The Impact of Ethnicity on Objectively Measured Physical Activity in Children

Obesity and obesity-related diseases (cardiovascular disease/metabolic risk factors) are experienced differently in individuals from different ethnic backgrounds, which originate in childhood. Physical activity is a modifiable risk factor for obesity and related diseases. Both physical activity and metabolic risk factors track to adulthood, and thus understanding the physical activity patterns in children from different ethnic backgrounds is important. Given the limitations of self-report measures in children, this study provides a review of studies which have objectively measured physical activity patterns in children from different ethnic backgrounds. From a total of 16 studies, it can be concluded that physical activity does seem to vary amongst the ethnic groups especially South Asian and Black compared to White EU (European Union). The findings are less consistent for Hispanic/Mexican American children. However, there are several methodological limitations which need to be considered in future studies. Firstly, there is a need for consistency in the measurement of physical activity. Secondly, there are a range of complex factors such as socioeconomic status and body composition which affect both physical activity and ethnicity. Studies have failed to account for these differences limiting the ability to generalise that ethnicity is an independent risk factor for physical activity.


Background
Physical inactivity is the fourth leading risk factor for global mortality [1]. It is well understood that engaging in regular physical activity (PA) has bene�cial physiological and psychological effects on health and well being in adults and children [2][3][4][5]. Speci�cally, the protective mechanism of PA for the primary and secondary prevention of obesity-related diseases including cardiovascular disease [6,7] and type 2 diabetes mellitus (T2DM) [2,8] is well established. is is because PA results in numerous physiological responses which improve cardiovascular �tness and health in adults [1,7] and children [9,10], healthy and those with cardiovascular disease [11].
Large observational studies show that physical inactivity is related to the progression to T2DM from normal glucose tolerance [12][13][14]. Engaging in regular PA may be able to delay the progression from impaired glucose tolerance to T2DM [15]. A systematic review and metaanalysis of randomised control trials concluded that exercise (150 mins/week) improved diabetes control with a reduction in HBA1c concentrations in adults with T2DM [16]. Daily walking alone has been associated with improvements in insulin sensitivity and glucose tolerance in adults [17].
Several recent reviews have also concluded that increasing PA can reduce metabolic risk factors in children. Andersen et al. [18] concluded that PA results in a lowering of blood pressure and healthy lipid pro�les in children. A metaanalysis of 14 studies found that accelerometer-measured PA was associated with cardiometabolic risk factors [19]. us, children engaging in more moderate and vigorous physical activity (MVPA) have better cardiometabolic risk factors [19,20]. Gutin and Owens [20] concluded that PA interventions in obese children (150-180 mins/week MVPA) were associated with favourable biomarker change. A review by Tompkins et al. [21] supports the role of PA in reducing the incidence of T2DM without weight loss. PA is thus described as an independent risk factor for metabolic syndrome in children independent of other factors [22].

ISRN Obesity
Metabolic disease and risk factors for metabolic disease vary by ethnic group and show higher prevalence in non-European groups [23]. Compared with White EU, people from African Caribbean backgrounds evidence lower cardiovascular disease mortality rates but experience higher risk of hypertension and T2DM [24][25][26][27][28]. British South Asian (SA) adults when compared to White EU also experience an increased risk of T2DM which is associated with increasing cardiovascular disease death rates [29][30][31][32].
In the last 10 years, growing evidence has shown that early markers of metabolic disease risk emerge in childhood which are strongly related to ethnicity and obesity [33][34][35][36]. Ethnic differences have been found in body composition between White EU compared to Black [37,38] and SA [39,40]. Research also reports that SA and Black children have higher levels of fasting insulin than those of White EU children [41]. Ethnic differences in metabolic risk factors between Black, Hispanic, and White were also reported by Casazza et al. [42]. is is important because Camhi and Katzmarzyk's [43] review has shown tracking of risk factors track from childhood into adulthood.
In adults, studies in the UK and USA have shown that PA patterns are different amongst ethnic groups but highest in White adults [44,45]. Fischbacher et al. 's [46] review of 12 studies suggests that SA adults are less active, with the majority of this information deriving from self-report measures. Self report in young children is not valid or reliable because children lack the ability to accurately recall PA patterns [47][48][49]. Questionnaires may also be interpreted differently across cultures. A review in the measurement of PA in children using indirect and direct measures of PA evidenced that there are large discrepancies in children [50]. Further reviews in the measurement of PA consistently report that objective assessment in children is the most accurate and reliable source of information [51][52][53][54][55][56].
PA patterns in adults have also been directly compared to metabolic risk. A study found that Black adults in the highest quartile for PA had 34% lower risk of developing hypertension over 6 years. Pitanga et al. 's [57] review also suggests that 185 minutes of PA for Black men and 215 minutes for Black women were the best cut-points for predicting the absence of diabetes. Williams et al. 's [32] study in SA adults showed that 21% difference in cardiovascular disease could be explained by PA.
Ethnic differences in adult metabolic disease and PA patterns are apparent. It is also known that PA can explain large proportions of ethnic differences in metabolic disease. Given that the development of metabolic diseases emerges in childhood and tracks to adulthood, it is imperative that a review is conducted to understand ethnic differences in childhood PA. Some studies have considered PA in ethnic groups, but there are differences in the measuring instruments of PA and interpretation of the data, and so no clear consensus on PA between ethnic groups is known. To provide the most accurate and reliable information on PA in ethnic groups of children, this paper will consider only objectively measured PA between ethnic groups of children. Knowing PA differences between children from different ethnic groups could inform ethnic-targeted interventions to increase PA and reduce ill health associated with lack of PA in ethnic groups. e purpose of this paper was to examine objectively measured PA in Black, Mexican American/Hispanic, and SA children and compare them to White EU to establish whether there are differences in physical activity patterns based on ethnic groups to inform physical activity interventions which will improve metabolic health.

Method
An electronic search was conducted from January 1990 until July 2012 using Pubmed, Embase, and Cochrane Database of Systematic Reviews. A child was de�ned as a subject who was under the age of 18 years in accordance with the UN convention on the right of children (UNICEF, 1989). e following search terms were used: PA, inactivity, sedentary time, exercise, ethnicity, race, ethnic minority, racial, race, ethnic, ethnicity, and children. Searches were limited to those published in the English language.

Selection.
Studies were included if PA was assessed objectively and if these �ndings were reported separately for ethnic groups. Forty-four studies compared PA between ethnic groups: of these, 28 measured physical activity subjectively and thus were removed ( Figure 1). Full-text articles were obtained for a total of 16 studies, and the following data were extracted from each paper: study design, country of origin, sampling method, instrument for gathering ethnic information, population characteristics, measuring instrument, outcome measures, and results (Table 1)

Results
A total of 16 studies were included, 13 studies were from America, two studies from the UK, and one from New Zealand (Table 1). e sample sizes varied widely from 169 to 3381 participants. Seven of the studies included a sample population above 1000 subjects. e studies reported using mainly a cross-sectional design. Results are presented under ethnic grouping.

Study Information.
A total of 13 studies assessed PA in Black children compared to White EU. Twelve of these were conducted in the USA with only one conducted in the UK [66]. Seven of these studies recruited from multicentres [62,[66][67][68][69][70][71]. A total of 9 studies included participants from Mexican American/Hispanic descent [42, 63, 67-71, 73, 74]. All these studies were conducted in the United States. Of these, six studies included multicentres of recruitment [62,[67][68][69][70][71]. Despite SAs experiencing the highest risk of metabolic disease, there were few studies examining PA patterns in this ethnic group. Two studied a mix of Indian, Pakistani, or Bangladeshi [65,66], and the other study included predominantly Indian (91.5%), Sri Lankan, Bangladeshi, and Nepalese [75]. ese studies combined ethnic subgroups. Owen et al. [66] considered the effect of ethnic subgroups but did not report statistical information for this.

Age and Gender.
For Black, three studies used girls only [62,69,71], and the remaining studies used mixed groups of boys and girls. Two studies assessed PA in preschool children (3-5 years) [58,72], Five studies included children aged 11 years or younger [59,66,68,70,73], and two studies included young people aged 12 or more [62,69]. e remaining studies used a wide age range of participants including both pubertal and prepubertal subjects [42,67,71,74].
For Mexican American/Hispanic, studies recruited participants from 6-18 years: of these studies, two examined prepubertal children [63,68], two examined children aged 12 years or above [69,71], and �ve reported combined aged groups [42,67,70,73,74]. Two out of the ten studies used girls only [69,71]. Studies in SA children used mixed samples of boys and girls. Two studies focused on SA children 9-10 years [65,66], and the remaining study included SA children aged from 6 to 16 years [75].  [42,66,69]. e remaining studies failed to describe how they gathered information on ethnic background. In Mexican American, �ve studies reported the method of gathering ethnic information and reported their criteria for ethnic reporting. is was either parental or self-reported [42,63,67,71,74].

Collection of Ethnic
3.4. e Measurement of PA. PA was assessed in 16 studies of which eleven used Actigraph accelerometers and described the cut-points used to determine time spent in activity such as sedentary or moderate and vigorous [42,59,66,67,[69][70][71][72][73][74]. e most commonly reported threshold criteria related to Trost et al. [60] or Treuth et al. [61] validation studies. One study used a Caltrac accelerometer [62]. PA was assessed by a pedometer in three studies [65,68,75] and observational analysis in one [58].

PA: Activity Counts/
Steps. Considering pedometer determined PA, one study reported that Black children were less active, engaging in fewer recorded steps than White children [68]. However, studies using Actigraph accelerometers evidenced that Black children recorded more counts per minute. is was apparent in prepubertal children [59,66,73] and for children ranging from prepubertal to pubertal (6-18 years old) [67]. Newton et al. 's study [59] was the only one of these studies to take into account differences in socioeconomic status (SES) evidencing that low socioeconomic Black girls engage in more counts/min. However, White and Jago [62] found that Black girls had less counts per day than White girls.
ere were two studies which examined PA in Mexican American/Hispanic as counts per minute or total activity using Actigraphs [63,74], and one study assessed PA using a pedometer as steps per day [68]. ese studies were all in children under 12 years. Johnson et al. 's [68] study examined three days of weekday pedometer data and found that Mexican American/Hispanic children aged from 10 to 11 years accumulated less steps per day than White children. ey also reported that region affected accumulated steps: children from urban areas accumulated fewer steps than their rural or suburban peers. However, this study did not match ethnic groups from each built environmental area to compare ethnic differences, and this may have affected the outcomes.
Two smaller studies (147 to 169 children) [63,74] and one larger study [73] assessed PA using the Actigraph accelerometer. All studies showed no ethnic differences in total PA. Byrd-Williams et al. 's [63] study of children (9.4 years study) reported that 24.5% were overweight and 15.4% obese. However, the effects of weight status were not controlled for during analysis of total PA between ethnic groups. Casazza et al. [42] also reported no difference in total PA or sedentary PA in 7-to 12-year-old children. is study also reported that Mexican American/Hispanic children had higher BMI, increased body adiposity, and lower socioeconomic status than White peers. Again, these variables were not controlled for during the analysis. �iven the in�uence of other variables in the previous studies that affect activity but were not controlled for, it is not possible to draw conclusions on Hispanic children's PA patterns assessed as activity counts or steps.
All the studies that measured PA in SA children found that they had a lower PA count or achieved fewer steps per day [65,66,75]. In addition, Owen et al. [66] considered within-ethnic differences in the SA sample, and although SAs did have lower levels of PA as a whole, there were no differences between Indians, Pakistanis, and Bangladeshis. Duncan et al. [65] controlled for BMI, hours of daylight, and age and still reported ethnic differences. ey also reported that socioeconomic status affected total activity steps and that this relationship was different on weekdays and weekends. However, ethnicity and socioeconomic status were not considered together. Duncan et al. [75] also reported signi�cant differences in steps with SES as well as age but failed to account for these differences. Duncan et al. [65] was the only study that considered body composition differences and accounted for them.

Sedentary Time.
A large multicentre study considered sedentary time in 6-11-year-old boys and girls and found that Black girls spent fewer hours a day in sedentary activity than White [70]. However, two fairly small studies in prepubertal boys and girls found no signi�cant difference in time spent in sedentary activity [42,59]. Two large multicentre studies found that Black children were more sedentary than White. Belcher et al. [67] reported that Black children aged from 6 to 11, 12 to 15, and 16 to 19 were more sedentary than White. Owen et al. [66] similarly reported, in children aged from 9 to 10, that Black children were more sedentary.
Studies used an activity count of less than 100 to determine sedentary time and thus are comparable based on activity. e results are equivocal, and this may relate to some differences in methodology. However, the large multicentre studies considering boys' and girls' sedentary behaviour together suggest that Black children are more sedentary than White children across age groups. However, Kelly et al. [69] reported that White girls, but not boys, were more sedentary. To understand sedentary behaviour in Black and White children more accurately, future studies may need to consider age, gender, and ethnicity. ere were inconsistent �ndings regarding sedentary time in Mexican American/Hispanic. Firstly, Casazza et al. [42] found no difference in total daily minutes of PA versus sedentary activity between Mexican American/Hispanic and Caucasian groups. Matthews et al. [70] reported that time spent in sedentary activity (de�ned as <100 counts min) was lower in Mexican American/Hispanic groups than in Caucasians. is study used the Treuth et al. [61] [69]. e study also reported no difference in time spent in MVPA between Mexican American/Hispanic and White girls [69]. Another multicentre study reported that Black girls and Mexican American/Hispanic girls spend less time in MVPA than White (10-13 years) [71]. e differences may thus be explained by a number of reasons, such as different population characteristics in relation to age and ratio of White to Black children and the inclusion criteria for activity monitoring.
In addition, Kelly et al. [69] reported that Mexican American/Hispanic had higher BMI than White, but failed to control for this during analysis. Pate et al. [71] did not report BMI or any other measure of body composition. is study also considered three different cut-points and different sampling periods (30 and 60 minutes MVPA) and suggested that ethnic differences were dependent upon the cut-points applied. No signi�cant ethnic difference was found for 60 minutes MVPA describes using ≥4.6 METS, ≥3.8 METS, and ≥3.0 METS. However, ethnic differences were found for 30 minutes MVPA for ≥4.6 METS and ≥3.8 METS. e statistics were not displayed separately for White and Mexican American, but the means show that higher number of White children achieved these guidelines. Fewer children engage in large amounts of daily activity which may explain the lack of differences at 60 minutes. e study highlights the importance of accelerometer cut-points, but to date no broad consensus has been met on what to use.
For Black boys and girls combined, the �ndings were more consistent. One small study suggested that Black and White children engage in similar MVPA [42]. However, the remaining �ve studies found that Black children spend more minutes in MVPA [59,66,67,72,73]. is was found across a range of ages including preschool [72], prepubertal [59,66,73], and prepubertal and pubertal [67]. Of these �ve studies, four used large multicentre sampling. All �ve studies used Actigraph accelerometers to assess PA but employed different thresholds to determine MVPA. e results from larger studies of both boys and girls would, therefore, suggest that Black children spent more time in MVPA.
e �ndings in Mexican American/Hispanic boys and girls combined were less consistent. Two studies reported that Mexican American/Hispanic children engage in fewer minutes of MVPA than White EU [42,74]. However, two studies also reported no difference in mean percentage of time in MVPA [63,73]. Again, these studies used different accelerometer thresholds for MVPA. Casazza et al. [42] used the soware calculations to report time spent in different intensities but did not report how these are calculated with reference to MET levels or activity counts that translate to MVPA. Casazza et al. [74] de�ne MVPA as anything greater than 3 METS. Byrd-Williams et al. [63] used the Puyau et al. [64] accelerometer cut-points which de�nes MVPA as counts greater than or equal to 3000 or greater than/equal to 0.5 kcal/kg/min (equivalent to 3 METS or more), developed from data from 6-16-year-old children.
However, Byrd-Williams et al. [63] �rstly accounts for age by focusing on 9-10 years unlike the other two studies which include children aged from 7 to 12 years. ey report no differences in sex, height, weight, or age. Yet, almost half of the samples (43%) were overweight or obese. Body composition differences between ethnic groups were also reported [42,74], but none of the studies controlled for these. All of the three studies also had uneven samples of Mexican American/Hispanic and White, with more White people taking part, and were based on small samples (<250 in total). e small sample sizes also limited the analysis in a number of ways. Furthermore, Byrd-Williams et al. [63] report gender differences in MVPA, but the sample size restricted exploring for gender and ethnicity interaction effects as well as controlling for differences in BMI. Secondly, SES differences were reported [42,74], whereby Mexican American/Hispanic children had lower SES status, but again the small sample size limited the ability to assess these effects within ethnicity on PA counts, and thus it is not possible to state that ethnic differences are independent of other in�uences.
Large studies have shown that Mexican American/Hispanic children spend more minutes in MVPA [67]. Belcher et al. [67], using data from NHANES, also reported results for boys and girls separately but found no signi�cant difference in Hispanic/Mexican American girls' versus White girls' engagement in moderate to vigorous activity but found that Mexican American/Hispanic boys spent more time in MVPA. ey reported results separately for age groups as well suggesting that differences in activity patterns between Hispanic/Mexican American and White exist in 6-11-and 12-15-year-old children but that there were no differences found in children aged 16-19 years. e study also accounts for difference in weight status with ethnic groups PA. e only ethnic differences found were for overweight children aged [16][17][18][19] in which White were less active than Hispanic. Belcher et al. [67] used accelerometer thresholds by Trost et al. [60] which provide thresholds based on age. eir study highlights that ethnic differences in activity patterns are likely to be dependent on age, gender, and adiposity. Belcher et al. [67] also reported a three-way interaction between age, BMI, and ethnicity for minutes in MVPA, evidencing that overweight Hispanic/Mexican American boys and girls spent less minutes in MVPA than non-overweight. is relationship might be complex and might explain some of the inconsistencies in the �ndings, which makes it difficult to draw conclusions from the data.
For SA children, the �ndings consistently reported that they were less active than White children. Duncan et al. [65] used a pedometer and cut-points of 15,000 for boys and 12,000 for girls to determine whether children met PA guidelines (60 mins/day). eir study found that a lower proportion of SA children met the guidelines for health. Owen et al. [66] also reported that lower numbers of SA children met the guidelines. ey also found that SA children spent less time in MVPA. e �ndings in SA children would suggest that SA children are less active than White.

Discussion
ere are few studies that have examined ethnicity and objective PA in children. e majority of research is based on cross-sectional studies predominantly from the USA. ere is limited information about PA patterns from the UK and few studies considering SA ethnic children.
e results would suggest that Black children are more active based on activity counts and time spent in moderate and vigorous PA [59,66,67,73] but that they also spend more time in sedentary activity [66,67]. ese �ndings, however, might be different for Black girls, and more high quality studies are needed to understand gender, age, and ethnicity effects on PA especially sedentary activity in Black children.
For Hispanic/Mexican American children compared to Whites, the results are equivocal. is may be explained by methodological limitations and failure to account for in�uential factors such as age, gender, body composition, and SES which made it impossible to conclude whether there were genuine differences in PA between Hispanics and Whites. More high quality studies are needed in Hispanic/Mexican American populations which control for confounding factors in order to gain a clearer understanding of PA patterns.
ere were fewer studies in SA children, but the studies that were available consistently reported that SA children were less active engaging in fewer minutes of MVPA, more sedentary time, and being less likely to meet PA guidelines [65,66,75].

Methodological
Limitations. ere were a number of methodological differences among the studies which made it difficult both to analyse the results and also to be able to compare between study �ndings. 4.1.1. e Measurement of PA. e measurement of PA is a complex phenomenon, especially in children. Firstly, the objective measurement of PA is a strength of the review and in particular the studies used accelerometer data which is viewed as an accurate way to measure PA in children. Accelerometers make it possible to measure PA count as well as intensity and frequency of PA whilst eliminating human error. However, accelerometers do not capture all types of PA especially nonweight bearing activities like swimming or cycling [60]. When examining the intensity of PA, it is possible to assess whether children are ful�lling national/international guidelines for levels of PA which are described in terms of MVPA.
ere are also differences in the way PA has been reported. Pedometers consider total activity steps, but accelerometers consider mean activity counts per minute. For accelerometer data speci�cally, these mean activity counts daily are then converted to time spent in sedentary activity, light activity, or MVPA using accelerometer cut-points. However, not all studies report each of these elements. e accuracy of some of the conversions needs to be considered because different accelerometer cut-points are applied which have been validated on a wide range of children of differing age and gender, which has not been considered. ere are also no studies that consider ethnicity in the validation of these thresholds. Currently, there is no standardised measure of accelerometer cut-points which makes it difficult to compare between PA studies. is has made it difficult to determine whether PA differences do exist or whether differences are due to differences in the criteria used in each study. A recent systematic review of objective measures of PA measurement reanalysed all PA data in children using the three most popular cut-points [54]. ey found signi�cant differences in the amount of sedentary behaviour and MVPA when different cut-points were used suggesting that the assessment of engagement in PA is dependent on the cut-points applied to the data [54].
e studies also varied in the measuring time for periods of activity. Studies used mainly 60, 30, or 15 second intervals. e nature of children's PA is intermittent, and thus using longer epochs may fail to describe the true nature of these patterns. Reilly et al. 's [54] systematic review also reanalysed all PA data in children using 15,30,45, and 60 second intervals. ey found that there was a small but signi�cant difference in the results using different intervals but concluded that there was still limited evidence concerning the use of long or short measuring periods. e study made recommendations that moderate and vigorous PA should be analysed together in order to reduce misclassi�cation from epoch lengths. is paper considers moderate and vigorous PA together to minimise the effect of choice of epoch on PA.
e studies also varied in the number of days used for monitoring ranging from 1 day to 7 days. e period of data collection that has been included has also varied from 5 hours to less than 18 hours. is has important implications as data reliability is reported to increase with an increased number of days of monitoring. Seven days monitoring with inclusion criteria of 10 hours per day of data is reported to be the most reliable [76]. Out of the 13 studies, �ve reported inclusion criteria of 10 or more hours [59,66,67,70,71]. e remaining studies inclusion criteria were less than 10 hours or failed to report them. is may affect the validity and reliability of the results produced. Future studies should include a minimum of 10 hours.
It is also important for studies to gain a whole picture when examining PA patterns. For example, studies in Black children found that Black children were more sedentary. is would suggest that they are less active. However, studies have also reported that Black children were more active. us by considering only one component of PA, it can cause overgeneralisation about PA patterns which might not be a true indication of activity patterns. By gaining the whole picture, it is possible to conclude that Black children can be both more active and more sedentary, and thus their activity patterns are likely to lie at both extremes. In addition, the majority of studies have unequal, usually smaller numbers of ethnic groups, and do not report response rates in ethnic groups. It is not known whether these are representative samples of ethnic groups, which limits the ability to generalise the results.

Age/Gender
. e studies consider a range of ages from preschool, prepubertal, and adolescent and in some cases mixtures of age groups. Yet, during analysis, very few studies have accounted for age differences. However, looking at age differences separately, a study has found differences in PA in different age groups of children [67]. Age differences in PA have also been described in other studies [60,77]. Gender differences in PA patterns have also been described in studies [66,78]. e majority of studies in this paper sampled both boys and girls, combining groups to compare ethnic differences. A few studies considered girls only, but none considered only boys. From these studies, the interaction effects between ethnicity and gender on PA remain unclear. It is likely that reporting results separately for gender and age groups would yield different results.

Ethnicity.
Not all studies reported how they gathered ethnic information or how they de�ned ethnic groups. In studies that did report this information, ethnicity was gained from self-report measures. A few studies also de�ned how ethnicity was de�ned. ere are differences in assessing ethnicity, and there are likely to be differences in tracing ethnicity back more than 3 generations to current generations. In addition, the studies and this paper do not consider ethnic heterogeneity, providing data on ethnicity as a homogenous group. For instance, there are likely to be differences in SA subgroups such as Indian, Pakistani, and Bangladeshi. Only Owen et al. [66] considered this (but did not report the statistical results�. is does signi�cantly increase the size of the sample that needs to be examined in any given study. ere are also likely differences in "Mexican American born" compared to "immigrant born" Hispanic PA patterns. ere may also be differences in White ethnic groups such as those born in the UK, those from the USA or those with another European background. is paper also only makes comparisons between Black, Mexican American/Hispanic, or SA children compared to White children. e difference between other ethnic groups is not known due to paucity of data on these ethnic groups.
Gaining nationally representative information in ethnic groups is a challenge. Ethnic groups tend to cluster in sociodemographically deprived areas, and it becomes dif-�cult to generalise the data to a wider population. Some studies provided in this paper did consider SES and found differences for ethnic groups with White being less disadvantaged [42,59,74,75]. However, few studies accounted for SES during analysis. Newton et al. 's [59] study suggests a difference in PA between low SES Black and low SES White but no other differences relating to SES, ethnicity, and PA. Duncan et al. [65] reported differences in SES status on weekdays and weekends, suggesting that high SES engaged in less PA on weekends but more on weekdays. Given that ethnic groups tend to cluster in the most disadvantaged areas, SES is an important covariable. A study found that poverty was increased in non-White neighbourhoods [79].
Additional studies have reported that SAs are the most socioeconomically deprived and are likely to live in the most deprived areas [80]. A number of studies have also associated deprivation with physical activity [81][82][83][84][85][86][87]. e reasons for differences in patterns of PA may thus be more complex with environmental factors being a determinant for PA. For instance, the environment may affect the opportunities available to be physically active but this may depend on SES. Perceptions about safety, facilities, and spaces and neighbourhood walkability are just a few environmental factors associated with physical activity [88][89][90][91][92]. However, there appear to be no studies examining environmental in�uences that have considered ethnicity, and thus it is hard to ascertain whether PA differences by ethnic groups are representative of environmental and SES differences rather than ethnic differences per se.
Finally, the role of body composition on PA in ethnic groups needs to be considered. Ethnic differences in body composition have been reported [35,37,62,[93][94][95][96] as have the differences in PA with body composition [42]. In this paper, body composition was assessed by proxy using BMI which can be problematic in ethnic groups that have different amounts of �tness recommendations. is paper highlights a number of important recommendations. Firstly, studies need consistency in methodology especially the need for universal accelerometer cut-points thresholds. Secondly, future studies would bene�t from controlling for age, gender, SES, and body composition. irdly, there is a need for UK-based studies as the effect of ethnicity might be different depending on the country. Finally, future studies should also consider longitudinal and experimental designs to enable more inferences about PA and the development of metabolic disease. Understanding ethnic differences in PA will enable PA interventions to be focused at speci�c ethnic groups to increase PA.

Conclusion
is paper does suggest that there are ethnic differences in PA for Black and SA children, but �ndings are unclear for Mexican American/Hispanic. It is not clear whether these differences are due to physiological, cultural, or environmental differences or a combination of these factors. Future studies with improved methodology are necessary to examine the impact of low levels of PA on cardiometabolic risk factors in childhood and to develop effective ethnically sensitive interventions to promote physical activity in risk groups.