Health Reports
Associations between parent and child sedentary behaviour and physical activity in early childhood

by Valerie Carson, Kellie Langlois and Rachel Colley

DOI: https://www.doi.org/10.25318/82-003-x202000200001-eng

Physical activity has numerous physical, social, and cognitive health benefits in early childhood.^{Note 1} Conversely, sedentary behaviour, in particular screen time, is detrimentally associated with healthy growth and development in this age group.^{Note 2} The behavioural patterns formed in early childhood have implications throughout childhood.^{Note 3} More specifically, physical activity has been found to be moderately stable from early childhood to middle childhood, whereas, over the same time period, moderate to large stability has been observed for sedentary behaviour.^{Note 3} At present, only 15% of Canadian preschoolers (3-4-year-olds) meet both physical activity (⋝180 minutes/day of total physical activity, including ⋝60 minutes/day of energetic play) and screen time (⋜1 hour/day) recommendations, within 24-Hour Movement Guidelines.^{Note 4} A similar pattern has been observed in a regional sample of toddlers in Edmonton, Canada.^{Note 5}

From an ecological perspective, children’s sedentary behaviour and physical activity can be influenced by several correlates across multiple levels, such as intrapersonal, interpersonal, organizational, environmental, and policy.^{Note 6} The interpersonal level, which includes parental correlates (e.g., behaviours and practices), is thought to be particularly important for children’s physical activity and sedentary behaviours.^{Note 7} Specifically, parents profoundly influence children’s development and are gatekeepers for children’s participation in both sedentary behaviour and physical activity.^{Note 7} This is particularly true for young children who have less independence from their parents, compared to older children.^{Note 8} Therefore, identifying key parental correlates of sedentary behaviour and physical activity can inform interventions designed to increase the proportion of early years children who meet guidelines.

One potential mechanism to explain how parents influence children’s sedentary behaviour and physical activity is role modeling.^{Note 7} Previous literature focusing on role modeling has typically examined the associations between sedentary behaviour and physical activity of parents and children.^{Note 7Note 9} However, robust conclusions on parental role modeling of sedentary behaviour and physical activity are difficult to make in early childhood due to gaps and limitations in the current evidence base. For example, previous research is limited by convenience samples, small sample sizes, and/or subjective measures of sedentary behaviour and physical activity.^{Note 10} Additionally, few studies have examined whether associations between parent and preschoolers’ sedentary behaviours and physical activity differ by child sex, parental sex, or day of the week (e.g., weekdays, weekend days).^{Note 9Note 11} Thus, the objectives of this study were to: 1) examine the associations between total days, weekday, and weekend objectively-measured sedentary time and physical activity as well as total parent-reported screen time among parents and children in a large representative sample of Canadian 3-5-year-olds, 2) determine if associations differed between sons and daughters, and 3) determine if associations differed between mothers and fathers.

Methods

Participants

Participants were children aged 3 to 5 years and their biological parent from cycles 2 (2009–2011), 3 (2012–2013), 4 (2014-2015) and 5 (2016-2017) of the Canadian Health Measures Survey (CHMS).^{Note 12} The CHMS uses a repeated cross-sectional design and a three-stage sampling procedure to collect data from a sample representing 96% of Canadian 3- to 79-year-olds living in the provinces.^{Note 13} At each collection site, a respondent was selected from each dwelling, and when the respondent was aged 3-11 years, an older member from that dwelling was also selected (12-79 years).^{Note 13} Data was not included from cycle 1 in this analysis because the age group was not within scope (only 6- to 79-year-olds were sampled).

Data collection included a computer-assisted in-person interview at each dwelling and a physical health examination collected at a mobile examination centre (MEC). At the MEC, each participant was given an accelerometer with specific wear instructions. A total of 2,181 children aged 3-5 at the MEC visit participated in the CHMS along with an older member of their dwelling, and 1,792 of these pairs were deemed eligible for this study because the older participating member was a biological parent. Ethics approval for the CHMS was obtained from Health Canada and the Public Health Agency of Canada Research Ethics Board.^{Note 14} Written informed consent was obtained from the parent and assent was obtained from the child.^{Note 15} Detailed information about the CHMS is available elsewhere.^{Note 13}

Sedentary Behaviour and Physical Activity

Sedentary time, light-intensity physical activity (LPA), and moderate- to vigorous-intensity physical activity (MVPA) were objectively-measured in both parents and children with Actical accelerometers (Philips Respironics, Bend, OR, USA). Both parents and children wore the device over their right hip on an elasticized belt for 7 consecutive days. Data was collected in 60 second epochs for parents across all cycles. However, for children data was collected in 60 second epochs for cycle 2 and 15-second epochs for cycles 3-5. Due to the smaller epochs used in cycles 3-5, only 5.6 days of data was recorded due to the memory capacity of the device. Therefore, only the first 5 days were used for children in cycles 3-5. Accelerometer non-wear time in both parents and children were defined as ⋝60 consecutive minutes of zero counts (or ⋝240 intervals of 15 seconds of zero counts), with allowance for 2 minutes of counts between zero and 100 (or 30 seconds of counts between 0 and 25).^{Note 16Note 17} To be included in the analyses, parents were required to have ⋝4 valid days, defined as ⋝10 hours of wear time, and children were required to have ⋝4 valid days in cycle 2 and ⋝3 valid days in cycles 3-5, defined as ⋝5 hours of wear time.^{Note 16Note 18Note 19Note 20} For children, sedentary time was defined as <100 counts per minute (cpm, or <25 counts per 15 seconds),^{Note 21} LPA as 100–1149 cpm (or 25-278 counts per 15 seconds), and MVPA as ⋝1150 cpm (or ⋝288 counts per 15 seconds).^{Note 22} In adults, sedentary time was defined as <100 cpm,^{Note 21} LPA as 100–1534 cpm, and MVPA as ⋝1535 cpm.^{Note 23} Average minutes/day of sedentary time, LPA, and MVPA were calculated for parents and children for all valid days, valid weekdays, and valid weekend days. Correction factors were applied to children’s minutes/day data in cycle 2 so the data was comparable to cycles 3-5.^{Note 24} Finally, to adjust for wear time, standardized sedentary time and LPA variables were calculated for both parents and children.^{Note 25} MVPA was not adjusted for wear time because it was not highly correlated with wear time in children (r=0.28) or parents (r=0.17).

Average daily screen time of children and parents were assessed as part of the household interview. Parents reported the average hours per day their child spends 1) watching TV or videos or playing video games, and 2) on a computer. The response categories were slightly different between cycle 2 (none, <1, 1 to 2, 3 to 4, 5 to 6, ⋝7) and cycles 3–5 (none, <1, 1 to <3, 3 to <5, 5 to <7, ⋝7) for children. However, the same mid-points could be applied across cycles (i.e., 0, 0.5, 1.5, 3.5, 5.5 and 7). Children’s average daily screen time was calculated by summing the mid-points of the two screen time questions. For cycles 2-4, parents also reported the average time in a typical week outside of work or school (leisure time only) during the past three months they have spent: 1) watching television, DVDs or videos, 2) playing video games, and 3) on a computer. In cycle 5, parents reported the time during the last 7 days outside of work or school (leisure time only) they have spent: 1) watching TV, DVDs, movies, or internet videos, 2) playing video or computer games, 3) on a computer, tablet or smart phone. Across all cycles, parental average daily screen time was calculated by summing the three screen time questions and dividing by seven.

Covariates

Covariates included child and parent sex, child and parent age, single-child household, parental education, and survey cycle. Parental education was coded into two categories, post-secondary graduate and less than post-secondary graduate. A single-child household was defined as a dwelling with no other children younger than 18 living in it.

Statistical analyses

All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and SUDAAN version 11.0.3, using DDF=46 (denominator degrees of freedom) in the SUDAAN procedure statements. Descriptive statistics were calculated for the total sample and separately for males and females. T-tests were conducted to compare participant characteristics between those in the final sample and those with valid (child) accelerometer data but deemed ineligible because the older member selected from the dwelling was not a biological parent, the biological parent was pregnant or did not have valid accelerometer data, or the dyad was excluded due to missing covariate information (non-study sample; n=511). Sex differences in descriptive information were also examined using t-tests. To address objective one, correlations between child and parent sedentary behaviour and physical activity variables were performed, followed by linear regression models adjusted for child and parent sex, child and parent age, parental education, single-child household, and survey cycle. Separate models were run for each of the sedentary behaviour and physical activity variables. To address objectives two and three, correlations were conducted for child and parent sex sub-groups. To determine if sex-stratified analyses were needed, child sex or parent sex by parent sedentary behaviour or physical activity interaction terms were added to the linear regression models separately. To meet the assumption of normality for linear regression models, child MVPA was log transformed. The analyses were weighted using CHMS survey weights assigned to the child for combined cycles, 2-5.^{Note 26} Survey weights accounted for non-response and incomplete accelerometer data, to ensure the sample was representative of Canadian children aged 3-5 years. To account for survey design effects, 95% confidence intervals and significance testing were estimated with the bootstrap technique.^{Note 27Note 28} Statistical significance was set a priori at p <0.05.

Results

Of the 1,792 eligible parent-child pairs, 51 pairs were excluded because the parent was pregnant, 621 pairs were excluded because of a lack of valid accelerometer data for both the parent and the child, and 4 were excluded because of missing covariate data. Therefore, the final sample for this study was 1,116 parent-child pairs, and for analyses involving weekend days, the sample was 935 parent-child pairs. No significant differences in child age, sex, sedentary time, LPA, MVPA and proportion of single-child households were observed between those in the final sample and those in the non-study sample. However, children in the final sample had lower screen time compared to children in the non-study sample. Also, adults in the final sample were older compared to adults in the non-study sample, and there was a higher proportion of university/college graduates in the final sample compared to the non-study sample (data not shown).

Participant characteristics, stratified by sex, are provided in Table 1. The average age of children was 4 years and approximately half (49%) were female. Average screen time and sedentary time among children was 1.8 hr/day and 452 min/day, respectively, whereas average LPA and MVPA for total days was 208 min/day and 73 min/day, respectively. Regardless of the day of the week, female children engaged in significantly more sedentary time and significantly less MVPA than male children. The average age of parents was 36.6 years and approximately half (52%) were female. Among parents, average screen time and sedentary time was 2.3 hr/day and 567 min/day, respectively; average LPA and MVPA for total days was 259 min/day and 23 min/day, respectively. Male parents had significantly higher average total and weekday wear time and engaged in significantly more total and weekday MVPA than female parents.

Correlations between parent-child screen time, sedentary time, and physical activity are displayed in Table 2. In the overall sample, all of the parental behaviours were significantly correlated with the children’s behaviours. For the accelerometer data, this was consistent for total days, weekdays, and weekend days. However, correlations were small in magnitude,^{Note 29} ranging from 0.08-0.20. The strength of correlation differed across the various sex combinations of the parent-child dyads. For instance, the MVPA of fathers tended to have stronger correlations with children’s MVPA compared to mothers. The same findings were observed for sedentary time for sons only. In most instances, correlations tended to be stronger for same sex parent-child dyads compared to opposite sex parent-child dyads.

The linear regression associations between parent-child screen time, sedentary time, and physical activity are displayed in Table 3. Child sex by parent physical activity or sedentary behaviour interaction terms and parent sex by parent physical activity or sedentary behaviour interaction terms were not significant so sex-stratified analyses were not conducted. All of the parental behaviours were significantly associated with the children’s behaviours. For the accelerometer data, this was consistent for total days, weekdays, and weekend days. However, for sedentary time and LPA, beta coefficients were larger on the weekend days compared to total days and weekdays. Overall the effect sizes were small. For example, every additional hour of parental screen time was associated with approximately 8 minutes higher screen time in children. For total days, every additional 60 minutes per day of parental sedentary time and LPA was associated with approximately 7 minutes higher sedentary time and 6 minutes higher LPA in children. After back-transforming MVPA on total days from the log scale, every additional 20 minutes per day of parental MVPA was associated with approximately 4% higher MVPA in children (equivalent to approximately 3 minutes per day).

Discussion

The objectives of this study were to examine the associations between parent and child sedentary behaviour and physical activity in early childhood, and determine if associations differed by the sex of the parent or child. Higher parental screen time, sedentary time, LPA and MVPA was significantly associated with higher screen time, sedentary time, LPA, and MVPA in this large representative sample of Canadian 3-5-year-olds. However, effect sizes were small. Additionally, associations did not differ between weekdays and weekend days, sons and daughters or mothers and fathers.

In line with the present study, a meta-analysis published in 2015 on the parental correlates of physical activity in children and adolescents between the ages of 2.5 and 18 years reported that parental modeling, operationalized as parental physical activity, was weakly associated with children’s physical activity.^{Note 9} Age group (i.e., 2-5.4, 5.5-12.4, 12.5-19 years) was not a significant moderator in this meta-analysis, though point estimates in random effect models were larger for younger age groups,^{Note 9} supporting the notion that parents have stronger influences on younger children compared to older children.^{Note 8} Overall, the present study strengthens the evidence base on the association between parental role modeling and physical activity in early childhood by addressing previous limitations, including small sample sizes and subjective measures.^{Note 9} Additionally, associations were not only observed for MVPA but also for LPA, which has rarely been examined in any group.^{Note 9}

Compared to physical activity, little is known on the association between parent and child objectively measured sedentary time in early childhood, highlighting another novel aspect of the present study.^{Note 30Note 31} Study findings are consistent with a previous study in a representative sample of Canadian 6-11-year-olds from the CHMS, where parental sedentary time was also found to be weakly associated with children’s sedentary time.^{Note 32} More research has examined the association between parent and child screen time in early childhood. For example, a systematic review published in 2012 on correlates of energy balance-related behaviours reported a positive association between parental television viewing and children’s screen time in both of the included studies.^{Note 33} Since that review was published, two Canadian studies with regional samples have also reported an association between higher parental screen time and higher screen time among children aged 0-5 years^{Note 11} and 1-2 years.^{Note 34} However, effect sizes were stronger in these previous studies compared to the present study. Overall, the findings of this study confirm that parental role modeling is a consistent correlate of screen time in early childhood.

The present study made an important contribution to the literature base on parental modeling of sedentary behaviour and physical activity in early childhood by examining the moderating effects of sex in both children and parents. In contrast to the present study, the meta-analysis on parental correlates of physical activity found that parental sex moderated the association between parental and children’s physical activity in boys only.^{Note 9} Specifically, effect sizes were larger among father-son dyads compared to mother-son dyads.^{Note 9} However, none of the studies included in this sub-analyses focused on young children.^{Note 9} Correlations in the present study were also stronger for father-son dyads compared to mother-son dyads for MVPA, however in linear regression models, interaction terms were not found to be statistically significant. In the previous study of 6-11-year-olds from the CHMS, it was also found that parental sex did not moderate the association between parental and children’s sedentary behaviour and physical activity.^{Note 11Note 32} However, some child sex differences were observed. For instance, parental MVPA was associated with girls’ MVPA but not boys’ MVPA for total days.^{Note 32} In line with the present study, children’s sex was also not a significant moderator in the previous study of 0-5-year-olds.^{Note 11} Therefore, it is possible that the moderating effects of parent and child sex is not as strong in younger age groups compared to older age groups.

Parental modeling is thought to be an important correlate of children’s physical activity, in particular for younger age groups, because of the social norms that it creates within the family unit.^{Note 9} A similar rationale can apply to sedentary behaviour. Additionally, for younger children, in particular, parental modeling is thought to also encompass co-participation in sedentary behaviour and physical activity.^{Note 9} One potential reason for the small effect sizes observed in the present study and others^{Note 9} could be the crude measures used. For example, while average durations of parental and children’s behaviours are likely capturing habitual levels,^{Note 19} it is not possible to know from these measures when parental and children’s behaviours are occurring independently or concurrently. Therefore, future research examining parental modeling of sedentary behaviour and physical activity should consider methodologies that provide more detailed information on parental modeling. For example, some newer models of accelerometers have proximity tagging features where sedentary time and physical activity can be objectively measured in multiple people while simultaneously recording when they are in close proximity.^{Note 35Note 36} This feature has been validated in parent-young child dyads.^{Note 36} To date, a study using this method in a small sample of mother-young child dyads has found dyads mostly spend their time together engaged in sedentary time or LPA but rarely MVPA.^{Note 35} This proximity tagging method could be expanded to include multiple caregivers (e.g., mother and father) along with siblings to better understand behavioural influences of the entire family unit.

The present study had a number of strengths including the large representative sample and the objective measures of sedentary time and physical activity in both parents and children. Additionally, the examination of whether associations differed by the sex of the child and the parent addressed a gap in the evidence base for this age group. The study also had some limitations. Given the repeated-cross-sectional design of the CHMS, observed associations cannot be interpreted as causal. Some differences (screen time, parental education) were noted between children included in the analysis (i.e., that had a biological parent and complete data) and those excluded from the analysis due to ineligibility. These differences would be accounted for in the survey weights; however, some bias could still exist in the findings observed. Screen time was parent-reported and therefore more prone to biases (e.g., recall and social desirability). Data was only available for one biological parent, however the participating parent was randomly selected by study personnel based on the dwelling roster and not by the family. Some methods differed between cycles, however, survey cycle was adjusted for in linear regression models. Finally, residual confounding may have occurred. For instance, it was not possible to adjust for children’s childcare status because this variable was not available in all cycles. In addition, children are influenced by other adults in addition to their parents (e.g., childcare providers, grandparents) and there was no mechanism within the present analysis to address this. Though analyses were stratified by weekdays and weekend days, with the weekend typically being when children are with their parents, regardless of their childcare status.

Conclusions

Findings from this large representative sample of Canadian 3-5-year-olds and their biological parent suggest that parental behaviours may be one target for family-based sedentary behaviour and physical activity interventions and public health messaging in early childhood. This appears relevant regardless of the sex of the parent or child. Given the small effect sizes observed, future research using more detailed measures of parental modeling is needed, along with the consideration of additional intervention targets.

Acknowledgements

Valerie Carson is supported by a Canadian Institutes of Health Research (CIHR) New Investigator Salary Award.