Study design and participants
This study includes data from the Early Movers project, which used a longitudinal study design. Participants were parents/guardians (parents thereafter) and their infants who were recruited while attending routine 2-month immunization appointments at one of five Public Health Centres in Edmonton, Canada that serve diverse populations. The provincial health authority (Alberta Health Services) helped to facilitate recruitment in the waiting rooms of these centres. The uptake of 2-month immunizations in the population of infants residing in Edmonton, Canada is approximately 85% (Personal Communication, Alberta Health Services, May 8, 2017). Recruitment took place between March, 2018 and November, 2019. Eligibility criteria for the Early Movers project has been previously reported [23, 24].
It is often challenging in movement behaviour research to use measures with high precision in large and diverse samples [25, 26]. The Early Movers project tried to address this challenge by enrolling participants in different groups or sub-studies based on measurement precision and burden. Further details on the group structure of the Early Movers project has been previously published [24]. Briefly, for the purpose of this study, all participants were enrolled in the main study, which included the completion of lower-burden questionnaire measures. Also, a sub-sample of participants agreed to be enrolled in a time-use diary sub-study that included additional measures with higher precision and burden (i.e., time-use diaries, physical therapist- assessed gross motor development). A total of 808 families were recruited for the Early Movers project. Ethics approval was obtained from the University of Alberta Research Ethics Board (Project # 00,078,438). Written informed consent was obtained from all participating parents. Details on the apriori power calculations for the main study and the time-use diary sub-study have been previously published [24].
Procedures
Contacts at each health centre informed research staff of scheduled 2-month immunization appointments. Research staff visited the waiting room when multiple appointments were scheduled and spoke to parents before or after their appointments. Families are required to stay in the waiting room for 15 min after their infant’s immunization for safety reasons. Interested and eligible parents completed a consent form, contact information form, and a parental questionnaire at the health centre using the secure web application REDCap [27] or on a hard paper copy. Parents were also given a gross motor milestone questionnaire to take home. Participants were then emailed a survey link to the follow-up questionnaires via REDCap or were mailed a hard copy of the questionnaires to be returned via mail when their infants turned 4 and 6 months of age. Next, participants were contacted monthly regarding the achievement of gross motor milestones until their child had reached all the milestones or was 18 months of age [28].
Parents who agreed at the immunization appointment to participate in the time-use diary sub-study were also asked to complete a 3-day/night time-use diary when their infant was 2, 4, and 6 months of age. Participants received a hard copy of the diary at the immunization appointment, along with verbal and written instructions. Subsequent time-use diaries were mailed to participants and returned via mail when infants turned 4 and 6 months of age. Additionally, when infants were between the ages of 6 months 0 days and 6 months 7 days, a physical therapist made a home visit to assess infants’ gross motor development. Gift cards were mailed to participants upon completion of data collection. Participants who enrolled in the main study only were eligible for a gift card of $25 CAD in value. Participants who enrolled in the main study and the time-use diary sub-study were eligible for a gift card of $35 CAD in value, given the extra burden of measures associated with the sub-study. Pro-rated gift card amounts were provided if participants withdrew early from the study.
Measures
Questionnaire-measured movement behaviours
Infant movement behaviours were measured using the parental questionnaire when infants were 2, 4, and 6 months of age. For physical activity, tummy time was measured with one question asking parents to report the typical time per day their child spends awake on their stomach when they are free to move. For sedentary behaviour, reading time was measured with one question asking parents to report the typical time per day their child spends reading/looking at books with the parent or another child/adult. Additionally, screen time was measured with two separate questions asking parents to report the typical time per day their child spends: 1) watching/looking at the television and 2) watching/looking at a cell phone/tablet. Responses were summed across screen time questions. Finally, for sleep, sleep time was measured with two separate questions asking parents to report the typical time their child: 1) usually sleeps in total per night at the moment (not including time spent feeding) and 2) naps in total during the day at the moment. Responses were summed across sleep questions. Detailed information on the psychometric properties of the questionnaire-measured movement behaviours in the Early Movers project have been previously reported [23, 24]. Briefly, these movement behaviour questions have been adapted from previous studies, where test re-test reliability has been reported (Intraclass correlation coefficient: ICC = 0.20 to 0.86) [29, 30]. Within sub-samples of the Early Movers participants, concurrent validity for the tummy time measure (against an accelerometer; rs = 0.60, p < 0.05) and all movement behaviour measures (against the time-use diary described in the next section; rs = 0.30–0.56; p < 0.05) have also been reported [24, 31].
Time-use diary-measured movement behaviours
Infant movement behaviours were measured using the time-use diary when infants were 2, 4, and 6 months of age. Over three 24 h periods, parents recorded in 5 min intervals their infant’s main activity from a list of 17 options and their infant’s position from a list of 10 options. For physical activity, tummy time included the average time across valid days that infants spent on their front/tummy or army/commando crawling (i.e., infant on tummy with some movement; 6 months: n = 7) position while awake. For sedentary behaviour, screen time included the average time across valid days that infants’ main activity was TV and/or cell phone/tablet. Additionally, reading time included the average time across valid days that infants’ main activity was reading (by an adult or another child). Finally, restrained time bouts were the average time across valid days that infants’ main activity was stroller ride, car ride, carrier, indoor swing, or other restricted activity (e.g., high chair, car seat) for a consecutive period greater than 1 h (while awake). For sleep, sleep time included the average time across valid days that infants’ main activity was sleeping. Further details regarding this time-use diary, including psychometric properties, have been described in detail elsewhere [31]. Briefly, the time-use diary was adapted from a previous study in adults, where test re-test reliability (ICC of 0.50 to 0.55) was reported [32]. Within a sub-sample of the Early Movers project (n = 26), concurrent validity for the tummy time measure (against an accelerometer; rs = 0.80, p < 0.05) has also been previously reported [31].
Guideline adherence
Both the questionnaire-measured and time-use diary-measured movement behaviour variables were categorized as meeting versus not meeting the infant recommendations within the Canadian 24-Hour Movement Guidelines for the Early Years [1]. Specifically, to meet the physical activity recommendation, infants had to engage in at least 30 min of tummy time per day. To meet the sedentary behaviour recommendation, two definitions were developed. For definition 1, infants had to engage in no screen time and some reading time. For definition 2, which could only be assessed in the time-use diary sub-sample, infants had to engage in no screen time, some reading time, and no restrained bouts greater than 1 h. To meet the sleep recommendation, infants had to sleep 14 to 17 h per 24-h period when they were 2 months of age and 12 to 16 h per 24-h period when they were 4 and 6 months of age. Meeting the overall guidelines was defined as meeting the physical activity, sedentary behaviour, and sleep recommendations.
Development
The Early Movers project included several measures of development. Communication, fine motor, gross motor, personal-social, problem solving and total development were measured at 2, 4, and 6 months of age with the Ages and Stages Questionnaire (ASQ-3) [33]. Specifically, at each time point, 30 items, specific to the age group, were included in the parental questionnaire with three response options (yes, sometimes, not yet). Each area of development was given a score between 0 and 60 and these scores were summed for the total development score, with higher scores indicating more advanced development [33]. Further details on the scoring of the ASQ-3 can be found elsewhere [24]. Validity for the ASQ-3 tool has previously been reported (Criterion validity against Battelle Developmental Inventory-II: Percent agreement at 2 months = 100%; 4 months = 83.3%; 6 months = 85.7%) [33].
The dates children acquired six gross motor milestones (independent sitting, hands and knees crawling, assisted standing, assisted walking, independent standing, and independent walking) in the first 18 months of life were reported by parents in a separate questionnaire. The questionnaire included detailed instructions and pictures from the World Health Organization (WHO) on how to determine if the milestones were achieved [28]. Parents also recorded whether the dates provided was exact or approximate. Further details on how milestone data were cleaned for the Early Movers project has been previously published [24]. Children were only followed up to 18 months as World Health Organization reference data indicates that 99% of children typically acquire these six milestones by this age [28].
In the time-use diary sub-sample, gross motor development was also directly observed by a physical therapist when infants were 6 months of age using the Alberta Infant Motor Scale (AIMS) [34]. Specifically, a total of 58 items were scored across four postural positions (Prone: 0–21, Supine: 0–9, Sitting: 0–12, Standing: 0–16), and a total score was calculated by summing the prone, supine, sitting and standing subscale scores [34]. Additionally, infants were assigned a percentile score between 0 and 100 based on their total score and age. The percentile score is based on normative data of 2,220 infants from Alberta in 1990–1992. Higher AIMS scores indicate more advanced gross motor development. Reliability (test–retest: r = 0.96–0.99) and validity (Concurrent validity: Bayley Scales of Infant Development motor scales: r = 0.98 and Peabody Development Motor Scales: r = 0.97) have previously been reported for the AIMS tool [34].
Covariates
Several infant and parental demographic characteristics that were measured in the parental questionnaires were considered as covariates based on previous research [35, 36]. Infant age, expressed as days, was calculated at each time point based on the date of questionnaire completion and the birth date reported at baseline. Non-parental care time (hours per week) was also reported at each time point. Infant sex, race/ethnicity and number of siblings were reported at baseline only. Infant sex had two response options (male, female), and response options for race/ethnicity (Caucasian, other) and number of siblings (zero, one, two or more) were collapsed from the original scales due to frequency distributions. Parental age, expressed as years, was calculated at each time point in conjunction with the infant age calculation. Mean imputation was performed for missing parental age data at baseline for one participant. Parental martial status (married/living common-law, not married/ living common-law), education (below bachelor level, bachelor’s degree, above bachelor level), and country of birth (Canada, other) were reported at baseline only, and were collapsed from the original scales due to frequency distributions.
Statistical analysis
Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC) and SPSS version 26.0 (SPSS Inc., Chicago, IL, USA). Descriptive statistics were calculated for demographic characteristics and for guideline adherence. To address objective 1, generalized estimating equations (GEE) were performed to calculate exponentiated longitudinal tracking coefficients (odds ratio [OR]). Specifically, individual recommendation or overall guideline adherence (meeting vs. not meeting) at 2 months was regressed on the corresponding longitudinal individual recommendation or overall guideline adherence from 4 to 6 months. Time point (4 and 6 months) was included in the model as a within-subject variable and an unstructured correlation structure was used for all models.
To address objective 2, continuous adherence variables were calculated based on the number of recommendations met across the three time-points. For individual recommendation adherence variables (i.e., tummy time, screen time, reading time, restrained time bouts, sedentary behaviour definition 1 [no screen time, some reading time], sedentary behaviour definition 2 [no screen time, some reading time, no restrained bouts ≥ 1 h], sleep time) the possible range was 0 to 3. Similarly, for overall guideline adherence variables (i.e., physical activity + sedentary behaviour definition 1 + sleep or physical activity + sedentary behaviour definition 2 + sleep) the possible range was also 0 to 3. A final continuous guideline adherence variable was calculated based on the total number of recommendations. The possible range for this variable was 0 to 9, given there are three movement behaviours (i.e., physical activity, sedentary behaviour (definition 1 or 2), sleep) and three time points (2 months, 4 months, 6 months). To examine the associations of recommendation and overall guideline adherence over time with ASQ-3 outcomes over time, linear mixed models were conducted. Separate models were run for each combination of guideline adherence variable and ASQ-3 outcome variable. In all models, time was included as a repeated and fixed effect and covariates were included as fixed effects. Analyses were conducted for both the questionnaire data and time-use diary data. Assumptions for linear mixed models were checked through visual inspection of residuals and all assumptions were met. The unstandardized beta coefficient can be interpreted as the pooled within- and between-individual differences in the ASQ-3 outcome variable for each additional time a recommendation or the overall guidelines are met across the three points.
To examine the associations of recommendation and overall guideline adherence over time with gross motor milestone and AIMS outcomes, linear and logistic regression models were conducted. The analyses for the AIMS outcomes were only conducted with the time-use diary measured movement behaviours as this outcome was only measured in the time-use diary sub-study. A logistic regression model was only implemented for the AIMS stand variable because a non-normal distribution was observed for this variable when the distributions of the model residuals were visually checked. As a result, this variable was dichotomized (value = 1 [score = 2, reference group]; value = 0 [score > 2]). All other linear regression model assumptions were met across models. All covariates were included in all models, except infant age, which was excluded from gross motor milestone models, since age was the unit of the gross motor milestone variables. Sensitivity analyses was conducted examining the associations between movement behaviours over time and gross motor milestones in those where exact milestone dates were reported. Findings from the sensitivity analysis were then compared with the findings from the main analysis where exact and approximate milestone dates were included. For the linear regression models, the unstandardized beta coefficient can be interpreted as the mean difference in the gross motor milestone or AIMS outcome variable for each additional time a recommendation or the overall guidelines are met across the three points. For the logistic regression model, the odds ratio can be interpreted as the likelihood of achieving a higher AIMS stand score, compared to a lower score, for each additional time a recommendation or the overall guidelines are met across the three points.
Additional analyses were conducted to address objective 2 by calculating categorical adherence variables based on participants who consistently met or did not meet the recommendations across all three time points. These analyses were not conducted for overall guidelines adherence outcomes due to low adherence. Linear mixed models and linear and logistic regression models were repeated as described for the continuous adherence variables above. For all analyses, participants with observations for all variables of interest were included. Statistical significance was defined as p < 0.05 for all analyses.