hours of adolescents sports activities per week, hours of nonsports extracurricular activities of adolescents per week, ie, foreign language classes), and adolescents’ history of breastfeeding during infancy.
Physical Activity Evaluation Information on the frequency and duration of a variety of physical activities of the adolescents was retrieved and recorded in a validated special and reliable questionnaire, by the participants with the presence of a researcher.19–22 In particular, the questionnaire solicited information for a typical week and mainly for the past 7 days. The questions assessed the participation of the students in free time activities such as going out with friends, social visits, going to the movies or theater, on the amount of time spent on sedentary activities (eg, watching television, working on a computer, playing video games), and the participation in physical activity and sports. Physical activity and sports questions evaluated modes of transportation, physical activity and sports during school, participation in recreational physical activity, participation in sporting teams and private gyms in the free time after school, and physical activity and sports during weekends. The information retrieved concerned the type of activity, the time spent, and the intensity of each activity and also its weekly frequency. In addition, adolescents evaluated their daily physical activity during weekdays and weekends on a 5-point scale and were categorized according to their reported activity score during a particular day of the last week (Monday, Tuesday, etc. including weekend). If they scored 3 or more in the corresponding day they were classified as following a pattern of increased physical activity during that particular day and if they scored 2 or less they were classified as following a pattern of decreased physical activity during that particular day of the last week. The physical activity questionnaire has been previously validated in metabolic equivalence (MET, 1 MET = 3.5 ml O2/kg/min) by the Department of Sports Science of the Democritus University of Thrace and was found to be a reliable and valid instrument of the estimation of physical activity in children for use in epidemiological studies, compared with other physical activity questionnaires and accelerometer measurements.19–22 According to the evaluation of the PALQ for the last 7 days in daily energy expenditure (kcal×kg–1×day–1), all participating adolescents were classified as nonactive (≤37 kcal×kg–1×day–1) and active (>37 kcal×kg–1×day–1).
Statistical Analysis Continuous variables are presented as mean ± standard deviation (SD), while categorical variables are presented as absolute and relative frequencies. Contingency tables with calculation of chi-square test and Fisher’s exact test
Physical Activity Patterns and Obesity 635
(when appropriate) evaluated associations between the categorical variables. Relationships between categorical and continuous variables were tested by student’s t test (for normally distributed variables) and Mann-Whitney criterion for skewed variables. To obtain physical activity patterns the Principal Components Analysis (PCA) was used. The PCA is a multivariate technique that evaluates the intercorrelations between the initial physical activity variables. Thus, using this method, physical activity patterns are revealed in an uncorrelated way. Moreover the extracted component acts as a score variable concerning the corresponding pattern. All the physical activity variables entered in PCA were statistically significant associated with the activity status of the adolescents (P < .05). The correlation matrix of the investigated variables showed that there were several correlation coefficients r > 0.4, indicating that a PCA could be effective for assessing physical activity patterns. In addition, the KMO measure of sampling adequacy (a measure of the association of categorical variables) was 0.71, which is close to 1 and implies high interrelationships between the variables. The orthogonal rotation (varimax option) was used to derive optimal noncorrelated components (physical activity patterns). To decide the number of the extracted components that should be retained from the PCA, the Kaiser criterion, a statistical criterion which indicates that the number of components that should be retained from PCA is equal to the number of their corresponding eigenvalues that are greater than 1, was used. In particular, 8 out of 22 components were retained and explained the 60% of the total variability. The components (patterns) were named according to the scores of the variables that
correlated most with the component (ie, those who had absolute scores > 0.4). Then gender-specific multiple logistic regression analysis determined to what extent the evaluated factors were associated with the prevalence of obesity status (the overweight and obese categories were combined because of the small number of participants in obese group). All the aforementioned patterns were entered simultaneously in the model and their effect was adjusted for the following epidemiological known obesity confounders: birth weight, history of breastfeeding, and parents’ obesity status and level of education. Deviance residuals were calculated to evaluate the model’s goodness-of-fit. All reported probability values (P-values) are based on 2-sided tests. All statistical analyses performed with SPSS v13.0 (SPSS Inc., Chicago Il, USA).
Results Table 1 presents the prevalence of overweight/obesity and several lifestyle characteristics (having their own bedroom, participation in sports, hours of homework per day, hours of TV viewing and videogame per day etc.) of the adolescents according to their physical activity status. More active boys than their nonactive counterparts had their own bedroom (P = .039) while active girls engaged more in sports activities (P = .007) and for longer mean hours of participation per week (P < .001) than the nonactive counterparts. Table 2 presents the prevalence of several physical activities in a typical week of the participating adolescents by gender and physical activity status. Active boys and girls engaged themselves significantly more times
Table 1 Anthropometric and Lifestyle Characteristics of the Participants According to Their Physical Activity Status Non-active (n = 139)
Active (n = 170)
Overweight/obese, (%)
46.5
40.7
0.37
Own bedroom, Yes (%)
68
79.3
0.039
Boys
Participation in Sports, Yes (%)
P
78.2
81.1
0.65
Hours of homework/day (Mean (SD))
2.31 (1.03)
2.37 (1.06)
0.62
Hours of TV/videogames/day (Mean (SD))
2.34 (1.58)
2.30 (1.19)
0.81
Hours of sports activities/week (Mean (SD))
3.90 (3.30)
4.47 (3.37)
0.19
Hours of non sport extracurricular activities/week (Mean (SD))
3.78 (2.25)
4.32 (2.43)
0.094
Girls
Nonactive (n = 195)
Active (n = 169)
P
Overweight/obese, (%)
28.0
32.1
0.519
Own bedroom, Yes (%)
71.1
75.7
0.381
Participation in Sports, Yes (%)
46.5
61.3
0.007
Hours of homework/day (Mean (SD))
2.77 (1.15)
2.76 (1.16)
0.95
Hours of TV/videogames/day (Mean (SD))
2.19 (1.26)
2.12 (1.29)
0.625
Hours of sports activities/week (Mean (SD))
2.58 (1.7)
3.74 (3.05)
90 min No. of trainings/week causing puffing for at least 20 min
18.9
29.9
16.4
32.9
No training
15.0
8.2
7.3
8.0
No puffing
42.5
31.3
53.6
48.0
1–2 times
25.0
27.1
26.1
25.3
3–4 times
5.0
14.6
1.4
6.7
Every time No. of recreational activities/week causing puffing for at least 20 min
12.5
18.8
11.6
12.0
No recreational activity
10.4
2.5
13.1
5.6
No puffing
27.2
20.1
39.3
29.8
1–2 times
40.8
37.7
31.7
38.5
3–4 times
11.2
16.4
4.4
10. 6
Every time No. of participation in physical activities/ weekend for at least 20 min
10.4
23.3
11.5
15.5
No participation for the last week
11.1
4.2
23.1
6.5
0.387
0.625
0.002
0.007
0.013
6 times
12.7
22.0
5.4
13.1
636
P
Physical Activity Patterns and Obesity 637
in recreational activities causing puffing for at least 20 minutes per week (P = .002 and P = .007, respectively) and particularly on weekends (0.013 and < 0.001, respectively). In addition, active girls spent significantly more minutes per day in walking or cycling for transportation (P = .015) and participated more times per week in physical education class (P = .026) than their nonactive girls of the study. In Table 3, the physical activity patterns extracted from PCA are reported. The first principal component is most associated with the variables referring to the increased physical activity from Monday to Friday and expresses the “increased activity in weekdays” pattern. The second component represents the “increased activity in weekend” pattern. The third component is associated
with “sports physical activity” and the 4th component is the “moderate physical activity” pattern. The fifth component is most related with physical activity from private gym training. The other patterns identified from the PCA are “physical activity in school breaks and active transportation,” “low physical activity,” and “vigorous physical activity.” Furthermore, to examine the relationship of the aforementioned activity patterns with the obesity status of the participating adolescents, sex-specific multiple logistic regression analysis was applied, adjusting for known epidemiological obesity confounders such as parents’ obesity status and educational level, birth weight, and history of breastfeeding. Table 4 presents the results from the logistic regression model. The adjusted odds
Table 3 Results From Principal Components Analysis That Was Applied in Physical Activity Variables of the PANACEA Study Database* Variable
Comp1
Comp2
Comp3
Comp4
Comp5
Comp6
Comp7
Comp8
Explained variability
14%
8%
7%
6%
5%
5%
5%
5%
Listen music at home
0.040
0.086
–0.069
0.041
0.071
–0.033
0.817
–0.119
Participation in athletic activities for fun
0.118
0.247
0.169
0.436
–0.095
0.238
0.076
0.218
Duration of walking/bicycling per day for transportation
0.041
0.081
–0.101
0.185
0.060
0.625
–0.101
0.144
Physical activity in school breaks
0.153
–0.104
0.383
–0.209
–0.034
0.602
0.113
–0.007
Member of a school sports team
0.041
0.173
0.612
0.178
–0.167
–0.124
–0.112
–0.012
Member of a sports team
0.0004
0.007
0.701
0.031
0.180
0.197
0.008
–0.046
No of trainings/week for sports team
–0.001
0.100
0.465
–0.028
0.366
–0.165
0.056
0.268
Duration of training in sports team
0.019
0.041
–0.030
0.012
0.037
0.173
–0.127
0.824
Member of a private gym
0.057
–0.099
–0.042
0.085
0.712
–0.032
0.085
0.057
No. of trainings/week for private gym
0.053
0.060
0.098
–0.005
0.736
0.066
–0.054
–0.055
No. of walking for fun in the past 7 days for > 20 min
–0.010
0.121
0.004
0.178
–0.053
0.453
0.447
0.038
No. of cycling for fun in the past 7 days for > 20 min
0.102
–0.091
–0.103
0.610
0.039
0.159
–0.131
–0.206
No. of running for fun in the past 7 days for > 20 min
–0.018
0.109
0.025
0.589
0.078
0.081
0.095
0.144
No. of playing basketball in the past 7 days for > 20 min
0.060
0.372
0.295
0.155
–0.082
0.146
–0.088
–0.117
No. of playing volleyball in the past 7 days for > 20 min
0.064
–0.028
0.223
0.557
0.008
–0.187
0.111
–0.061
Increased physical activity in Monday
0.663
0.169
0.012
0.032
0.182
0.150
–0.194
–0.204
Increased physical activity in Tuesday
0.691
–0.028
0.124
0.147
–0.134
–0.115
0.259
0.256
Increased physical activity in Wednesday
0.616
0.130
–0.037
0.110
0.099
0.083
–0.096
–0.166
Increased physical activity in Thursday
0.713
–0.094
0.040
–0.029
–0.042
–0.156
0.116
0.226
Increased physical activity in Friday
0.586
0.368
0.007
–0.007
0.057
0.107
0.036
–0.019
Increased physical activity in Saturday
0.164
0.802
0.083
0.015
0.042
–0.036
0.111
0.080
Increased physical activity in Sunday
0.077
0.833
0.029
0.014
–0.027
0.031
0.063
0.027
* With bold characters are coefficients greater than 0.25, which corresponding variables represent better than the component.
638 Antonogeorgos et al
Table 4 Results From Gender Specific Multiple Logistic Regression Analysis That Evaluated Physical Activity Factors Associated With Obesity Status in Adolescents Age 10–12 Years From Athens Boys
Girls
Adjusted OR*
95% CI
Adjusted OR*
95% CI
Comp1: “Increased physical activity in weekdays”
0.89
0.67–1.19
1.16
0.86–1.56
Comp2: “Increased physical activity in weekend”
0.65
0.48–0.90
1.01
0.73–1.40
Comp3: “Sports physical activity”
1.00
0.73–1.36
0.91
0.67–1.25
Comp4: “Moderate physical activity”
1.31
0.94–1.84
1.28
0.97–1.69
Comp5: “Gym training physical activity”
0.93
0.67–1.28
1.06
0.80–1.41
Comp6: “School breaks—transportation physical activity”
1.03
0.76–1.39
1.19
0.87–1.63
Comp7: “Low physical activity”
0.94
0.69–1.26
1.29
0.95–1.75
Comp8: “Vigorous physical activity”
0.66
0.49–0.88
0.93
0.68–1.27
* Adjusted OR: Adjusted odds ratio for birth weight, history of breastfeeding, parental BMI, and educational level.
ratio for the second principal component for the boys is 0.65 (95% CI: 0.48–0.90), meaning that increased physical activity during weekends is negatively associated with obesity. Similarly, vigorous physical activity (ie, Comp8) also decreased the odds of the boys being overweight or obese. On the other hand, for the girls of the study, moderate physical activity was marginally positively associated with obesity (adjusted odds ratio: 1.28; 95% CI: 0.97–1.69).
Discussion We assessed the patterns of physical activity in a typical week of adolescents age 10 to 12 years living in a metropolitan city in Greece. Multivariate analysis transformed the initial set of variables into a reduced set of uncorrelated physical activity-related patterns. The 2 main patterns of adolescents’ physical activity are increased physical activity during weekdays and increased physical activity during weekends. Principal component analysis also revealed 6 more patterns, identifying physical activity derived from participating in sports teams and during free time activities, as well as patterns of vigorous and low physical activity. We also evaluated the relation of the extracted patterns with the obesity status (overweight/ obese) of the boys and girls of the study, adjusting the results for parental BMI and educational level, birth weight, and history of breastfeeding. Increased physical activity in the weekend and vigorous physical activity were inversely associated with overweight/obesity in the boys of the study, while for the girls moderate physical activity was marginally associated with increased odds of overweight/obesity. In our study the 2 main patterns of physical activity identified were the increased PA in the weekdays and increased PA in the weekends. There are several studies
in the literature that also report different patterns of PA between weekends and weekdays. In a study by Duncan et al23 assessing ambulatory PA in school children from central England, results indicated higher activity levels during weekdays than weekends. Rowlands et al using accelerometry in 84 children age 9 to 11 years, also reported increased frequency and duration of activity bouts in weekdays.24 Similar findings are presented for Swedish, Australian, American, and New Zealand school age children.25–27 Increased physical activity in weekends was significantly negatively associated with overweight/obesity only in the participating boys. This may be explained by the increased level of physical activity for boys in the weekends, which leads to overall greater energy expenditure. Ziviani et al28 examined physical activity patterns using a pedometer for 2 weekdays and 2 weekends and note that boys took more steps than girls per day in weekends. They also reported an inverse relationship between body mass index and number of steps taken per day. Duncan et al23 and Rowlands et al29 reported higher mean steps/day for boys and accelerometer counts/day in both weekdays and weekends. Several studies supports the above findings.30,31 Vigorous physical activity pattern was negatively associated with being overweight or obese in the boys of our study. Rush et al32 estimated the physical activity levels for 79 boys and girls age 5 to 14 years using the energy expenditure/resting metabolic rate ratio and their body fat, and found an inversely associated significant relationship between them only for boys. This finding extends the age limits to a prior work by Ball et al33 who reported the same inverse association for boys age 6 to 9 years old. On the other hand, in a large cross sectional study of 5500 12-year-old children using accelerometry for assessing total and moderate/vigorous physical activity (MVPA) and x-ray emission absorptiometry for
Physical Activity Patterns and Obesity 639
fat mass and obesity, no gender differences were found in the negative effect of MVPA to obesity, adjusted for the effect of total PA.31 No gender differences were also observed for the effect of vigorous physical activity pattern in obesity in numerous other studies.34,35 Our study is a cross sectional study and shares the limitations of observational studies, like absence of causal relationships and generalization of the results. Although the sample size is relatively small, compared with other international studies, the statistical power for all analyses was high (ie, >80%) to evaluate standardized differences between various groups greater than 0.5 at probability level
