- Email: [email protected]
Correlates of Preschool Children's Physical Activity
Correlates of Preschool Children’s Physical Activity Trina Hinkley, PhD, Jo Salmon, PhD, Anthony D. Okely, EdD, Kylie Hesketh, PhD, David Crawford, PhD Background: Physical activity is important for children’s health, and identifying factors associated with their physical activity is important for future interventions and public health programs.
Purpose: This study sought to identify multidimensional correlates of preschool children’s physical activity.
Methods: The social– ecological model (SEM) was used to identify constructs potentially associated with preschool children’s physical activity. Data were collected from 1004 preschool children, aged 3–5 years, and parents in 2008 –2009, and analyzed in 2010 –2011. Physical activity was measured over 8 days using ActiGraph accelerometers. Parents completed a comprehensive survey. Generalized linear modeling was used to assess associations between potential correlates and percentage of time spent in physical activity.
Results: Correlates of physical activity were found across all the domains of the SEM and varied between boys and girls and week and weekend days. Age was the only consistent correlate, with children spending approximately 10% less time in physical activity for each advancing year of age. Some modifıable correlates that were related to more than one physical activity outcome were rules restricting rough games inside and usual daily sleep time for boys. For girls, a preference to play inside/draw/do crafts rather than be active, and child constraints, was associated with more than one of the physical activity outcomes. A novel fınding in this study is the counterintuitive association between parental rules restricting rough games inside and boys’ higher physical activity participation levels. Conclusions: Potential strategies for promoting children’s physical activity should seek to influence children’s preference for physical activity and parent rules. Gender-specifıc strategies also may be warranted. (Am J Prev Med 2012;43(2):159 –167) © 2012 American Journal of Preventive Medicine
Introduction
P
hysical activity is an important contributor to health in adults1 and early childhood.2,3 Health behaviors such as physical activity track from early childhood into later childhood and adolescence.4,5 Establishing and maintaining adequate physical activity levels in young children is an important public health issue. From the Centre for Physical Activity and Nutrition Research (Hinkley, Salmon, Hesketh, Crawford), Faculty of Health, Deakin University, Burwood, Victoria; and the Interdisciplinary Educational Research Institute (Okely), Faculty of Education, University of Wollongong, Wollongong, New South Wales, Australia Trina Hinkley was on the Faculty of Education, Interdisciplinary Educational Research Institute, University of Wollongong, Wollongong, New South Wales, Australia, at the time some of the analyses and write-up of this paper occurred. Address correspondence to: Trina Hinkley, PhD, Center for Physical Activity and Nutrition Research, Deakin University, 221 Burwood Hwy, Burwood, Victoria 3125, Australia. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2012.04.020
Recent research shows that preschool children spend only a small proportion of their time being active,6,7 and compliance with physical activity recommendations is low.8,9 It is therefore important to identify the factors that may support or constrain physical activity. The social– ecologic model (SEM)10,11 provides a framework through which potential correlates of preschool children’s physical activity may be conceptualized. The SEM posits that the correlates of physical activity are multidimensional and operate across individual, social, and physical environment domains. Research to date has explored a range of potential correlates across those domains, and variables in each have been associated with preschool children’s physical activity.12 However, studies primarily have investigated only a narrow range of correlates, generally in only one domain, and there is little insight into the impact of the broader contexts in which those correlates exist. Further, as the majority of studies have used relatively small samples, it has not been possible to investigate
© 2012 American Journal of Preventive Medicine • Published by Elsevier Inc.
Am J Prev Med 2012;43(2):159 –167 159
160
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
differences in correlates between the genders, despite established knowledge that boys are more active than girls12 and correlates differ between school-aged boys and girls.13,14 Research15,16 also has shown that physical activity may vary between week and weekend days. However, the published literature has not reported possible differences in potential correlates between week and weekend days when differences in preschoolers’ environments may exist. The aim of the current study was to investigate potential correlates of preschoolers’ physical activity across all levels of the SEM, for boys and girls separately, and to investigate potential differences between correlates for weekday and weekend physical activity.
Methods Recruitment and Participants Methods of the current study have been published previously.9 Data were drawn from the baseline period of the Healthy Active Preschool Years Study when children were aged 3–5 years. Six local government areas (LGAs) in the Melbourne metropolitan area were selected randomly, two each from the lowest, middle, and highest SES quintiles as identifıed from the 2001 Socio-economic Indexes for Areas17 Index of Relative Socio-economic Advantage and Disadvantage. The fırst randomly ordered eight child care centers and eight preschools were selected in each of the two lowest-SES quintile LGAs, and the fırst fıve of each in each of the four LGAs in the middle and highest-SES quintiles. As participants from low-SES areas are often under-represented in research,18,19 they were oversampled in the present study. If a center declined to participate, the next center on the randomly ordered respective LGA preschool or child care center list was approached. Recruitment and data collection occurred in two phases: July to November 2008; and May to October 2009. In total, 156 child care centers and 137 preschools were approached. The fınal sample consisted of 71 (46%) child care centers and 65 (47%) preschools. Parents (9794) of children aged 3–5 years at each participating center were invited to participate. In total, 1036 parents provided written, informed consent (11%); four children were aged ⬎5 years and 28 withdrew before data collection, leaving a fınal sample of 1004. Ethical approval was provided by the Deakin University Human Research Ethics Committee and the Victorian Department of Education and Early Childhood Development.
Measures and Data Management Accelerometry. An ActiGraph GT1M accelerometer on an elastic belt was fıtted to each child. Accelerometers record date– time stamped information regarding the magnitude of movement, and are thus able to measure the frequency, intensity, and duration of physical activity. ActiGraph has demonstrated validity, reliability and utility in preschool children.20 Participants were instructed to wear accelerometers during waking hours at the right iliac crest for an 8-day period. Data were collected in 15-second epochs.20,21 Monitoring start times were identifıed as the beginning of the fourth complete minute of the appearance of counts above zero after 4:00AM, with a tolerance of four epochs (1 minute) of zero counts. Nonwear time was determined as ⱖ10 minutes of consec-
utive zero counts. Given the sporadic and constant nature of young children’s physical activity, it would be diffıcult for a child this age to be motionless for this long.21 Days with ⱖ18 hours of recorded data were excluded as being improbable. Age-specifıc cut points21 were applied to the data to distinguish the various intensities. Physical activity was operationalized as total physical activity (light, moderate, and vigorous intensities combined), consistent with recent recommendations22 and studies.8,9 The current study included three primary outcome variables: weekly, weekday, and weekend-day physical activity. The proportion of time (possible range 0 –1) the child spent in total physical activity on any given day was calculated to account for differences in wear time between children and days (reported as percentage for ease of interpretation and understanding). Any given day was considered valid if ⱖ7 hours of data were recorded. Average daily physical activity across the week included data from at least 3 valid weekdays and 1 valid weekend day (referred to as weekly physical activity). Weekday physical activity included data from at least 3 valid weekdays. Weekend-day physical activity included data from at least 2 valid weekend days. Reliability analyses, using repeated measures ANOVA, intraclass correlation (ICC), and Spearman-Brown prophecy formula, showed that this amount of weekly and weekday data achieved a reliability of 0.7 (2.9 weekdays achieved ICC⫽0.7; 3.9 days [weekday and weekend days combined] achieved ICC⫽0.7). Three weekend days would have been required to achieve reliability at this level for weekend days (2.9 weekend days achieved ICC⫽0.7). As only three children in the sample had 3 weekend days with suffıcient data, 2 weekend days were used with reliability of 0.61. Time of year was determined as the month during which the accelerometer commenced recording.
Parent survey. Parents completed a survey covering all domains of the SEM10,11 during the week their child wore the accelerometer. Items for inclusion in the survey were identifıed from a comprehensive review of the literature12 and focus groups with mothers of preschool children.23 Individual-level correlates included biological/demographic variables (e.g., parent’s age, gender, country of birth, education; child age, gender, disability, siblings, and night and day sleep time); child behavioral variables (e.g., participation in active transport, nonorganized or organized activities, and screen-based entertainment); and psychological variables (e.g., child requests for physical activity, child constraints to physical activity, and child preferences for physical activity). Social-level correlates included parental correlates (e.g., parental concerns, preferences, constraints, rules and regulations) and broader social correlates (e.g., social gatherings, provision of logistic and emotional support, role-modeling, and physical activity interaction). Environment-level correlates included variables in the home (e.g., yard size, indoor play spaces, availability of sports and electronic equipment, and number of TVs) and neighborhood correlates (e.g., availability and suitability of playgrounds, constraints to active transport, and frequency of visits to active play spaces). Items were considered reliable and included in the analyses if they met the following criteria: for test–retest reliability, categoric or dichotomized items had Kappa ⱖ0.6 or percentage agreement ⱖ60% and continuous items and scales had intraclass correlation ⱖ0.50; for internal reliability, items in a scale had a Cronbach’s alpha ⱖ0.7. A separate sample of 47 parents completed the survey on two occasions 14 –50 days apart (mean 24 days).24 Appendix A (available online at www.ajpmonline.org) provides www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167 details of constructs and items included in the HAPPY Study parent survey with their response options.
Anthropometrics. Child height and weight were measured by a trained researcher using a Wedderburn Seca portable rigid stadiometer, Wedderburn Tanita portable digital scales, and standardized measurement procedures.25,26 Parents self-reported their and their partner’s (where applicable) height and weight. BMI was calculated by standard formula and was determined for children and parents. Child and parent BMI and child weight status (BMI adjusted for age and gender27) were included in analyses as potential correlates.
Data Analysis Descriptive statistics were used to describe characteristics of the sample and the child’s physical activity. T-tests were used to assess differences in physical activity between boys and girls and between weekdays and weekend days. Generalized linear modeling (GLM) with a binomial family and logit transformation28,29 was used to examine associations between potential correlates and outcome variables. Each explanatory variable was entered individually into a GLM analysis with each of the physical activity outcome variables for boys and girls separately. The resultant ORs were interpreted as a percentage increase or decrease in the percentage of time spent in the physical activity outcome variable being examined relative to each unit increase in the explanatory variable or compared to the referent category of the explanatory variable.29,30 For each of the three outcome variables, the explanatory variables that showed an association (pⱕ0.01) in bivariable analyses were tested for collinearity by estimating the individual variance inflation factor (VIF) and tolerance for each item, and the mean VIF. Items with a VIF ⬎10 and tolerance less than 0.1 were removed.31,32 Where the mean VIF was signifıcantly ⬎133 (operationalized as ⱖ2) but no variables had an individual VIF ⬎10, the variable with the highest VIF was removed (variables identifıed where applicable in Tables 1 and 2). All remaining explanatory variables for each of the outcome variables were entered simultaneously into multivariable GLM analyses to determine the signifıcance (pⱕ0.05) of each of the variables when in combination with other signifıcant correlates. Model fıt statistics (Akaike information criterion [AIC]) and model specifıcation (data not reported)29 were examined to ensure each model was appropriately specifıed. AIC values ⬍1 indicate an acceptable model.29 Clustering by center (preschool/child care) of recruitment was controlled for in all analyses because of children being nested within centers. Analyses were undertaken in 2010 and 2011.
Results Data were collected from 1004 children and their parents. In total, 943 parents completed surveys, of which 93.7% were female respondents. Respondent parents had an average age of 37.3 years (95% CI⫽36.9, 37.6) and 69.8% were born in Australia. More than half of mothers (55%) reported tertiary education level; 34% had completed Year-12, trade, apprenticeship, or diploma as their highest educational qualifıcation; and 11% reported Year-10 or equivalent education. Children (n⫽939) had a mean age of 4.5 years (95% CI⫽4.5, 4.6). The fınal sample included 705 children who met the criteria for suffıcient August 2012
161
weekly accelerometry data (55% boys; mean wear time 655.0 [SD⫽64.3, range⫽478.4 –1019.7] minutes/day); 773 with suffıcient weekday accelerometry data (54% boys; mean wear time 649.5 [SD⫽64.3, range 440.9 – 1019.0] minutes/day); and 605 with suffıcient weekendday accelerometry data (56% boys; mean wear time 666.1 [SD⫽89.9, range 456.5–995.6] minutes/day). Those with suffıcient accelerometry data did not vary from the total sample by gender, maternal education, or child weight status; except boys were slightly more likely to have suffıcient weekend accelerometry data than girls (p⫽0.049) and children of mothers with university education were more likely to have suffıcient weekend accelerometry data than children of mothers with Year 12 or equivalent education (p⫽0.030). However, older children were more likely to have suffıcient accelerometry data for all three outcome variables (p⬍0.05 for all) than younger children. Possibly, this is a consequence of younger children sleeping more than older children (p⬍0.001). Boys spent a signifıcantly higher mean percentage of their daily time across the entire week (17.3% vs 15.4%); on weekdays (17.0% vs 15.3%); and on weekend days (18.2% vs 15.8%) in physical activity than did girls (p⬍0.001 for all). Boys and girls spent a greater mean percentage of their daily time in physical activity on weekend days than on weekdays (boys: 18.2% vs 16.7%, p⬍0.001; girls: 15.8% vs 15.1%, p⫽0.004). Results for associations between correlates across all domains of the SEM and boys’ mean percentage of daily time across the entire week, on weekdays and weekend days in physical activity are presented in Table 1. The only consistent correlate across all three outcome variables was age; for each additional year of age, boys spent 11%, 13%, and 10% less time in weekly, weekday, and weekend-day physical activity, respectively. Other variables associated with a lower percentage of time in physical activity included weekly computer-use time, parent housework, parent preference for their child to do the same activities as older siblings, and owning a desktop computer. Correlates associated with an increased percentage of time in physical activity included time spent outside on weekends, child’s total daily sleep time, child being active on his own, rules restricting rough games inside, the frequency the child’s mother and other children were active with the preschool child, and the number of weekly visits to active play spaces. Table 2 presents the results of the multivariable model analyses of correlates of mean percentage of daily time across the entire week, on weekdays, and weekend days in physical activity for girls. Age was consistently associated with all three physical activity outcome variables; for each additional year of age, girls spent 11%, 12%, and 15% less time in weekly, weekday, and weekend-day physical
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
162
Table 1. Multidimensional correlates of boys’ weekly, weekday, and weekend-day physical activity,a OR (95% CI) Multivariable associations Weekly physical activity (n⫽366, AIC⫽0.69)
Weekday physical activity (n⫽399, AIC⫽0.68)
Weekend-day physical activity (n⫽298, AIC⫽0.73)
0.89 (0.85, 0.92)*
0.87 (0.83, 0.91)*
0.90 (0.85, 0.96)*
1.04 (1.02, 1.07)*
1.04 (1.01, 1.06)*
—d
—d
—d
0.98 (0.96, 1.01)
—d
0.98 (0.96, 1.00)*
—d
Do not agree (ref)
—d
—d
1.00
Agree
—
—
1.13 (1.01, 1.27)*
Do not agree (ref)
—d
1.00
—d
Agree
—
0.91 (0.85, 0.97)*
—
1.00
1.00
—d
0.94 (0.88, 0.99)*
0.92 (0.87, 0.97)*
—
1.00
—d
1.00
1.06 (1.01, 1.12)*
—
1.11 (1.02, 1.20)*
1.01 (1.00, 1.03)*
—d
—d
—d
—d
1.04 (1.01, 1.07)*
Never (ref)
—d
—d
1.00
Once a month or less
—
—
—
Once every 2 weeks
—
—
1.05 (0.91, 1.21)
Once a week
—
—
0.99 (0.85, 1.14)
Two or more times a week
—
—
1.03 (0.89, 1.20)
Correlate INDIVIDUAL Biological and demographic Child age (years)b b
Total daily sleep (hours)
Number of cars in the homeb Behavioral Weekly time using computer (hours)b Psychological Preschool child is active by himselfc
SOCIAL Parental Time parent spends doing housework stops him/her from supporting preschool child to be activec
Parent prefers child to do same activities as older childrenc Do not agree/not applicable (ref) Agree Preschool child is not allowed to play rough games inside the housec Do not agree (ref) Agree Other social Frequency mother is active with preschool childb Frequency other children are active with preschool childb Frequency of attendance at social gatheringsc
(continued on next page)
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
163
Table 1. (continued) Multivariable associations
Correlate
Weekly physical activity (n⫽366, AIC⫽0.69)
Weekday physical activity (n⫽399, AIC⫽0.68)
Weekend-day physical activity (n⫽298, AIC⫽0.73)
—d
—d
1.00
Frequency of no one being active at social gatheringsc Not often (ref) Often
0.92 (0.83, 1.02)
PHYSICAL ENVIRONMENT Home and neighborhood —d
—d
1.01 (1.00, 1.02)*
1.01 (1.00, 1.03)*
—d
1.01 (1.00, 1.02)*
1.01 (1.00, 1.02)*
—d
Do not agree (ref)
—d
—d
1.00
Agree
—
—
0.86 (0.78, 0.95)*
Frequency of visits to shopping centers (per week)b
1.01 (0.98, 1.04)
Total frequency of visits to active play spaces (per week)b Time outside on weekends (hours)b Have desktop computer in homec
a
Generalized linear modeling analyses were adjusted for clustering by center of recruitment. OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for every one-unit increase in the correlate. c OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for the comparison group(s) compared with the referent group. d Variable was not associated with physical activity outcome in specific column. *p⬍0.05 AIC, Akaike information criterion b
activity, respectively. Girls’ preference to play inside/ draw/do crafts, child constraints, maternal work, frequency of visits to active play spaces, and lack of footpaths were associated with a lower percentage of time in physical activity for girls. The number of siblings, paternal provision of logistic support, and paternal time in moderate physical activity were associated with girls spending a higher percentage of time being active.
Discussion It was found that 20 correlates were associated with at least one outcome variable for either boys or girls: eight individual, seven social, and fıve physical environment. For boys, correlates across all domains were associated with all the physical activity outcomes. For girls, individual correlates were associated with all the outcome variables, and social and physical environment correlates were associated with weekly and weekend-day physical activity. The present study identifıes a number of previously unreported correlates of preschoolers’ physical activity. These include sleep, maternal work status, child being active on his/her own, parent preference for child to do the same activities as older siblings, child’s preference to August 2012
play inside/draw/do crafts rather than be active, child constraints, and lack of footpaths. Although each of the domains was represented in most of the multivariable models, there was little consistency between correlates for boys and girls, or between weekdays and weekend days. A novel fınding in the current study is the counterintuitive inverse association between parental restriction of rough play/games inside the home and higher levels of physical activity in boys. It is possible that parents who have such rules do so because their boys are intrinsically more active, thus necessitating the imposition of the rule. Alternatively, parents with such rules may make additional efforts to support their son being active by taking him outside more frequently or to active play spaces. However, such suggestions are conjecture and this association requires further investigation. Two of the most commonly studied correlates are time spent outside and watching TV.12 Time outside is commonly found to be positively associated with physical activity in preschool12 children. In the present study, boys’ time outside on weekend days was associated with their weekly and weekend-day physical activity, with each additional hour outside representing an additional 1% more time being
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
164
Table 2. Multidimensional correlates of girls’ weekly, weekday, and weekend-day physical activity,a OR (95% CI) Multivariable associations
Correlate
Weekly physical activity (n⫽262; AIC⫽0.69)
Weekday physical activity (n⫽258; AIC⫽0.68)
Weekend-day physical activity (n⫽196; AIC⫽0.83)
0.89 (0.84, 0.93)*
0.88 (0.84, 0.93)*
0.85 (0.79, 0.91)*
INDIVIDUAL Biological and demographic Child’s age (years)b c
Maternal work status
Home duties full-time (ref)
—d
1.00
1.00
Working part-time
—
0.92 (0.86, 0.97)*
0.99 (0.90, 1.10)
—
0.93 (0.83, 1.05)
0.84 (0.74, 0.97)*
Married/partnered (ref)
—e
—e
—d
Not married/partnered
—
—
—
Working full-time c
Parent’s marital status
b
Total number of siblings
1.06 (1.01, 1.11)*
d
—
1.02 (0.96, 1.09)
Psychological Preschool child is more likely to play inside/draw/do crafts than be activec Do not agree (ref) Agree Child constraints (e.g., lack of time, energy)b
1.00
1.00
1.00
0.90 (0.82, 0.99)*
0.89 (0.81, 0.97)*
0.88 (0.77, 1.00)*
0.99 (0.99, 1.00)*
0.99 (0.98, 1.00)
0.99 (0.98, 1.00)*
SOCIAL Other social Frequency of attendance at social gatheringsc Never (ref)
—d
—d
1.00
Once a month or less
—
—
—
Once every 2 weeks
—
—
0.94 (0.78, 1.14)
Once a week
—
—
0.94 (0.77, 1.13)
Two or more times a week
—
—
1.04 (0.87, 1.24)
d
Paternal provision of logistic supportb
1.01 (0.99, 1.03)
—
1.03 (1.00, 1.05)*
Maternal time in vigorous physical activity per week (hours)b
1.01 (0.99, 1.02)
1.00 (0.99, 1.02)
—d
—d
—d
—e
Paternal time in moderate physical activity per week (hours)b
1.01* (1.00, 1.02)
1.01 (1.00, 1.02)
1.01 (1.00, 1.02)
Frequency child sees mother being physically activeb
—e
1.00 (0.98, 1.03)
—d
Frequency child sees father being physically activeb
1.01 (0.99, 1.03)
1.01 (0.98, 1.03)
1.02 (1.00, 1.05)
Paternal time in total physical activity per week (hours)b
(continued on next page)
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
165
Table 2. (continued) Multivariable associations Weekday physical activity (n⫽258; AIC⫽0.68)
Weekend-day physical activity (n⫽196; AIC⫽0.83)
1.02 (0.99, 1.05)
1.01 (0.98, 1.04)
1.01 (0.98, 1.04)
Frequency child sees parents being physically activeb
—e
—e
—d
Frequency of proximal active role modelsb
—e
—e
—e
Frequency of all visible active role modelsb
—e
—e
—e
—d
—d
0.98 (0.96, 0.99)*
0.81 (0.71, 0.93)*
—d
—d
1.00
—d
—d
0.81 (0.71, 0.93)*
—
—
Correlate Frequency child sees other adults being physically activeb
Weekly physical activity (n⫽262; AIC⫽0.69)
Physical environment Total frequency of visits to active play spaces per weekb There are no footpaths in neighborhood for child and parents to usec Do not agree (ref) Agree Time outside on weekend days (hours)b
d
d
—
—
1.00 (0.98, 1.02)
June (ref)
—d
—d
1.00
July
—
—
1.02 (0.87, 1.20)
August
—
—
1.29 (1.07, 1.57)*
September
—
—
1.01 (0.86, 1.18)
October
—
—
1.13 (0.97, 1.32)
November
—
—
0.99 (0.86, 1.14)
December
—
—
1.13 (0.88, 1.45)
Month physical activity data measured
a
Generalized linear modeling analyses was adjusted for clustering by center of recruitment. OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for every one-unit increase in the correlate. c OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for the comparison group(s) compared with the referent group. d Variable was not associated with physical activity outcome in specific column. e Variable was not entered in multivariable analyses because of collinearity. *p⬍0.05 AIC, Akaike information criterion b
active. Such small differences may not be biologically meaningful. In bivariable analyses, time outside on weekend days was associated with boys’ weekend physical activity and with all of the girls’ physical activity outcomes (p⬍0.05). However, time outside on weekdays was not associated with any of the physical activity outcomes even in bivariable analyses. It may be that parents are not the best source for reporting children’s time outside on weekdays, when children may be cared for by others. August 2012
Prior research34 has found that TV-viewing time is associated with preschool children’s physical activity; however, the current study and others15,35 have found no such association. This may suggest that the notion of TV viewing displacing physical activity is not well supported, particularly at this young age, and that targeting a reduction in TV viewing as a means to increase physical activity may not be a relevant strategy. Nonetheless, targeting reductions in TV viewing
166
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
and other screen-related behaviors is warranted for health and developmental reasons.36 –38 Most notably, the current study shows that the correlates of physical activity vary between boys and girls. When viewed collectively with previous research,13,14,39 these fındings suggest that future studies and interventions may need to recruit a suffıciently large sample to ensure adequate power to stratify analyses by gender. Future interventions also should consider genderspecifıc strategies to address such differences and increase their chances of successful outcomes. The present study examined a large number and broad range of potential correlates, across all levels of the SEM, although some potentially important correlates may have not been included. Further, parents may not be the best source of data collection for items such as time outside if their child is being cared for by others. Although an objective, reliable, and validated measure of physical activity was used, accelerometers are unable to detect certain activities and may therefore slightly underestimate total physical activity in some circumstances. Further, lack of agreement exists as to the most appropriate cut points to use when interpreting data. As the cut points used have a high threshold between sedentary and light physical activity, they may underestimate the time spent in light (and in this study therefore total) physical activity. The low reliability level for weekend-day data is also a limitation of the current study. However, this study used a large, heterogeneous sample and was able to investigate correlates for boys and girls, and weekdays and weekend days, separately. Finally, the instrument used to measure correlates of physical activity has established reliability.
Conclusion Future studies should investigate correlates separately for boys and girls as such correlates vary between the genders. Intervention studies and public health campaigns should ensure strategies address the multidimensional nature of physical activity correlates in preschool children, and include gender-specifıc strategies to increase the likelihood of success. Interventions may need to be delivered during the fırst 2–3 years of a child’s life to maximize their physical activity levels, as well as during the preschool period to stem declines in physical activity. TH was supported by a Deakin University APA PhD Scholarship during the fırst half of data collection. JS is supported by a National Heart Foundation of Australia Career Development Award and Sanofı-Aventis. KH is supported by a National Heart Foundation of Australia Career Development Award. DC is supported by a Victorian Health Promotion Foundation Senior Research Fellowship.
The project was funded by Deakin University. No fınancial disclosures were reported by the authors of this paper.
References 1. DHHS U. Physical activity and health: a report of the Surgeon General. Atlanta GA: DHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, 1996. 2. Jiménez-Pavón D, Kelly J, Reilly JJ. Associations between objectively measured habitual physical activity and adiposity in children and adolescents: systematic review. Int J Pediatr Obes 2010;5(1):3–18. 3. Janz KF, Burns TL, Torner JC, et al. Physical activity and bone measures in young children: the Iowa Bone Development Study. Pediatrics 2001;107(6):1387. 4. Kelly LA, Reilly JJ, Jackson DM, Montgomery C, Grant S, Paton JY. Tracking physical activity and sedentary behavior in young children. Pediatr Exerc Sci 2007;19(1):51– 60. 5. Taylor RW, Murdoch L, Carter P, Gerrard DF, Williams SM, Taylor BJ. Longitudinal study of physical activity and inactivity in preschoolers: the FLAME study. Med Sci Sports Exerc 2009;41(1):96 –102. 6. Oliver M, Schofıeld G, Kolt G. Physical activity in preschoolers: understanding prevalence and measurement issues. Sports Med 2007; 37(12):1045–70. 7. Tucker P. The physical activity levels of preschool-aged children: a systematic review. Early Childhood Res Q 2008;23(4):547–58. 8. Beets MW, Bornstein D, Dowda M, Pate RR. Compliance with national guidelines for physical activity in U.S. preschoolers: measurement and interpretation. Pediatrics 2011;127(4):658 – 64. 9. Hinkley T, Salmon J, Okely AD, Crawford DA, Hesketh KD. Preschoolers’ physical activity, screen time and compliance with recommendations. Med Sci Sports Exerc 2012;44(3):458 – 65. 10. Davison KK, Birch LL. Childhood overweight: a contextual model and recommendations for future research. Obes Rev 2001;2(3):159 –71. 11. Bronfenbrenner U. The ecology of human development: experiments by nature and design. Cambridge: Harvard University Press, 1979. 12. Hinkley T, Crawford D, Salmon J, Okely AD, Hesketh K. Preschool children and physical activity: a review of correlates. Am J Prev Med 2008;34(5):435– 41. 13. Patnode CD, Lytle LA, Erickson DJ, Sirard JR, Barr-Anderson D, Story M. The relative influence of demographic, individual, social, and environmental factors on physical activity among boys and girls. Int J Behav Nutr Phys Act 2010;7:79. 14. Trost SG, Pate RR, Ward DS, Saunders R, Riner W. Correlates of objectively measured physical activity in preadolescent youth. Am J Prev Med 1999;17(2):120 – 6. 15. Burdette HL, Whitaker RC. A national study of neighborhood safety, outdoor play, television viewing, and obesity in preschool children. Pediatrics 2005;116(3):657– 62. 16. Cardon G, De Bourdeaudhuij I. Are preschool children active enough? Objectively measured physical activity levels. Res Q Exerc Sport 2008;79(3):326 –32. 17. Australian Bureau of Statistics. Census of population and housing: socio-economic indexes for areas (SEIFA) 2033.8.55.001. Canberra: Australian Bureau of Statistics, 2001. 18. Telford A, Salmon J, Timperio A, Crawford D. Examining physical activity among 5- to 6- and 10- to 12-year-old children: the Children’s Leisure Activities Study. Pediatr Exerc Sci 2005;17(3):266 – 80. 19. Madigan MP, Troisi R, Potischman N, et al. Characteristics of respondents and non-respondents from a case-control study of breast cancer in younger women. Int J Epidemiol 2000;29(5):793– 8. 20. Cliff DP, Reilly JJ, Okely AD. Methodological considerations in using accelerometers to assess habitual physical activity in children aged 0-5 years. J Sci Med Sport 2009;12(5):557– 67.
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167 21. Sirard JR, Trost SG, Pfeiffer KA, Dowda M, Pate RR. Calibration and evaluation of an objective measure of physical activity in preschool children. J Phys Act Health 2005;2(3):345. 22. Department of Health and Aging (DoHA). Get up and grow: healthy eating and physical activity for early childhood. www.health.gov.au/ internet/main/publishing.nsf/Content/EDFEDB588460BCE3CA2576 2B00232A13/$File/gug-directorscoord-2.pdf. 23. Hinkley T, Salmon J, Okely AD, Crawford D, Hesketh K. Influences on preschool children’s physical activity: exploration through focus groups. Fam Community Health 2011;34(1):39 –50. 24. Hinkley T, Salmon J, Okely AD, Crawford D, Hesketh K. The HAPPY Study: development and reliability of a parent survey to assess correlates of preschool children’s physical activity. J Sci Med Sport 2012. 25. Australian Council for Health Physical Education and Recreation. Australian health and fıtness survey 1985. Adelaide: ACHPER Publications, 1985. 26. Wake M, Salmon L, Waters E, Wright M, Hesketh K. Parent-reported health status of overweight and obese Australian primary school children: a crosssectionalpopulationsurvey.IntJObesRelatMetabDisord2002;26(5):717–24. 27. Cole T, Bellizi M, Flegal K, Dietz W. Establishing a standard defınition for child overweight and obesity worldwide: international survey. BMJ 2000;320(7244):1240 –3. 28. Stata FAQ: How does one do regression when the dependent variable is a proportion? www.ats.ucla.edu/stat/stata/faq/proportion.htm. 29. Hardin JW, Hilbe JM. Generalized linear models and extensions. College Station TX: Stata Press, 2007. 30. Hailpern SM, Visintainer PF. Odds ratios and logistic regression: further examples of their use and interpretation. Stata J 2003;3(3):213–25. 31. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis. Englewood Cliffs NJ: Prentice Hall, 1998.
August 2012
167
32. Kutner MH, Nachtsheim CJ, Neter J, Li W. Applied linear statistical models. 5th ed. New York: McGraw-Hill Irwin, 2005. 33. Hamilton LC. Statistics with stata. Belmont CA: Brooks/Cole, Cengage Learning, 2009. 34. Burdette HL, Whitaker RC, Daniels S. Parental report of outdoor playtime as a measure of physical activity in preschool-aged children. Arch Pediatr Adolesc Med 2004;158(4):353–57. 35. Sallis JF, Nader PR, Broyles SL, et al. Correlates of physical activity at home in Mexican-American and Anglo-American preschool children. Health Psychol 1993;12(5):390 – 8. 36. Mistry KM, Minkovitz CS, Strobino DM, Borzekowski DLG. Children’s television exposure and behavioural and social outcomes at 5.5 years: does timing of exposure matter? Pediatrics 2007;120(4):762–9. 37. Hancox RJ, Milne BJ, Poulton R. Association of television viewing during childhood with poor educational achievement. Arch Pediatr Adolesc Med 2005;159(7):614 – 8. 38. Zimmerman FJ, Christakis DA. Children’s television viewing and cognitive outcomes: a longitudinal analysis of national data. Arch Pediatr Adolesc Med 2005;159(7):619 –25. 39. Yamamoto S, Becker S, Fischer J, De Bock F. Sex differences in the variables associated with objectively measured moderate-to-vigorous physical activity in preschoolers. Prev Med 2011;52(2):126 –9.
Appendix Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.amepre.2012.04.020.
Purpose: This study sought to identify multidimensional correlates of preschool children’s physical activity.
Methods: The social– ecological model (SEM) was used to identify constructs potentially associated with preschool children’s physical activity. Data were collected from 1004 preschool children, aged 3–5 years, and parents in 2008 –2009, and analyzed in 2010 –2011. Physical activity was measured over 8 days using ActiGraph accelerometers. Parents completed a comprehensive survey. Generalized linear modeling was used to assess associations between potential correlates and percentage of time spent in physical activity.
Results: Correlates of physical activity were found across all the domains of the SEM and varied between boys and girls and week and weekend days. Age was the only consistent correlate, with children spending approximately 10% less time in physical activity for each advancing year of age. Some modifıable correlates that were related to more than one physical activity outcome were rules restricting rough games inside and usual daily sleep time for boys. For girls, a preference to play inside/draw/do crafts rather than be active, and child constraints, was associated with more than one of the physical activity outcomes. A novel fınding in this study is the counterintuitive association between parental rules restricting rough games inside and boys’ higher physical activity participation levels. Conclusions: Potential strategies for promoting children’s physical activity should seek to influence children’s preference for physical activity and parent rules. Gender-specifıc strategies also may be warranted. (Am J Prev Med 2012;43(2):159 –167) © 2012 American Journal of Preventive Medicine
Introduction
P
hysical activity is an important contributor to health in adults1 and early childhood.2,3 Health behaviors such as physical activity track from early childhood into later childhood and adolescence.4,5 Establishing and maintaining adequate physical activity levels in young children is an important public health issue. From the Centre for Physical Activity and Nutrition Research (Hinkley, Salmon, Hesketh, Crawford), Faculty of Health, Deakin University, Burwood, Victoria; and the Interdisciplinary Educational Research Institute (Okely), Faculty of Education, University of Wollongong, Wollongong, New South Wales, Australia Trina Hinkley was on the Faculty of Education, Interdisciplinary Educational Research Institute, University of Wollongong, Wollongong, New South Wales, Australia, at the time some of the analyses and write-up of this paper occurred. Address correspondence to: Trina Hinkley, PhD, Center for Physical Activity and Nutrition Research, Deakin University, 221 Burwood Hwy, Burwood, Victoria 3125, Australia. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2012.04.020
Recent research shows that preschool children spend only a small proportion of their time being active,6,7 and compliance with physical activity recommendations is low.8,9 It is therefore important to identify the factors that may support or constrain physical activity. The social– ecologic model (SEM)10,11 provides a framework through which potential correlates of preschool children’s physical activity may be conceptualized. The SEM posits that the correlates of physical activity are multidimensional and operate across individual, social, and physical environment domains. Research to date has explored a range of potential correlates across those domains, and variables in each have been associated with preschool children’s physical activity.12 However, studies primarily have investigated only a narrow range of correlates, generally in only one domain, and there is little insight into the impact of the broader contexts in which those correlates exist. Further, as the majority of studies have used relatively small samples, it has not been possible to investigate
© 2012 American Journal of Preventive Medicine • Published by Elsevier Inc.
Am J Prev Med 2012;43(2):159 –167 159
160
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
differences in correlates between the genders, despite established knowledge that boys are more active than girls12 and correlates differ between school-aged boys and girls.13,14 Research15,16 also has shown that physical activity may vary between week and weekend days. However, the published literature has not reported possible differences in potential correlates between week and weekend days when differences in preschoolers’ environments may exist. The aim of the current study was to investigate potential correlates of preschoolers’ physical activity across all levels of the SEM, for boys and girls separately, and to investigate potential differences between correlates for weekday and weekend physical activity.
Methods Recruitment and Participants Methods of the current study have been published previously.9 Data were drawn from the baseline period of the Healthy Active Preschool Years Study when children were aged 3–5 years. Six local government areas (LGAs) in the Melbourne metropolitan area were selected randomly, two each from the lowest, middle, and highest SES quintiles as identifıed from the 2001 Socio-economic Indexes for Areas17 Index of Relative Socio-economic Advantage and Disadvantage. The fırst randomly ordered eight child care centers and eight preschools were selected in each of the two lowest-SES quintile LGAs, and the fırst fıve of each in each of the four LGAs in the middle and highest-SES quintiles. As participants from low-SES areas are often under-represented in research,18,19 they were oversampled in the present study. If a center declined to participate, the next center on the randomly ordered respective LGA preschool or child care center list was approached. Recruitment and data collection occurred in two phases: July to November 2008; and May to October 2009. In total, 156 child care centers and 137 preschools were approached. The fınal sample consisted of 71 (46%) child care centers and 65 (47%) preschools. Parents (9794) of children aged 3–5 years at each participating center were invited to participate. In total, 1036 parents provided written, informed consent (11%); four children were aged ⬎5 years and 28 withdrew before data collection, leaving a fınal sample of 1004. Ethical approval was provided by the Deakin University Human Research Ethics Committee and the Victorian Department of Education and Early Childhood Development.
Measures and Data Management Accelerometry. An ActiGraph GT1M accelerometer on an elastic belt was fıtted to each child. Accelerometers record date– time stamped information regarding the magnitude of movement, and are thus able to measure the frequency, intensity, and duration of physical activity. ActiGraph has demonstrated validity, reliability and utility in preschool children.20 Participants were instructed to wear accelerometers during waking hours at the right iliac crest for an 8-day period. Data were collected in 15-second epochs.20,21 Monitoring start times were identifıed as the beginning of the fourth complete minute of the appearance of counts above zero after 4:00AM, with a tolerance of four epochs (1 minute) of zero counts. Nonwear time was determined as ⱖ10 minutes of consec-
utive zero counts. Given the sporadic and constant nature of young children’s physical activity, it would be diffıcult for a child this age to be motionless for this long.21 Days with ⱖ18 hours of recorded data were excluded as being improbable. Age-specifıc cut points21 were applied to the data to distinguish the various intensities. Physical activity was operationalized as total physical activity (light, moderate, and vigorous intensities combined), consistent with recent recommendations22 and studies.8,9 The current study included three primary outcome variables: weekly, weekday, and weekend-day physical activity. The proportion of time (possible range 0 –1) the child spent in total physical activity on any given day was calculated to account for differences in wear time between children and days (reported as percentage for ease of interpretation and understanding). Any given day was considered valid if ⱖ7 hours of data were recorded. Average daily physical activity across the week included data from at least 3 valid weekdays and 1 valid weekend day (referred to as weekly physical activity). Weekday physical activity included data from at least 3 valid weekdays. Weekend-day physical activity included data from at least 2 valid weekend days. Reliability analyses, using repeated measures ANOVA, intraclass correlation (ICC), and Spearman-Brown prophecy formula, showed that this amount of weekly and weekday data achieved a reliability of 0.7 (2.9 weekdays achieved ICC⫽0.7; 3.9 days [weekday and weekend days combined] achieved ICC⫽0.7). Three weekend days would have been required to achieve reliability at this level for weekend days (2.9 weekend days achieved ICC⫽0.7). As only three children in the sample had 3 weekend days with suffıcient data, 2 weekend days were used with reliability of 0.61. Time of year was determined as the month during which the accelerometer commenced recording.
Parent survey. Parents completed a survey covering all domains of the SEM10,11 during the week their child wore the accelerometer. Items for inclusion in the survey were identifıed from a comprehensive review of the literature12 and focus groups with mothers of preschool children.23 Individual-level correlates included biological/demographic variables (e.g., parent’s age, gender, country of birth, education; child age, gender, disability, siblings, and night and day sleep time); child behavioral variables (e.g., participation in active transport, nonorganized or organized activities, and screen-based entertainment); and psychological variables (e.g., child requests for physical activity, child constraints to physical activity, and child preferences for physical activity). Social-level correlates included parental correlates (e.g., parental concerns, preferences, constraints, rules and regulations) and broader social correlates (e.g., social gatherings, provision of logistic and emotional support, role-modeling, and physical activity interaction). Environment-level correlates included variables in the home (e.g., yard size, indoor play spaces, availability of sports and electronic equipment, and number of TVs) and neighborhood correlates (e.g., availability and suitability of playgrounds, constraints to active transport, and frequency of visits to active play spaces). Items were considered reliable and included in the analyses if they met the following criteria: for test–retest reliability, categoric or dichotomized items had Kappa ⱖ0.6 or percentage agreement ⱖ60% and continuous items and scales had intraclass correlation ⱖ0.50; for internal reliability, items in a scale had a Cronbach’s alpha ⱖ0.7. A separate sample of 47 parents completed the survey on two occasions 14 –50 days apart (mean 24 days).24 Appendix A (available online at www.ajpmonline.org) provides www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167 details of constructs and items included in the HAPPY Study parent survey with their response options.
Anthropometrics. Child height and weight were measured by a trained researcher using a Wedderburn Seca portable rigid stadiometer, Wedderburn Tanita portable digital scales, and standardized measurement procedures.25,26 Parents self-reported their and their partner’s (where applicable) height and weight. BMI was calculated by standard formula and was determined for children and parents. Child and parent BMI and child weight status (BMI adjusted for age and gender27) were included in analyses as potential correlates.
Data Analysis Descriptive statistics were used to describe characteristics of the sample and the child’s physical activity. T-tests were used to assess differences in physical activity between boys and girls and between weekdays and weekend days. Generalized linear modeling (GLM) with a binomial family and logit transformation28,29 was used to examine associations between potential correlates and outcome variables. Each explanatory variable was entered individually into a GLM analysis with each of the physical activity outcome variables for boys and girls separately. The resultant ORs were interpreted as a percentage increase or decrease in the percentage of time spent in the physical activity outcome variable being examined relative to each unit increase in the explanatory variable or compared to the referent category of the explanatory variable.29,30 For each of the three outcome variables, the explanatory variables that showed an association (pⱕ0.01) in bivariable analyses were tested for collinearity by estimating the individual variance inflation factor (VIF) and tolerance for each item, and the mean VIF. Items with a VIF ⬎10 and tolerance less than 0.1 were removed.31,32 Where the mean VIF was signifıcantly ⬎133 (operationalized as ⱖ2) but no variables had an individual VIF ⬎10, the variable with the highest VIF was removed (variables identifıed where applicable in Tables 1 and 2). All remaining explanatory variables for each of the outcome variables were entered simultaneously into multivariable GLM analyses to determine the signifıcance (pⱕ0.05) of each of the variables when in combination with other signifıcant correlates. Model fıt statistics (Akaike information criterion [AIC]) and model specifıcation (data not reported)29 were examined to ensure each model was appropriately specifıed. AIC values ⬍1 indicate an acceptable model.29 Clustering by center (preschool/child care) of recruitment was controlled for in all analyses because of children being nested within centers. Analyses were undertaken in 2010 and 2011.
Results Data were collected from 1004 children and their parents. In total, 943 parents completed surveys, of which 93.7% were female respondents. Respondent parents had an average age of 37.3 years (95% CI⫽36.9, 37.6) and 69.8% were born in Australia. More than half of mothers (55%) reported tertiary education level; 34% had completed Year-12, trade, apprenticeship, or diploma as their highest educational qualifıcation; and 11% reported Year-10 or equivalent education. Children (n⫽939) had a mean age of 4.5 years (95% CI⫽4.5, 4.6). The fınal sample included 705 children who met the criteria for suffıcient August 2012
161
weekly accelerometry data (55% boys; mean wear time 655.0 [SD⫽64.3, range⫽478.4 –1019.7] minutes/day); 773 with suffıcient weekday accelerometry data (54% boys; mean wear time 649.5 [SD⫽64.3, range 440.9 – 1019.0] minutes/day); and 605 with suffıcient weekendday accelerometry data (56% boys; mean wear time 666.1 [SD⫽89.9, range 456.5–995.6] minutes/day). Those with suffıcient accelerometry data did not vary from the total sample by gender, maternal education, or child weight status; except boys were slightly more likely to have suffıcient weekend accelerometry data than girls (p⫽0.049) and children of mothers with university education were more likely to have suffıcient weekend accelerometry data than children of mothers with Year 12 or equivalent education (p⫽0.030). However, older children were more likely to have suffıcient accelerometry data for all three outcome variables (p⬍0.05 for all) than younger children. Possibly, this is a consequence of younger children sleeping more than older children (p⬍0.001). Boys spent a signifıcantly higher mean percentage of their daily time across the entire week (17.3% vs 15.4%); on weekdays (17.0% vs 15.3%); and on weekend days (18.2% vs 15.8%) in physical activity than did girls (p⬍0.001 for all). Boys and girls spent a greater mean percentage of their daily time in physical activity on weekend days than on weekdays (boys: 18.2% vs 16.7%, p⬍0.001; girls: 15.8% vs 15.1%, p⫽0.004). Results for associations between correlates across all domains of the SEM and boys’ mean percentage of daily time across the entire week, on weekdays and weekend days in physical activity are presented in Table 1. The only consistent correlate across all three outcome variables was age; for each additional year of age, boys spent 11%, 13%, and 10% less time in weekly, weekday, and weekend-day physical activity, respectively. Other variables associated with a lower percentage of time in physical activity included weekly computer-use time, parent housework, parent preference for their child to do the same activities as older siblings, and owning a desktop computer. Correlates associated with an increased percentage of time in physical activity included time spent outside on weekends, child’s total daily sleep time, child being active on his own, rules restricting rough games inside, the frequency the child’s mother and other children were active with the preschool child, and the number of weekly visits to active play spaces. Table 2 presents the results of the multivariable model analyses of correlates of mean percentage of daily time across the entire week, on weekdays, and weekend days in physical activity for girls. Age was consistently associated with all three physical activity outcome variables; for each additional year of age, girls spent 11%, 12%, and 15% less time in weekly, weekday, and weekend-day physical
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
162
Table 1. Multidimensional correlates of boys’ weekly, weekday, and weekend-day physical activity,a OR (95% CI) Multivariable associations Weekly physical activity (n⫽366, AIC⫽0.69)
Weekday physical activity (n⫽399, AIC⫽0.68)
Weekend-day physical activity (n⫽298, AIC⫽0.73)
0.89 (0.85, 0.92)*
0.87 (0.83, 0.91)*
0.90 (0.85, 0.96)*
1.04 (1.02, 1.07)*
1.04 (1.01, 1.06)*
—d
—d
—d
0.98 (0.96, 1.01)
—d
0.98 (0.96, 1.00)*
—d
Do not agree (ref)
—d
—d
1.00
Agree
—
—
1.13 (1.01, 1.27)*
Do not agree (ref)
—d
1.00
—d
Agree
—
0.91 (0.85, 0.97)*
—
1.00
1.00
—d
0.94 (0.88, 0.99)*
0.92 (0.87, 0.97)*
—
1.00
—d
1.00
1.06 (1.01, 1.12)*
—
1.11 (1.02, 1.20)*
1.01 (1.00, 1.03)*
—d
—d
—d
—d
1.04 (1.01, 1.07)*
Never (ref)
—d
—d
1.00
Once a month or less
—
—
—
Once every 2 weeks
—
—
1.05 (0.91, 1.21)
Once a week
—
—
0.99 (0.85, 1.14)
Two or more times a week
—
—
1.03 (0.89, 1.20)
Correlate INDIVIDUAL Biological and demographic Child age (years)b b
Total daily sleep (hours)
Number of cars in the homeb Behavioral Weekly time using computer (hours)b Psychological Preschool child is active by himselfc
SOCIAL Parental Time parent spends doing housework stops him/her from supporting preschool child to be activec
Parent prefers child to do same activities as older childrenc Do not agree/not applicable (ref) Agree Preschool child is not allowed to play rough games inside the housec Do not agree (ref) Agree Other social Frequency mother is active with preschool childb Frequency other children are active with preschool childb Frequency of attendance at social gatheringsc
(continued on next page)
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
163
Table 1. (continued) Multivariable associations
Correlate
Weekly physical activity (n⫽366, AIC⫽0.69)
Weekday physical activity (n⫽399, AIC⫽0.68)
Weekend-day physical activity (n⫽298, AIC⫽0.73)
—d
—d
1.00
Frequency of no one being active at social gatheringsc Not often (ref) Often
0.92 (0.83, 1.02)
PHYSICAL ENVIRONMENT Home and neighborhood —d
—d
1.01 (1.00, 1.02)*
1.01 (1.00, 1.03)*
—d
1.01 (1.00, 1.02)*
1.01 (1.00, 1.02)*
—d
Do not agree (ref)
—d
—d
1.00
Agree
—
—
0.86 (0.78, 0.95)*
Frequency of visits to shopping centers (per week)b
1.01 (0.98, 1.04)
Total frequency of visits to active play spaces (per week)b Time outside on weekends (hours)b Have desktop computer in homec
a
Generalized linear modeling analyses were adjusted for clustering by center of recruitment. OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for every one-unit increase in the correlate. c OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for the comparison group(s) compared with the referent group. d Variable was not associated with physical activity outcome in specific column. *p⬍0.05 AIC, Akaike information criterion b
activity, respectively. Girls’ preference to play inside/ draw/do crafts, child constraints, maternal work, frequency of visits to active play spaces, and lack of footpaths were associated with a lower percentage of time in physical activity for girls. The number of siblings, paternal provision of logistic support, and paternal time in moderate physical activity were associated with girls spending a higher percentage of time being active.
Discussion It was found that 20 correlates were associated with at least one outcome variable for either boys or girls: eight individual, seven social, and fıve physical environment. For boys, correlates across all domains were associated with all the physical activity outcomes. For girls, individual correlates were associated with all the outcome variables, and social and physical environment correlates were associated with weekly and weekend-day physical activity. The present study identifıes a number of previously unreported correlates of preschoolers’ physical activity. These include sleep, maternal work status, child being active on his/her own, parent preference for child to do the same activities as older siblings, child’s preference to August 2012
play inside/draw/do crafts rather than be active, child constraints, and lack of footpaths. Although each of the domains was represented in most of the multivariable models, there was little consistency between correlates for boys and girls, or between weekdays and weekend days. A novel fınding in the current study is the counterintuitive inverse association between parental restriction of rough play/games inside the home and higher levels of physical activity in boys. It is possible that parents who have such rules do so because their boys are intrinsically more active, thus necessitating the imposition of the rule. Alternatively, parents with such rules may make additional efforts to support their son being active by taking him outside more frequently or to active play spaces. However, such suggestions are conjecture and this association requires further investigation. Two of the most commonly studied correlates are time spent outside and watching TV.12 Time outside is commonly found to be positively associated with physical activity in preschool12 children. In the present study, boys’ time outside on weekend days was associated with their weekly and weekend-day physical activity, with each additional hour outside representing an additional 1% more time being
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
164
Table 2. Multidimensional correlates of girls’ weekly, weekday, and weekend-day physical activity,a OR (95% CI) Multivariable associations
Correlate
Weekly physical activity (n⫽262; AIC⫽0.69)
Weekday physical activity (n⫽258; AIC⫽0.68)
Weekend-day physical activity (n⫽196; AIC⫽0.83)
0.89 (0.84, 0.93)*
0.88 (0.84, 0.93)*
0.85 (0.79, 0.91)*
INDIVIDUAL Biological and demographic Child’s age (years)b c
Maternal work status
Home duties full-time (ref)
—d
1.00
1.00
Working part-time
—
0.92 (0.86, 0.97)*
0.99 (0.90, 1.10)
—
0.93 (0.83, 1.05)
0.84 (0.74, 0.97)*
Married/partnered (ref)
—e
—e
—d
Not married/partnered
—
—
—
Working full-time c
Parent’s marital status
b
Total number of siblings
1.06 (1.01, 1.11)*
d
—
1.02 (0.96, 1.09)
Psychological Preschool child is more likely to play inside/draw/do crafts than be activec Do not agree (ref) Agree Child constraints (e.g., lack of time, energy)b
1.00
1.00
1.00
0.90 (0.82, 0.99)*
0.89 (0.81, 0.97)*
0.88 (0.77, 1.00)*
0.99 (0.99, 1.00)*
0.99 (0.98, 1.00)
0.99 (0.98, 1.00)*
SOCIAL Other social Frequency of attendance at social gatheringsc Never (ref)
—d
—d
1.00
Once a month or less
—
—
—
Once every 2 weeks
—
—
0.94 (0.78, 1.14)
Once a week
—
—
0.94 (0.77, 1.13)
Two or more times a week
—
—
1.04 (0.87, 1.24)
d
Paternal provision of logistic supportb
1.01 (0.99, 1.03)
—
1.03 (1.00, 1.05)*
Maternal time in vigorous physical activity per week (hours)b
1.01 (0.99, 1.02)
1.00 (0.99, 1.02)
—d
—d
—d
—e
Paternal time in moderate physical activity per week (hours)b
1.01* (1.00, 1.02)
1.01 (1.00, 1.02)
1.01 (1.00, 1.02)
Frequency child sees mother being physically activeb
—e
1.00 (0.98, 1.03)
—d
Frequency child sees father being physically activeb
1.01 (0.99, 1.03)
1.01 (0.98, 1.03)
1.02 (1.00, 1.05)
Paternal time in total physical activity per week (hours)b
(continued on next page)
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
165
Table 2. (continued) Multivariable associations Weekday physical activity (n⫽258; AIC⫽0.68)
Weekend-day physical activity (n⫽196; AIC⫽0.83)
1.02 (0.99, 1.05)
1.01 (0.98, 1.04)
1.01 (0.98, 1.04)
Frequency child sees parents being physically activeb
—e
—e
—d
Frequency of proximal active role modelsb
—e
—e
—e
Frequency of all visible active role modelsb
—e
—e
—e
—d
—d
0.98 (0.96, 0.99)*
0.81 (0.71, 0.93)*
—d
—d
1.00
—d
—d
0.81 (0.71, 0.93)*
—
—
Correlate Frequency child sees other adults being physically activeb
Weekly physical activity (n⫽262; AIC⫽0.69)
Physical environment Total frequency of visits to active play spaces per weekb There are no footpaths in neighborhood for child and parents to usec Do not agree (ref) Agree Time outside on weekend days (hours)b
d
d
—
—
1.00 (0.98, 1.02)
June (ref)
—d
—d
1.00
July
—
—
1.02 (0.87, 1.20)
August
—
—
1.29 (1.07, 1.57)*
September
—
—
1.01 (0.86, 1.18)
October
—
—
1.13 (0.97, 1.32)
November
—
—
0.99 (0.86, 1.14)
December
—
—
1.13 (0.88, 1.45)
Month physical activity data measured
a
Generalized linear modeling analyses was adjusted for clustering by center of recruitment. OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for every one-unit increase in the correlate. c OR relates to an X% increase or decrease in the percentage of time in the physical activity outcome variable for the comparison group(s) compared with the referent group. d Variable was not associated with physical activity outcome in specific column. e Variable was not entered in multivariable analyses because of collinearity. *p⬍0.05 AIC, Akaike information criterion b
active. Such small differences may not be biologically meaningful. In bivariable analyses, time outside on weekend days was associated with boys’ weekend physical activity and with all of the girls’ physical activity outcomes (p⬍0.05). However, time outside on weekdays was not associated with any of the physical activity outcomes even in bivariable analyses. It may be that parents are not the best source for reporting children’s time outside on weekdays, when children may be cared for by others. August 2012
Prior research34 has found that TV-viewing time is associated with preschool children’s physical activity; however, the current study and others15,35 have found no such association. This may suggest that the notion of TV viewing displacing physical activity is not well supported, particularly at this young age, and that targeting a reduction in TV viewing as a means to increase physical activity may not be a relevant strategy. Nonetheless, targeting reductions in TV viewing
166
Hinkley et al / Am J Prev Med 2012;43(2):159 –167
and other screen-related behaviors is warranted for health and developmental reasons.36 –38 Most notably, the current study shows that the correlates of physical activity vary between boys and girls. When viewed collectively with previous research,13,14,39 these fındings suggest that future studies and interventions may need to recruit a suffıciently large sample to ensure adequate power to stratify analyses by gender. Future interventions also should consider genderspecifıc strategies to address such differences and increase their chances of successful outcomes. The present study examined a large number and broad range of potential correlates, across all levels of the SEM, although some potentially important correlates may have not been included. Further, parents may not be the best source of data collection for items such as time outside if their child is being cared for by others. Although an objective, reliable, and validated measure of physical activity was used, accelerometers are unable to detect certain activities and may therefore slightly underestimate total physical activity in some circumstances. Further, lack of agreement exists as to the most appropriate cut points to use when interpreting data. As the cut points used have a high threshold between sedentary and light physical activity, they may underestimate the time spent in light (and in this study therefore total) physical activity. The low reliability level for weekend-day data is also a limitation of the current study. However, this study used a large, heterogeneous sample and was able to investigate correlates for boys and girls, and weekdays and weekend days, separately. Finally, the instrument used to measure correlates of physical activity has established reliability.
Conclusion Future studies should investigate correlates separately for boys and girls as such correlates vary between the genders. Intervention studies and public health campaigns should ensure strategies address the multidimensional nature of physical activity correlates in preschool children, and include gender-specifıc strategies to increase the likelihood of success. Interventions may need to be delivered during the fırst 2–3 years of a child’s life to maximize their physical activity levels, as well as during the preschool period to stem declines in physical activity. TH was supported by a Deakin University APA PhD Scholarship during the fırst half of data collection. JS is supported by a National Heart Foundation of Australia Career Development Award and Sanofı-Aventis. KH is supported by a National Heart Foundation of Australia Career Development Award. DC is supported by a Victorian Health Promotion Foundation Senior Research Fellowship.
The project was funded by Deakin University. No fınancial disclosures were reported by the authors of this paper.
References 1. DHHS U. Physical activity and health: a report of the Surgeon General. Atlanta GA: DHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, 1996. 2. Jiménez-Pavón D, Kelly J, Reilly JJ. Associations between objectively measured habitual physical activity and adiposity in children and adolescents: systematic review. Int J Pediatr Obes 2010;5(1):3–18. 3. Janz KF, Burns TL, Torner JC, et al. Physical activity and bone measures in young children: the Iowa Bone Development Study. Pediatrics 2001;107(6):1387. 4. Kelly LA, Reilly JJ, Jackson DM, Montgomery C, Grant S, Paton JY. Tracking physical activity and sedentary behavior in young children. Pediatr Exerc Sci 2007;19(1):51– 60. 5. Taylor RW, Murdoch L, Carter P, Gerrard DF, Williams SM, Taylor BJ. Longitudinal study of physical activity and inactivity in preschoolers: the FLAME study. Med Sci Sports Exerc 2009;41(1):96 –102. 6. Oliver M, Schofıeld G, Kolt G. Physical activity in preschoolers: understanding prevalence and measurement issues. Sports Med 2007; 37(12):1045–70. 7. Tucker P. The physical activity levels of preschool-aged children: a systematic review. Early Childhood Res Q 2008;23(4):547–58. 8. Beets MW, Bornstein D, Dowda M, Pate RR. Compliance with national guidelines for physical activity in U.S. preschoolers: measurement and interpretation. Pediatrics 2011;127(4):658 – 64. 9. Hinkley T, Salmon J, Okely AD, Crawford DA, Hesketh KD. Preschoolers’ physical activity, screen time and compliance with recommendations. Med Sci Sports Exerc 2012;44(3):458 – 65. 10. Davison KK, Birch LL. Childhood overweight: a contextual model and recommendations for future research. Obes Rev 2001;2(3):159 –71. 11. Bronfenbrenner U. The ecology of human development: experiments by nature and design. Cambridge: Harvard University Press, 1979. 12. Hinkley T, Crawford D, Salmon J, Okely AD, Hesketh K. Preschool children and physical activity: a review of correlates. Am J Prev Med 2008;34(5):435– 41. 13. Patnode CD, Lytle LA, Erickson DJ, Sirard JR, Barr-Anderson D, Story M. The relative influence of demographic, individual, social, and environmental factors on physical activity among boys and girls. Int J Behav Nutr Phys Act 2010;7:79. 14. Trost SG, Pate RR, Ward DS, Saunders R, Riner W. Correlates of objectively measured physical activity in preadolescent youth. Am J Prev Med 1999;17(2):120 – 6. 15. Burdette HL, Whitaker RC. A national study of neighborhood safety, outdoor play, television viewing, and obesity in preschool children. Pediatrics 2005;116(3):657– 62. 16. Cardon G, De Bourdeaudhuij I. Are preschool children active enough? Objectively measured physical activity levels. Res Q Exerc Sport 2008;79(3):326 –32. 17. Australian Bureau of Statistics. Census of population and housing: socio-economic indexes for areas (SEIFA) 2033.8.55.001. Canberra: Australian Bureau of Statistics, 2001. 18. Telford A, Salmon J, Timperio A, Crawford D. Examining physical activity among 5- to 6- and 10- to 12-year-old children: the Children’s Leisure Activities Study. Pediatr Exerc Sci 2005;17(3):266 – 80. 19. Madigan MP, Troisi R, Potischman N, et al. Characteristics of respondents and non-respondents from a case-control study of breast cancer in younger women. Int J Epidemiol 2000;29(5):793– 8. 20. Cliff DP, Reilly JJ, Okely AD. Methodological considerations in using accelerometers to assess habitual physical activity in children aged 0-5 years. J Sci Med Sport 2009;12(5):557– 67.
www.ajpmonline.org
Hinkley et al / Am J Prev Med 2012;43(2):159 –167 21. Sirard JR, Trost SG, Pfeiffer KA, Dowda M, Pate RR. Calibration and evaluation of an objective measure of physical activity in preschool children. J Phys Act Health 2005;2(3):345. 22. Department of Health and Aging (DoHA). Get up and grow: healthy eating and physical activity for early childhood. www.health.gov.au/ internet/main/publishing.nsf/Content/EDFEDB588460BCE3CA2576 2B00232A13/$File/gug-directorscoord-2.pdf. 23. Hinkley T, Salmon J, Okely AD, Crawford D, Hesketh K. Influences on preschool children’s physical activity: exploration through focus groups. Fam Community Health 2011;34(1):39 –50. 24. Hinkley T, Salmon J, Okely AD, Crawford D, Hesketh K. The HAPPY Study: development and reliability of a parent survey to assess correlates of preschool children’s physical activity. J Sci Med Sport 2012. 25. Australian Council for Health Physical Education and Recreation. Australian health and fıtness survey 1985. Adelaide: ACHPER Publications, 1985. 26. Wake M, Salmon L, Waters E, Wright M, Hesketh K. Parent-reported health status of overweight and obese Australian primary school children: a crosssectionalpopulationsurvey.IntJObesRelatMetabDisord2002;26(5):717–24. 27. Cole T, Bellizi M, Flegal K, Dietz W. Establishing a standard defınition for child overweight and obesity worldwide: international survey. BMJ 2000;320(7244):1240 –3. 28. Stata FAQ: How does one do regression when the dependent variable is a proportion? www.ats.ucla.edu/stat/stata/faq/proportion.htm. 29. Hardin JW, Hilbe JM. Generalized linear models and extensions. College Station TX: Stata Press, 2007. 30. Hailpern SM, Visintainer PF. Odds ratios and logistic regression: further examples of their use and interpretation. Stata J 2003;3(3):213–25. 31. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis. Englewood Cliffs NJ: Prentice Hall, 1998.
August 2012
167
32. Kutner MH, Nachtsheim CJ, Neter J, Li W. Applied linear statistical models. 5th ed. New York: McGraw-Hill Irwin, 2005. 33. Hamilton LC. Statistics with stata. Belmont CA: Brooks/Cole, Cengage Learning, 2009. 34. Burdette HL, Whitaker RC, Daniels S. Parental report of outdoor playtime as a measure of physical activity in preschool-aged children. Arch Pediatr Adolesc Med 2004;158(4):353–57. 35. Sallis JF, Nader PR, Broyles SL, et al. Correlates of physical activity at home in Mexican-American and Anglo-American preschool children. Health Psychol 1993;12(5):390 – 8. 36. Mistry KM, Minkovitz CS, Strobino DM, Borzekowski DLG. Children’s television exposure and behavioural and social outcomes at 5.5 years: does timing of exposure matter? Pediatrics 2007;120(4):762–9. 37. Hancox RJ, Milne BJ, Poulton R. Association of television viewing during childhood with poor educational achievement. Arch Pediatr Adolesc Med 2005;159(7):614 – 8. 38. Zimmerman FJ, Christakis DA. Children’s television viewing and cognitive outcomes: a longitudinal analysis of national data. Arch Pediatr Adolesc Med 2005;159(7):619 –25. 39. Yamamoto S, Becker S, Fischer J, De Bock F. Sex differences in the variables associated with objectively measured moderate-to-vigorous physical activity in preschoolers. Prev Med 2011;52(2):126 –9.
Appendix Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.amepre.2012.04.020.