Comparison of accelerometer measured levels of physical activity and sedentary time between obese and non-obese children and adolescents: a systematic review

This systematic literature review was performed in accordance with the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines [22]. The review protocol was registered on PROSPERO (registration number CRD42015026882), the international prospective register for systematic reviews (http://​www.​crd.​york.​ac.​uk/ NIHR_PROSPERO).

The literature search was conducted searching for English language peer-reviewed studies using the five most relevant electronic databases from 2000 up to March 2015 (accelerometry became more widely used in research from the early 2000’s): MEDLINE OVID; Cochrane library; EMBASE; SPORTSDiscus and CINAHL by AM. The literature search in the Cochrane Central Register of Controlled Trials is shown in Table 1, and was adapted as required for the other databases. Full literature search details are available from the corresponding author on request. The electronic search was complemented by reference citation tracking (forward and backward) of the included studies and of previous reviews.

To be eligible for inclusion in the review, papers had to meet all of the following criteria as per the PICOS principles: Population: children and adolescents aged from 0 to 19 years as defined by the WHO; Intervention or exposure: children or adolescent classified as obese. Obesity had to be defined using an acceptable objective method, e.g. defined as having a body mass index (BMI) ≥ 95th percentile for children of the same sex and age, or defined as the equivalent of 30 kg/m² International Obesity Task Force (IOTF definition), or defined as obese relative to World Health Organization (WHO) BMI for age and sex charts; Comparison: habitual amount of time spent in MVPA and/or ST of non-obese children and adolescents; Outcomes: habitual amount of time spent in MVPA and/or ST measured by accelerometer and reported in the form of minutes/day of MVPA or ST; MVPA and its relationship to the 60 min/day recommended. All study designs were considered eligible: cross-sectional, longitudinal, case-control studies and intervention studies were eligible if pre-intervention data could be extracted.

We excluded studies that included only overweight participants, combined overweight and obese groups, or included participants with any known barrier or limitation to physical activity (e.g. physical disability). Studies that used subjective methods, objective (e.g. doubly labelled water) or direct observation methods apart from accelerometer measurements were excluded.

Since the aim of the review was to examine habitual levels of MVPA and ST, studies that measured these variables for less than 6 h per day or over 2 days or less were excluded. Recommendations currently exist for habitual (overall) MVPA rather than MVPA during specific domains (e.g. the after school period) and so studies that focused only on specific periods of the day (e.g. school activity only, or outdoor activity only, or weekend activity only, or weekday activity only, or after-school only) were also excluded. A detailed description of the eligibility criteria is given in [Additional file 1].

Titles, abstracts, and full-text articles were screened in duplicate for eligibility by RE and JYP and disagreements were resolved through discussions with other reviewers when required. Reference lists of eligible studies were examined for potentially eligible studies, and studies that cited eligible studies were identified and tested for eligibility. The reviewers were not blinded to authors or journal of publication. Reasons for exclusion are summarised in the study flow diagram (Fig. 1) and available in details from the corresponding author on request.

A standardised data extraction form was used to populate the evidence tables by RE and repeated by JJR and JYP. The extracted items were: first author, publication year, country, study design, sample group, comparison group-if applicable, accelerometer type, cut points for MVPA and ST, finding of MVPA (minutes/day) and ST (minutes/day or %) data, summary and author conclusions. International recommendations are usually for the achievement of at least 60 min of MVPA every day, but in the eligible studies the achievement of MVPA recommendations was never operationalised in this way. In most studies that referred to the achievement of MVPA recommendations, the mean or median daily MVPA (minutes/day) was provided, and so this was used as a proxy for achievement of recommendations in the present study.

We considered the data for meta-analysis but identified a substantial level of statistical heterogeneity between the studies (I² statistic > 70%) that led to the decision not to present the combined results of individual studies. Hence, we performed a narrative synthesis of the data and present the findings in tabular, textual and graphical form. Data were synthesised by the age and sex of the subgroups as those are factors known to be strongly associated with both the exposure variable, obesity, and the outcomes, MVPA and ST, and so might explain some of the observed findings. The age subgroup was categorised according to the WHO definition of children and adolescence, i.e. as children aged 0 to 9.9 years old and adolescents aged 10.1 to 19 years old. Data for boys, girls and mixed-sex studies are reported separately where possible.

Eligible articles were assessed for methodological quality using a 15-item quality assessment scale as shown in [Additional file 2], collapsed to 6 items for scoring, with higher scores suggesting higher study quality. Each eligible study was assessed by RE, and disagreements were resolved by discussion with JJR and JYP. The quality assessment scale was modified from the methodological quality assessment scale of Tooth et al. [23]. This is a reliable and valid tool for assessing the quality of observational studies. It was considered initially for use in its original form, which consists of over 30 items. The modifications to the original scale were made to focus quality assessment on issues of particular importance to accelerometry measurement of physical activity. The modified Tooth et al. tool has been used in several recent systemic reviews of physical activity, all of which have reduced the number of items in the quality assessment to 8 to 17 items, which make up the quality score [24‐28].

The PRISMA flow diagram with the numbers of included and excluded articles at each step of the review process is provided in Fig 1. Tables 2 and 3 provide a brief summary of all studies included in this systematic review. Of 1503 papers identified in the initial review of the five databases, 467 were selected for full-text screening and of these, 22 met the inclusion criteria. A further four eligible studies were identified from searching reference of included studies and of previous reviews, giving a total of 26 studies which met the inclusion criteria.

Of the 26 included studies: six studies involved children, 18 studies involved adolescents and two studies involved both children and adolescents. Further, 22/26 compared MVPA data in those with obesity with a non-obese peers, while 13/26 studies also provided data on accelerometer measured ST; 10/13 studies compared ST data in those with obesity with non-obese peers.

The ActiGraph was the most common accelerometer type used to measure habitual MVPA and/or ST, used in 20/26 studies, though with a variety of different ActiGraph models and approaches to data collection and reduction. Of the remaining six studies: three used the Actical accelerometer [29‐31]; two the Triaxial Research Tracker (RT3) accelerometer [32, 33]; and one the Actiwatch accelerometer [34].

Eight eligible studies involved obese children, with a total sample size of 2138 children (478 with obesity; 131 boys, 136 girls and 211 no sex specified). Two of the eligible studies were clinical samples with study participants recruited from outpatient clinics. Eligible studies were from different nations with one study from Asia [35], three from Canada and USA [36‐38] and four from Europe [32, 39‐41], with the study characteristics summarized in Table 2. In four studies, MVPA data of boys and girls were reported separately while in the other four studies MVPA data were reported as mixed sex. 7/8 of eligible studies reported mean daily time spent in MVPA in minutes; in four studies mean time spent in MVPA was < 60 min/day. Furthermore, in 2/7 of the eligible studies, children with obesity reached or exceeded 60 min of MVPA per day [37, 38], while in one study they came close to a mean of 60 min/day of MVPA [36]. In all cases time spent in MVPA in the children who were obese was compared to the comparison group (non-obese peers). In only one study was the mean time spent in MVPA similar in both groups [37]; in three studies, time spent in MVPA was significant lower in children with obesity than in the comparison group [35, 36, 39], while in two studies time spent in MVPA of children with obesity was lower than the comparison group but differences were not significant [38, 40]. In the other 2 studies, time spent in MVPA of children with obesity was different in terms of gender compared to the comparison group: Hussey et al. reported that mean MVPA was significantly lower in boys with obesity but not in girls [32]; while Vale et al. reported that mean time spent in MVPA was significantly lower in girls with obesity but not in boys [41] compared to the comparison groups.

With respect to ST, 4/8 eligible studies reported on accelerometer-measured time spent in SB of children with obesity with a total sample size of 536 children (191 with obesity; 28 boys, 32 girls and 131 no sex specified). In one study, ST data of boys and girls was reported separately while in other the data were reported as mixed sex. Across all four eligible studies, mean time spent in SB was > 70% of waking time [32, 35, 39, 40]. In 3/4 of the studies ST was significantly higher in the obese than the non-obese groups, although, in one study it was significantly higher in boys with obesity but not in girls [32]. In one study ST was similar in both groups [39].

Twenty of the eligible studies involved adolescents, with a total sample size of 12,601 adolescents (3045 with obesity; 1615 boys, 1575 girls and 195 no sex specified). Four of the eligible studies were clinical samples with participants recruited from outpatient clinics. Eligible studies were from different nations with one study from Asia [42], 11 from Canada and the USA, and eight from Europe, with the study characteristics summarized in Table 3. In 12/20 studies, MVPA data of boys and girls were reported separately; in 6/20 studies MVPA data were reported as mixed sex, while the other two studies involved only adolescent girls. All 20 eligible studies reported mean daily time spent in MVPA in minutes and in these studies it ranged from a low of 16 (SD 4) minutes/day [36] to a high of 140 (SD 47) minutes/day [43]. In only 2/ 20 studies did daily time spent in MVPA reach an average of at least 60 min [43, 44] in the adolescents who were obese. A total of 16/20 eligible studies compared time spent in MVPA of those with obesity with a comparison group: in 3/16 time spent in MVPA was similar between obese and non-obese groups, while in 10/16 mean time spent in MVPA was significantly lower in adolescents with obesity than in non-obese peers.

In regard to time spent in SB, nine out of the 20 eligible studies reported on accelerometer measured ST in adolescents with obesity with a total sample size of 5484 adolescents (1101 with obesity; 546 boys and 555 girls), as summarised in Table 3. In 8/9 studies, ST data of boys and girls were reported separately and 1/9 study involved only adolescent girls. In 7/9 studies, mean daily ST was reported in minutes and in these studies it ranged from a low of 345 (SD 122) minutes/day [34] to a high of 731 (SD 110) minutes/day [30]. In 6/9 studies there was a comparison group; in 2/6 studies mean daily ST was similar in obese and non-obese groups [42, 45]; in 2/6 studies ST was significantly higher in those with obesity than in the non-obese comparison groups [34, 46], while in the other 2/6 studies it was higher in the adolescents with obesity, but not significantly so [47, 48].

A graphical synthesis of the mean differences and 95% CI of time spent in MVPA by sex for both children and adolescents with obesity and non-obese groups, is shown in Fig. 2. A summary of the mean differences and 95% CI of time spent in SB by sex for both children and adolescents with obesity and non-obese groups, is shown in Fig. 3.

Study quality assessment summaries are given in [Additional file 3]. One study scored 4/6 12 studies scored 5/6 while 13 scored 6/6 on study quality.

We believe that the present study is the first systematic review to ask whether or not levels of accelerometer measured MVPA are adequate in children and adolescents with obesity, and whether time spent in MVPA and ST differed between obese and comparison groups based on accelerometer data. There are therefore no directly comparable studies. However, our findings are consistent with some studies on the correlates and determinants of objectively MVPA [52, 53], and consistent with a growing belief that obesity is associated with reduced MVPA and that low MVPA could be both a cause of obesity and a consequence of obesity, i.e., “bidirectional causation” [20, 54, 55].

The evidence considered by our review had a number of strengths. Firstly, it investigated the accelerometer-measured time spent in MVPA and ST of children and adolescents with obesity, with clear definitions of obesity so that samples included in the review were not contaminated by the inclusion of overweight but non-obese individuals. Secondly, there are several methodological strengths to this study. Studies were identified from an extensive search of the published literature conducted in a range of databases, over the last 15 years, covering the time when accelerometers started to become available and popular in PA research and, more recently, sedentary behavior research. The broad definition of search terms applied across multiple databases enabled the searching and identification across many potential studies with no limitations on place of publication, sample size or country of origin. Restricting eligibility to studies using accelerometry was important in increasing confidence in the measurement of MVPA [19, 56‐58]. The included studies were in general rated as being of high or very high methodological quality with respect to their accelerometry methods. Also, in some cases the eligible studies were based on large nationally representative samples or surveys another strength in terms of generalizability.

There were some sources of weakness in our systematic review. Firstly, since studies had to be published in peer-reviewed journals in English, this may have excluded some relevant evidence. The generalisability of review results is subject to certain limitations; for instance, eligible studies in our systematic review were from high-income nations, and we lacked data from low-middle income countries. Most of the included studies were based on relatively small samples of obese children and adolescents with a total (n = 14,739 participants; n = 3523 with obesity) and their power to estimate habitual MVPA might have been limited, and thus the extent to which the results observed are generalizable to the general obese paediatric population is unclear. The method of quality assessment in our review, in which the original 15 items scale has been collapsed to 6 items, has been used in several accelerometry systematic reviews [24‐28]. However, the process of collapsing 15 items to a six-item might have reduced the possibility of identifying differences in quality between studies.

Eligible studies generally obtained MVPA and ST data using the ActiGraph accelerometer, but methods used varied between studies. Methodological differences include: the definition of epoch, the number of hours and days of data constituting a valid/acceptable data set, MVPA and ST cut-points, and the choice of non-wear criteria. These methodological variation tends to produce meaningful differences in MVPA and ST estimates between studies [16] and also make it difficult to compare across studies. However, while there were multiple differences between studies in accelerometry methodology (e.g. in epochs, cut-points, handling of non-wear time, duration of accelerometry monitoring), in all cases the methods were the same within studies between the obese and non-obese comparison groups, so these methodological differences probably had limited effect on the ability of studies to identify differences in MVPA and ST between the obese and non-obese. Finally, the validity of accelerometry (in particular hip-worn accelerometry, the method in almost all eligible studies) to determine ST is less well established than the validity of this placement for measurement of MVPA [59]. Hip-worn accelerometers are not designed to measure posture, and devices such as inclinometers may provide improved measurement. Accelerometers are used widely to measure ST though, and there is some evidence of validity for group-comparisons as here (obese vs non-obese comparisons) [19, 16].