Physical activity in children and adolescents with CHD: review from a measurement methodological perspective

© The Author(s), 2021. Published by Cambridge University Press. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited. Physical activity in children and adolescents with CHD: review from a measurement methodological perspective

with more severe heart defects. Lower peak oxygen uptake is associated with lower cardiovascular health, academic achievement, and well-being. 9 Concerns have been raised about the increased risk of being overweight and having additional cardiometabolic disease later in life in patients with CHD and the risk is even higher with more severe CHD. [10][11][12] Due to limitations in physical capacity, 6,7 but also because of restrictions from parents and caregivers and low selfefficacy, 8,13 it may be assumed that children and adolescents with CHD are less physically active than children and adolescents in general. This creates a need to focus on aspects of health-related quality of life, physical activity, and prevention of acquired cardiovascular disease in this group of patients. As physical activity, sports participation, and aerobic fitness have been acknowledged as crucial for health and development in children and adolescents, their promotion has been emphasised by international cardiology associations. 12,14,15 A recent review conducted by Caterini et al 16 especially argued for the importance of actively promoting physical activity in the younger CHD population for fostering a healthy, active lifestyle and also highlights the existing evidence gap in lack of models for implementing strategic physical activity in CHD populations as well as reliable and valid wearable technology for increasing and measure physical activity.
Acosta-Dighero et al 17 and Van Deutekom and Lewandowski 18 provided recent reviews of original studies using either subjective or objective methods to assess physical activity in children and adolescents with CHD. These reviews suggest similar physical activity level in children and adolescents with CHD compared to healthy controls, or in relation to the severity of the CHD, although several inconsistencies between studies were reported. This finding is somewhat unexpected, considering the physical limitations and other restrictions reported in children and adolescents with CHD. However, a deeper and more critical analysis of the measurement methodological limitations was missing from these review studies. This knowledge is crucial for explaining the unexpected finding, in order to determine the methodological progression required in future assessment of physical activity in clinical research.
Assessment of physical activity is mainly divided into two areas: subjective and objective measures. Former research of physical activity was primarily conducted using subjective methods like interviews and questionnaires as they are considered to be cost efficient, easily administrated, accessible, and is claimed to put little strain on patients. 19 However, extensive methodological limitations such as recall ability, memory, age, language, perception, understanding, and overestimation of both quantity and intensity of the performed physical activity have been identified, especially in the younger populations, causing poor reliability and validity. 15,[19][20][21][22][23][24] In 2013, the American Heart Association stated: " : : : use of self-reports is recommended only when more objective measures cannot be obtained". 15 Thus, the quantification of physical activity is now merely performed using objective measures.
Objective devices to assess physical activity involve pedometers, accelerometers, heart rate monitors, multisensors (e.g. acceleration, heart rate, heat, sweat), indirect calorimetry, and doubly labelled water. Indirect calorimetry and doubly labelled water are considered golden standards of objective physical activity measures. Nevertheless, they are expensive and resource-intensive, thus not very convenient in most physical activity studies. 25 As an alternative, accelerometers are considered as cheap, well developed, and easy to use, showing greater validity than subjective measures. 22,26,27 However, even if being the most frequently used and evaluated objective method for assessing physical activity, 28 the use of accelerometers to assess physical activity holds certain limitations. Generally, there is a lack of knowledge regarding the field and function of accelerometers, how the specific settings and data management affect the outcome, and how these may be the source of measurement errors. Many accelerometer-based studies also fail to provide a transparency in the settings and data processing used, preventing others to uncover impacts on the measurement outcome or possible underlying measurement errors. In addition, the methodological transparency by the manufacturer may be quite low. Epoch lengths (time resolution of physical activity measures investigated), cut-points (threshold markers for the classification of physical activity intensity categories), and raw data filtration method represent three major issues that are seen to especially affect the outcome of the accelerometer-assessed physical activity. 25,29 Thus, even if the results are presented equally, the parameters may imply dissimilar aspects of the assessed physical activity, complicating comparisons of the physical activity measure. 30 The first objective of this study was to compile and organise the existing studies assessing physical activity in children and adolescents with CHD by subjective and objective (accelerometers) methods. The second objective was to critically evaluate the physical activity measurement methodology in the accelerometer studies and the consequences on results and conclusions. An important outcome from the second objective was to provide guidelines on the assessment of physical activity using accelerometers, in order to improve clinical research.

Search strategy
A literature search and a data extraction were performed between April 2020 and October 2020 in PubMed database. Two separate searches were conducted. The first search concerned subjectively assessed physical activity and the second search concerned accelerometer-assessed physical activity. We combined the following search terms in search one: Congenital heart disease OR defect, acquired heart defect OR disease, physical activity, exercise, children OR adolescents OR youth, surveys and questionnaires OR self-scattered OR self-reported OR subjective OR questionnaire; and search two: Congenital heart OR disease OR defect, acquired heart defect OR disease, physical activity, exercise, children OR adolescents OR youth, accelerometer OR accelerometry.

Inclusion/exclusion criteria
Inclusion and exclusion criteria are listed in Table 1. We included original articles studying children or/and adolescents with treated CHD, written in English, and published in peer-reviewed journals with a quantified physical activity outcome. Articles published in 2000 and later were included for subjectively assessed physical activity and 2009 and later for accelerometer-assessed physical activity due to developmental aspects in accelerometers. The patient group was set to children and adolescents with CHD between 3 and 20 years. Articles studying patient groups with known extensive health issues other than CHD were excluded. the articles were included for both searches as they used both subjectively and accelerometer-assessed physical activities. Figure 1 presents the article extraction. Two researchers reached consensus regarding the included articles. An overview of the included articles from the first and the second search is presented in Tables 2 and 3.
The study populations in the included studies differ substantially. Physical activity results are reported from both populations with mixed CHD, 31-36 from children and adolescents with one specific diagnosis, 37 three specific diagnoses, 38 from CHD as divided by mild to severe CHD group (different specifications), [39][40][41] between specific CHD diagnosis 42 or between specific CHD diagnosis and as varied CHD group, 43 restricting comparisons and generalisation of the results.

Included articles, subjectively assessed physical activity
Six studies using subjectively assessed physical activity were included. Two articles reported physical activity data in CHD children, 31,36 one article compared different types of CHD and controls, 43 two studies compared different types of CHD with control samples from external databases, 40,41 and one article compared children with Fontan circulation to healthy controls. 37 All of the six articles used different physical activity assessment questionnaires: Physical Activity Questionnaire for Older Children, 43 Youth Risk Behavior Survey, 41 International Physical Activity Questionnaire-short version, 31 "The New South Wales Schools Fitness and physical activity Survey" 40 , one question assessment, 36 and self-reported physical activity regarding organised physical activity. 37

Cardiology in the Young 521
When looking at the comparison of children and adolescents with CHD towards healthy controls, contrasting results were present. Ray and Henry 41 reported no significant difference in physical activity between patient and control groups in children with mild, moderate, and surgically treated CHD. Relatedly, Lunt et al 40 reported lower physical activity in male patients with mild and severe CHD and a similar trend in females, while as Zaqout et al 43 presented the oppositea higher physical activity in the overall CHD patient group (ventricular septal defect, coarctation of aorta, transposition of great arteries, and tetralogy of Fallot) when compared to controls. Hedlund et al 37 presented significantly lower physical exercise and significantly lower average intensity on Borg scale in patients with Fontan circulation than in healthy controls.
When comparing patients with different severity of CHD, two studies reported no significant differences in physical activity, 40,43 whereas one failed to report the results for physical activity divided by the different severity groups. 41 In the studies providing informative physical activity data, Schaan et al 31 reported low levels of patients in the "very active" (24%) and "active" (32%) output variables. Brudy et al 36 found that the patients reported themselves as active 4.7 days/week, generally not meeting the World Health Organisation recommendations for children and adolescents in 60 minutes of moderate-tovigorous-physical-activity a day. 44 These results are similar to those reported by Ray and Henry 41 who reported that only 38% of the patients were physically active for at least 60 minutes and 7 days a week. Similarly, Hedlund et al 37 reported a mean value of self-reported physical exercise of 135 minutes/week, indicating that the Fontan circulation patients generally fail to meet the World Health Organisation recommendations of physical activity. In contrast to these findings, both Zaqout et al 43 and Lunt et al 40 stated that most of the children and adolescents with CHD did meet the World Health Organisation recommendations of physical activity.

Included articles, accelerometer-assessed physical activity
In the accelerometer-based studies, nine articles were included (Table 4), whereas six studies compared children and adolescents with CHD towards healthy controls, [32][33][34][35]37,38 two studies compared types of CHD, 39,42 and one study compared children and adolescents with CHD to a healthy control group and type of CHD. 43 Epoch lengths of 60 seconds were reported in three of the articles, 35,37,38 30 seconds epoch lengths were listed in one study, 32 15 seconds in two of the studies, 39,42 and a 3 seconds epoch length in one study. 34 Two of the included articles using accelerometerassessed physical activity did not report the epoch length used in the study. 33,43 Eight articles reported the use of hip-worn accelerometers, [32][33][34][35]38,39,42,43 whereas one reported using wrist-worn accelerometers. 37 Considering the cut-points used for defining the physical activity intensity categories, three articles [37][38][39]43 reported using Evenson 45 cut-points for classification of physical activity intensity categories, one article 34 reported using Pate's 46 cut-points, one article 32 reported using age-appropriate intensity levels from a metabolic equivalent prediction equation 47,48 for generating cutpoints, one article 42 reported using Puyau 49 cut-points but did not submit how the threshold of moderate-to-vigorous-physicalactivity > 1600 counts per minute was calculated, one article 35 reported using the ActiReg monitor with calibration equation    50 whereas one reported transferred accelerometer (RT3) originated energy expenditure as intensity measure. 33 Similar physical activity levels were found between the patient group and the healthy controls in the majority of the included accelerometer-based studies. [32][33][34][35]37,39,42,43 In contrast, White et al 38 reported that the CHD group spent significantly more time in light physical activity and less moderate-to-vigorous-physical-activity than the healthy controls, typically engaging in more sporadic bouts (<5 minutes), fewer short (5-10 minutes) and medium-to-long (>10 minutes) bouts of moderate-to-vigorous-physical-activity than the healthy controls. Similarly, Kao et al 33 observed significantly lower levels of total energy expenditure in boys with CHD compared to healthy controls, even if the reported moderate-to-vigorousphysical-activity was similar between the groups.
The majority of the accelerometer-based studies reported that the patients with CHD generally failed to meet the World Health Organisation recommendations of physical activity in children and adolescents. 32,34,35,39,43 Banks et al 42 recognised that the majority of the atrial septal defect patients met the recommendations of physical activity, but not the transposition of the great arteries, tetralogy of Fallot, or single ventricle patients; however, they did not state the proportions. Hedlund et al 37 studied the physical activity in Fontan circulation patients and observed an average moderate-to-vigorous-physical-activity of 148 minutes/day in the patient group, stating that they meet the World Health Organisation recommendations of physical activity.

Discussion
The main observation from all of the included studies was the contradicting finding of similar physical activity levels in children and adolescents with CHD compared to healthy controls, or due to the severity of CHD. These results are in line with the previous two review studies. 17,18 Van Deutekom and Lewandowski 18 raised the concern about the low level of physical activity in the general population, which could affect the possibility of detecting different physical activity behaviour in children and adolescents with CHD. Although this might be true, we argue for that the contradicting findings and absence of group difference in physical activity are largely explained by the methodological variations and limitations in the assessment of physical activity. In order to comply with the second objective of this study, and as it is recommended to use objective methods before subjective methods for obtaining more reliable assessment of physical activity, 15 the discussion will mainly focus on the accelerometer-based studies included.

Subjectively assessed physical activity
As measurement errors in subjectively assessed physical activity have been stated by earlier research, demonstrating poor reliability and validity, especially in children, 15,19,20 a potential misclassification and a variation among the results are seen as probable. Reports of both no significant differences towards healthy controls, 41 significantly lower physical activity, 37,40 and significantly higher physical activity in the patient group 43 were found. The inconsistent and widely divergent findings may be a result of all six articles using different questionnaires for the assessment of physical activity. Dissimilar properties of physical activity are thereby captured. Thus, stating general conclusions or even comparing the physical activity outcome is seen as inappropriate. A similar verdict was reported by Acosta-Dighero et al. 17 and by Van Deutekom and Lewandowski. 18

Accelerometer-assessed physical activity
Concerning the accelerometer-based studies, the overall results suggest that there are no differences in the physical activity levels between children and adolescents with CHD compared to healthy controls. [32][33][34][35]37,39,42,43 Similar finding was reported by Acosta-Dighero et al 17 and by Deutekom and Lewandowski. 18 Two of the studies showed contradicting results. Kao et al. 33 reported lower total energy expenditure in boys with CHD, indicating that they move less than their healthy controls even if the reported moderate-to-vigorous-physical-activity was similar between the groups, whereas White et al. 38 reported the patient group as spending less time in moderate-to-vigorous-physical-activity and engaging in smaller bouts of moderate-to-vigorous-physical-activity than the healthy controls. As the physical activity outcome is highly dependent on the distinct settings made, 28,30,51 a likely explanation of the many cases of the unexpected "no-difference in PA" between patients and healthy controls can be related to the irregularities in the methodologies used in and between the studies.
Common variances in accelerometer-based studies regard matters such as device placement, raw data processing, epoch lengths, value calibrations, altered use of pre-calibrated cut-points, weekend-weekday-criteria, number of valid days/hours-a-day/week, and handling of sleep-time and non-wear time. The handling of these parameters is often poorly described or completely lacking, making comparisons between studies even more problematic. The lack of methodological consensus within accelerometry makes it difficult, or even impossible, to generalise and compare the results. 28,30,51 In the following section, we will go through the methodological issues of the included studies and their consequences on the results. This will be performed by considering each issue separately.

Device placement
As the activities of the arm not necessarily reflect the movements of the rest of the body, the registered data differ between the hip-and wrist-worn placement sites. 52 Hip-worn sensors typically capture movements that better reflect the whole-body energy consumption, 28 while wrist-worn sensors have been shown to be disposed for misclassifying seated behaviours that are involving high levels of upper body movement. 52 One of the accelerometer-based studies used a wrist-worn accelerometer but applied physical activity intensity cut-points developed from a hip-worn accelerometer. 37 The use of wrist-worn accelerometers and the application of hip-worn accelerometer cut-points to wrist data reduce both study validity and comparability to the other studies included in this review.

Epochs
The epoch lengths are ranging from 3 to 60 seconds in the included studies, with two of the studies failing to report the epoch length used. The occurring epoch-length variances (and in many cases long epoch lengths) may result in different estimates of physical activity within the studied populations and thereby lead to distorted interpretations. Previous research has stated significant variation in physical activity volume and intensity using various epoch lengths, showing a progressive decrease in time spent in moderate-to-vigorous-physical-activity with longer epoch lengths. 29,[53][54][55] Also, with the movement pattern of children being highly intermittent, shorter epoch lengths have been recommended as longer epochs fail to capture the executed physical activity. 29,56,57 To demonstrate, in the study by White et al 38 29 showed that with 60-second epochs, a large proportion of SED would be misclassified as light physical activity, and vigorous physical activity would be misclassified as light physical activity or moderate physical activity, but the total physical activity would not be affected. Hence, the 60-second epochs will not capture the variation in physical activity in children and would distort/reduce expected group differences.  37 Voss et al. 39 White et al, 38 and Zaqout et al 43 used the Evenson cut-points 45 of >2296 counts per minute for the same physical activity intensity category. With different cutpoints being applied, the easiness to accumulate moderate-to-vigorous-physical-activity differs between the studies, not displaying the same variations of physical activity within each physical activity intensity category even if results are presented in the same terminology. Pre-calibrated cut-points are population specific as body mass and age are important factors for the calculated mechanical energy used in accelerometry. Thus, when applying the cut-points, it is recommended to follow the same data gathering and processing criteria upon the same age group that was utilised in the original calibration study. In many of the included studies using pre-calibrated cut-points, the age group is not compatible with the age group used in the original calibration study. Only one of the studies used the same age-criteria. 34 Instead, Banks et al 41 applied cut-points calibrated for 7-18 year old on a population of 4-12 year old, and the Evenson 45 cut-points calibrated for 5-8 year old, were applied to populations of 8-20 year old, 37 8-19 year old, 39 8-18 year old, 38 and 6-14 year old. 43 Consequently, with children generally moving with a bigger effort and a higher energy cost at a given activity than older (and taller) individuals, 58 the calibrated cut-points used indistinguishably between the age groups will cause false estimates of physical activity categorisation.

Pre-calibrated cut-points
Modifications, or scaling, of pre-calibrated cut-points to fit the chosen epoch length are frequently seen in accelerometer-based research. However, a modification of cut-points alters physical activity estimates. 56,59 Only one of the accelerometer-based articles used the same epoch setting as in the original calibration study. 39  Taken the age, epoch length, and device placement criteria for the usage of pre-calibrated cut-points all together, the divergent settings will affect the between-study comparisons and decline the internal validity of the measure within each study.
Calibration of cut-points and inclusion of whole-spectrum physical activity Metabolic equivalent of task (VO 2total /VO 2rest ) is a frequently used criterion measure of absolute physical activity intensity and is a way of expressing the energy cost of task-specific physical activity relative to body mass. Two articles reported calibrating cut-points via energy expenditure equations of MPA corresponding to 3 metabolic equivalent of task 35 or 4 metabolic equivalent of task. 32 Validating cut-points against indirect calorimetry can be problematic as metabolic equivalent of task-calibrated cut-points are not comparable between ages. Typically, a 3 metabolic equivalent of task value is set for calibration of moderate physical activity threshold in both children, adolescents, and adults. However, when children walk at a speed of 5.6 km/hour, a metabolic equivalent of task value of 4.3 is reached, 60 whereas adults typically reach a metabolic equivalent of task of 5.0 at the same speed. 61 However, the internal effort and energy cost of a child are greater. A child will consume more O 2 per kg body weight, 58 moving with higher step frequencies than taller individuals at a given speed. 62,63 Adding the fact that the resting energy expenditure decreases with age 64 and metabolic equivalent of task values for specific activities typically increase with age (even more distinct at higher intensities), 60,65 the usage of metabolic equivalent of task as a measure of effort across age groups when calibrating cut-points can be distortive.
To enable direct comparison of age groups and achieve ageequivalent measures of physical activity intensity categories (e.g. light physical activity, moderate physical activity, vigorous physical activity, moderate-to-vigorous-physical-activity), the VO 2net (VO 2total -VO 2stand , ml/kg/minute) has been recommended to use as criterion measure for calibration of accelerometers. 58 When a child (shorter) and an adolescent (taller) attain the same VO 2net , they will move with the same metabolic effort, but with the child generating less acceleration (or less mechanical work) as moving at a slower speed than the adolescent. Hence, minutes spent in, for example, vigorous physical activity will be directly comparable between age groups from a metabolic effort perspective. In contrast, when a child and an adolescent move at the same speed, they will generate similar acceleration (or mechanical work) but with different metabolic efforts.
Nevertheless, the crude classification of physical activity used in all studies included might cause a potential loss of information from the collected acceleration data. Aadland et al 57 showed how presenting physical activity as a high-resolution physical activity intensity spectrum provides more comprehensive information regarding the physical activity behaviour. By presenting physical activity as an intensity spectrum, the concern with studies using different cut-points is resolved.

526
P. Skovdahl et al. Device placement • Device placements of both hip, wrist, and tight occur (back and chest are also seen, but infrequently used). Thighand hip-worn sensors generally assess activities overall reflecting the energy consumption of the whole body and are seen as good at stating the physical activity dimensions volume, frequency, duration, and intensity. Thigh placement also enables the assessment of body position (e.g. sit, stand, lay down) and type of activity performed (e.g. biking, walk, run), but demands good tape solutions for attachment of the accelerometer as it has shown a tendency to be worn of during activity and change of clothes 25,28 • The wrist placement captures different movement patterns compared to the other placements, typically misclassifying seated behaviours involving high levels of upper body movement 52 • A concise instruction to wear the sensor 24 hours/day is seen to contribute to similar wear compliance between the placement sites 25 • Comparisons should only be considered when using the same measurement location Data collection • Clear instructions to patients with pictures displaying body attachment facilitate good wear compliance. 25 Registration period protocols at 24 hours/day show higher wear time compliance compared to waking-hour protocols 73 • 7-day protocols are commonly used in physical activity research and are seen as sufficient for capturing normal variation in physical activity. If more accuracy on individual level is required, more days should be included. 25 • Sampling frequencies should be sufficiently high to cover the movement frequencies, 30-100 Hz has been recommended. 28 A higher sampling frequency will limit the number of days to record physical activity as it requires more memory • A sampling amplitude at 8 g covers most human activities 25 Processing Epoch length Raw data filtration method • Frequency filters are commonly applied in order to reduce noise • The narrow raw data filtration method used (ActiGraph counts) is seen to acquire misclassifications of >90% when compared to wider filters at the higher intensity spectra. 70 Interpretation of ActiGraph counts filtered acceleration is therefore highly deceptive • A wider filter is recommended to better capture the physical activity performed, for example, frequency extended method (FEM) 69,70 Value calibration and calibration of cut-points • Value calibration against a reference method (e.g. indirect calorimetry (VO 2 )) is performed in order to translate the accelerometer measure to more established measures of physical activity intensity (e.g. EE, MET). Cut-points are defined to create physical activity intensity categories. • If measured VO 2 is used as reference, it is recommended to apply VO 2net (VO 2gross -VO 2stand , ml/kg/minute) as it provides an age-equivalent measure of metabolic intensity in order to compare physical activity between age groups 58 • It is recommended to follow the same data collection protocol, sample characteristics, and processing criteria as used in the original calibration study when applying cut-points to a specific data set. 28 • Scaling of pre-calibrated cut-points to fit the chosen epoch length is seen to alter physical activity estimates 56,59 Inclusion criteria • Including too few measurement days will decrease the chance of capturing the individual physical activity behaviour, differences between groups and relationships between physical activity behaviours and health aspects, as great within-individual variances in habitual physical activity exist 25,76 • Even if measured for several days, great variances typically occur in wear compliance, both in regard to whole days and in weekdays/weekend days. Studies should state inclusion criteria in regard to wear time (8-10 hours/day), valid day criteria (minimum of 4 days, >3weekdays, and >1 weekend day) as well as non-wear definition to enable tracking of time that should not be involved in the activity analysis. 76 For higher precision, more days and hour/days should be included. • Non-wear time (when the accelerometer is taken off) should be defined and sorted out to enable discrimination between sedentary behaviour and when the accelerometer is not in use. A non-wear time of at least 60 minutes of zero values is commonly used 28

Management of outcome parameters
• Outcome/results are often presented as time spent in crude physical activity intensity categories (SED, LPA, MPA, VPA, and VVPA) 28 • It has been suggested to present and analyse the physical activity as a high-resolution physical activity intensity spectrum 74,75,77,78 • Even if higher intensities appear particularly beneficial to health, 74,77,78 they only account for a fraction of time spent in physical activity when measured in minutes per day, even in very active individuals. The physical activity intensity spectrum allows more detailed inspection of the physical activity and may reveal physical activity patterns otherwise hidden when applying crude physical activity intensity categories • With cut-points varying greatly among studies, and scaling of cut-points to fit the wanted epoch length being performed, the risk of confusion is assumed to be highly present when comparing and contrasting the results

Cardiology in the Young 527
Raw data filtration Even if seen as more reliable than subjective measures, accelerometers have been shown to possess difficulties in capturing intermittent and high intensity physical activity, generating a decrease in counts even if activity is increasing (e.g. the "plateau" effect). 66,67 The "plateau" occurs mainly as a result of the raw data frequency filtration in the original, most commonly used ActiGraph counts. 66 This is believed to be particularly applicable in the measurement of children as their general movement pattern is sporadic and highly intermittent, 53,57,68 moving with a higher step frequency at a given speed, 25,57,68 consequently reducing the acceleration signal ever further. Processing the acceleration through a wider filter improves the assessment of physical activity and reduces the age variances in gait patterns. 69,70 Notably, when compared towards broader filters, the ActiGraph counts showed a misclassification of >90% in the higher intensity spectra. 70 Consequently, the capture of physical activity intensities is considerably more accurate when processing the acceleration through a wider filter. With the previous raw data filtration being insufficient, it is likely that the included studies failed to capture the variance in physical activity at higher intensity levels as interpretations of ActiGraph filtrated moderate-tovigorous-physical-activity are highly unreliable. Therefore, a possible larger difference in physical activity might be present between the studied groups than the included studies imply, as they rely on the narrow ActiGraph raw data filtration. • It is important to keep in mind that accelerometers primarily record mechanical, absolute work, translated to meaningful terms via value calibration. The relative intensity measured in individuals is thereby not accounted for • Absolute intensity can be seen as especially problematic when used for specific groups who possess biological or physiological limitations, like the children and adolescents with CHDs. For example, running at a certain speed will potentially be of greater effort for individuals with reduced capacity in the cardiovascular system compared to healthy individuals, even if the same work is performed EE, energy expenditure; MET, metabolic equivalent of task (EE total /EE rest ); crude intensity categories; SED, sedentary behaviour (<1.

Fulfillment of World Health Organisation physical activity recommendation
When it comes to fulfilling the World Health Organisation recommendation of ≥60 minutes per day of moderate-to-vigorous-physical-activity, 44 the included articles point towards an agreement: children and adolescents with CHD generally fail to meet the recommended amount of physical activity, a result similar to that of Acosta-Dighero et al 17 and by Van Deutekom and Lewandowski. 18 However, the World Health Organisation guidelines are based upon subjective measures of physical activity. The results from the accelerometer-based studies regarding physical activity recommendations should therefore be interpreted with caution as they are based on different, incomparable methods. Further, the chosen cutpoints and epoch lengths will affect the ratio of individuals reaching the physical activity recommendations as lower cut-points and shorter epochs will accumulate more moderate-to-vigorousphysical-activity; conversably, higher set cut-points and longer epochs will accumulate less moderate-to-vigorous-physicalactivity. 28,29,57 The fulfillment of the physical activity recommendation is also dependent on how strict the criteria are, that is, if attaining 60 minutes of moderate-to-vigorous-physical-activity on most days or as a daily average. For example, Voss et al 39 applied Evenson cut-points on 15-second epoch data and the stricter criteria of fulfilling the physical activity recommendation on most days; only 8% of the patients were sufficiently physically active. Further, difficulties may arise when attempting to implement the World Health Organisation recommendations of physical activity for healthy individuals on children and adolescents with physical restrictions. The included articles contain a range of different types and severities of CHD, with various physician subscribed (former) recommendations for engagement in physical activity. Accelerometers measure absolute intensity, regardless of the intensity relative to fitness level. A moderate intensity level for an individual with a severe CHD might differ from the same physical activity level for a healthy individual. In addition, the cut-point for moderate intensity level might be set too low in general, corresponding to normal walking speed rather than to brisk walking speed in accordance with the original definition of moderate-intense physical activity. 71 In summary, the methodological challenges, variances, and limitations might explain why we generally fail to see a difference between the groups even when a larger difference in physical activity is to be expected. We need to control for the discussed parameters and strive towards agreement between the methods used for enabling future comparisons and interpretation of the results. As long as there is no consensus concerning accelerometer protocols and settings, research protocols will be designed unequally, making between-study comparisons highly questionable. To enable further research on the effect of interventions, strategies, and models for promoting physical activity in CHD populations, valid and reliable baseline measures of physical activity patterns in CHD populations are needed. Interdisciplinary collaborations are advantageous when implementing accelerometry into clinical research for assessing valuable and accurate assessments of physical activity and thereby improve the quality of clinical physical activity research. Figure 2 (a-f) presents a brief overview of methodological considerations for tailoring a physical activity measure using accelerometers together with a case scenario with implementation of the methodological steps of the accelerometer study protocol. More detailed information, concepts, and guidelines are provided in Table 4. Complementary information about existing methods may be provided in the work by Voss and Harris. 72

Conclusion
Previous research has been unable to establish whether the physical activity patterns in children with CHD differ to healthy controls, or due to the severity of CHD. These results are largely explained by methodological variation and limitations in the assessment of physical activity. This review provides methodological knowledge and guidelines for improved assessment of physical activity using accelerometers in clinical research.