Descriptive and factor analysis of the Developmental Coordination Disorder Questionnaire (DCDQ‘07) in a population-based sample of children with and without Developmental Coordination Disorder

Authors

  • L. Rivard,

    Corresponding author
    1. School of Rehabilitation Science, McMaster University, Hamilton, ON, Canada
    2. CanChild Centre for Childhood Disability Research, McMaster University, Hamilton, ON, Canada
    • Correspondence:

      Lisa Rivard, School of Rehabilitation Science, Institute for Applied Health Sciences, Room 408, McMaster University, 1400 Main St. West, Hamilton, Ontario, Canada L8S 1C7

      E-mail: lrivard@mcmaster.ca

    Search for more papers by this author
  • C. Missiuna,

    1. School of Rehabilitation Science, McMaster University, Hamilton, ON, Canada
    2. CanChild Centre for Childhood Disability Research, McMaster University, Hamilton, ON, Canada
    Search for more papers by this author
  • D. McCauley,

    1. School of Rehabilitation Science, McMaster University, Hamilton, ON, Canada
    2. CanChild Centre for Childhood Disability Research, McMaster University, Hamilton, ON, Canada
    Search for more papers by this author
  • J. Cairney

    1. CanChild Centre for Childhood Disability Research, McMaster University, Hamilton, ON, Canada
    2. Family Medicine and Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, ON, Canada
    Search for more papers by this author

Abstract

Background

The Developmental Coordination Disorder Questionnaire (DCDQ‘07) discriminates children with Developmental Coordination Disorder (DCD) from their peers. Studies employing the DCDQ have typically used clinical samples. To further validate the DCDQ'07, this study: (1) described its distributions in a population-based sample, and a sample of children with DCD; (2) explored sex and age differences at important cut-points; and (3) examined its factor structure.

Methods

This secondary analysis of data collected from 23 schools (n = 3151) included a sample of 3070 children (1526 boys, 1544 girls) and a sample of 122 children (73 boys, 49 girls) who met DCD diagnostic criteria. DCDQ‘07 distributions were described by age and sex. Chi-square analyses were conducted using three clinically important percentile ranges; a factor analysis explored the construct validity of DCDQ scores.

Results

Parents of 3070 children (97.4%) completed the questionnaire independently. Significant sex differences were noted in both samples. Significant differences in proportions by sex, and DCDQ means by age were found in the population sample. A three-factor solution was found, accounting for 70.3% of the variance.

Conclusions

This is one of the largest studies using the DCDQ'07 with a non-clinical sample. The three-factor solution, including item loading, was consistent with previous research. When using DCDQ cut-offs it is important to consider sex and age.

Introduction

Developmental Coordination Disorder (DCD) is a common neuro-developmental disorder affecting 5–6% of school-aged children (American Psychiatric Association (APA) 2000). By definition, children presenting with this disorder have substantial difficulties learning and performing motor tasks in their daily lives, and their difficulties are not related to any specific medical or neurological deficit (APA 2000). As a result of their primary motor impairment, children with DCD are at significant risk of developing poor physical fitness, obesity and cardiovascular health (Cairney et al. 2005, 2007; Rivilis et al. 2011), as well as anxiety, depression, low self-worth and social isolation (Schoemaker & Kalverboer 1994; Piek et al. 2000, 2005, 2008; Smyth & Anderson 2000; Skinner & Piek 2001; Rose & Larkin 2002; Francis & Piek 2003). Although their difficulties persist into adolescence and adulthood (Losse et al. 1991; Cousins & Smyth 2003; Fitzpatrick & Watkinson 2003), children with DCD can benefit immensely from supportive environments where significant others are provided with ways to manage their disorder, thereby preventing secondary consequences (Missiuna et al. 2004, 2006, 2007, 2008; Rivard et al. 2011). For this reason, it is of the utmost importance that children with DCD be identified as early as possible, using sound measurement tools, to initiate intervention, provide accommodations, and set realistic expectations to ensure they are successful (Missiuna et al. 2003).

Widely used for both research and clinical purposes, the Developmental Coordination Disorder Questionnaire (DCDQ) (Wilson et al. 2000) is a parent-report identification tool used to screen children for the presence of motor impairments. The DCDQ asks parents to compare their child's performance in everyday tasks with that of their typically developing peers. The revised DCDQ'07 can be completed by parents of children aged 5 to 15 years (Wilson et al. 2009).

The DCDQ (and DCDQ'07) are often evaluated using clinical samples (Green et al. 2005; Loh et al. 2009). Recommended cut-points indicating the presence or risk of DCD have varied in the literature [e.g. 5th percentile (DCD) and 15th percentile (at risk) (Wilson et al. 2009), or <10th percentile (DCD) and 25th percentile (at risk) (Wilson et al. 2000)] and are based on clinical samples of children with motor difficulties. It is important to describe the characteristics of the DCDQ'07 in a large, non-clinical sample, to further explore the validity of DCDQ scores. The primary purpose of this study is to describe the DCDQ'07 distributions in: (1) a population-based sample of children aged 8–15 years; and (2) a sample of children who met diagnostic criteria for DCD (APA 2000) diagnostic criteria for DCD. The secondary aims of this study are to explore potential proportional sex and age differences at the lower percentile ranges of the DCDQ'07 and to re-examine its factor structure using our population-based sample.

Methods

Participants

This is a secondary analysis of data collected from two public school boards (n = 23 schools) in Ontario, Canada, where 3151 school-aged children (1590 boys, 1561 girls) were screened for motor co-ordination difficulties (Missiuna et al. 2011). Ethics approval was obtained from both school boards and the joint Hamilton Health Sciences McMaster University Research Ethics Board prior to data collection; consent was received from all participating students and parents. Parents of 3084 (98%) of the children completed the DCDQ'07; 3070 (97.4%) had at least 13 of 15 items completed on the DCDQ, and comprise the current study sample. The ratio of boys to girls was roughly equal (1:1) with 1526 boys and 1544 girls, aged 8–15 years (mean age 11.7 years; SD: 1.5) and the sample was representative of the socio-demographic characteristics of the population.

From the large sample, 122 children (mean age 11.4 years; SD: 1.5) were identified through child and parent surveys with scores indicating a likelihood of DCD, as determined by very conservative cut-offs (at or below the 5th percentile on one or both of the child and parent surveys). Of the sample of 122 children, 62 (51%) were identified through the DCDQ'07. In order to be included, children were assessed individually to determine whether they met the inclusion/exclusion criteria for DCD [DSM-IV-TR (APA 2000)] (Table 1). The ratio of boys to girls in the DCD sample was 1.5:1, with 73 boys and 49 girls.

Table 1. Diagnostic criteria and measures to confirm Developmental Coordination Disorder (DCD)
Diagnostic criteria for DCD (APA 2000)Measures to confirm DCD
A) Performance in daily activities that require motor co-ordination is substantially below that expected, given the person's chronological age and measured intelligence.The Movement Assessment Battery for Children (Henderson & Sugden 1992), a measure of severity of motor impairment, was administered by a registered occupational therapist (OT). Children scoring at or below the 15th percentile were deemed to have significant motor impairment.
B) The co-ordination difficulties and/or motor delays significantly interfere with academic achievement or activities of daily living (self-care activities).A structured clinical interview (Missiuna et al. 2008) was conducted by the OT with the parents to confirm whether or not the motor impairment impacted on the child's ability to perform academic and/or self-care activities.
C) The difficulties are not due to a general medical condition (e.g. cerebral palsy, hemiplegia, or muscular dystrophy) and do not meet criteria for pervasive developmental disorder.We were not able to have a physician review each child; hence, parental report of child's medical history and therapist observation of child's physical status were used to determine whether other medical conditions were present.
D) If mental retardation is present, the motor difficulties are in excess of those usually associated with it.The Kaufman Brief Intelligence Test-2 (K-BIT2) (Kaufman & Kaufman 2004) was administered by the OT to screen for the possible presence of cognitive deficits. Children scoring less than 70 on the KBIT-2 were excluded from the DCD group.

Measure

Developmental Coordination Disorder Questionnaire (DCDQ)

Using a 5-point unipolar scale, the DCDQ instructs parents to rate their child's motor proficiency compared with a child of his or her age. Questionnaire items include everyday activities (e.g. catching a ball, running, printing/writing) that children typically perform and that are often of concern to parents of children with motor difficulties.1 Individual item scores are summed to give a total score ranging from 15 to 75 (higher scores indicate better motor co-ordination), which is then used to determine if the score suggests the presence, risk or absence of DCD.

Originally developed in 2000, the tool was revised as the DCDQ'07 (Wilson et al. 2009). The DCDQ'07 contains 15 items and is intended for parents of children aged 5 to 15 years. The 15 items are grouped into three distinct factors: ‘Control During Movement’, ‘Fine Motor/Handwriting’, and ‘General Coordination’. Internal consistency, test–retest reliability, construct and concurrent validity of DCDQ scores, as well as high sensitivity in identifying children at risk for DCD have been noted in the literature (Wilson et al. 2000; Crawford et al. 2001; Green et al. 2005; Schoemaker et al. 2006; Civetta & Hillier 2008). Additional studies have confirmed the strong psychometric properties of the revised version of the tool (Cairney et al. 2008; Keijsers et al. 2009; Wilson et al. 2009; Tseng et al. 2010). For the remainder of this paper, the DCDQ'07 will be referred to as the DCDQ for the sake of simplicity.

Statistical analysis

IBM spss Statistics Version 20.0 was used for all statistical analyses (IBM Corporation, New York).

Descriptive analysis

Questionnaires with at least 13 of 15 items completed (n = 3070; 99.5%) were included in the descriptive analysis. Missing data (4%) were imputed by using the average of valid responses across individual children's scores. DCDQ distributions were described for both samples by sex and age grouping. Age groupings of less than 10 years, 10 to less than 12 years, and greater than or equal to 12 years were chosen to approximate the age groupings found in the existing literature for comparison. Independent sample t-tests were used to test for significant differences in means by sex. Analyses of variance were conducted to determine differences in means by age groupings. Tukey's post hoc tests were used where statistically significant differences were found.

Chi-square analyses

Using the large sample, a chi-square analysis (Pearson chi-square test) was conducted to compare differences in the proportions of boys and girls identified at three clinically important percentile ranges: 0 to 5th, 6th to 15th and 16th to 25th. A separate chi-square analysis using the three age groupings outlined above tested for differences in proportions of children of different ages identified at the three percentile ranges.

Factor analysis

A factor analysis of DCDQ items was conducted with the large sample using a principal components analysis using varimax with Kaiser normalization rotation methods, to replicate methods previously used in the literature (Wilson et al. 2009).

Results

Descriptive analysis

The DCDQ total score distribution means and SDs by sex and age grouping are found in Table 2 (population sample) and Table 3 (DCD sample).

Table 2. Population sample DCDQ total scores [mean (SD)] by sex and age
 Boys (n = 1526)Girls (n = 1544)Total (n = 3070)
  1. ***P < 0.001; **P < 0.01.
  2. Significant differences also noted across the three age groupings (P < 0.05).
  3. Missing age data: 61 cases (2%).
  4. DCDQ, Developmental Coordination Disorder Questionnaire.
Total sample63.87 (11.07)66.39 (9.03)65.14*** (10.17)
Age (years)   
<10 (n = 503; 246 M, 247 F)62.53 (11.12)65.09 (9.13)63.84** (10.22)
10 to <12 (n = 1213; 600 M, 613 F)63.67 (10.93)66.61 (8.73)65.15*** (9.99)
≥12 (n = 1293; 644 M, 649 F)64.55 (11.11)66.64 (9.29)65.60*** (10.29)
Table 3.  DCD sample DCDQ total scores [mean (SD)] by sex and age
 Boys (n = 73)Girls (n = 49)Total (n = 122)
  1. ***P < 0.001.
  2. Missing age data: 4 cases (3%).
  3. DCD, Developmental Coordination Disorder; DCDQ, Developmental Coordination Disorder Questionnaire.
Total sample44.18 (12.33)52.78 (12.49)47.64*** (13.05)
Age (years)
<10 (n = 25)
  • No significant differences found by age
  • Sub-analyses by sex were not conducted due to the small sample size
46.89 (12.42)
10 to <12 (n = 45)47.47 (13.45)
≥12 (n = 48)48.22 (13.48)

For the population sample, statistically significant sex differences in mean DCDQ scores were found for the total sample (t = −6.930, d.f. = 3068, P < 0.001) (Table 2). Girls had higher mean scores than boys in all cases. Differences in mean DCDQ scores by age were also significant (F (2,3006) = 5.43, P < 0.05), with older children demonstrating higher means than younger children. A Tukey's post hoc test indicated that the mean for the younger group (less than 10 years) was significantly different than the means for both the 10 to less than 12 year group (P < 0.05) and the greater than or equal to 12 year group (P < 0.01). There was no statistically significant difference between the means of the two older groups (P > 0.5).

For the DCD group, significant sex differences in mean DCDQ scores were found (t = −3.755, d.f. = 120, P < 0.001) (Table 3). Higher mean scores were found for the girls: the sex difference in means was much greater than that found in the population sample. The analysis of scores by age grouping revealed non-significant results (F (2,115) = 0.09, P > 0.05) (Table 3). Further sub-analyses were not conducted in the DCD group due to the small sample size.

Chi-square analyses

Sex

Table 4 shows sex differences across clinically meaningful percentile ranks of the DCDQ. At the 0 to 5th percentile, the ratio of boys to girls identified was approximately 2:1; at the 6th to 15th percentile it was approximately 1.6:1, and at the 16th to 25th percentile the ratio was 1:1. While there are significant sex differences across these percentile ranges overall (χ2 = 10.84, P < 0.01), it is not possible to determine specifically where the significant difference lies with this analysis type (i.e. there is no post hoc test for chi-square). However, visual inspection of the cross-tabulation results clearly shows that sex differences are most pronounced between the 0 to 5th and 6th to 15th percentiles, and this likely accounts for the overall significant effect observed.

Table 4. Population sample observed frequencies by sex and percentile (n = 767)
PercentileChild's sexTotal
BoysGirls
0 to 5th10251153
6th to 15th188119307
16th to 25th159148307
Total449318767

Age

Table 5 shows the number of children in each age grouping identified at the three percentiles reported. There were no statistically significant differences in the proportions of children of different ages identified at these three percentile ranges (χ2 = 6.33, P > 0.05).

Table 5. Population sample observed frequencies by age grouping and percentile (n = 754)
PercentileAge grouping (years)Total
<1010 to <12≥12
0 to 5th276161149
6th to 15th58120124302
16th to 25th77124102303
Total162305287754

Factor analysis

A three-factor solution was found, accounting for 70.3% of the variance (Table 6) with Eigenvalues of 7.8, 1.7 and 1.0 for Factors 1, 2 and 3. Factor 1 explained 32.8% of the total variance; Factors 2 and 3 explained 24.7% and 12.9% of the variance, respectively. Seven items loaded cleanly on Factor 1 ‘Control During Movement’ (‘throws’, ‘catches’, ‘hits’, ‘jumps’, ‘runs’, ‘likes participating in sports’, ‘learns’); four items on Factor 2 ‘Fine Motor/Handwriting’ (‘writes fast’, ‘writes legibly’, ‘effort/tension’, ‘cuts’) and two items on Factor 3 ‘General Coordination’ (‘not a “bull in a china shop” ’, ‘does not fatigue’). Two items showed factorial complexity, with roughly equal factor loadings on Factors 1 and 2 (‘plans’, ‘quick/competent’).

Table 6. Population sample factor analysis (n = 3070)
DCDQ'07 questionnaire itemFactor
Control During MovementFine Motor/HandwritingGeneral Coordination
  1. DCDQ, Developmental Coordination Disorder Questionnaire.
 1. Throws ball accurately0.814  
 2. Catches ball0.807  
 3. Hits ball accurately0.797  
 4. Jumps easily over obstacles0.721  
 5. Runs as fast as, and in a similar way to0.690  
 6. Plans motor activity0.5510.429 
 7. Writes fast 0.833 
 8. Writes legibly 0.868 
 9. Writing effort/tension is appropriate 0.838 
10. Cuts pictures/shapes accurately 0.763 
11. Likes participating in sports/active games0.762  
12. Learns new motor tasks easily0.746  
13. Quick and competent0.4080.566 
14. Not a ‘bull in a china shop’  0.852
15. Does not fatigue/slouch/fall out of chair  0.804

Discussion

Findings from this study inform our understanding of the use of the DCDQ in a population-based sample of children. This was one of the largest, non-clinical samples ever to be screened using the DCDQ, with an extremely high rate of completion (97.4%) among those parents who consented to participate. The authors are aware of only one other study describing the characteristics of the DCDQ in a population-based sample of a size comparable to the present study. While our study was being conducted, Nakai and colleagues translated the DCDQ and administered it in Japan to parents of 6330 children varying in age from 4 to 15 years (Nakai et al. 2011).

In our study, performance of the population sample on the DCDQ was as expected: a high mean score indicated that a large number of children without co-ordination difficulties scored close to the highest possible total score. The means and SDs reported for the large sample by age grouping (Table 2) were slightly higher than those reported previously for the DCDQ (B.N. Wilson, pers. comm., 19 October 2010) and much higher than those reported by Nakai and colleagues, particularly with the two younger age groups (Nakai et al. 2011). The age range of children in the Japanese study was broader, with approximately one-third of the children less than 8 years of age. It is possible that our inclusion of school-aged children who were 8 to 15 years of age resulted in higher means. Nonetheless, given the discriminative purpose of the DCDQ, and the prevalence of DCD at approximately 5–6% (APA 2000), we would expect the majority of the children in a population sample to score in the higher range on a screening tool, yielding a relatively high mean, with few children indicating possible motor impairment.

The DCDQ mean reported for the DCD group as a whole was slightly higher than reported previously in the literature (Schoemaker et al. 2006). The difference in scores may reflect differences in the co-ordination profiles of children who are identified via different identification processes (clinically referred sample with motor difficulties versus identified from a population sample). In fact, the group means found for our total DCD sample, and by sex, were more comparable to those of Wilson and colleagues (2000), where children with DCD were identified from a group of children with learning and attention disorders but not motor difficulties specifically.

Both the large sample (containing roughly equal numbers of boys and girls) and DCD sample (with a slightly greater number of boys), demonstrated significant sex differences with girls scoring consistently higher than boys. Our study findings support the work of Nakai and colleagues who found a significant main effect of sex (P < 0.01), with girls demonstrating higher scores (Nakai et al. 2011). Both studies differ from previous research conducted with a population-based sample of 608 Dutch children (Schoemaker et al. 2006) where sex did not impact the scores of children aged 8 to 14 years. However, Schoemaker and colleagues did note a significant sex influence among 4-to 8-year-old children, with boys demonstrating significantly lower scores than girls (Schoemaker et al. 2006). Reasons for the sex difference observed in these studies are unclear and may reflect a developmental disparity in the rate of attainment and/or quality of performance of certain motor skills by boys and girls, an observation that has been borne out in literature on typically developing children (Thomas & French 1985). At all ages, boys have been shown to be more proficient than girls of the same age at throwing, as well as running and jumping (Butterfield & Loovis 1993; Krombholz 1997; Davies & Rose 2000; Thomas et al. 2010); girls have tended to demonstrate more well-developed fine motor skills, including handwriting, in comparison with boys of the same age (Cohen 1997; Krombholz 1997), suggesting that it is important to consider sex differences when assessing motor skills (Davies & Rose 2000). Sex differences have also been noted on another DCD measure [Movement Assessment Battery for Children (Henderson & Sugden 1992)], where 7- to 8-year-old typically developing boys were shown to be more proficient at ball skills and same-aged girls to have better fine motor skills (Junaid & Fellowes 2006). However, given that the items on the DCDQ are divided relatively equally between gross and fine motor items, with no weighting of particular items or groups of items in the final score, performance on the questionnaire should not bias towards either boys or girls. Nonetheless, the sex differences noted in this study and others would suggest that separate norms by sex for the DCDQ may be warranted.

Our finding that older children have significantly higher mean scores than younger children is not surprising in a population sample where the majority of children will not demonstrate motor impairment, where children are likely to become more proficient in abilities with increasing age (Davies & Rose 2000), and where a ceiling effect is possible. Our results are in line with Nakai and colleagues (2011) who found a similar significant age trend and Schoemaker and colleagues (2006) who noted a non-significant age trend for the younger group of children in their study. To account for age differences, the DCDQ scoring manuals (Wilson et al. 2000, 2009) provide separate age norms. Results of the current study confirm that this convention remains appropriate, especially for younger children.

In the population sample, the statistically significant finding regarding the proportion of boys and girls identified at the lower percentiles indicates that the proportion of boys and girls being identified is not as would be expected. Taken in combination with the fact that girls in this study had significantly higher means than boys at all ages, this may suggest that, if separate norms are not used, girls with motor impairment could be under-identified, and girls who are identified may have more severe difficulties; conversely, boys might be over-identified.

In summary, findings from the current research and the literature point to the importance of taking into account both sex and age when using recommended cut-offs for DCDQ scores. Further research is needed to explore and quantify potential sex differences in scoring, which would assist in establishing appropriate screening cut-points. As the revised DCDQ is now suitable for children up to age 15, it would also be important to investigate if previous age norms continue to be appropriate.

Several recent studies have examined the factor structure of the DCDQ (Schoemaker et al. 2006; Cairney et al. 2008; Wilson et al. 2009; Tseng et al. 2010). We compared our findings with those of Wilson and colleagues (2009), as their study examined the revised DCDQ and used identical methods, allowing for a direct comparison of results. A three-factor solution was noted in both studies, explaining a large proportion of the variance, indicating factor stability. In both studies, items 1 through 5 loaded heavily on Factor 1 ‘Control During Movement’; items 7 through 10 loaded on Factor 2 ‘Fine Motor/Handwriting’, and item 6 (‘plans’) loaded roughly equally on Factors 1 and 2. Examining item 6 more closely, several activity examples are provided for parents to consider when scoring this item (i.e. ‘moving on playground equipment’ and ‘using craft materials’). The example of ‘moving on playground equipment’ seems to fit well with the other items loading on the Factor ‘Control During Movement’, whereas ‘using craft materials’ appears to fit best with the Factor ‘Fine Motor/Handwriting’. The heterogeneity of children with DCD, and the range of responses that could be elicited based on parent interpretation using these examples, may have led to the complex loading of this item. This same reasoning may also explain the findings for item 13 (‘quick/competent’) which loaded on both Factors 1 and 2 in our study, and loaded only on Factor 3 in the study by Wilson and colleagues (2009). Another inconsistency relates to item 12 (‘learns’) which loaded on Factor 1 in our study and on Factor 3 in Wilson and colleagues (2009). Item 12 includes the examples of ‘swimming’ and ‘rollerblading’. Focusing on these examples may have led parents to narrow their interpretation of the item, resulting in the variable factor loading. Applying this reasoning does not, however, explain a similar inconsistency in loading found with item 11 (‘likes participating in sports’) where examples are not provided, but loading on different factors was again evident. Finally, Wilson and colleagues (2009) found that five items – 11 through 15 loaded on a third factor, called ‘General Coordination’. In the current study, only two items – 14 and 15 loaded on this factor. The current study findings would suggest that this factor may be weak, as there were only two items loading on it. Interestingly, both items are negatively worded and while ‘Not a “bull in a china shop” ’ might fit with ‘General Coordination’ it is more difficult to see how ‘Does not fatigue/slouch/fall out of chair’ would relate to a factor named in this way.

Limitations

This study described a secondary analysis that was purely descriptive in nature; as such, sensitivity and specificity of the DCDQ could not be determined.

Key messages

  • The DCDQ is a screening tool that can easily be completed by parents of children with or without motor co-ordination difficulties.
  • Study findings suggest that current recommendations regarding cut-offs for the DCDQ may not reflect sex and age differences among either typically developing children or those with motor co-ordination difficulties; there may be a need for separate sex and age norms.
  • Results of this study suggest that modifications to the wording of specific items on the DCDQ may improve the validity of test scores.

Acknowledgements

The authors are grateful to Ms Brenda Wilson for her encouragement and to the research team of the STACK study (CIHR Grant # MOP 81120) for permitting this secondary analysis. At the time of writing, Ms Rivard was supported by a Doctoral Fellowship through the McMaster Child Health Research Institute and an Ontario Graduate Scholarship and Dr Missiuna held a Rehabilitation Scientist award from the Ontario Ministry of Health and Long Term Care and the Ontario Neurotrauma Foundation. Ms Rivard is currently supported through a CIHR Banting and Best Doctoral Scholarship. Dr Cairney holds the McMaster Family Medicine Professorship in Child Health Research. This work was presented as a poster at The 9th International Conference on Developmental Coordination Disorder held in Lausanne Switzerland in June 2011.

  1. 1

    The complete questionnaire and rating scale are available at: http://www.dcdq.ca.

Ancillary