Objectively measured physical activity among treatment seeking children and adolescents with severe obesity and normal weight peers

Abstract Background Treatment seeking children and adolescents with severe obesity often experience barriers to physical activity. Studies objectively measuring physical activity in this group and investigating explanatory factors for physical activity levels could inform clinical practice. Objectives This study aimed to compare objectively measured physical activity levels among treatment seeking children and adolescents with severe obesity and normal weight peers, and to investigate explanatory factors for time spent in moderate physical activity and vigorous physical activity among children and adolescents with severe obesity. Methods Children with severe obesity (n = 85) were matched 1:1 by age, gender, and the season for accelerometer measurements with normal weight peers (n = 85). Children wore accelerometers for seven consecutive days, yielding measures of physical activity, sleep duration and timing. Parents reported on screen time, parental body mass index and participation in organized sports. Results Children and adolescents with severe obesity spent significantly less time in moderate physical activity (12 min, p < 0.001) and vigorous physical activity (21 min, p < 0.001) per day compared to normal weight peers. No difference for time spent in sedentary activity was found between groups. For participants with severe obesity, age ≤12 years (p = 0.009) and participation in organized sports (p = 0.023) were related to more moderate physical activity, while age ≤12 years (p = 0.038) and early sleep timing (p = 0.019) were related to more vigorous physical activity. Conclusion Children and adolescents with severe obesity were less physically active than their normal weight peers. Factors related to more moderate and vigorous physical activity in children with severe obesity were lower age, participation in organized sports and earlier sleep timing.


| INTRODUCTION
Modification of physical activity behavior is a key component in lifestyle interventions for childhood obesity. 1 However, sustainable, healthy changes in activity levels are seldom achieved. 2 Increased understanding of physical activity behavior and explanatory factors for activity levels in treatment seeking children with severe obesity is an important next step to improve the physical activity component in lifestyle treatment programs currently delivered to this patient group.
National guidelines for physical activity commonly recommend 60 min of moderate-to-vigorous (MVPA) physical activity as a daily minimum for school-aged children and adolescents. 3 In a study of Norwegian 9-year-old children, 72.1% of the girls and 89.2% of the boys with normal weight reached these recommendations, while for children with overweight the numbers were 61.0% and 70.2, respectively. 4 These numbers are in agreement with a majority of previous studies concluding that children with obesity are less physically active than normal weight children. [5][6][7] In addition, emerging differences in MVPA from 6 years of age to middle adolescence (ages 14-17) are reported. 8 However, studies reporting on differences in physical activity levels amongst children with obesity and normal weight are mainly based on community samples, with few studies conducted with a clinical sample of children and adolescents with severe obesity. 9 This treatment seeking group of children may have more and other barriers for physical activity than children with less severe forms of obesity. Further, more studies using objective measures of physical activity are still needed, as self-reported data on activity levels often are found to overreport time spent in MVPA. 10 Explanatory factors for the identified difference in physical activity levels between children with normal weight and obesity are not fully understood. Several explanatory factors for time spent in MVPA for children across all weight categories have been put forward during the last decades. Level of physical activity have often been investigated in relationship to screen time and sleep duration, with mixed findings. [11][12][13][14][15][16] Some studies find a relationship between low levels of physical activity and more screen time, 14,15 while others find no association. 16 For sleep duration the findings are similar, with some studies reporting an association between short sleep duration and decreased physical activity levels and others no association. [11][12][13][17][18][19] Interestingly, later timing of sleep has recently been suggested as more strongly linked to lower physical activity levels in children and adolescents than sleep duration. 20, 21 Olds et al. 22 found late sleep and rise time in adolescents to be associated with more screen time, less MVPA and increased obesity independent of sleep duration. Another suggested explanatory factor for time spent in physical activity is participation in sports, 23,24 with children engaged in organized sports being more physical active. 23 Nevertheless, more investigation of behavioral, environmental, interpersonal and individual factors for time spent in physical activity among children and adolescents with obesity is needed. This could be particularly informative for clinical practice related to children with severe obesity that often report more barriers to physical activity. 5 The aim of the present study was therefore to compare objectively measured physical activity levels among treatment seeking children and adolescent with severe obesity and a matched group of normal weight peers. Further, to investigate explanatory factors (age, sex, sleep timing and duration, screen time, parental body mass index (BMI; kg/m 2 ) and participation in organized sports), for time spent in moderate physical activity (MPA) and vigorous physical activity (VPA) among children and adolescent with severe obesity. The hypotheses were that children and adolescents with severe obesity spent less time in physical activity compared to normal weight children and adolescents, and that increased age, female gender, more screen time, as well as shorter sleep and later sleep timing would be associated with less MPA and VPA.

| Study design and participants
In this case-control study 170 (100 girls) children and adolescents between 5.8 and 17.1 years were included. Participants aged ≤12 years were defined as children (n = 78) and participants aged >12 years as adolescents (n = 92). Eighty-five children and adolescents with severe obesity participated prior to attending a familybased behavioral treatment of childhood and obesity (FABO) program. 25 The FABO-study was conducted at a specialist department for treatment of severe pediatric obesity at the Obesity Outpatient Clinic, Haukeland University Hospital, Bergen, Norway. The department has catchment area responsibilities for all children and adolescents with severe obesity that are entitled to publicly financed treatment. The inclusion criteria were BMI above the International Obesity Task Force (IOTF) cut-off ≥ 35, or BMI ≥ IOTF 30 26 with obesity related comorbidity (e.g., psychosocial problems or emergence of cardio-metabolic risk factors). 25 Participants were excluded if they were enrolled in any other treatment targeting weight reduction or had severe somatic or psychiatric illness that could affect full participation in the FABO program. All referred children fitting these criteria and living within an hour drive from the hospital were consecutively invited to participate in the FABO-study and the 85 first were included in this study.

-
The comparison group consisted of 85 children and adolescents with normal weight (BMI ≤ IOTF 25) recruited from randomly selected schools in Bergen municipality. Stratified random sampling ensured that the obesity and normal weight group were matched 1:1 on age, sex, and season for data collection (April-September vs. October-March). Seasonal matching was conducted to prevent bias in the comparison on physical activity data, due to large seasonal differences in weather and hours of daylight in Norway. Data were collected over a period of 4 years (2014-2018).
Participation in the study was voluntary. An informed written consent was obtained prior to inclusion in the study. For the participating children with obesity, other possible treatment programs were discussed before inclusion. The study was approved by the Norwegian Regional Committee for Medical and Health Research Ethics (2013/1300) and the treatment study registered at http:// clinicaltrials.gov (NCT02687516).

| Measures
For the group of children and adolescents with obesity, baseline measurements from the FABO-study were used in the analyses. 25 The primary outcome in the current study was accelerometer measured physical activity, and included explanatory factors were anthropometric measures, sleep measures, screen time and participation in organized sports and demographic information. 25

| Demographic information
The parents/carers self-reported on a questionnaire about demographic information, which included information about family structure, parental education level and weight status, child participation in organized sports and daily screen time. The questions "Are both parents living together?" and "Does the child live together with siblings?" were used to evaluate family structure. Parental education level was categorized as either low (≤3 years after secondary school) or high (>4 years after secondary school). Participation in organized sports was reported as "1 = yes" or "2 = no" and daily screen time was rated on a scale (1 = never, 2 = less than 30 min, 3 = ½-1 h, 4 = 2-3 h, 5 = 3-4 h, 6=>4 h of screen time). The questionnaire was completed at the Obesity Outpatient Clinic by the parents/carers of the participants with severe obesity and sent/and returned by mail to the parents of the normal weight participants.

| Anthropometric measures
Height and weight were measured by trained personnel at the Obesity Outpatient clinic for the children with obesity and by a trained physiotherapist at the school nurse's office for the normal weight children. The participants were wearing light indoor clothing and no shoes during measurements. For the children and adolescents with severe obesity, InBody 720 (Biospace) was used to measure weight to the nearest 0.1 kg and height was measured by using a wall-mounted electronic stadiometer (Seca 264, Seca) to the nearest 0.1 cm. For the normal weight children and adolescents, Harpeden portable stadiometer (Crosswell) was used to measure height to the nearest 0.1 cm and weight was measured using a calibrated Seca persona digital scale (Hamburg, Germany) to the nearest 0.1 kg. BMI was calculated using the formula kg/m 2 and converted to standard deviation scores (SDS) by adjusting for age and sex 27 using Norwegian growth references. 28

| Physical activity measures
Physical activity was measured using Actiwatch 2 (Philips, Respironics, BEND, OR) for seven consecutive days. Actiwatch 2 is a wristworn accelerometer that records all uni-axial movement over 0.05 G in periods of 30 seconds. The device was worn on the non-dominant hand and the registrations were retrieved between 8 AM and 9 PM.
Wrist-worn accelerometers are validated for measuring physical activity in children, 29 and were chosen to maximize compliance. 30 Activity data was downloaded using Respironics Actiware software version 6.0.9 and transferred to Microsoft Excel 2016.
Metabolic equivalent of task (MET) was chosen to define intensity level due to its qualities of predicting energy expenditure across body composition. All activity data was reported as counts per

| Sleep measures
Sleep pattern was objectively measured using Actiwatch 2 (Philips, Respironics, BEND, OR). Wrist-worn accelerometers have been validated for measuring sleep in children. 31,32 As for physical activity, data was collected in 30 seconds epochs. Using a medium sensitivity threshold, the epochs were sorted as either "wake" or "sleep". This threshold was chosen as it has proven to be the least biased estimates of wakefulness, total sleep time and wake after sleep onset in school-aged children. 33 To ensure correct sleep measurements, the participants were instructed to press the event marker when turning of light at night and when waking up in the morning. To be included in the analysis, data for at least 5 days, including at least three school nights and two weekend nights, had to be recorded.
Time spent in bed was defined as the rest interval. Sleep time within the main rest interval was further detected by a standard default algorithm in the Respironics Actiware software version 6.0.9.
To yield more precise rest intervals, data was also manually scored according to a standardized scoring protocol. 27 Two independent observers scored 30.0% of the recordings twice, to ensure inter-rater reliability. The scoring was compared on total time in bed and total sleep time, and had a 99.6% and 99.9% agreement, respectively. Sleep timing was operationalized as mid-sleep time defined by the formula:

| Statistics
Descriptive statistics (mean, percentage, and standard deviation (SD)) was used to describe the study population, including physical activity and sleep measures. Comparison of continuous variables across groups was analyzed by using independent sample t-tests and categorical variables by using chi-square tests. Cohens d effect sizes for group differences on continuous variables were calculated. An effect size of 0.2 is considered a small, 0.5 a medium effect and 0.8 a large effect size. 34 The relationship between demographic data, anthropo- Estimated regression coefficients are presented with 95% confidence interval (CI) and p-values. The significance level was set to 0.05. All statistical analyses were performed with SPSS Statistics 26 (SPSS Inc.) after being controlled for prerequisites needed to run these analyses.

| Power estimates
To determine required sample size, G*Power version 3.1.3 was used.
The statistical significance was determined by α = 0.05 (two-tailed) and power (1-β) of 0.8 (80%). To detect a medium effect size (Cohen's d = 0.5) for the group comparisons, 64 individuals in both groups was required. Eighty-five children and adolescents were recruited to each group as this was the sample size recruited to the FABO-study at the time the matched comparison group was collected.

| Missing data
Out of 170 included participants, 154 children and adolescents (90.5%) provided valid accelerometer measured physical activity data and were included in the activity analyses. In total 16 (9.4%) of the participants (12 children and adolescents with severe obesity) were excluded from the activity analysis due to less than 10 h of wear time

| RESULTS
Characteristics of the participants (n = 170) are presented in Table 1.
Summarized, the participants' mean age was 12 � 3 years, ranging 56.5%, p < 0.001), as well as having parents with a higher education level (mother p < 0.001, father p = 0.003) compared to the children and adolescents with severe obesity.

| Physical activity and sleep measurements in children and adolescent with obesity and normal weight
Physical activity and sleep measurements are presented in Table 2.

| Physical activity levels and explanatory factors for children and adolescents with severe obesity
The relationship between MPA, VPA and explanatory variables are presented in Table 3.
More time in MPA was related to age and participation in orga- More time in VPA was related to age and sleep timing. Children

| DISCUSSION
The main findings of the study were that children and adolescents with severe obesity spent less time in moderate and vigorous physical DANIELSEN ET AL.
activity compared to normal weight peers, while there was no difference for time spent in light physical activity or sedentary activity between the groups. Higher physical activity level in the group of children with severe obesity was associated with lower age (≤12 years), participation in sports and earlier sleep timing.
In line with most previous studies, the group of children with severe obesity spent significantly less time in moderate and vigorous physical activity than their normal weight peers. [35][36][37][38] In the current study, children and adolescents with severe obesity and children and adolescents with normal weight spent 72 min/day and 105 min/day in MVPA, respectively, somewhat more time than demonstrated in most comparable studies. 9 This might indicate that Norwegian children in general are more physically active than children in many other countries, due to less unsafe areas, more outdoor leisure time facilities and more government funded organized sports. Another factor explaining these differences could be that wrist worn accelerometers, as used in the current study, is found to be more sensitive to physical activity in low active children than hip-worn accelerometers used in several other studies. 9,27 In the current study, mean physical activity levels in both children with severe obesity and normal weight are in line with international guidelines that recommend 60 min of daily MVPA for  The results from the current study showed a similar decrease in physical activity between children and adolescents (Figure 1), in both individuals with severe obesity and with normal weights, indicating that other factors might be of more importance than weight status in explaining this trend. Plausible explanatory factors for the decrease in PA by age could be changes in daily activities, including school hours, homework, travel distance to school, and a change in pleasure activities toward more sedentary activities. 40 As expected, and observed in previous studies, children with severe obesity below the age of 12 had more MPA and VPA than children over the age of 12. 41 Further, MPA time was strongly associated with participation in organized sports. In general, it appears that sport participation contributes to higher levels of MVPA in children and adolescents of normal weight, but in children with obesity the findings are inconclusive. 42 Only 31.8% of the children and adolescents with obesity in the current study engaged in organized sports. A previous review of qualitative studies has reported that children with severe obesity may experience more barriers to and less support for participation in organized sports than their normal weight peers. 5 A review of physical activity interventions for childhood obesity also concluded that close supervision and support during the activities predicted better weight related outcomes. 43 Even though active play is found to be at least as effective as organized sports in accumulating MVPA in children, 44 it might seem that the children not participating in  -807 organized sports also engage less in spontaneous active play. 43 46 and possibly to less MVPA, but this needs to be examined in further studies.
Assessing physical activity and sleep with objective measures over seven consecutive days, is an advantage of the present study as well as high compliance, low percentage of missing data and a matched comparison group. Registration of physical activity was performed using ActiWatch wrist-worn accelerometer, while in other studies hip-worn accelerometers have most often been used. 35,47 ActiWatch was preferred in the current study because of its ability to measure both physical activity and sleep. However, in the study of Ekblom et al., 27 ActiWatch was reported to have a tendency to register lower levels of physical activity in highly active children and higher levels in low-active children compared to the hip-worn Acti-Graph. 27 This could influence the total amount of physical activity reported in the low-active children. Comparison of ActiWatch placed on hip and wrist show that total measured activity was lower on the hip than on the wrist, 48  Qualitative studies exploring barriers to engaging in organized physical activity for children with obesity reported barriers such as lack of family and peer support, perceptions of being negatively judged by others or not experiencing competence compared to others when participating in sports. 5 Fatigue and physical discomfort, as well as dislike of being "visible", and problems with finding appropriate clothing for sports were also reported as barriers. 5 This knowledge might guide clinicians or sports coaches on how to support children with obesity in organized sports for example, adults taking the responsibility for choosing teams, demonstrating a zero tolerance for negative comments or providing team clothing in all sizes. In treatment of childhood obesity providing a setting for organized physical activity with peers of similar weight status might be a good idea. Further, to encourage families to participate in sports where a larger body size and muscular strength is an advantage in order to enhance sports related self-efficacy (e.g., American football or downhill skiing). In the current study late sleep timing emerges as a factor related to lower physical activity level. Other studies have also found sleep timing to be related to more screen time and higher intake of energy dense food. 50,51 Addressing sleep habits might be a good starting point for lifestyle intervention, as it might be easier to change than food and activity habits and might create more energy to engage in physical activity.

| CONCLUSION
Children and adolescents with severe obesity were less physically active than normal weight peers. Higher levels of physical activity for

ACKNOWLEDGMENTS
The staff at Obesity Outpatient Clinic, Haukeland University Hospital, for data collection efforts. To all the participating children and their parents for contributing with information making is possible to perform this study. University of Bergen (PhD-grant), Western Norway Regional Health Authority, The Norwegian Competency Centre for Sleep Disturbances.