- Email: [email protected]
Tracking cardiorespiratory fitness and physical activity in children with and without motor coordination problems
Accepted Manuscript Title: Tracking cardiorespiratory fitness and physical activity in children with and without motor coordination problems Author: John Cairney Scott Veldhuizen Sara King-Dowling Brent E. Faught John Hay PII: DOI: Reference:
S1440-2440(16)30185-2 http://dx.doi.org/doi:10.1016/j.jsams.2016.08.025 JSAMS 1385
To appear in:
Journal of Science and Medicine in Sport
Received date: Revised date: Accepted date:
9-3-2016 17-8-2016 21-8-2016
Please cite this article as: Cairney John, Veldhuizen Scott, King-Dowling Sara, Faught Brent E, Hay John.Tracking cardiorespiratory fitness and physical activity in children with and without motor coordination problems.Journal of Science and Medicine in Sport http://dx.doi.org/10.1016/j.jsams.2016.08.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Self-reported physical activity does not explain differences in cardiorespiratory fitness between children with and without motor coordination problems: a longitudinal analysis Tracking cardiorespiratory fitness and physical activity in children with and without motor coordination problems
John Cairney, PhDa,b,e* ##Email##[email protected]##/Email##, Scott Veldhuizen MA , PhD Candidatea,e, Sara King-Dowling, B.Sc.Kin, PhD Candidatec,e, Brent E. Faught, PhDd, John Hay, PhDd aDepartment of Family Medicine, McMaster University, Hamilton, Canada bFaculty of Kinesiology & Physical Education, University of Toronto, Toronto, Canada, Department of Psychiatry & Behavioural Neurosciences, CanChild Centre for Studies in Childhood Disability, Offord Centre for Child Studies, McMaster University, Hamilton, Canada cDepartment of Kinesiology, McMaster University, Hamilton, Canada dDepartment of Health Sciences, Brock University, St. Catharines, Canada eINfant and Child Health (INCH) Lab, McMaster University, Hamilton, Canada *
Corresponding author at: Faculty of Kinesiology & Physical Education, University of Toronto; Departments of Family Medicine, Psychiatry and Behavioural Neuroscience; CanChild Centre for Childhood Disability Research and Offord Centre for Child Studies, McMaster University; Address: 100 Main St West, 5th Floor, Hamilton, Ontario, Canada, L8P 1H6; Tel.: 905-525-9140 ext: 28506; Fax: 905527-4440.Abstract Objectives Previous research has shown children with Developmental Coordination Disorder (DCD) have lower cardiorespiratory fitness (CRF) than typically developing (TD) children. This has been hypothesized to be due to an activity deficit, whereby poor motor functioning discourages children from participating in physical activities, but this hypothesis has not been directly tested. In this study, we use longitudinal data to measure the extent to which physical activity explains differences in CRF between children with and without motor coordination deficits. Design Longitudinal observational study
Methods The study sample is an open cohort of children, numbering 2278 at baseline (age 9-10), that was followed for up to 5 years (to age 13-14). Motor skills were assessed once over the study period. Children scoring at or below the 5th percentile (n=103) on the Bruininks-Oseretsky Test of Motor Proficiency-Short Form were considered to have possible DCD (pDCD). CRF (estimated peak VO2) was estimated from performance on the Léger 20m shuttle run test, and physical activity was measured with the Participation Questionnaire. Both fitness and physical activity were measured up to 7 times over the study period. Results Children with pDCD had significantly lower CRF than their TD peers at each time point. CRF declined for both groups, but this decline was steeper for children with pDCD. Physical activity explained only a small part of the difference in CRF. Conclusions The activity deficit did not contribute to the persistent and gradually widening gap in CRF between children with and without possible DCD. Possible reasons for this and future directions are discussed. Keywords: Developmental Coordination Disorder; motor skills; motor skills disorders; physical fitness; physical activity; youth
1
Introduction Developmental coordination disorder (DCD) is a chronic neurodevelopmental condition affecting between 2 and 6 percent of all children1,2. The functional presentation of the disorder is heterogeneous, but involves deficits in gross and/or fine motor skills1. Existing studies show that children with DCD are at greater risk for a number of physical and mental health-related problems, including poor physical fitness3,4 (particularly cardiorespiratory fitness (CRF)5,6), overweight and obesity7, and depression and anxiety8. Studies have also shown that children with DCD are less physically active than their typically developing peers9,10. Lower levels of CRF are one of the more concerning health outcomes associated with DCD4. Improving CRF produces physiological adaptations that improve the efficiency of the respiratory and circulatory oxygen transport system, and these adaptations confer numerous health benefits, including reduced risk for cardiovascular disease and reduced all-cause mortality11. The activity deficit hypothesis describes connections between poor motor coordination, physical inactivity and CRF (as well as overweight/obesity) as a negative feedback loop4,5,9. It is hypothesized that poor motor coordination discourages children from participating in physical activities, which negatively affects the development of physical fitness and increases the risk for unhealthy weight gain. Poor health-related fitness, in turn, leads to further reductions in physical activity4. In this way, poor motor coordination becomes a risk factor for negative health outcomes later in life12. Evaluating the activity deficit hypothesis requires longitudinal data with repeated assessments of physical activity (PA) and fitness. Few studies have tracked changes in CRF in children with DCD 3,13,14, and these studies are limited by small samples3,14 and/or by relatively short follow-up periods3,13. More importantly, no attempts have been made to test the activity-deficit hypothesis with these data. If an activity deficit is responsible for poor fitness among children with DCD, then a measure of physical activity should explain a substantial part of the association between motor proficiency and fitness. In a previous paper, we examined changes in CRF over a 2.5-year period, and showed that both boys and girls with possible DCD (pDCD) have poorer fitness levels than typically developing children13. Since this analysis was conducted, data collection for this study cohort has been completed. The final dataset includes an additional 2 years of follow-up, making it possible to track CRF from age 9 to age 14. In this report, we also examine the associations between self-reported participation in organized and active free play and CRF in children with and without pDCD. Methods The Physical Health Activity Study Team (PHAST) project was a longitudinal prospective cohort study of children designed to examine change over time in PA and health-related fitness in children with and without poor motor coordination. Details of the design have been described in previous publications10,15. Data collection commenced during the 2004-2005 school year. The target population was all children in the fourth grade who were enrolled in the public school system in the Niagara region of Ontario, Canada. We approached 92 schools, of which 75 (83%) agreed to participate. At baseline, we obtained written informed consent from the parents of 2278 of 2378 (96%) children. Our sample includes the 1954 (86%) of these participants who had a complete motor assessment and data on PA and CRF, as well as a further 163 children who entered the study in later waves (total n=2117). The first formal data collection effort (wave 1) occurred in May/June of the grade 4 school year (age 9-10). For the next two school years (waves 2 through 5), data collection took place in the fall (Sept/Oct) and again in the spring (May/June). Beginning in wave 6, data were collected once per year, always in the spring (May/June). Data collection was done in school classes. New students sometimes joined participating schools during the study period. As excluding these children would have served no obvious purpose, and would have required the provision of alternative activities during assessments, new students were enrolled in the study. As a result, PHAST was an open cohort, with a total sample size sometimes rising over time. The sample decreased
2
sharply in size after wave 5 (Table 1), because many students transferred to different schools to begin grade 7 (age 11-12). All data were collected by trained research assistants and supervised by the core investigators. Ethics approval was obtained from the District School Board of Niagara and the research ethics board at Brock University. To assess Cardiorespiratory Fitness (CRF) during each wave, we used the Léger 20m shuttle run test16, which is a well-established field measure used to estimate maximal VO2 in children that correlates well with lab-based tests (r = 0.72 17). The shuttle run test is well-suited to school-based assessments, where laboratory testing is not feasible. It involves running back and forth between two lines set 20m apart in synchrony with a sound signal16. At each successive stage, the time between signals or beeps decreases, requiring the participant to run faster. When a child fails to cover the 20m distance before the beep sounds on two successive occasions, the last stage reached is recorded and the test is terminated. This information is then used to determine the maximal speed obtained, which is entered into an equation to calculate estimated relative peak VO218: Estimated Peak VO2 = 31.025 + 3.238 (maximal speed) – 3.248 (age) + 0.1536 (speed x age). Participants performed the test in groups of up to 15, under the supervision of trained research assistants, who also assisted the children with pacing and encouragement. Children were classified as having ‘possible’ Developmental Coordination Disorder (pDCD) based on the results of their performance on the short form of the Bruininks-Oseretsky Test of Motor Proficiency 1st Edition (BOT-SF)19. The BOT-SF uses a subset of core items from the full measure to evaluate different components of motor coordination (fine motor ability; gross motor ability; balance; reaction time). The BOT-SF has been validated against the full BOT in children ages 8 to 14 years, with correlations of 0.90 to 0.9119. In PHAST, most children’s motor coordination was assessed only once. Schools were randomly divided into 3 groups, and motor testing occurred separately for each group over three different waves of data collection. Children were not all tested at the same wave because of time and cost considerations. To evaluate the longitudinal stability of our measure of pDCD, a random subset of participants (n=77) were re-tested on the BOT-SF approximately 2 years after their original assessment. The correlation between tests was .707. Trained research assistants administered the test individually to each child in the school’s gymnasium during regular school hours. To the best of our ability, assessments were conducted in quiet places where distractions could be minimized and privacy respected. Children scoring at or below the 5th percentile on the BOT-SF were identified as having poor motor functioning. Although we excluded children with known learning or physical disabilities, we did not assess the full DSM criteria for DCD (i.e. daily impact and clinician motor assessment)1. We therefore use the term ``possible DCD'' (pDCD). This descriptor is frequently used in the literature (e.g. 5,7,8,10,12,13,15). To assess the validity of our evaluation of motor skills, we conducted a sub-study on a random subset of 24 children we identified as pDCD. Twenty-one (88%) scored at or below the ``at risk'' cutoff for DCD (16th percentile) on the Movement Assessment Battery for Children20, which was administered by a pediatric occupational therapist who was blind to the results of the BOT-SF15. This high positive predictive value supports the validity of the original assessment. To measure physical activity (PA) participation, children completed the Participation Questionnaire (PQ21). The PQ is a self-report questionnaire that asks about the frequency of current participation in organized activities (e.g., team sports) and active free play (e.g., bike riding). Scores range from 0 to 20 for free play and from 0 to 29 for organized activities, with higher scores indicating higher levels of activity. Validation studies have shown that the PQ is associated with body fat, aerobic capacity, teacher ratings of physical ability21, gender, and urban/rural residence10,21. It was developed for children in grades
3
4 to 9 (ages 8-16), the two-week test-retest reliability has been measured as 0.81 for elementary school and 0.89 for secondary school students21. We used mixed effects modeling to fit growth curves representing change over time in CRF. This method allows us to generate estimates using all time points available for each student. We included a random intercept at the child level, a random slope for age, and used an unstructured covariance matrix. To allow for the possibility of a non-linear relationship between age and the outcome, we fit a series of models with fractional polynomial transformations for the sample as a whole and for boys and girls separately. This demonstrated that the relationship was essentially linear, and we therefore proceeded without transforming child age. In order to reflect known sex differences in CRF, we included age*sex interaction terms in all models. This approach allows for different slopes and intercepts for boys and girls, while making it possible to reflect a common effect for pDCD status. To check the appropriateness of this approach, we tested a 3way interaction between age, sex, and pDCD. We did not attempt to standardize our measures using agespecific norms. Although issues of scaling make it difficult to draw conclusions about the nature of agerelated change in CRF, this analysis is concerned with group differences. We first fit a model with age, sex, pDCD status, and their interactions only (Model 1), and then added self-reported PA (Model 2). For both sets of results, we calculated expected group differences by age, and calculated slopes for the 4 groups of individuals produced by the combination of sex and DCD status. Other work on this sample has shown that children with pDCD had higher body mass index (BMI) values than other participants22. To evaluate the possibility that BMI modified the effect of pDCD on CRF, we fit a further model including an interaction between these variables. As the design was an open cohort, attrition typically resulted from movement between schools, and the analysis chosen makes use of all available measurement points, we left missing data as missing. We used Stata 14 for all analyses. Results Descriptive statistics are presented in Table 1. At baseline, there were 97 children (5.0%) with pDCD. Of these, 56 (58%) were girls and 41 (42%) boys. As the study was an open cohort, the number of children with pDCD rises to 103 when participants joining in later waves are included. The analysis included a total of 12,338 follow-ups on 2117 children. [Insert Table 1 about here] The number of boys and girls in the cohort were roughly equal, and the average age of children at the start of the study was 9.9 years (SD=0.4). On average, the pDCD group reported less engagement in both organized and free play activity at each wave. Crude means show slight declines over time for CRF and organized PA, and a slight increase for free PA, as measured by the self-report Participation Questionnaire. At all time points, and for both sexes, children with DCD had substantially lower CRF. This gap was somewhat smaller among girls. Our measure of CRF (estimated relative peak VO2) decreased over time in all groups, with the slowest decline among typically developing boys and more rapid decline among both boys and girls with pDCD (Fig 1). [insert Figure 1 about here] Subtracting the derived slopes for groups with and without pDCD shows that CRF declined among children with pDCD by an extra 0.53 (95% CI = 0.24 to 0.82) units per year before self-reported activity is taken into account. As we found no 3-way interaction between age, sex, and DCD status, these results are the same for males and females. This value does not change substantially when the PQ subscales are included as covariates (estimate = 0.55, 95% CI = 0.26 to 0.84) (Table 2). The main effect of pDCD status is also only slightly attenuated in Model 2. Neither the persistent gap between groups nor its gradual widening, therefore, can be attributed to recent self-reported PA. A final model exploring the effects of BMI yielded significant main effects for BMI (estimate = -0.49, p<0.01) and pDCD status (estimate = -10.8, p<0.01), as well as a significant interaction (estimate = 0.30,
4
p<0.01). The interaction indicates that the effect of BMI is weaker for children with pDCD, which, along with the increased size of the main effect for pDCD status, is consistent with the view that children with pDCD have poor fitness irrespective of BMI. The adjusted absolute difference between children with and without pDCD did not change in this model (in Model 2, AAD = 5.08, 95% CI = 4.35-5.82; in the BMI model, AAD = 5.11, 95% CI = 5.47-5.75). [Insert Table 2 about here]
Discussion Performance on the shuttle run in our sample was very similar to that reported in a recent meta-analysis23, and showed closely comparable patterns of age and sex differences: The mean stage reached increased for both boys and girls, while body mass-scaled VO2 peak decreased. Consistent with previous work reporting on smaller samples, our results also show that children with pDCD, regardless of sex, have poorer CRF than typically developing children6,24. Moreover, differences in CRF between pDCD and typically developing children appear to emerge early – they are present at the first wave of data collection in our study – and not only persist3, but increase slightly through childhood to adolescence14. A major focus of the present study was to test existing theories concerning the activity-deficit hypothesis in children with DCD5,9. While our results show that participation in organized and active free play are positively associated with CRF, these measures explain only a small part of the overall association between CRF and pDCD, and essentially none of the difference in rates of change. Given the intuitive appeal of the activity-deficit hypothesis as an explanation for poorer CRF in children with DCD, this result is somewhat surprising. As we did observe a group difference in CRF, some other explanation must be sought. One possibility is that our data on CRF, PA, or both, were affected by measurement issues. Self-reported measures of PA have been shown to agree poorly with objective measures25, which may therefore have reduced our ability to measure effect modification. It is also possible that our measure, which was concerned largely with the frequency of PA, did not adequately measure intensity of effort. If participants in the pDCD group exercised at lower intensity levels, this would help to explain why frequency of PA fails to explain differences in CRF. Another measurement-related issue concerns the use of field-based protocols to estimate CRF. It is conceivable that running gaits were less efficient in the pDCD group, or that difficulty with rapid stopping and turning, or differences in motivation or self-confidence, led pDCD participants to drop out of the test prematurely. This is also a possible explanation of the effect we noted for BMI, as pDCD children may have had difficulty with the motor demands of the test (principally, rapid turning) irrespective of body mass. Cairney and colleagues6 however, have shown that the difference in CRF between children with and without pDCD remained constant regardless of whether a field-based protocol or direct laboratory measurement (with cycle ergometers) was used. Another possibility is that PA in our sample was insufficiently intense to produce changes in CRF. Armstrong and colleagues note that, although aerobic fitness and PA are two of the most studied variables in pediatric exercise science, over three decades of observational research have failed to confirm any meaningful association between them26. An analysis of 15 years of data from the Amsterdam Growth and Health Study, for example, found that a 30% increase in self-reported habitual PA resulted in only a 2-5% increase in VO2 maximum27. Even when studies that use both objective measures of PA (e.g. accelerometry) and direct measures of peak VO2 are considered on their own, there are no strong relationships between habitual activity and aerobic fitness in children and youth26. On the other hand, numerous studies of youth athletes and of aerobic fitness responses to exercise training (e.g.28) show clear positive effects of both sport participation and structured exercise training on peak VO2. One obvious explanation for these discrepancies is that children in general-population studies are insufficiently physically active with regard to intensity and duration in everyday, habitual PA to increase peak VO2.
5
This may account for the generally weak associations in our sample: even though children with DCD are less active than others, neither group is active enough to maintain CRF, with pDCD children further below the necessary level. Poor physical fitness has been linked to many long-term health consequences11, and this association begins early in life. CRF is associated with cardiovascular disease risk factors in children and adolescents 29 , and lower aerobic fitness in youth is associated with increased adiposity and higher cardiovascular disease risk in adulthood30. Since CRF tracks moderately well from childhood to adulthood30, it is plausible that low levels of CRF in children with pDCD may increase their risk of cardiovascular disease and other illnesses later in life. One implication of these results is that targeted interventions might usefully aim not only to improve motor abilities, but to increase CRF. This may require intensities or volumes of PA greater than those observed in our sample. Our findings on gender differences, which are largely consistent with results from general-population samples, also imply that the design of interventions in this context should give careful consideration to the fitness gap between boys and girls in this population. Future longitudinal research intended to elucidate the association between DCD and CRF should also begin with younger children. The fact that group differences are present at the beginning of our study makes it impossible to clearly establish precedence and to determine whether data are consistent with a causal hypothesis. Research should also consider additional variables (e.g. motivation, PA intensity, self-efficacy) that might mediate the relationship between motor coordination and fitness. Conclusion The purpose of this study was to evaluate the activity deficit hypothesis in a large cohort of children by testing whether levels of measured physical activity accounted for poor fitness among children with pDCD. Our results show that they do not: Although we have suggested possible explanations for the absence of effect modification by PA in our models, we have found no good evidence that PA explains overall group differences in CRF. Although this difference increases with age, it is substantially present at baseline, when most children were 9-10 years old. This suggests than factors operating in early childhood, which may include an activity deficit, produce differences in CRF that then persist into adolescence. As the effects of poor motor functioning on CRF begin early in life, greater efforts should be made to track the early development of fitness in children with and without poor motor coordination. Future studies using objective PA assessment with larger samples of children with DCD would be useful in either detecting an effect of PA on CRF differences or in ruling one out. Practical Implications - Children with motor coordination problems have poorer cardiorespiratory fitness than typically developing children. - This fitness difference is present at age 9 and increases somewhat through age 14. - The gap in fitness is not explained by differences in self-reported physical activity at these ages.Acknowledgements Financial support for this study was provided by the Canadian Institute of Health Research, Award No. 66959; Dr. Cairney is supported by an endowed professorship through the Department of Family Medicine at McMaster University. References
6
[1] ;1; American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (5th Ed.).Arlington, VA: American Psychiatric Publishing ; 2013. [2] Blank R, Smits-Engelsman B, Polatajko H, et. al.;1; European Academy for Childhood Disability (EACD): Recommendations on the definition, diagnosis and intervention of developmental coordination disorder (long version)*. Dev Med Child Neurol. 2012;54(1):54-93. [3] Haga M.;1; Physical fitness in children with high motor competence is different from that in children with low motor competence. Phys Ther. 2009;89(10):1089-1097. 10.2522/ptj.20090052. [4] Hands B, Larkin D.;1; Physical fitness and developmental coordination disorder. In: Cermak S, Larkin D, eds. Developmental Coordination Disorder. Albany, NY : Delmar; 2002:172-184. [5] Cairney J, Hay JA, Wade TJ, et. al.;1; Developmental coordination disorder and aerobic fitness: is it all in their heads or is measurement still the problem? Am J Hum Biol. 2006;18(1):66-70. 10.1002/ajhb.20470. [6] Cairney J, Hay J, Veldhuizen S, et. al.;1; Comparison of VO2 maximum obtained from 20 m shuttle run and cycle ergometer in children with and without developmental coordination disorder. Res Dev Disabil. 2010;31(6):1332-1339. 10.1016/j.ridd.2010.07.008. [7] Cairney J, Hay J, Veldhuizen S, et. al.;1; Trajectories of relative weight and waist circumference among children with and without developmental coordination disorder. Can Med Assoc J. 2010;182(11):1167-1172. 10.1503/cmaj.091454. [8] Missiuna C, Cairney J, Pollock N, et al.;1; Psychological distress in children with developmental coordination disorder and attention-deficit hyperactivity disorder. Res Dev Disabil. 2014;35(5):1198-1207. 10.1016/j.ridd.2014.01.007. [9] Bouffard M, Watkinson EJ, Thompson LP, et. al.;1; A test of the activity deficit hypothesis with children with movement difficulties. Adapt Phys Act Q. 1996;13(Journal Article):61-73. [10] Cairney J, Hay JA, Veldhuizen S, et. al.;1; Developmental coordination disorder, sex, and activity deficit over time: a longitudinal analysis of participation trajectories in children with and without coordination difficulties. Dev Med Child Neurol. 2010;52(3):e67-e72. 10.1111/j.1469-8749.2009.03520. x. [11] Kodama S, Saito K, Tanaka S, et al.;1; Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: a meta-analysis. JAMA. 2009;301(19):2024-2035. 10.1001/jama.2009.681. [12] Faught BE, Hay JA, Cairney J, et. al.;1; Increased risk for coronary vascular disease in children with developmental coordination disorder. J Adolesc Health. 2005;37(5):376-380. 10.1016/j.jadohealth.2004.09.021. [13] Cairney J, Hay J, Veldhuizen S, et. al.;1; Trajectories of cardiorespiratory fitness in children with and without developmental coordination disorder: a longitudinal analysis. Br J Sports Med. 2011;45(15):1196-1201. 10.1136/bjsm.2009.069880. [14] Hands B.;1; Changes in motor skill and fitness measures among children with high and low motor competence: a five-year longitudinal study. J Sci Med Sport. 2008;11(2):155-162. [15] Cairney J, Hay J, Veldhuizen S, et. al.;1; Comparing probable case identification of developmental coordination disorder using the short form of the Bruininks-Oseretsky Test of Motor Proficiency and the Movement ABC. Child Care Health Dev. 2009;35(3):402-408. 10.1111/j.1365-2214.2009.00957. x. [16] Léger LA, Lambert J.;1; A maximal multistage 20-m shuttle run test to predict VO2 max. Eur J Appl Physiol. 1982;49(1):1-12. 7
[17] Liu NY, Plowman SA, Looney MA.;1; The reliability and validity of the 20-meter shuttle test in American students 12 to 15 years old. Res Q Exerc Sport. 1992;63(4):360-365.10.1080/02701367.1992.10608757. [18] Léger L, Lambert J, Goulet A, et. al.;1; [Aerobic capacity of 6 to 17-year-old Quebecois--20 meter shuttle run test with 1 minute stages]. Can J Appl Sport Sci. 1984;9(2):64-69. [19] Bruininks RH.;1; Bruininks-Oseretsky Test of Motor Proficiency: Examiner’s Manual. Circle Pines, Minn.: American Guidance Service; 1978. [20] Henderson SE, Sugden DA.;1; Movement Assessment Battery for Children. London: Psychological Corporation ; 1992. [21] Hay JA.;1; Adequacy in and Predilection for Physical Activity in Children. Clin J Sport Med. 1992;2(3). 10.1097/00042752-199207000-00007. [22] Joshi D, Missiuna C, Hanna S, et. al.;1; Relationship between BMI, waist circumference, physical activity and probable developmental coordination disorder over time. Hum Mov Sci. 2015;40:237-247. 10.1016/j.humov.2014.12.011. [23] Olds T, Tomkinson G, Léger L, et. al.;1; Worldwide variation in the performance of children and adolescents: An analysis of 109 studies of the 20-m shuttle run test in 37 countries. J Sports Sci. 2006;24(10):1025-1038. 10.1080/02640410500432193. [24] Farhat F, Masmoudi K, Cairney J, et. al.;1; Assessment of cardiorespiratory and neuromotor fitness in children with developmental coordination disorder. Res Dev Disabil. 2014;35(12):3554-3561. 10.1016/j.ridd.2014.08.028. [25] Ekelund U, Tomkinson G, Armstrong N.;1; What proportion of youth are physically active? Measurement issues, levels and recent time trends. Br J Sports Med. 2011;45(11):859-865. 10.1136/bjsports-2011-090190. [26] Armstrong N, Tomkinson G, Ekelund U.;1; Aerobic fitness and its relationship to sport, exercise training and habitual physical activity during youth. Br J Sports Med. 2011;45(11):849858. 10.1136/bjsports-2011-090200. [27] Kemper HCG.;1; Amsterdam Growth and Health Longitudinal Study (AGAHLS): A 23Year Follow-up from Teenager to Adult about the Relationship Between Lifestyle and Health. Karger Medical and Scientific Publishers; 2004. [28] Armstrong N, Barker AR.;1; Endurance training and elite young athletes. Med Sport Sci. 2011;56:59-83. 10.1159/000320633. [29] Machado-Rodrigues AM, Leite N, Coelho-e-Silva MJ, et al.;1; Independent association of clustered metabolic risk factors with cardiorespiratory fitness in youth aged 11-17 years. Ann Hum Biol. 2014;41(3):271-276. 10.3109/03014460.2013.856471.
8
[30] Twisk JW, Kemper HC, van Mechelen W.;1; Tracking of activity and fitness and the relationship with cardiovascular disease risk factors. Med Sci Sports Exerc. 2000;32(8):14551461.
TableTable 1: Participant Characteristics Wave 1Table 2. Mixed effects models predicting estimated relative peak VO2 (mL/kg/min), with random effects and derived group-specific slopes. Model 1 Estimate (L95%, U95%) Fixed effects Age Female Female*Age pDCD pDCD*Age pDCD*Female PQ organized PQ free play Intercept
S1440-2440(16)30185-2 http://dx.doi.org/doi:10.1016/j.jsams.2016.08.025 JSAMS 1385
To appear in:
Journal of Science and Medicine in Sport
Received date: Revised date: Accepted date:
9-3-2016 17-8-2016 21-8-2016
Please cite this article as: Cairney John, Veldhuizen Scott, King-Dowling Sara, Faught Brent E, Hay John.Tracking cardiorespiratory fitness and physical activity in children with and without motor coordination problems.Journal of Science and Medicine in Sport http://dx.doi.org/10.1016/j.jsams.2016.08.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Corresponding author at: Faculty of Kinesiology & Physical Education, University of Toronto; Departments of Family Medicine, Psychiatry and Behavioural Neuroscience; CanChild Centre for Childhood Disability Research and Offord Centre for Child Studies, McMaster University; Address: 100 Main St West, 5th Floor, Hamilton, Ontario, Canada, L8P 1H6; Tel.: 905-525-9140 ext: 28506; Fax: 905527-4440.
1
Introduction Developmental coordination disorder (DCD) is a chronic neurodevelopmental condition affecting between 2 and 6 percent of all children1,2. The functional presentation of the disorder is heterogeneous, but involves deficits in gross and/or fine motor skills1. Existing studies show that children with DCD are at greater risk for a number of physical and mental health-related problems, including poor physical fitness3,4 (particularly cardiorespiratory fitness (CRF)5,6), overweight and obesity7, and depression and anxiety8. Studies have also shown that children with DCD are less physically active than their typically developing peers9,10. Lower levels of CRF are one of the more concerning health outcomes associated with DCD4. Improving CRF produces physiological adaptations that improve the efficiency of the respiratory and circulatory oxygen transport system, and these adaptations confer numerous health benefits, including reduced risk for cardiovascular disease and reduced all-cause mortality11. The activity deficit hypothesis describes connections between poor motor coordination, physical inactivity and CRF (as well as overweight/obesity) as a negative feedback loop4,5,9. It is hypothesized that poor motor coordination discourages children from participating in physical activities, which negatively affects the development of physical fitness and increases the risk for unhealthy weight gain. Poor health-related fitness, in turn, leads to further reductions in physical activity4. In this way, poor motor coordination becomes a risk factor for negative health outcomes later in life12. Evaluating the activity deficit hypothesis requires longitudinal data with repeated assessments of physical activity (PA) and fitness. Few studies have tracked changes in CRF in children with DCD 3,13,14, and these studies are limited by small samples3,14 and/or by relatively short follow-up periods3,13. More importantly, no attempts have been made to test the activity-deficit hypothesis with these data. If an activity deficit is responsible for poor fitness among children with DCD, then a measure of physical activity should explain a substantial part of the association between motor proficiency and fitness. In a previous paper, we examined changes in CRF over a 2.5-year period, and showed that both boys and girls with possible DCD (pDCD) have poorer fitness levels than typically developing children13. Since this analysis was conducted, data collection for this study cohort has been completed. The final dataset includes an additional 2 years of follow-up, making it possible to track CRF from age 9 to age 14. In this report, we also examine the associations between self-reported participation in organized and active free play and CRF in children with and without pDCD. Methods The Physical Health Activity Study Team (PHAST) project was a longitudinal prospective cohort study of children designed to examine change over time in PA and health-related fitness in children with and without poor motor coordination. Details of the design have been described in previous publications10,15. Data collection commenced during the 2004-2005 school year. The target population was all children in the fourth grade who were enrolled in the public school system in the Niagara region of Ontario, Canada. We approached 92 schools, of which 75 (83%) agreed to participate. At baseline, we obtained written informed consent from the parents of 2278 of 2378 (96%) children. Our sample includes the 1954 (86%) of these participants who had a complete motor assessment and data on PA and CRF, as well as a further 163 children who entered the study in later waves (total n=2117). The first formal data collection effort (wave 1) occurred in May/June of the grade 4 school year (age 9-10). For the next two school years (waves 2 through 5), data collection took place in the fall (Sept/Oct) and again in the spring (May/June). Beginning in wave 6, data were collected once per year, always in the spring (May/June). Data collection was done in school classes. New students sometimes joined participating schools during the study period. As excluding these children would have served no obvious purpose, and would have required the provision of alternative activities during assessments, new students were enrolled in the study. As a result, PHAST was an open cohort, with a total sample size sometimes rising over time. The sample decreased
2
sharply in size after wave 5 (Table 1), because many students transferred to different schools to begin grade 7 (age 11-12). All data were collected by trained research assistants and supervised by the core investigators. Ethics approval was obtained from the District School Board of Niagara and the research ethics board at Brock University. To assess Cardiorespiratory Fitness (CRF) during each wave, we used the Léger 20m shuttle run test16, which is a well-established field measure used to estimate maximal VO2 in children that correlates well with lab-based tests (r = 0.72 17). The shuttle run test is well-suited to school-based assessments, where laboratory testing is not feasible. It involves running back and forth between two lines set 20m apart in synchrony with a sound signal16. At each successive stage, the time between signals or beeps decreases, requiring the participant to run faster. When a child fails to cover the 20m distance before the beep sounds on two successive occasions, the last stage reached is recorded and the test is terminated. This information is then used to determine the maximal speed obtained, which is entered into an equation to calculate estimated relative peak VO218: Estimated Peak VO2 = 31.025 + 3.238 (maximal speed) – 3.248 (age) + 0.1536 (speed x age). Participants performed the test in groups of up to 15, under the supervision of trained research assistants, who also assisted the children with pacing and encouragement. Children were classified as having ‘possible’ Developmental Coordination Disorder (pDCD) based on the results of their performance on the short form of the Bruininks-Oseretsky Test of Motor Proficiency 1st Edition (BOT-SF)19. The BOT-SF uses a subset of core items from the full measure to evaluate different components of motor coordination (fine motor ability; gross motor ability; balance; reaction time). The BOT-SF has been validated against the full BOT in children ages 8 to 14 years, with correlations of 0.90 to 0.9119. In PHAST, most children’s motor coordination was assessed only once. Schools were randomly divided into 3 groups, and motor testing occurred separately for each group over three different waves of data collection. Children were not all tested at the same wave because of time and cost considerations. To evaluate the longitudinal stability of our measure of pDCD, a random subset of participants (n=77) were re-tested on the BOT-SF approximately 2 years after their original assessment. The correlation between tests was .707. Trained research assistants administered the test individually to each child in the school’s gymnasium during regular school hours. To the best of our ability, assessments were conducted in quiet places where distractions could be minimized and privacy respected. Children scoring at or below the 5th percentile on the BOT-SF were identified as having poor motor functioning. Although we excluded children with known learning or physical disabilities, we did not assess the full DSM criteria for DCD (i.e. daily impact and clinician motor assessment)1. We therefore use the term ``possible DCD'' (pDCD). This descriptor is frequently used in the literature (e.g. 5,7,8,10,12,13,15). To assess the validity of our evaluation of motor skills, we conducted a sub-study on a random subset of 24 children we identified as pDCD. Twenty-one (88%) scored at or below the ``at risk'' cutoff for DCD (16th percentile) on the Movement Assessment Battery for Children20, which was administered by a pediatric occupational therapist who was blind to the results of the BOT-SF15. This high positive predictive value supports the validity of the original assessment. To measure physical activity (PA) participation, children completed the Participation Questionnaire (PQ21). The PQ is a self-report questionnaire that asks about the frequency of current participation in organized activities (e.g., team sports) and active free play (e.g., bike riding). Scores range from 0 to 20 for free play and from 0 to 29 for organized activities, with higher scores indicating higher levels of activity. Validation studies have shown that the PQ is associated with body fat, aerobic capacity, teacher ratings of physical ability21, gender, and urban/rural residence10,21. It was developed for children in grades
3
4 to 9 (ages 8-16), the two-week test-retest reliability has been measured as 0.81 for elementary school and 0.89 for secondary school students21. We used mixed effects modeling to fit growth curves representing change over time in CRF. This method allows us to generate estimates using all time points available for each student. We included a random intercept at the child level, a random slope for age, and used an unstructured covariance matrix. To allow for the possibility of a non-linear relationship between age and the outcome, we fit a series of models with fractional polynomial transformations for the sample as a whole and for boys and girls separately. This demonstrated that the relationship was essentially linear, and we therefore proceeded without transforming child age. In order to reflect known sex differences in CRF, we included age*sex interaction terms in all models. This approach allows for different slopes and intercepts for boys and girls, while making it possible to reflect a common effect for pDCD status. To check the appropriateness of this approach, we tested a 3way interaction between age, sex, and pDCD. We did not attempt to standardize our measures using agespecific norms. Although issues of scaling make it difficult to draw conclusions about the nature of agerelated change in CRF, this analysis is concerned with group differences. We first fit a model with age, sex, pDCD status, and their interactions only (Model 1), and then added self-reported PA (Model 2). For both sets of results, we calculated expected group differences by age, and calculated slopes for the 4 groups of individuals produced by the combination of sex and DCD status. Other work on this sample has shown that children with pDCD had higher body mass index (BMI) values than other participants22. To evaluate the possibility that BMI modified the effect of pDCD on CRF, we fit a further model including an interaction between these variables. As the design was an open cohort, attrition typically resulted from movement between schools, and the analysis chosen makes use of all available measurement points, we left missing data as missing. We used Stata 14 for all analyses. Results Descriptive statistics are presented in Table 1. At baseline, there were 97 children (5.0%) with pDCD. Of these, 56 (58%) were girls and 41 (42%) boys. As the study was an open cohort, the number of children with pDCD rises to 103 when participants joining in later waves are included. The analysis included a total of 12,338 follow-ups on 2117 children. [Insert Table 1 about here] The number of boys and girls in the cohort were roughly equal, and the average age of children at the start of the study was 9.9 years (SD=0.4). On average, the pDCD group reported less engagement in both organized and free play activity at each wave. Crude means show slight declines over time for CRF and organized PA, and a slight increase for free PA, as measured by the self-report Participation Questionnaire. At all time points, and for both sexes, children with DCD had substantially lower CRF. This gap was somewhat smaller among girls. Our measure of CRF (estimated relative peak VO2) decreased over time in all groups, with the slowest decline among typically developing boys and more rapid decline among both boys and girls with pDCD (Fig 1). [insert Figure 1 about here] Subtracting the derived slopes for groups with and without pDCD shows that CRF declined among children with pDCD by an extra 0.53 (95% CI = 0.24 to 0.82) units per year before self-reported activity is taken into account. As we found no 3-way interaction between age, sex, and DCD status, these results are the same for males and females. This value does not change substantially when the PQ subscales are included as covariates (estimate = 0.55, 95% CI = 0.26 to 0.84) (Table 2). The main effect of pDCD status is also only slightly attenuated in Model 2. Neither the persistent gap between groups nor its gradual widening, therefore, can be attributed to recent self-reported PA. A final model exploring the effects of BMI yielded significant main effects for BMI (estimate = -0.49, p<0.01) and pDCD status (estimate = -10.8, p<0.01), as well as a significant interaction (estimate = 0.30,
4
p<0.01). The interaction indicates that the effect of BMI is weaker for children with pDCD, which, along with the increased size of the main effect for pDCD status, is consistent with the view that children with pDCD have poor fitness irrespective of BMI. The adjusted absolute difference between children with and without pDCD did not change in this model (in Model 2, AAD = 5.08, 95% CI = 4.35-5.82; in the BMI model, AAD = 5.11, 95% CI = 5.47-5.75). [Insert Table 2 about here]
Discussion Performance on the shuttle run in our sample was very similar to that reported in a recent meta-analysis23, and showed closely comparable patterns of age and sex differences: The mean stage reached increased for both boys and girls, while body mass-scaled VO2 peak decreased. Consistent with previous work reporting on smaller samples, our results also show that children with pDCD, regardless of sex, have poorer CRF than typically developing children6,24. Moreover, differences in CRF between pDCD and typically developing children appear to emerge early – they are present at the first wave of data collection in our study – and not only persist3, but increase slightly through childhood to adolescence14. A major focus of the present study was to test existing theories concerning the activity-deficit hypothesis in children with DCD5,9. While our results show that participation in organized and active free play are positively associated with CRF, these measures explain only a small part of the overall association between CRF and pDCD, and essentially none of the difference in rates of change. Given the intuitive appeal of the activity-deficit hypothesis as an explanation for poorer CRF in children with DCD, this result is somewhat surprising. As we did observe a group difference in CRF, some other explanation must be sought. One possibility is that our data on CRF, PA, or both, were affected by measurement issues. Self-reported measures of PA have been shown to agree poorly with objective measures25, which may therefore have reduced our ability to measure effect modification. It is also possible that our measure, which was concerned largely with the frequency of PA, did not adequately measure intensity of effort. If participants in the pDCD group exercised at lower intensity levels, this would help to explain why frequency of PA fails to explain differences in CRF. Another measurement-related issue concerns the use of field-based protocols to estimate CRF. It is conceivable that running gaits were less efficient in the pDCD group, or that difficulty with rapid stopping and turning, or differences in motivation or self-confidence, led pDCD participants to drop out of the test prematurely. This is also a possible explanation of the effect we noted for BMI, as pDCD children may have had difficulty with the motor demands of the test (principally, rapid turning) irrespective of body mass. Cairney and colleagues6 however, have shown that the difference in CRF between children with and without pDCD remained constant regardless of whether a field-based protocol or direct laboratory measurement (with cycle ergometers) was used. Another possibility is that PA in our sample was insufficiently intense to produce changes in CRF. Armstrong and colleagues note that, although aerobic fitness and PA are two of the most studied variables in pediatric exercise science, over three decades of observational research have failed to confirm any meaningful association between them26. An analysis of 15 years of data from the Amsterdam Growth and Health Study, for example, found that a 30% increase in self-reported habitual PA resulted in only a 2-5% increase in VO2 maximum27. Even when studies that use both objective measures of PA (e.g. accelerometry) and direct measures of peak VO2 are considered on their own, there are no strong relationships between habitual activity and aerobic fitness in children and youth26. On the other hand, numerous studies of youth athletes and of aerobic fitness responses to exercise training (e.g.28) show clear positive effects of both sport participation and structured exercise training on peak VO2. One obvious explanation for these discrepancies is that children in general-population studies are insufficiently physically active with regard to intensity and duration in everyday, habitual PA to increase peak VO2.
5
This may account for the generally weak associations in our sample: even though children with DCD are less active than others, neither group is active enough to maintain CRF, with pDCD children further below the necessary level. Poor physical fitness has been linked to many long-term health consequences11, and this association begins early in life. CRF is associated with cardiovascular disease risk factors in children and adolescents 29 , and lower aerobic fitness in youth is associated with increased adiposity and higher cardiovascular disease risk in adulthood30. Since CRF tracks moderately well from childhood to adulthood30, it is plausible that low levels of CRF in children with pDCD may increase their risk of cardiovascular disease and other illnesses later in life. One implication of these results is that targeted interventions might usefully aim not only to improve motor abilities, but to increase CRF. This may require intensities or volumes of PA greater than those observed in our sample. Our findings on gender differences, which are largely consistent with results from general-population samples, also imply that the design of interventions in this context should give careful consideration to the fitness gap between boys and girls in this population. Future longitudinal research intended to elucidate the association between DCD and CRF should also begin with younger children. The fact that group differences are present at the beginning of our study makes it impossible to clearly establish precedence and to determine whether data are consistent with a causal hypothesis. Research should also consider additional variables (e.g. motivation, PA intensity, self-efficacy) that might mediate the relationship between motor coordination and fitness. Conclusion The purpose of this study was to evaluate the activity deficit hypothesis in a large cohort of children by testing whether levels of measured physical activity accounted for poor fitness among children with pDCD. Our results show that they do not: Although we have suggested possible explanations for the absence of effect modification by PA in our models, we have found no good evidence that PA explains overall group differences in CRF. Although this difference increases with age, it is substantially present at baseline, when most children were 9-10 years old. This suggests than factors operating in early childhood, which may include an activity deficit, produce differences in CRF that then persist into adolescence. As the effects of poor motor functioning on CRF begin early in life, greater efforts should be made to track the early development of fitness in children with and without poor motor coordination. Future studies using objective PA assessment with larger samples of children with DCD would be useful in either detecting an effect of PA on CRF differences or in ruling one out. Practical Implications - Children with motor coordination problems have poorer cardiorespiratory fitness than typically developing children. - This fitness difference is present at age 9 and increases somewhat through age 14. - The gap in fitness is not explained by differences in self-reported physical activity at these ages.
6
[1] ;1; American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (5th Ed.).
[17] Liu NY, Plowman SA, Looney MA.;1; The reliability and validity of the 20-meter shuttle test in American students 12 to 15 years old. Res Q Exerc Sport. 1992;63(4):360-365.
8
[30] Twisk JW, Kemper HC, van Mechelen W.;1; Tracking of activity and fitness and the relationship with cardiovascular disease risk factors. Med Sci Sports Exerc. 2000;32(8):14551461.
Table