In this longitudinal study, trajectories of total and intensity-specific PA and SED were examined in Norwegian children aged 3 to 9 years. Boys generally had more favourable PA levels and trajectories than girls, including a steeper increase and a later plateau and decline in PA of moderate and vigorous intensities. Trajectories were generally similar across weekdays and weekend days but differed for preschool/school hours and after school hours. Epoch length had a major influence on levels and trajectories for all PA intensities and SED. Being one of a small number of studies to follow a relatively large sample of young children from the preschool to the school setting, these findings provide further understanding of how PA develops with age across the transition from preschool to school, where prior studies suggest the well-known decline in PA during childhood begins.
While evidence generally support an increase in SED from the age of 3 onwards [5, 11, 12, 16, 20], the evidence on trajectories of PA across childhood is equivocal [5, 7, 8, 10, 12,13,14, 17, 18, 25, 34]. Part of this uncertainty results from the use of different accelerometer data reduction methodologies across studies [7, 21, 23]. Most studies show an increase in PA with age during preschool years (3–4/5 years of age) [5, 9, 16,17,18] and a decline during school years (from age 5/6) [5, 7, 12,13,14,15, 19]. The evidence on the timing of the peak PA level as determined by longitudinal studies is, however, conflicting and varies from 3 to 11 years depending on sex and intensity [7,8,9,10,11,12,13]. Few studies have followed large samples of young children over several years capturing the transition from preschool to school. One study that captured trajectories with multiple measurements across this transition, showed a U-shaped development from ages 3 to 7 among 242 children from New Zealand [8], which is contrary to trajectories described by the present study and the prevailing longitudinal evidence [7, 9, 10, 12, 13]. Our findings are consistent with most previous studies [7, 10, 12, 13], showing that PA levels (total day) peak at 3 to 8 years depending on sex, intensity, and, to some extent, epoch length. While LPA peaks at age 3–4 and total PA level (cpm) peaks at age 5 for both boys and girls, both MPA, VPA, and MVPA appear to peak at age 7 in girls and age 8 in boys in the present study. Interestingly, the peaks for moderate and vigorous intensities are considerably more pronounced for a 60-s epoch, for which we found clear increases in PA levels from the age of 3, than for a 1-s epoch, for which we found minor changes after age 4 in girls and age 5 in boys. SED increases from age 3 in girls and age 4 in boys. Later peaks for boys than for girls and for VPA than for MPA support previous longitudinal evidence [7, 13].
Our findings show smaller declines with age than Norwegian surveillance studies [19] and international data from the ICAD [5, 14]. Based on representative samples of 4406 Norwegian 6- and 9-year-olds in 2011 and 2018, Steene-Johannessen [19] showed significant declines for both total PA (100–132 cpm) and MVPA (8–12 min/day) between children aged 6 and 9 using a 10-s epoch. For comparison, we found smaller declines for total PA (< 94 cpm), while MVPA declined by 5 min/day using a 1-s epoch and increased by 3–6 min/day using a 60-s epoch. These findings further contrast findings from the ICAD, showing declined MVPA [5] and VPA [14] from the age of 5–6 years based on a 60-s epoch, though MVPA declined less than 1% until age 9–10. We have no clear explanation for the apparent discrepancies across studies and study designs beyond inherent variation among populations, possible methodological differences in handling accelerometry data, and possible secular trends influencing the comparison of our data collected 2015–2019 with older ICAD data.
While ICAD data suggest the transition from the preschool to the more formal school setting had a negative influence on PA trajectories, our findings, consistent with other longitudinal studies [10, 12, 13], suggest that PA of moderate and vigorous intensities do not decline when enrolled in school. However, previous longitudinal studies have not provided analyses for separate settings (i.e., preschool/school hours versus after school hours) across this transition. Our findings show that preschools and schools are important environments for children’s movement opportunities, given the higher PA levels during weekdays than weekend days and during preschool/school hours than the after school hours on weekdays, as also shown previously for the larger PRESPAS sample [38]. We found that MVPA increased by 15 min/day in boys and 10 min/day in girls on weekdays, compared to 2 min/day in boys and 0 min/day in girls on weekend days from age 3 to 9, reinforcing that children have the most positive trajectories over this age span on weekdays. This finding is consistent with the findings by the meta-analysis by Farooq et al. [7] that show a more pronounced decline in MVPA with age during weekends than weekdays. A closer look at trajectories during preschool/school hours showed that being introduced to formal schooling seems to affect PA levels negatively. We observed earlier peaks in MVPA (age 5 in both boys and girls) during preschool/school hours than for the total day (age 7–8). MVPA decreased by 4–5 min/day during school time but increased by 1 min/day during after school hours from ages 6 to 9. The positive trajectories during after school hours compared to preschool/school hours is, however, explained by the increased wear time of 66–67 min/day during after school hours, contrasting the stable (and restricted) wear time during preschool/school hours. Thus, when expressed as the proportion of wear time, MVPA levels are stable in both boys (8.9–9.2%) and girls (8.3–8.1%) from ages 3 to 9 during after school hours. Given that decreased sleep time [36] and thus increased opportunities for movement is a natural part of a child’s development, we did not adjust for wear time in our analyses.
Consistent with previous studies [5, 7, 16, 19, 37], we found that boys had more positive PA trajectories than girls from a young age, had a later peak for moderate- and vigorous intensities, and were more active throughout the follow-up period. Of particular interest, boys appeared to have more pronounced positive MVPA trajectories than girls during preschool/school hours (increase of 14 min/day for boys versus 8 min/day for girls from ages 3 to 5 and 8 min/day for boys versus 3 min/day for girls from ages 3 to 9, respectively) compared with after school hours (increase of 5 min/day for boys versus 4 min/day for girls from age 3 to 5 and 7 min/day for boys versus 5 min/day for girls from age 3 to 9, respectively). These findings support those by Nilsen et al. [38], showing that boys, older children, and highly active children benefit most from the preschool environment with regards to MVPA. This pattern likely results from different play preferences at preschool [38, 39], but also suggests the preschool movement environment and pedagogical approach may suit boys better than girls and that initiatives specifically should address girls’ movement needs and preferences. Thus, preschool interventions should be designed to specifically promote PA in girls by designing PA programs meeting girls’ preferences and by raising the awareness of these gendered activity patterns by preschool staff through professional development. Such early initiatives could hopefully have immediate and long-lasting effects on PA levels in girls throughout childhood.
It is well known that the epoch length is fundamental for the capture of intensity-specific PA using accelerometry in children due to children’s sporadic and intermittent bursts of activity [23, 24, 40]. Summation of PA over longer periods leads to underestimation of time spent in the lower (SED) and the higher (VPA, MVPA) end of the intensity spectrum as compared to summation over shorter periods [23, 41, 42]. This effect clearly suggests that a short epoch is needed to capture children’s time spent across the intensity spectrum correctly, as supported by studies showing that a short epoch can capture information that is lost when applying longer epoch length [23, 43]. To the best of our knowledge, the present study is the first to compare PA trajectories in childhood using two different epoch lengths. Our findings show not only differences in SED and PA levels across epoch lengths, but also different trajectories across epochs for SED and all PA intensities. The most pronounced differences were found for MPA and MVPA, where steeper increases and larger differences over time were found for a 60-s (increase of 25 and 33 min/day from age 3 to 7, respectively) than a 1-s (increase of 5 and 15 min/day from age 3 to 7, respectively) epoch length, expectedly resulting from children participating in less sporadic and more continuous PA patterns with older age. This is consistent with more favourable changes in 1-min bouts of MPA than for total MPA (15-s epoch) from age 4.6 to 10.6 in a large Australian sample [44], and may be related to increased participation in organized sport with older age [45]. Given that previous longitudinal studies including multiple timepoints have applied 15- [8, 10, 15] or 60-s [12, 13] epoch lengths, the ICAD [5, 14] applies a 60-s epoch and the meta-analysis by Farooq [7] included studies with various epoch lengths, different epoch lengths is clearly a candidate for inconsistency in findings across studies. Consistent with previous studies showing that a 1-s epoch can capture information of relevance for metabolic health that 10- or 60-s epochs cannot [23, 43], we suggest that using a 1-s epoch in children is appropriate to more accurately capture time spent across the PA intensity spectrum.
We found that tracking coefficients (ICCs) across the 5 timepoints ranged from 0.47 to 0.59 for the total day. These estimates are in line with a previous systematic review [46] and a recent Australian study with 3 follow-up timepoints over 6 years [44]. Thus, our findings add to the literature showing moderate stability of PA and SED from early to mid childhood, reinforcing the importance of efforts to establish favourable activity levels during the early years to support long-term healthy movement behaviours. Interestingly, the ICCs observed over 5–6 years of follow-up in the present study and the study by Downing et al. [44] are similar to ICCs previously shown over multiple timepoints within a year in children [29, 47, 48]. This similarity of ICCs over the short and long term, raise the question what should be considered measurement error and actual behaviour change in assessment and understanding of habitual PA levels. If variation is similar over weeks or months (which principally could be considered measurement error) and years (which principally could be considered actual behaviour change), it could be argued that stability in PA over time is higher than previously believed, given that much of the variability over the long term could be considered random fluctuations and thus measurement error. The somewhat lower stability for weekdays, weekend days, preschool/school hours, and after school hours as compared to PA over the total day, probably reflect shorter monitoring periods (hours/day and/or days/week) and thus more variability. Particularly for weekend days, for which the lowest ICCs were found, this finding is expected and consistent with a previous study in children [49], given the inclusion of only 1–4 days of monitoring for each timepoint.
Strengths and limitations
The main strength of this study was the inclusion of a relatively large sample of children who was followed annually for 5 years to establish PA trajectories for children aged 3 to 9 years. The study was specifically designed to capture the transition from preschool to school, having multiple timepoints of assessment both prior to and after the start of school to provide detailed evidence of shifts in activity levels across various settings. Children wore accelerometers for 14 consecutive days during the same period each year (autumn) to minimize variability in assessments, though such an extended monitoring period (i.e., ≥ 7 days) may be of minor importance for reliability of accelerometer-determined PA [29].
Though we invited all children from the included preschools in 3 municipalities that participated in PRESPAS [16] and the response rate was high, our findings are limited by not being a representative sample of children from a larger geographical area and with a more diverse sociodemographic background. We also found a higher parental education level in the follow-up sample compared to the larger PRESPAS sample. The lack of statistical significance testing of changes in PA and SED with age (i.e., 6 versus 5 years, 5 versus 4 years, etc.) might be considered a limitation, since we were not able to conclude with respect to whether differences between specific age groups were statistically significant. However, we did not put forward any hypotheses about specific changes between age groups, leaving such testing less meaningful. Thus, we have described trajectories and tested overall trends with age, including possible interactions between age and sex, type of day, time of day, and epoch length. We used a waking accelerometer monitoring protocol. Wear time was restricted to the hours between 06:00 and 23:59 to remove invalid wear time if children wore the accelerometer overnight. Given the age of the children was from 3 to 9 years, we believe that this restriction is appropriate to balance the risk of including invalid wear time if wearing the accelerometer during the night and the risk of missing wear time during the daytime. Still, this removal may have underestimated PA and/or SED for some children if they went to bed very late or got up very early in the morning. In general, little agreement exists on the most approrpiate data reduction protocols of accelerometry data in children (21) and alternative choices, beyond the influence of epoch length as shown herein, could affect our findings. PA may also be underestimated because accelerometers are not able to correctly capture activities like bicycling, swimming, and movements involving the upper body.