Physical activity , psychosocial and perceived environmental factors in adolescents from Northeast Brazil

The aim of this study was to evaluate the association between levels of physical activity, psychosocial and perceived environmental factors in adolescents from Northeastern Brazil. A cross-sectional epidemiologic study was conducted with 2,859 adolescents enrolled in secondary schools (57.8% females; mean = 16.5 years; SD = 1.2) in the city of João Pessoa, Paraíba State, Brazil. The following physical activity correlates were measured: attitude, self-efficacy, social support from friends and parents, and perceived environmental characteristics. Physical activity was measured using a questionnaire. Multivariable ordinal logistic regression with proportional odds model analysis showed that the following factors are positively related to physical activity levels in adolescents: attitude, self-efficacy, as well as social support from parents and friends. Physical activity intervention programs should increase self-efficacy and social support from parents and friends, as well as a positive attitude toward physical activity. Motor Activity; Social Support; Adolescent Resumo Este estudo analisou a associação do nível de atividade física com fatores psicossociais e ambientais em adolescentes do Nordeste do Brasil. Estudo epidemiológico transversal desenvolvido com 2.859 adolescentes (57,8% do sexo feminino, média de idade: 16,5 anos; DP = 1,2) no Município de João Pessoa, Paraíba, Brasil. Os seguintes fatores associados à atividade física foram mensurados: atitude, autoeficácia, apoio social dos pais e amigos e características do ambiente percebido. O nível de atividade física foi mensurado por questionário. Os resultados da análise multivariável por regressão logística ordinal, com modelo de odds proporcionais, identificaram os seguintes fatores positivamente associados ao nível de atividade física dos adolescentes: atitude, percepção de autoeficácia, apoio social dos pais e amigos. Programas de promoção da atividade física devem desenvolver ações para aumentar a autoeficácia dos adolescentes e o apoio social fornecido pelos pais e amigos, bem como desenvolver e reforçar atitudes positivas em relação à atividade física. Atividade Motora; Apoio Social; Adolescente 941 ARTIGO ARTICLE http://dx.doi.org/10.1590/0102-311X00010813 Farias Júnior JC et al. 942 Cad. Saúde Pública, Rio de Janeiro, 30(5):941-951, mai, 2014 Introduction Regular physical activity engaged in by adolescents has beneficial effects on cardiorespiratory fitness, muscle strength, bone density, obesity, blood pressure, lipoprotein levels, metabolic syndrome, and mental health 1,2. Furthermore, physical activity behaviors are usually maintained from adolescence to adulthood, that is, adolescents who are currently inactive tend to retain these behaviors as they grow older 3. However, the prevalence of adolescents who do not perform enough physical activity to result in health benefits (for example, at least 60 minutes per day of moderate-to-vigorous physical activity 4) is unacceptably high 5 in a number of countries, including Brazil 6. Efforts to increase physical activity levels among young people have expanded 7,8. However, despite progress in evaluating the effectiveness of physical activity intervention programs, their effects are considered non-significant and much remains unknown about their long-term effectiveness 7,8. The identification of factors that may influence physical activity levels in adolescents, particularly those that could be potentially modified, has been considered an essential requirement in developing effective intervention programs 9,10,11. Physical activity in adolescents has been shown to be associated with factors such as demography, biology, psychology, behaviour, and socio-cultural and environmental factors, indicating complex and multidimensional behaviour 10,11,12. Thus, social-ecological models are considered the most suitable for studying correlates or determinants of physical activity, since they indicate that different factors (demographic, biological, psychological, behavioural, social and environmental) influence the physical activity of adolescents 11,13,14. The social-ecological model in the present study was conceptualised based on psychological (attitude and self-efficacy), social (social support from parents and friends), and environmental factors (characteristics of the neighbourhood where the adolescents lived) 14. Previous studies have demonstrated that attitude, self-efficacy, social support from parents and friends and environmental characteristics are important factors associated with physical activity in adolescents 10,11,15. Few studies have analysed factors from different levels of analyses that may influence physical activity in this population, especially those described previously. Moreover, most studies on the subject generally used poor methodology (the most common shortcomings were reduced age group with non-representative samples, lack of validity and reliable measures of both correlates or determinants and physical activity practice) 12, and were conducted with North-American or European adolescents 10,15,16, limiting extrapolation of the results to lowand middle-income countries, due to their particular socioeconomic, cultural and environmental characteristics 10. In the Lancet series on physical activity, Bauman et al. 10 emphasises that understanding the correlates and determinants of physical activity, especially in lowand middle-income countries, could reduce the effect of future inactivity epidemics and contribute to effective global prevention of non-communicable diseases. To address this research gap, this study examined the association of physical activity and psychosocial (attitude, self-efficacy, social support) and environmental characteristics (characteristics of the neighbourhood) factors in adolescents from Northeastern Brazil.

The aim of this study was to evaluate the association between levels of physical activity, psychosocial and perceived environmental factors in adolescents from Northeastern Brazil.A cross-sectional epidemiologic study was conducted with 2,859 adolescents enrolled in secondary schools (57.8% females; mean = 16.5 years; SD = 1.2) in the city of João Pessoa, Paraíba State, Brazil.The following physical activity correlates were measured: attitude, self-efficacy, social support from friends and parents, and perceived environmental characteristics.Physical activity was measured using a questionnaire.Multivariable ordinal logistic regression with proportional odds model analysis showed that the following factors are positively related to physical activity levels in adolescents: attitude, self-efficacy, as well as social support from parents and friends.Physical activity intervention programs should increase self-efficacy and social support from parents and friends, as well as a positive attitude toward physical activity.
Motor Activity; Social Support; Adolescent

Introduction
Regular physical activity engaged in by adolescents has beneficial effects on cardiorespiratory fitness, muscle strength, bone density, obesity, blood pressure, lipoprotein levels, metabolic syndrome, and mental health 1,2 .Furthermore, physical activity behaviors are usually maintained from adolescence to adulthood, that is, adolescents who are currently inactive tend to retain these behaviors as they grow older 3 .However, the prevalence of adolescents who do not perform enough physical activity to result in health benefits (for example, at least 60 minutes per day of moderate-to-vigorous physical activity 4 ) is unacceptably high 5 in a number of countries, including Brazil 6 .
Efforts to increase physical activity levels among young people have expanded 7,8 .However, despite progress in evaluating the effectiveness of physical activity intervention programs, their effects are considered non-significant and much remains unknown about their long-term effectiveness 7,8 .The identification of factors that may influence physical activity levels in adolescents, particularly those that could be potentially modified, has been considered an essential requirement in developing effective intervention programs 9,10,11 .
Physical activity in adolescents has been shown to be associated with factors such as demography, biology, psychology, behaviour, and socio-cultural and environmental factors, indicating complex and multidimensional behaviour 10,11,12 .Thus, social-ecological models are considered the most suitable for studying correlates or determinants of physical activity, since they indicate that different factors (demographic, biological, psychological, behavioural, social and environmental) influence the physical activity of adolescents 11,13,14 .The social-ecological model in the present study was conceptualised based on psychological (attitude and self-efficacy), social (social support from parents and friends), and environmental factors (characteristics of the neighbourhood where the adolescents lived) 14 .Previous studies have demonstrated that attitude, self-efficacy, social support from parents and friends and environmental characteristics are important factors associated with physical activity in adolescents 10,11,15 .Few studies have analysed factors from different levels of analyses that may influence physical activity in this population, especially those described previously.
Moreover, most studies on the subject generally used poor methodology (the most common shortcomings were reduced age group with non-representative samples, lack of validity and reliable measures of both correlates or determinants and physical activity practice) 12 , and were conducted with North-American or European adolescents 10,15,16 , limiting extrapolation of the results to low-and middle-income countries, due to their particular socioeconomic, cultural and environmental characteristics 10 .In the Lancet series on physical activity, Bauman et al. 10 emphasises that understanding the correlates and determinants of physical activity, especially in low-and middle-income countries, could reduce the effect of future inactivity epidemics and contribute to effective global prevention of non-communicable diseases.To address this research gap, this study examined the association of physical activity and psychosocial (attitude, self-efficacy, social support) and environmental characteristics (characteristics of the neighbourhood) factors in adolescents from Northeastern Brazil.

Design and sampling
This study is a cross-sectional epidemiological study, with a target population of both public and private secondary school students in the city of João Pessoa, capital of Paraíba State, in the Northeast of Brazil.João Pessoa has a hot, humid inter-tropical climate with a population of 723,515 inhabitants, 99.6% of whom are in urban areas, and medium human development index (HDI = 0.783).The study was approved by the Federal University of Paraíba Ethics Research Committee (n.0062/2009).All adolescents aged under 18 who enrolled in this study were authorised to do so by their parents or guardians.
The following parameters were considered for sample size calculation: estimated prevalence of 50% (≥ 300 minutes per week of physical activity 4 ); acceptable error of 3 percentage points; 95% confidence interval (95%CI); design effect = 2; 30% increase for possible losses and refusals, with a resulting sample of 2,686 students.Of the 3,220 adolescents answering the questionnaires, 361 were excluded (271 individuals aged < 14 or > 19; 65 who did not provide their age; 5 with some physical or mental limitation; and 25 who left several questions unanswered).The final sample comprised 2,859 adolescents aged 14 to 19.The sample had 80% power to detect as statistically significant odds ratios ≥ 1.40 for female adolescents and 1.50 for male adolescents 17 .
The sample was selected by conglomerates in two phases (1 st phase: schools; 2 nd phase: classes).In the first phase, secondary schools were stratified into four municipal regions (north, south, east and west) and ordered, from the lowest to the highest, by number of students, separately for type (public and private).Next, 30 schools were systematically selected (16 public and 14 private), proportionally distributed according to their size (measured as number of students) and the four municipal regions.In the second phase, we randomly selected 135 secondary school classes stratified by school year (1 st , 2 nd , 3 rd ) and school shift [day (morning + afternoon), night].This strategy produced a representative sample of high school students who were living in neighbourhoods with different physical (infrastructure) and socioeconomic characteristics.This is important for studies on factors associated with physical activity, especially those involving the environment (perceived characteristics of the neighbourhood where the adolescents lived).

Data collection and measures
Data collection took place between May and September 2009 by a trained team, composed of six Physical Education university students supervised by the head researcher.Questionnaires were filled out by the adolescents themselves in the classroom, according to instructions provided during the interviews.
The following data were collected: socio-demographic (gender, age, time spent living in the neighbourhood, economic class), physical activity and psychosocial and perceived environmental factors.The economic classes were based on Brazilian Economic Classification Criteria of the Brazilian Association of Research Companies (ABEP, acronym in Portuguese) 18 .These criteria include a series of questions related to household items such as number of bathrooms, owning a radio, television, washing machine, car, employing a full-time housekeeper, and parents' schooling, classifying people into the following economic classes: A (wealthiest), B, C, D and E (poorest).
Physical activity was measured using a previously tested questionnaire, adapted from a Self-Administered Physical Activity Checklist (test-retest reliability -ICC = 0.88; 95%CI: 0.84-0.91;concurrent validity -comparing the results of four 24-hour physical activity recalls: Spearman's rho = 0.62; p < 0.01; kappa = 0.59) 19 .The adolescents reported the frequency (days/weeks) and duration (minutes per day) of the physical activity engaged in for at least 10 minutes during the week previous to the collection, considering a list of 24 moderate to vigorous activities.The adolescents were instructed to consider the physical activities in different contexts: leisure, travel, and at school, considering only real-time activities.
The level of physical activity was calculated as the sum of time spent in each activity multiplied by its corresponding frequency, resulting in a score in minutes per week.This score was categorised into four levels (ordinal outcome) as follows: 0 minutes per week, 10-149 minutes per week, 150-299 minutes per week and ≥ 300 minutes per week.This categorisation was necessary for statistical analyses and allows separating adolescents into different levels of physical activity and not just into two groups: meets vs. does not meet physical activity recommendations.
The following psychosocial and perceived environmental factors were measured: psychosocial -attitude, self-efficacy, social support from parents and friends; perceived environmental factors -safe physical activities, accessibility and attractiveness of physical activity sites, infrastructure and general maintenance of the neighbourhood where the adolescents lived.The scales used to measure the psychosocial and perceived environmental factors are described in Table 1.Detailed information about the development and validation process of these scales are described in Farias Júnior et al. 20 .Scores for each of the psychosocial and perceived environmental factors were determined from the sum of points attributed to each answer.

Data analysis
The chi-square test was used to compare proportions and the independent t-test to compare means by gender.All continuous variables exhibited normal distribution (p < 0.05).Ordinal logistic regression with a proportional odds model was used to analyse the association between physical activity level (ordinal outcome categorised into four levels: 0 minutes per week = 0; 10-149 minutes per week = 1; 150-299 minutes per week = 2; ≥ 300 minutes per week = 3), psychosocial factors (attitude, self-efficacy, social support from parents and friends) and perceived environmental characteristics (access and attractiveness of the physical activity sites, safety, and neighbourhood structure).For the purpose of analysis, tertiles were calculated for each of the factors, with the first tertiles representing the highest values for each psychosocial and environmental factor.
Ordinal logistic regression produces an odds ratio that estimates the chance of an increase in one unit of the dependent variable for each unit rise in the independent variable.The Brant test 17 was used to verify the premise of proportional odds ratios for each of the variables, both individually and in the final models.All the independent variables were considered to construct the multiple model, excluding those with p-value Cad.Saúde Pública, Rio de Janeiro, 30(5):941-951, mai, 2014

Results
The final sample comprised 2,859 adolescents aged 14 to 19 (mean = 16.5 years; SD = 1.2; 57.8% female); 65.9% of males and 69.3% of females were between 14 and 16 years old, 53.2% be-longed to middle and low economic classes and approximately 50% were children of parents who completed, at least, secondary school (Table 2).
Approximately half of the adolescents fulfilled the recommended 300 minutes per week of physical activity (50.2%; 95%CI: 47.3-53.1).Males were more physically active than females (66.3 vs. 38.5%;p < 0.001; data not shown in tables).In relation to the psychosocial and perceived environmental factors, the females had lower average values for the safe physical activity score compared to the males (p < 0.001).Other factors showed no significant difference between sexes (Table 2).
Tables 3 and 4 present the results for crude and adjusted ordinal logistic regression analysis.In crude analysis, the following variables (Table 3) were directly associated to the level of physical activity in males: attitude, self-efficacy, social support from friends and parents, access and attractiveness of physical activity sites.The same variables were identified in females (Table 4), with the addition of variable structure and general maintenance of the neighbourhood.
In adjusted analysis (Table 3 and 4), we observed that adolescents with more positive attitudes towards physical activity, more self-efficacy and social support from friends were more likely to be physically more active than those with lower scores.The supposition for proportional odds ratio was met by all variables in the final crude and adjusted models, with p-value > 0.05 (Brant test).

Discussion
The results of the present study showed that adolescents with positive attitudes towards physical activity, higher values of perceived self-efficacy and social support from both parents and friends are more likely to be physically active.
Strengths of the study include: the simultaneous analysis of multiple factors (individual, socio-cultural and perceived environmental factors), the inclusion of the main confounders in all analyses, the representative sample of adolescents in terms of sample size and selection, the satisfactory reliability and validity of the instruments used and their application by a trained team that followed the previously established protocol.
There was a positive association between attitude and physical activity level.In fact, as the attitude score increased, the chance of the a dolescents' reaching a category of higher physical activity level also rose for both males and females.A systematic review demonstrated that adolescents with more positive attitudes towards physical activity were physically more active 15 .Haerens et al. 21observed a positive and significant relationship between attitude score and leisure physical activity levels in females, but not males, and Nelson et al. 22 found a significant association between negative attitudes and physical activity.However, Lee et al. 23 found no significant association between adolescents' attitude and physical activity level.This may be partially due to the oversimplified conceptualisation and measurement of attitudes and the tendency to focus only on the role of positive attitudes 22 .It is also important to consider that the influence of attitude on the physical activity levels of adolescents seems to vary with the type of attitude indicator (eg, instrumental, affective) and physical activity measured (global physical activity, structured physical activity, and active transport).Moreover, this association may be moderated by physical activity level and other variables such as sex and age.
There was a positive association between self-efficacy and physical activity level in adolescents.For instance, as the self-efficacy score increased, the likelihood of adolescents' reaching a category of higher physical activity level also rose.These results are similar to those reported in other studies 23,24,25,26 .The consistency of the relationship between physical activity and self-efficacy in a number of studies with adolescents suggests that the latter is an important predictor of physical activity level.Self-efficacy may influence motivational and self-regulatory processes, such as the type, magnitude and persistence of behaviour when facing obstacles 27 .Lubans et al. 28 found that self-efficacy was the most commonly assessed mediator and received the strongest support for mediating the relationship between theory-based interventions and physical activity in young people.Dishman et al. 29 observed that females who exhibited greater perception of social support during high school had less decline in physical activity, but only if they also had high self-efficacy.Furthermore, self-efficacy is a mediator of the relationship between the influence of social support and physical activity level in adolescents 30 .Adolescents with higher levels of social support from parents and friends tend to report higher perception of self-efficacy for overcoming barriers to physical activity.This construct has been shown to be associated with more adolescent participation in physical activity 30 .
There was a positive and significant relationship between social support from friends and parents and physical activity levels.Adolescents that received more social support from their parents were about 50% more likely to reach higher levels of physical activity than those receiving less social support.For social support from friends, the chance of an adolescent being physically more active was almost three times higher among those reporting more social support.These results are similar to those obtained by other studies, showing that adolescents who had more social support from their parents 23,25 and/or friends 24,23,30,31 were physically more active.However, some studies identified no significant relationship between social support from parents 32 and/or friends 21,32 and physical activity level.Social support for physical activity can come in different forms, such as social integration or companions for physical activity; psychological or emotional support, provision of information on physical activity; and transport to the physical activity site 33 .Previous studies found that social support from parents and friends was inversely related to declines in physical activity in adolescents 29,33 .
In relation to the perception of environmental characteristics, the access and attractiveness of the physical activity site score was directly associated to the physical activity levels of adoles-cents, in crude analysis, but was less significant (p > 0.05) in adjusted analysis.Other authors also found similar results to those obtained here 21,23 .This variable significance loss occurred after adjusting for self-efficacy, suggesting that possible relationships between access and attractiveness of physical activity sites and physical activity levels may be mediated by self-efficacy.This might be a plausible explanation for studies that did not identify significant relationships between these variable 34,35 .Motl et al. 36 observed that relationships between sites and equipment used for physical activity and the physical activity levels of adolescents were totally mediated by self-efficacy.These results were similar to those obtained in other research using cross-sectional 37 and longitudinal data analysis 38 .It is also important to underscore that the relationship between environment characteristics and physical activity may be moderated by perceptions of self-efficacy.Deforche et al. 9 observed that self-efficacy appeared to moderate the link between active transportation and certain perceived physical environmental factors.Lower perceived safety and poorer access to recreational facilities was only associated with lower active transport in adolescents with lower self-efficacy (p < 0.05).
The absence of a direct relationship between environmental characteristics and physical activity levels in the present investigation was also described in other studies 23,34,35 .However, some research has demonstrated that adolescents living in neighbourhoods with sites and equipment available for physical activity were physically more active 39,40 .Systematic reviews have demonstrated an inconsistent relationship between physical activity level and environment characteristics in adolescents 10,15,16 .Most recently, Ding et al. 13 summarised findings from studies on the relationship between neighbourhood environment and physical activity in young people.Objectively measured, environmental attributes were much more consistently related to physical activity.There is a correlation between physical activity and residential density as well as land use mix; however, there is no correlation between physical activity and proximity to parks, recreational facilities and street connectivity.This may be explained by more accurate environment objective measures 13 .Another aspect to consider is that environmental characteristics may differ in terms of degree of influence, depending on the outcome or measure of physical activity.Mota et al. 41 showed that there are differences in the extent to which environmental characteristics can influence the participation of adolescents in structured and non-structured physical activities.
Cad. Saúde Pública, Rio de Janeiro, 30(5):941-951, mai, 2014 The present study has the following limitations: since this was a cross-sectional study, it is not possible to establish cause-effect relationships.Measuring levels of physical activity using a questionnaire is always a limitation because of recall bias, reliability and validity of those being questioned.However, the reliability and validity (despite the limitation of the method of reference used in the present study) of the instrument used in the present study were satisfactory.Finally, perceptions of the environment were measured rather than actual characteristics; however, the measures were reliable and valid for the sample.
The present study shows that psychosocial factors such as attitude, self-efficacy and social support from parents and friends were positively associated with physical activity.These findings have very important implications for phys-ical activity promotion programmes.Craggs et al. 11 showed that higher levels of self-effıcacy and social support were associated with smaller declines in physical activity compared to lower levels of self-effıcacy and support for physical activity, indicating that these factors are also very important for maintaining physical activity levels during adolescence.Recently, priorities have been established to promote physical activity in Brazil, including the need for identifying physical activity correlates across different age groups 42 .The results of this study suggest that self-efficacy and social support from parents and friends are important components of strategies aimed at promoting physical activity in this group.Moreover, these strategies could produce better results if reinforcement of positive attitudes towards physical activity is encouraged.

Table 1
Description of scales used to measure psychosocial and perceived environmental factors related to the practice of physical activity in adolescents.
α: Cronbach's alpha; ICC: intra-class correlation coefficient.* To answer these questions, the adolescents were oriented to consider a distance up to 10 minutes from their homes.

Table 2
Socio-demographic and psychosocial and perceived environmental characteristics related to the practice of physical activity in adolescents.

Table 3
Ordinal logistic regression for psychosocial and perceived environmental factors and levels of physical activity in male adolescents.

Table 4
Ordinal logistic regression for psychosocial and perceived environmental factors and levels of physical activity in female adolescents.