Describing objectively measured physical activity levels, patterns, and correlates in a cross sectional sample of infants and toddlers from South Africa

Background: Physical activity is considered to have health benefits across the lifespan but levels, patterns, and correlates have not been well described in infants and toddlers under the age of two years. Methods: This study aimed to describe objectively and subjectively measured physical activity in a group of South African infants aged 3- to 24-months (n=140), and to investigate individual and maternal correlates of physical activity in this sample. Infants’ physical activity was measured using an Axivity AX3 wrist-worn accelerometer for one week and the mean vector magnitude was calculated. In addition, mothers reported the average amount of time their infant spent in various types of activities (including in front of the TV), their beliefs about infants’ physical activity, access to equipment in the home environment, and ages of motor development milestone attainment. Analysis of variance (ANOVA) and pair-wise correlations were used to test age and sex differences and associations with potential correlates. Results: There were significant age and sex effects on the distribution of time spent at different physical activity intensities (Wilks’ lambda=0.06, p<0.01). In all cases, the trend was for boys to spend more time in higher intensity physical activity and less time in lower intensity activity than girls; and for time spent in higher intensity activities to be higher in older children. Time spent outside was higher in boys, and this reached significance at 18-months (F=3.84, p=0.02). Less concern around floor play was associated with higher physical activity at 12-months in females only (p=0.03, r=0.54), and no other maternal beliefs were correlated with physical activity. The majority (94%) of children were exceeding TV time recommendations. When controlling for age and sex, overall TV time was positively associated with BMI z-score (β=0.01, p=0.05). Conclusion: This study is the first to show sex and age differences in the patterns of physical activity, and to report on objectively measured and maternal reported physical activity and sedentary behaviour in the first two years of life in South Africa infants. Infants and toddlers should be provided with as many opportunities to be active through play as possible, and TV time should be limited. Participants (convenience sample of 152 mother and child pairs) were recruited from the Chris Hani Baragwanath Academic Hospital, MRC/Wits Developmental Pathways for Health Research Unit in Soweto, South Africa. Soweto is over 200km2 in size with over 1.2 million inhabitants. Fathers in Soweto often have little to no involvement in childcare [22], and therefore only mothers were considered in this study. In most cases, mothers had been recruited from the Chris Hani Baragwanath Academic Hospital (which is the largest public hospital in South Africa and is thus likely to provide a sample of mothers representative of the Soweto community) for participation in a pregnancy and infancy study (M130905) at the research unit, and were invited telephonically to participate in the current study if their child was within 2 weeks of 3-, 6-, 12-, 18-, or 24-months of age (information obtained via pregnancy study birth records). For the purposes of this paper, children of all ages from this sample will be referred to as “infants”. Participants were excluded if infants had any diagnosed developmental abnormalities that may affect normal movement or development. Mothers were required to read and sign informed consent and assent documents for themselves and on behalf of their infant, and were free to withdraw from the study at any time. Ethical approval for this study was provided by the University of the Witwatersrand Human Research Ethics Committee (M150632). Participants were required to visit the research unit twice. At the first visit all demographic information was collected, questionnaires completed, measurements taken, and infants were fitted with an accelerometer and mothers were provided with a sleep diary to complete for their infant. At the second visit, one week later, participants were required to return the accelerometers and sleep diaries. If participants were unable to return the devices, researchers collected the devices from participants’ homes. Infants’ physical activity was objectively measured using an Axivity AX3 wrist worn accelerometer (Axivity Ltd., Newcastle-upon-Tyne, UK). These monitors allow for transparent data processing analysis due to the open-source software platform and unforced sampling of raw measurement data [23, 24]. Furthermore, they are small enough to be worn on the wrist of infants from 3 months. Monitors were initialised (using the open source OmGui software, Open Movement, UK) to capture triaxial acceleration data at 100 Hz with a dynamic range of +/−8 g. The monitors were worn in a specially designed fabric band (Open Lab, Newcastle, UK), which is waterproof, breathable, flexible and safe. This band was designed and piloted by the authors and Open Lab, and has been successfully used and tested for feasibility in a South African sample of infants (manuscript under review [25]). Mothers were instructed not to remove the band from their infant’s wrist for a seven-day period. Following download (OmGui, Open Movement, UK), the raw acceleration data were autocalibrated to gravity, using methods described elsewhere [26]; and vector magnitude (expressed in mg) was calculated representing the magnitude of acceleration over the summarised time period [23]. A high-pass (0.2 Hz) and low-pass (20 Hz) frequency filter was then applied to the data in order to remove gravity, as well as high frequency noise [27]. Thus, the remaining data provided an approximation of acceleration due to human movement alone. Thereafter, non-wear was identified based on the deviation of each axis below 13 mg for >1 h, and removed [23]. Remaining data was summarised to generate average acceleration (mg), as well as time (hourly and weekly) spent in intensity thresholds (10 mg increments from 0 to 400 mg). During summation, non-wear periods were taken into account in order to minimize potential diurnal bias [28]. At least 45 h of wear time (equating to three days of at least 15 h data per day [29]) spread equally across the 24-h day (at least 4 h each day during morning, noon, afternoon, and night) were required for inclusion in the analysis. Mothers were asked to report their date of birth and their infants’ date of birth and sex. Infant age was calculated and categorised to the nearest month. Mothers’ height was taken to the nearest 1 mm using a wall-mounted stadiometer (Holtain, UK). Infant length was measured to the nearest 1 mm using an infantometer (Chasmors Ltd., UK). Weight of mothers and infants was measured to the nearest 0.1 kg using a digital scale (Dismed, USA). All anthropometry measurements were taken twice, by trained research staff according to standardised procedures, and the average of the two values was reported. Mother and infant body mass index (BMI) was calculated as (weight(kg)/height(m)2). Infant weights, lengths and BMIs were converted to age-specific z scores according to the 2006/2007 World Health Organisation (WHO) growth standards [30] using the WHO Anthro software [31]. Mothers were asked to report how many hours per day they looked after their infant, and were considered primary caregivers if they reported looking after their infant for 8 or more hours per day. All questionnaires used in this study were based on previously developed and reliability tested items [19], which were then adapted for language and setting based on pilot work conducted on mothers from Soweto. These adaptations mainly consisted of rewording sections that were not well understood in a South African context (i.e.: “…a baby that is very active” was changed to “…a baby that moves a lot”). Infant activities were assessed at 3-, 6-, 12-, and 18-months. Mothers reported the amount of time (minutes) their infant spent in various activities, on an average day. Activities were age-specific, but included tummy time (only 3- and 6-months), time spent playing games with an adult, time spent being physically active with the mother, time spent on the floor, time spent with other babies of a similar age, time spent with older toddlers or children, time spent outside, time spent in front of a television (TV), time spent strapped to the mother’s back (only 3-, 6- and 12-months), and time spent restrained (in a highchair, car chair, pushchair etc). Activities were pooled to determine total time restrained (i.e. sum of time spent strapped to mother’s back, time spent in a highchair, car chair, pushchair, or cot), and total time free to play (i.e. sum of tummy time, time spent playing games with an adult, time spent being physically active with mother, time spent on the floor, time spent with other babies of a similar age, time spent with older toddlers or children, time spent outside). All activities are presented as minutes per day. Maternal reported time that infants spent in front of a TV was extracted from this questionnaire and used as proxy for TV time in order to examine potential correlates thereof. During the measurement period, mothers were asked to complete sleep diaries recording the time at which their infant was put to bed at night, and the time at which they were picked up from bed in the morning in order to provide an estimate of time in bed. If sleep diary data were missing, values were imputed based on trends for other infants’ days and times. Maternal beliefs about their infant’s physical activity, and attitudes and intentions around their infant’s physical activity and TV viewing, were assessed at 6- and 12-months using a 24-item questionnaire with a 4-point likert-type scale and response options: 1 = strongly agree to 4 = strongly disagree. Based on previous studies [19], 7 factor variables were created, including physical activity knowledge (7 questions examining the importance of physical activity for infant health and development), positive views about physically active children (3 questions examining maternal perceptions of active children), negative views about physically active children (4 questions examining maternal fears and concerns about physically active children), physical activity optimism (3 questions examining the anticipated ease of engaging children in physical activity), self-efficacy for promoting physical activity (3 questions examining mothers’ confidence for promoting physical activity), future expectations around infant’s physical activity and TV viewing (2 questions examining maternal expectations of infant’s future physical activity and TV viewing behaviours), and floor play concerns (2 questions examining perceptions of safety of floor play). Scores for each factor variable were then generated by averaging the item scores within each factor, where a higher score indicates lower agreement with the factor variable. The internal consistency in the present sample (mean Cronbach alpha = 0.60) for these factor variables was acceptable and comparable to previous studies [19], with only one factor variable (positive views about physical activity) scoring below 0.60. Mothers of infants aged 3-, 6-, 12-, and 18-months were asked to report, using a 4 point likert-type scale, whether their child had access to, or was likely to have access to age-specific toys and equipment (such as balls, push toys, bicycles, etc) in the home or at nearby facilities within the next year. Each response was coded as “3” if mothers responded that they had the equipment already, “2” if they were very likely to get it within the next year, “1” if mothers responded they would possibly get the equipment in the next year, and “0” if mothers responded that they were unlikely to get the equipment within the next year. Thereafter a sum of the responses for each item was created, with a higher score indicating better access to equipment or intention to provide access, and a lower score indicating no access and/or less intention to provide access. Mothers of infants aged 3-, 6-, 12-, and 18-months were asked to report whether their infant had attained specific motor development milestones according to the age of the child (based on the CDC recommendations: www.cdc.gov/ActEarly). Milestones included: holding up head at 3-months; sitting, crawling, and rolling to both sides at 6-months; crawling, pulling up to stand, and starting to walk at 12-months; and walking and running at 18-months. Milestones were recorded as attained/not attained. Thereafter, infants were categorised as either not yet mobile (this assumed all 3-month infants, as well as those aged 12-months who were not yet crawling); crawling (if 18-month infants were not yet walking/running they were presumed to be crawling); or walking/running (all infants aged 24-months were presumed to be walking/running). Assumptions were based on WHO reported ages for milestone attainment [32]. All statistical analyses were conducted using Stata 13 for Mac. Participant characteristics and all questionnaire data were summarised by age and presented as mean (SD), n (%), or median (IQR). Since there is no consensus on intensity thresholds for sedentary, light, moderate, and vigorous activities in non-ambulatory infants (which made up the majority of the sample), objectively measured infant physical activity intensity distribution was presented as time spent in each systemic acceleration category from 0 to 400 mg by age and sex. Furthermore, hourly mean vector magnitude (diurnal patterns) of activity were compared between ages, sex, and developmental stages. These differences were tested using multivariate analysis of variance (ANOVA) through Wilks’ lambda test. Two-way ANOVAs, linear regressions, and students’ unpaired t-tests were used to test age and sex differences in the time spent in various activities as reported by mother, as well as for objectively measured physical activity. Thereafter, pairwise correlations were run by age for maternal variables (home environment, maternal BMI and age, maternal-reported infant activities, and maternal beliefs factor variables) on infant objectively measured physical activity. Regressions were also run for potential correlates of maternal reported infant TV time controlling for age and sex, and margins were calculated. Chi-squared tests were used to determine the proportion of infants meeting TV time guidelines according to developmental stage. A p < 0.05 was considered significant in all cases.