Individual and day‐to‐day differences in domain‐specific physical activity of 10‐ to 11‐year‐old children in Denmark—Measured using GPS and accelerometry

Abstract

Objective

Physical activity (PA) and the achievement of 60 min of moderate‐to‐physical‐activity daily is declining in school‐aged‐children, and effective strategies to increase PA is needed. We aimed to examine the individual and day‐to‐day distribution of PA on schooldays among children aged 10–11 in 4 domains—school, home, transport, and other.

Methods

Data were collected from August to September 2020 using accelerometer and GPS data to measure daily PA‐levels and to locate in which domain PA occurs. Daily PA‐levels were assessed in each domain, and analyses of the individual and day‐to‐day differences in PA‐levels were calculated.

Results

The school domain contributed the most to children's daily MVPA with 47% of average MVPA, followed by the home domain with 26% of daily average MVPA, the other domain with 19% of daily average MVPA and the transport domain with 8% of daily average MVPA. Our results showed individual differences in where PA occurs, day‐to‐day differences in total MVPA and day‐to‐day differences in the MVPA‐levels across domains.

Conclusions

The school domain contributed the most to children's MVPA‐levels followed by the domains of home, other, and transport. Our study indicated that PA‐levels and the distribution of PA across domains differ from day‐to‐day. Future interventions should target more than one domain to accommodate these individual‐ and day‐to‐day differences in the goal of increasing PA‐levels and to reduce the decline in PA seen from childhood to adolescence.

1. INTRODUCTION

To encourage physical activity (PA), the World Health Organization (WHO) recommends that children and adolescents aged 5–17 should accumulate a minimum of 60 min daily of moderate‐to‐vigorous‐physical‐activity (MVPA). ¹ , ² It is shown that health benefits received from sufficient PA during childhood transfers into adulthood emphasizing the importance of effective strategies to increase PA. However, 81% of school‐aged adolescents worldwide do not achieve the recommended minutes of daily PA, ³ and this calls for effective strategies to increase PA in this age group.

School‐based interventions promoting PA for this age group have been conducted, but the effect is limited for some participants, especially girls, and older pupils. ⁴ , ⁵ , ⁶ , ⁷ In Denmark, a national policy implemented in 2014 mandates that schools are organized in such a way that children should on average achieve a minimum of 45 min of daily PA during school hours. Yet, only 58% of the school aged children met this threshold. ⁵ Another initiative enabling teachers to include PA in classroom teaching, has shown to be a feasible tool to increase boys' MVPA and girls' light‐physical‐activity (LPA), but did not increase PA in children with insufficient baseline PA. ⁴ Despite schools' potential to access and influence children, the newest research shows that the effects of school‐based PA‐interventions were not distributed equally across subpopulations. ⁶ For example, girls, older students, and students who were less active benefited less than boys, younger students and students with higher levels of baseline PA. ⁶ Interventions in sports clubs have likewise been investigated and have proven to be important for promoting PA, but the extent to which overall PA is increased through participating in sports club needs further investigation. ⁷ Knowledge of whether increased PA in one context (also called a domain) is related to a decrease in another context, or if more activity in one domain is likely to be associated with more activity in other domains needs to be further investigated. ⁸ It should also be subject of further investigation whether the increase or decrease of PA in one context/domains affects the total amount of daily PA.

Current research indicates that correlates of PA are specific to domains, ⁹ and the SLOTH model ¹⁰ describes four domains of PA (leisure‐time, occupation, transport, and home), as these locations constitute a central part of daily behavior and are related to individual PA. Physical behavior in domains can be investigated using device‐based measurements such as Global Positioning System (GPS) and accelerometry and several studies using these methods have been conducted in the past decade. ⁴ , ⁷ , ⁸ , ¹¹ , ¹² , ¹³ , ¹⁴ Although some knowledge exists on domain‐specific PA‐ levels, research in this area is still scarce. The potential of knowing the locations and levels of PA makes it easier to initiate interventions promoting PA and to target the right group. As children aged 10 and 11 (tweens) can be physically active in many ways and spend their time in different domains, interventions targeting one domain might only be relevant for some tweens. For tweens, studies have observed a steep decline in PA starting at age nine ¹⁵ making it even more important to understand in which domain PA is occurring and for which children.

Moreover, little research has been conducted on the reasons for the diverging results of PA‐interventions for children. ¹⁶ One possible explanation could be that PA is related to individual behavior, which differs across locations and from day‐to‐day. For example, it is likely that active transportation to school, which has shown to constitute a central part of daily PA‐levels, ¹⁷ can only be successful for those who live relatively close to school, but do not transport themselves actively already. Furthermore, individual PA differs from day‐to‐day, which is seen in a study where children appeared to compensate more PA on 1 day with more sedentary time (SED) on other days. ¹⁸ It seems that the expectation that school‐based interventions target children equally might not be true, but knowledge about individual and day‐to‐day differences occurring in PA‐patterns of children is scarce, limiting our ability to design and implement more effective interventions. Therefore, this study aims to examine the individual and day‐to‐day distribution of PA on schooldays among tweens aged 10–11 in four domains school, home, transport, and other. The knowledge of where tweens obtain daily PA and how it can differ individually and in a day‐to‐day perspective can help explain differences in PA distribution over different domains and provide important guidance for future interventions seeking to increase tweens' physical activity.

2. MATERIALS AND METHODS

2.1. Study setting and participants

This study included accelerometer and GPS data collected from August to September 2020. The Research Ethics Committee at the University of Southern Denmark (20/29520) and Legal Services, University of Southern Denmark (11.068) approved the study. Participants were tweens (children aged 10 and 11) attending 4th or 5th grade in the three public schools. The schools were selected based on factors known to influence PA ¹⁸ : geographical location, variation in outdoor PA opportunities, size of the school playground, and number of users. This study was a part of a larger study investigating PA behavior on different school playgrounds. Therefore, an elaboration on the selection criteria and in‐ and exclusion criteria for participants can be found in Amholt et al. ¹⁹ Researchers handed out GPS and accelerometer placed in an adjustable waist belt. Verbal and written instructions were provided on how to wear and remove the belt as well as how to charge the GPS. Participants were instructed to wear the belt on their hip during waking hours and to take off the belt when they were in contact with water. Two daily text‐message reminders were sent out to the children. One in the morning instructing them to put the belt on and one at night reminding them to charge the GPS. The children were asked to wear the belt for five school days.

The parents of participants were asked to consent to their child participating in the study, and after receiving parental consent, the participants were asked for their assent to participate. They were informed about the opportunity to withdraw from the study at any time. Participants and schools were anonymized and given an ID number prior to data analyses in this study.

2.2. Physical activity measurements

PA and location (domains) were assessed using the tri‐axial Axivity AX3‐accelerometer ²⁰ and the Qstarz BT‐Q1000XT GPS device. Accelerometer data were recorded at 100 hz and only data from the vertical axis were used. ²¹ The raw accelerometer data were converted to Actigraph activity counts using the package actiCounts in the statistical program R. ²² The GPS was set to measure locational data every 15 s. The good dynamic accuracy of the Qstarz device makes it commonly used in many health studies. ⁹ , ²³

2.3. GPS‐ and accelerometer protocol

The Human Activity Behavior Identification Tool and data Unification System (HABITUS) was used to aggregate and process accelerometer‐ and GPS data into 15 s epochs. ²⁴ PA‐levels were assigned using the Evenson cut points, which have been recommended to estimate physical intensities among children ²⁵ , ²⁶ and non‐wear time was defined as 60 min of continuous zeros. ²⁵ HABITUS identified invalid GPS‐data points using extreme speed (>130 km/h) or extreme changes in the distance (>1000 m) and elevation (>100 m) between two consecutive data points as criteria. ⁸ , ²⁷ The invalid data points were replaced with the last known valid point for up to 600 s. Trips were identified and categorized into walking, cycling or in a vehicle. If the 90th percentile for speed during a trip was below 10 km/h, a trip was classified as walking. From 10 km/h up to 35 km/h was classified as cycling, and over 35 km/h was classified as vehicle travel. A trip was defined as a continuous period of movement longer than 120 s with a length of a minimum of 100 m.

Stationary pauses of up to 180 s were considered as a pause in a trip and pauses longer than 180 s were considered as an ended trip. ⁸ GPS data were merged to accelerometer data based on the timestamps of each GPS point. ²³

Finally, each valid epoch was assigned to different domains using a PostgreSQL database combining the HABITUS dataset and location data from GIS. Domains were categorized according to the SLOTH‐model and included domains of school, home, transport, and others. The database was set up is such a way that each epoch was assigned to only one domain using a hierarchical decision tree.

1. Epochs within the participants' home (the location of the GPS signal at 3 am buffered with 100 m digitized in GIS) were assigned and aggregated as the home domain.

2. Epochs calculated within schoolgrounds (digitalized in GIS) were assigned and aggregated as the school domain.

3. Epochs classified as a trip were aggregated into the transport domain, which were later categorized into walking, cycling, and vehicle transport.

4. The remaining epochs not assigned to one of the three domains above were assigned and aggregated into the other domain.

Lastly, a PostgreSQL database was used to aggregate the context‐specific variables (in seconds) into wear time, time spent in each domain, and PA‐levels for all participants, per day. Information about gender and grades was added to the dataset. The final dataset could now be used for further descriptive analyses.

To exemplify the categorization of epochs into domains, Figure 1 was created in ArcGIS showing 1 day of data for one participant. Each dot is a GPS point representing 15 s, color‐coded for the four domains.

FIGURE 1.

Visualization of GPS points: GPS point of one participant divided into domains of school (blue), home (red), transport (yellow), and other (green).

2.4. Analysis of PA in schooldays, individual‐ and day‐to‐day differences

Valid participants were defined by having at least two valid schooldays of 10 h accelerometer wear time, which provides reliable estimates of daily PA. ²⁸ Descriptive statistics were calculated using frequency distribution of participants (n), and a total of valid days. The mean (range) of PA‐levels divided by gender and grades, and into domains (school, home, transport, and other) were calculated. Linear regression analyses were conducted to test differences of daily MVPA between genders and grades. Furthermore, analyses of the individual and day‐to‐day differences in PA‐levels were performed by detecting all participants who fulfilled the WHO recommendation of minimum 60 min of daily MVPA and from that list we chose five different participants who achieved the recommendations differently. The selection of participants was done with the criteria of finding one participant in each domain with high levels of MVPA and one participant who had MVPA across all domains. This case‐ based analysis was chosen with the aim of showing how differently MVPA can be distributed across domains for different children. Additionally, they were selected representing gender and grade differences.

3. RESULTS

A total of 436 participants were invited to participate in the study and 278 (63.76%) wore the devices. The participants were children in 4th and 5th grade, where the common age is 10–11 years of age. We merged GPS and accelerometer data from 261 participants and a total of 197 participants had sufficient valid data (at least 2 days with 10 h of accelerometer weartime per day) and were included in the analyses with a total of 578 valid school days (Table 1).

TABLE 1.

Overview of the participants divided by grade and gender.

			Total	4th grade	Girls	Boys	5th grade	Girls	Boys
N			197	111	55	56	86	51	35
Valid school days			578	334	168	166	244	147	97
Minutes per day
Total	Wear‐time	Mean	752.67	749.30	755.38	743.16	757.27	748.83	770.07
		Range	600.25–1440.50	602.75–1019.75	611.00–1019.75	602.75–884.25	600.25–1440.50	600.25–1255.25	600.25–1440.50
	MVPA	Mean	62.79	65.62	66.88	64.34	58.92	58.33	59.82
		Range	8.25–243.25	8.25–243.25	11.50–243.25	8.25–164.25	10.50–171.50	10.50–171.50	21.50–140.25
	LPA	Mean	318.80	321.19	322.60	319.77	315.53	309.36	324.88
		Range	162.50–583.50	162.50–524.25	168.50–524.25	162.50–448.50	163.75–583.50	190.25–453.00	163.75–583.50
	SED	Mean	371.08	362.50	365.90	359.05	382.82	381.14	385.37
		Range	159.00–820.00	159.25–605.25	183.25–595.50	159.25–605.25	159.00–820.00	159.00–820.00	180.75–782.00

Note: N is the total number of participants who met the inclusion criteria of two valid schooldays. Schooldays are the total number of at least two valid schooldays of 10 h, the average of valid days per participant was 2.9 days. In addition, the table shows the mean (range) daily wear time in minutes, and physical activity levels in minutes ‐moderate‐to‐vigorous‐physical‐activity (MVPA), light‐physical‐activity (LPA), and sedentary time (SED) in total and in domains.

3.1. Participants and PA‐levels

On average, the tweens spent 62.79 min in MVPA per day. The 4th graders spent significantly more average time on daily MVPA than 5th graders; 65.62 min versus 58.92 min (p = 0.010).

The average daily LPA was 318.80 min (5.3 h), and average daily sedentary was 371 min (6.1 h), where the home domain was the biggest contributor to sedentary behavior. Across all domains, 5th graders spent more time sedentary on average than 4th graders; 382.82 min versus 362.50 min (Table 1).

3.2. MVPA in the four domains

The share of daily MVPA in all four domains varied from 47% in the school domain, 26% in the home domain, 19% in the other domain, and 8% in the transport domain. The range of total daily MVPA varied from 8.25 to 243.25 min which seems a large range, indicating that there were differences between the days as well as between the participants. The same tendency was observed in each of the domains, which indicates a large variation between participants of how much MVPA was achieved in each of the four domains. (Table 2).

TABLE 2.

Physical activity levels in minutes calculated in mean (range) in total and in the four domains of school, home, transport, and other.

Domains		Total	School	Home	Transport	Other
Wear‐time	Mean	752.67	307.59	281.20	44.54	119.34
	Range	600.25–1440.50	0–613.0	0–1063.5	0–208.0	0–561.25
MVPA	Mean	62.79	28.36	15.98	4.84	11.32
	Range	8.25–243.25	0–116.25	0–97.00	0–40,0	0–91.0
LPA	Mean	318.80	147.66	94.77	21.84	38.90
	Range	162.50–583.50	0–310.25	0–97.00	0–310.25	0–91.0
SED	Mean	371.08	131.58	170.44	17.87	33.00
	Range	159.00–820.00	0–314.75	0–63.,5	0–98.0	0–356.5

The five selected tweens all met the recommended minimum of 60 min of MVPA for at least 1 day. Figure 2 illustrates that they gained their MVPA in very different ways. Tween A is gaining most of her MVPA in the home domain with 81.8% (94.3 min), followed by 16.9% (19.5 min) in the school domain. Tween B gains almost all of his MVPA in the school domain with 98% (108 min). Tween C gains daily MVPA minutes in all four domains with most MVPA in the transport domain 41.3% (26.3 min) followed by the other domain with 29.1% (18.5 min), the home domain 17.3% (11 min), and 12.1% (7.8 min) in the school domain. Tween D archives the PA recommendations primarily by MVPA in the other domain with 75% (68.3 min) and is further active in the transport domain with 20.9% (19 min). Finally, tween E distributes her minutes across all four domains, but is achieving most MVPA in the school domain with 46.5% (53 min), followed by the home domain with 21.3% (24.3 min) the other domain with 18.6% (21.3 min), the transport domain with 13.6% (15.5 min).

FIGURE 2.

Visualization of the percentage and absolute minutes (>1) of daily MVPA in five participants achieving the WHO‐recommendations of 60 min daily MVPA differently on a random school day in the four domains school, home, transport, and other.

Figure 3 shows the relative distribution in percentage and absolute minutes of daily MVPA levels for the five selected participants on their valid schooldays, meaning that the number of days can vary between two and four days. Tween A is still gathering most daily MVPA in the home domain and her data tells us that she is only achieving the recommendations of 60 min of MVPA daily on the first day displayed. Tween B fulfills the recommendations by gaining his daily MVPA in the school domain. Tween C shows to be more active in the school domain on the other days than displayed in Figure 2 but does not achieve the 60 min of daily MVPA the two following days. Tween D who was most active in the other domain in Figure 2 shows to be more active in the school domain when looking at the other days displayed. He only meets the recommendation the first 2 days. Tween E shows the same pattern of gaining her MVPA minutes daily spread across the domains. She is the only of the selected participants who meets the recommendations during all valid schooldays.

FIGURE 3.

Visualization of the percentage and absolute minutes (>1) of daily MVPA of the same five tweens showing their MVPA levels daily on weekdays in the four domains on all their valid schooldays.

4. DISCUSSION

This study investigated the distribution of PA on schooldays in tweens aged 10 and 11 in 4 domains quantifying the most important domains of MVPA using GPS and accelerometer measurements. We furthermore explored if there were individual‐ and day‐to‐day differences in tweens PA‐pattern, which is not well described in the existing literature. We found that the school domain was the biggest contributor to daily MVPA with 47% of daily average MVPA. We registered 26% of the daily average MVPA in the home domain, followed by the other domain with 19% of daily average MVPA, and the transport domain with 8% of daily average MVPA. Furthermore, our results show individual difference in where PA occurs, day‐to‐day differences in total MVPA, as well as day‐to‐day differences in the MVPA distribution across domains.

4.1. Physical activity in domains

On average, most daily MVPA was found in the school domain and interventions initiated in this domain are highly important as they might have a broad impact. However, the main focus of future interventions should be to target some of the least active tweens who we also observed in our results. Our findings, therefore support the Danish regulation of aiming at 45 min of MVPA in the school domain.

The home domain had the highest levels of SED‐behavior, which indicates that there is an obvious potential to increase PA in this domain. The past years, SED‐time has increased and studies attribute increased screen time after school, playing computer games, or using tablets. ²⁹ Previous research could indicate that other contexts such as family and siblings might also influence PA habits of children. ³⁰ Future research should examine which physical attributes the home domain consists of and how to promote PA in this domain as it appears unexplored.

In the other domain, we saw a very large individual differences, indicating that this domain was not necessarily represented in daily MVPA. Currently, we can only guess that this domain consists of, for example, visiting friends, participating in sports, etc. Our results could indicate that this domain has relatively little importance for achieving sufficient daily PA for some tweens, but is important for others. Although previous research has suggested that sports clubs are important for promoting PA, ⁷ this was not the case for all tweens in our study. There could be different explanations for this as our data do not explain the reason. One explanation could be a delayed or ongoing consequence of Covid, as many sports clubs were shut down during this period. Future research, should divide the other domain into subdomains to provide contexts for conducted activities, and explore if this domain could contribute to more PA for those with insufficient activity levels. Surprisingly, the transport domain did not play a central part of the tweens daily MVPA across schooldays in the current study. As cycling is known to be the predominant mode of travel for the Danish population, this result was unexpected. ³¹ In an earlier study, active transport to school accounted for approximately 20% of daily recommended MVPA, ³² more than twice what we found here, so this might be a domain where an intervention could be successful. A possible explanation of the low levels of MVPA in the transport domain could be due to geographical challenges with the tweens living too far away or does not have access to safe roads, why parents passive transport them, for example, in a vehicle. Another explanation could be due to underestimation of the cycling when using accelerometers. ³³

4.2. Individual differences

Our results illustrate that there are large individual differences in where tweens are active. A hypothesis could be that high daily MVPA in the school domain is compensated by lower daily MVPA in the other domains. A systematic review from Beck et al. found inconclusive results regarding daily potential compensations of PA‐levels across domains or during a time‐span in children and adolescents. In summary the synthesis of the included studies revealed a tendency for compensatory behavior in intervention studies, which was similar to our results. Furthermore, the study did not find indicators for compensatory behavior for children and adolescents who regularly participated in organized sports. ³⁴

Our results suggest that interventions only targeting one domain might not be effective for the whole target group. A more holistic approach including interventions in more than one domain is needed. Recently, a large whole‐system multicomponent intervention has been conducted in Denmark. It is based on the existing evidence and is developed in close collaboration with all actors surrounding the children, including the family, school, after‐school‐organization, municipality, supermarket etc. This could be a possible solution of intervention targeting the whole group of children across all domains. ³⁵ In general, from a public health preventive perspective, policymakers, practitioners, and evaluators should include more system thinking and system intervention approaches to identify, implement, and evaluate complex adaptive systems to, for example, increase the levels of MVPA in children. ³⁶ , ³⁷

4.3. Day‐to‐day differences

There were day‐to‐day differences for many tweens in their MVPA‐levels, meaning that they were more active 1 day compared to other days. A possible explanation to this could be that children compensate their PA‐levels and SED‐levels between days, ¹⁸ but it might also be due to differences in daily activities, such as PE or after school sports activities that do not happen every day. It was only tween E who achieved 60 min daily MVPA on all of her valid schooldays, which might be because of the distribution of MVPA across all domains. This could be a good example of why interventions should be initiated in more than one domain.

4.4. Strength and limitations

The use of GPS and accelerometer to access PA is not without limitations. ²¹ GPS signal loss can be misclassified inside buildings or in densely populated cities. ²³ Accelerometers underestimate PA when cycling, which most likely influenced our results. ³³ Adjustments for this could have been made, and adjustments could have increased levels of MVPA with up to 6 min per day. ³³ The included schools were not randomly selected, and the result may not be representative of the population and therefore limiting the generalizability of our findings. The inclusion criteria were based on the school playgrounds having facilities intended for tweens, which might have affected their behavior positively and caused them being more active in the school domain. We only had information on gender and grade which limited our possibilities for analyzing other cofounders that could explain individual and group level differences across days and domains. Data were collected during the summer season and the results are not comparable to winter months as weather and season are known to influence behavior. ³⁸ It is worth mentioning that a few months before the data collection process, Denmark was closed due to the COVID‐19 pandemic. Just before the data collection, the children were able to go back to school again, and this can possibly have affected their behavior.

One of the strengths of this study is this new way of presenting the data. It provides new insight which often can be overlooked when working with statistical analysis. Our data presentation sheds light on what the big challenges of health promotion really are. By presenting individual and differently distributed data, we highlight the large individual differences in PA behavior. This can add a new angle to the discussion on how to target children's PA behavior in future health promoting interventions. This study is an observational cross‐sectional study and cannot provide evidence of causality. Regardless, the result from this study can be used to generate hypotheses that can be tested in future intervention studies investigating PA across domains.

5. PERSPECTIVES

The decline we found in daily MVPA between 4th and 5th graders witnesses of the challenges that we are facing when trying to increase or maintain PA as children grow older. The promotion of PA should target some of the least active children, whom we also observed in our study having large individual differences and day‐to‐day differences. Future research should investigate the individual and day‐to‐day differences in more detail as our study indicates that this could be important for the effect of PA‐interventions. Interventions should focus on initiating efforts across different domains to target the most inactive children. Additionally, future interventions should try to suite both an upstream and a downstream approach, as we addressed in this study. Interventions trying to increase PA levels for all are often upstream initiatives from policy, for example, changing Danish school policy so that children on average should achieve a minimum of 45 min of daily PA during school hours, but in contrast downstream approaches focusing on individual levels of PA might target those more in risk, which is needed to raise the overall levels of PA.

The results of this study can help explain differences in PA distribution and provide guidance for researchers, politicians, schools and sports organizations seeking to increase PA in this age‐group. This study can draw attention to the diversity of PA seen in children and should be considered as a point of attention in the global political debate.

6. CONCLUSION

This study successfully examined the individual and day‐to‐day distribution of PA on schooldays among tweens aged 10–11 in the four domains school, home, transport, and others. Tweens achieved the largest proportions of daily MVPA in the school domain, followed by the home domain, the other domain, and the transport domain. Our study indicated that PA‐levels and the distribution of PA across domains differ from day‐to‐day. Future interventions should target more than one domain to accommodate these individual‐ and day‐to‐day differences in the goal of increased PA‐levels and to reduce the decline in PA from childhood to adolescence.

FUNDING INFORMATION

This work was supported by Innovation Fund Denmark [0153‐00012B].

CONFLICT OF INTEREST STATEMENT

Thea Toft Amholt was employed by KOMPAN A/S. The other authors declare that there is no conflict of interest.

Stage A, Amholt TT, Schipperijn J. Individual and day‐to‐day differences in domain‐specific physical activity of 10‐ to 11‐year‐old children in Denmark—Measured using GPS and accelerometry. Scand J Med Sci Sports. 2024;34:e14631. doi: 10.1111/sms.14631

DATA AVAILABILITY STATEMENT

Research data are not shared.