Article Text

Download PDFPDF

Developmental coordination disorder in “apparently normal” schoolchildren born extremely preterm
  1. T-A Goyen1,2,
  2. K Lui2
  1. 1
    Westmead Hospital, Westmead, New South Wales, Australia
  2. 2
    School of Women’s and Children’s Health, University of New South Wales, Sydney, Australia
  1. Dr K Lui, Department of Newborn Care, Royal Hospital for Women, Barker Street, Randwick, New South Wales, Australia 2031; kei.lui{at}


Aims: To determine the prevalence of developmental coordination disorder (DCD) in “apparently normal” extremely premature (<29 weeks) or extremely low birthweight (<1000 g) schoolchildren at 8 years of age and whether motor skill assessments at an earlier age could predict DCD.

Method: From a neonatal intensive care unit cohort, 50 of the 53 eligible children (IQ >84 and without disabilities at age 5 and residing in Sydney metropolitan) and full-term classroom controls matched for gender and age were assessed with the Movement Assessment Battery for Children (MABC) at school. Previous Griffith’s Scales (1 and 3 years) and Peabody Motor Scales (3 and 5 years) results were evaluated for prediction.

Results: The prevalence of DCD (MABC impairment scores >1 SD below the norm) was significantly higher in the study group than controls (42% vs 8%, respectively), and severe DCD (scores >1.5 SD) was also significantly higher (30% vs 0%). DCD was independently associated with prolonged rupture of membranes and retinopathy of prematurity but not with parental education or occupation. Motor assessment using Peabody Fine Motor Scales at 3 years with a cut-off of <27th centile was the best predictor of DCD (areas under curve 78%).

Conclusions: Apparently normal high-risk infants are at risk of motor dysfunction into their school years. Most of these could be identified at age 3.

View Full Text

Statistics from

Researchers have found motor problems to be common in children born extremely preterm (<29 weeks’ gestation) or with extremely low birth weight (ELBW, <1000 g).18 Even in groups of “apparently normal” extremely preterm/ELBW children who do not have cerebral palsy, have an average IQ and attend mainstream school, many experience subtle motor and learning difficulties and school failure.9 10 Despite this, relatively few studies have been conducted to specifically investigate the development of motor skills in this apparently normal group.3 1117 In a longitudinal study, we have shown that a considerable proportion have gross and fine motor problems that persist to school entry age.18 Subtle movement difficulties that can affect learning and participation in daily activities, as implied by the diagnostic label of developmental coordination disorder (DCD), may not be apparent until well into the school years.

What is already known on this topic

  • Children who were born very premature, but who do not have significant neurodevelopmental disabilities, are at risk of significant motor-based problems.

  • Developmental coordination disorder (DCD) may contribute to the learning difficulties and low self-esteem of affected schoolchildren.

What this study adds

  • At an early school age of 8 years, the prevalence of DCD is 42% in this “apparently normal” high-risk population compared with 8% for matched classroom full-term controls.

  • Motor assessment at 3 years of age using Peabody Fine Motor Scales is highly predictive of subsequent DCD at school age.

  • As resources for follow-up services are often limited, early prediction and identification of DCD would allow early intervention, before motor problems are compounded by secondary loss of self-esteem.

DCD refers to performance in daily activities requiring motor coordination that is markedly below the expected level, given the age and intellectual capacity. These motor coordination difficulties interfere with academic achievement and/or activities of daily living, and are not due to a known physical disorder, such as cerebral palsy, hemiplegia or muscular dystrophy.19 It is estimated that the incidence of DCD is 5–6% in the general population.19

Although many authors have cited poor motor coordination as an outcome of preterm birth,3 4 7 relatively few studies have examined motor dysfunction longitudinally into the school years or the prevalence of DCD in the high-risk population. DCD prevalence rates of 30.7%20 and 51%21 have been reported. Only one recent study22 has examined the prevalence of DCD in the most vulnerable preterm population, born <29 weeks or <1000 g, who were recipients of neonatal medicine in the post-surfactant era (ie, born in the 1990s). Their findings of only 10% with DCD, however, are considerably lower than other studies.

Serial motor assessments over time give the best indication of motor outcome.23 However, as resources for follow-up services are often limited and many are unable to continue surveillance to school age, it is useful to know whether an assessment before school age can detect those likely to have DCD. If an earlier assessment can be shown to be predictive of DCD, parents could be alerted to their child’s potential difficulties and the child targeted for early intervention.

The aims of this study were to: (1) determine the prevalence of DCD in a cohort of apparently normal extremely premature/ELBW schoolchildren at 8 years; (2) identify perinatal or social factors associated with DCD; (3) determine whether assessment of motor skills at 12 months, 3 or 5 years could be used to predict DCD in this population at 8 years.


The perinatal centre at Westmead Hospital is a tertiary referral hospital in Sydney receiving high-risk referrals from Sydney metropolitan and regional centres of the state of New South Wales. In this study of apparently normal high-risk children attending normal school, a matched case–control design was used in which the subjects were matched for age and gender with classroom controls. This enabled major confounding factors that can influence motor skills, namely age, gender and the school programme, to be controlled. It was not feasible for the researcher to conduct the study with controls from outside Sydney metropolitan, as New South Wales is geographically a vast state with a sparse population outside of Sydney (4 of the 6.7 million reside in Sydney metropolitan).

The study group of preterm children (cases) were recruited from a consecutive cohort of neonatal intensive care survivors who were born <1000 g or at <29 weeks’ gestation and prospectively enrolled in the growth and development clinic for regular follow-up with a multidisciplinary team. Participants were included if they resided in the Sydney metropolitan area, attended a regular school in mainstream classes, and were not identified to have a disability at their 5-year-old developmental follow-up visit (ie, full-scale IQ >84, no neurological abnormality, no visual impairment or hearing impairment requiring hearing aids).

The control group had been born at full term with normal birth weight, and were from the same class as the preterm children. The class teachers were asked to select a child of the same gender and closest birth date to each preterm child. Parents completed a form, so that any child who had been identified as having a disability, had been born preterm or had required intensive care at birth could be excluded.

Assessment of both cases and controls was conducted at the school. The Movement Assessment Battery for Children (MABC) was administered by a single examiner, an experienced occupational therapist (T-AG). The examiner was not blinded to the perinatal variables because these preterm children had previously been assessed by the examiner at 3 and 5 years of age. The MABC24 is a standardised and normative-referenced test that is designed to identify motor impairment in children aged 4–12 years. There are three subsections: manual dexterity, ball skills and static/dynamic balance. At this age, the manual dexterity subtest involves placing pegs into a pegboard, a lacing task, and tracing through narrow guidelines. The ball skills subtest involves performing a one-handed bounce and catch, and throwing a bean bag into a box. The balance subtest involves balancing on one foot, jumping into squares, and walking heel-to-toe along a line. The test was first standardised on 854 Canadian children in 1972 and revised in 1984 on 1200 children from the UK and Canada. The current MABC was further revised in 1992 on 1200 children from the USA. The total impairment score range is from 0 to 40. A higher score indicates poorer motor competence. A total impairment score above 13.5 indicates the lowest 5% in a normal distribution of scores, and represents a definite motor problem. A score of 10–13.5 reflects performance in the 6–15th centile of the range and suggests a degree of difficulty that is borderline. In this study, the 15th centile (−1 SD below the mean) was chosen as the cut-off point for indication of DCD, as recommended by others.21 2527 A score of >13 (−1.5 SD below the mean) has been regarded as severe DCD.20 22

A power analysis indicated that a sample size of 50 case–control pairs was required (odds ratio (OR) 4.5, α set at 0.05 and power at 0.80), assuming that the prevalence of DCD was 40% in the preterm group and 6% in the control group. Data were analysed using SPSS V10.1, and the p value was set at <0.05. The MABC scores were not normally distributed, and hence non-parametric analyses (Wilcoxon signed ranks test) were performed. Logistic regression analysis was used to examine perinatal/social factors independently associated with DCD. To determine whether DCD could be predicted earlier, receiver operator characteristic (ROC) curves for the Griffiths Scales (3 years) and Peabody Scales (3 and 5 years) were constructed. The area under the ROC curve was calculated to reveal the best predictor of DCD at 8 years.

Human research ethics approvals were obtained from the Scientific and Ethics Committee of Western Sydney Area Health Service (Westmead Hospital) and the governing research ethics committees of the schools attended by the children. These included the ethics committees of the NSW Department of Education (state schools) and the Catholic Dioceses of Sydney, Parramatta, Wollongong and Broken Bay (Catholic schools).


Participants were born from 1 January 1992 to 10 May 1995. There were 214 infants who were born at <29 weeks’ gestation or <1000 g and enrolled in the growth and development clinic for follow-up. By 5 years of age, 47 children (22%) were lost to follow-up, two had died, and 16 were enrolled in follow-up elsewhere. There were 149 children assessed at 5 years of age. Of these, 64 (43%) were excluded because they had a disability (eg, cerebral palsy, visual or hearing impairment) or IQ <85. Thirty-two were excluded as they resided outside the Sydney metropolitan area. Of the 53 eligible 5-year-old apparently normal children who resided in Sydney, a further two could not be traced at 8 years and were lost to follow-up. One child was excluded because parental consent for participation in the study was not given. A total of 50 subjects were enrolled.

To determine whether the study group was representative of apparently normal preterm children, neonatal variables and IQ at 5 years were compared with those of children excluded because they lived outside of the metropolitan area (n = 32), the two who were lost to follow-up after the 5-year assessment, and the one whose parents refused consent. No significant differences were found. The study group was also compared with those children lost to follow-up by the 5-year assessment (n = 47). A significant difference was only detected in gestation between the groups (included, 27.9 (1.7) weeks; excluded, 27.1 (1.5) weeks, p = 0.02). The neonatal characteristics of the group included in the study were 27.9 (1.7) week’s gestation, weighing 997 (194) g at birth and 25 (50%) were boys. Values are mean (SD). Neonatal morbidities included necrotising enterocolitis (five; 10%), grade 3–4 intraventricular haemorrhage (five; 10%), periventricular leucomalacia (one; 2%), stage 3 retinopathy of prematurity (ROP) (three; 6%) and chronic lung disease (12; 24%). The median duration of ventilation was 9.4 days (interquartile range 8–16 days)

Prevalence of DCD

The children in the study group were assessed at a mean (SD) age of 8.8 (0.3) years (controls at 8.8 (0.4) years). The median MABC total score for cases was 8.75 compared with 5 for controls (p<0.001; Wilcoxon signed ranks test for case–control pairs). The subtest scores for manual dexterity and balance were significantly different between the groups (table 1).

Table 1 Comparison of Movement Assessment Battery for Children (MABC) scores between high-risk preterm children and class-matched controls

Significantly more children met the criteria of DCD in the study group: 21 (42%) cases and four (8%) controls (p = 0.0001, OR 8.3, 95% CI 2.6 to 26.7). When a cut-off score of 13 is used (−1.5 SD below the mean), 15 (30%) of the cases and none of the controls met this criterion of severe DCD (p<0.0001, OR 44, 95% CI 2.6 to 761).

DCD was not more common in children with visual problems. Six of the 50 study children had visual problems: three needed glasses for myopia, one had strabismus, one had astigmatism, and one required glasses for reading. Of these, two (33%) had DCD. There was no difference in DCD between children with and without visual problems (2/6 vs 19/44; p = 0.50). Their MABC scores (median (interquartile range)) were also not significantly different: 7 (3.4, 17.4) for those with visual problems vs 8.8 (5.5, 13.5) for those with no visual problems (p = 0.65).

Perinatal and social factors associated with DCD

The three significant perinatal variables identified from the univariate analysis (table 2)—ie, prolonged rupture of membranes (PROM), ROP and Apgar score ⩽5—were entered into the logistic regression analysis model, and, as each variable was added, collinearity and stability were assessed.28 PROM and ROP were independently associated with DCD in the high-risk group (table 3), but when Apgar score ⩽5 was entered into the model, this variable was no longer significant (p = 0.08), and the model became unstable and indicated a degree of collinearity between PROM and Apgar score. In relation to the strength of association, analysis indicated that PROM and ROP explained 34% of the variation in DCD.

Table 2 Comparison of perinatal and social characteristics for high-risk preterm children with or without developmental coordination disorder (DCD)
Table 3 Multivariate analysis of risk factors for preterm children with developmental coordination disorder

Early prediction of DCD

Table 4 summarises results of the ROC curves of motor assessments at earlier ages in the prediction of DCD at 8 years. The 3-year fine and gross motor tests (27th and 41st centile, respectively) of the Peabody Scale had the largest area under the curve (0.78 for both tests) as predictors of DCD at 8 years. However, the fine motor assessment had a lower cut-off at the 27th centile than the gross motor assessment of the 41st centile. The 5-year Peabody Fine Motor Scale had the lowest cut-off of the 19th centile and yielded a similar result (area under the curve = 0.71).

Table 4 Receiver operator characteristic curve analysis for predicting developmental coordination disorder (defined as less than the 15th centile) using previous motor test results at earlier ages


Results of this study indicate that a significant proportion of apparently normal extremely preterm/ELBW children continue to have motor-based problems into the school years. The prevalence of DCD was 42% in the preterm/ELBW group, which is significantly higher than 8% in the control group, using a −1 SD cut-off score on the MABC. The MABC total impairment score was higher for the apparently normal preterm/ELBW group. More importantly, 30% had severe DCD compared with none in the controls.

Holsti et al21 used a −1 SD cut-off and reported a 51% prevalence in their group of <800 g infants and that the cohort was born into a considerably earlier era. Both these factors have been associated with poorer outcomes in the extreme preterm/ELBW population and could explain the higher rate of DCD.10 29 30 Powls et al,26 who used the MABC with preterm children, reported similar rates of 51% motor impairment in their cohorts.

Our findings are also comparable to those of Foulder-Hughes and Cooke,17 whose apparently normal preterm group had a median score of 8.5 on the MABC. These authors used a cut-off score of −1.5 SD and found a DCD prevalence of 30% for the extreme preterm/ELBW group and none for the controls, which is identical with our findings for severe DCD. However, one could question whether this is the most appropriate cut-off score to use, as it did not detect any of the controls with DCD. A prevalence of 5–6% of controls would be expected in the general population. The −1 SD cut-off point appears to more accurately reflect expectations in the general population and has been recommended by others.21 2527 31 32 Holsti et al21 argued that it is standard clinical practice to use the −1 SD cut-off score on standardised motor assessments and that “the use of a −1 SD cutoff captures those children who are performing below their peers” (p 12).

In contrast, Davis et al22 found a considerably lower prevalence of DCD (10%) in a comparable group of 8-year olds born <1000 g or <28 weeks. The MABC with a cut-off of −1.5 SD (<5th centile) was used in the Davis et al study. However, it does not explain the significantly lower rate of DCD. In fact, other studies, including ours, have found a prevalence rate three or more times that reported by Davis et al.22 The reason for this unique discrepancy from others is entirely unclear, as neonatal survival and morbidity are similar in the two concurrent Australian neonatal intensive care unit populations. Our study was performed by an experienced occupational therapist. Although assessments were not blinded to the group allocations, inter-rater reliability was validated, and results were consistent with other studies conducted by occupational therapists. The study of Davis et al incorporated assessment data from a large range of paediatricians and psychologists, who were administering a battery of other tests. The inter-rater reliability of the MABC assessors was not established. Of note, Davis et al found that the prevalence of DCD was five times higher in their high-risk group than in their control group (−1.5 SD cut-off; 10% vs 2%). Our results also showed a five times increase in DCD compared with our control children (−1 SD cut-off; 42% vs 8%).

Two perinatal variables, PROM and ROP, were independently associated with DCD, accounting for 34% of the variance. No other perinatal or social variables were significantly associated with DCD. However, it should be acknowledged that those with severe neonatal morbidity, such as severely abnormal head ultrasound finding and chronic lung disease, are likely to be excluded from the study of only apparently normal subjects.

The role of PROM in cerebral palsy and white matter injury in preterm infants is not clear because of the various definitions of PROM used, differences in gestation and paucity of studies available.33 However, it is likely that PROM, a variable linked with antenatal infection and cerebral palsy, is associated with DCD in high-risk children. In addition, ROP was also independently associated with DCD. Of those with no ROP, 27% were identified with DCD, whereas 46% with stage 1–2 ROP and 100% with stage 3 ROP had DCD. Even though there were only three subjects with stage 3 ROP in this study, these findings lend support to the view that the severity of ROP influences DCD. Contrasting results have been found.34 35 Cooke et al34 showed that, once gestation was controlled for, ROP did not influence DCD. However, the relationship between visual outcome and DCD has also been reported by the same authors.36 Previous studies have found a relationship between DCD and motor imagery, namely a deficit in visual spatial representations of intended movement.37 38 It is feasible that ROP may be linked with a motor imagery deficit in preterm children with DCD. If this is the case, treatment with visual imagery training37 could ameliorate motor dysfunction in this group and may be a future line of research. As visual outcome for the high-risk subjects was not specifically examined at 8 years in the present study, it is difficult to determine whether this was associated with DCD.

Our results indicate that the Peabody Gross and Fine Motor Scales at 3 years are good at detecting DCD at 8 years (area under the ROC curve 0.78). However, the 3-year fine motor assessment would be clinically more precise and relevant with a lower cut-off at the 27th centile, as compared with the 41st centile cut-off using the gross motor assessment (similarly with the Griffith’s Locomotor Scales). The 5-year Peabody Fine Motor Scale score of less than the 19th centile had a clinically more precise prediction cut-off and a similar, but slightly lower, yield (area under the ROC curve 0.71). Prediction of motor morbidity at a school age of 5 years was not surprising and may be considered a more precise assessment. However, prediction at an earlier age, ie, 3 years, may allow more opportunity for early intervention to improve fundamental motor skills. This study has established that an earlier motor assessment at 3 years of age is a good predictor of DCD at 8 years. Our findings, if confirmed with studies of larger samples and, more importantly, if they lead to successful intervention trials, will have a large impact on the outcome of these vulnerable children.

In summary, our findings reveal that motor dysfunction, as implied by the diagnostic label of DCD, is prevalent in apparently normal children born at extreme prematurity/ELBW. Perinatal and environmental factors provide limited insight into those children likely to have inferior motor skills at school age. PROM and ROP were significantly and independently associated with DCD, perhaps reflecting the effect of the antenatal infection process and visual development related to ROP on motor outcome. Extremely preterm/ELBW children who are apparently normal should be assessed at 3 years with a standardised motor assessment to identify those likely to have DCD. This is a particularly crucial finding for developmental follow-up services with constraints that do not allow surveillance into school age.


View Abstract


  • Competing interests: None.

  • Ethics approval: Obtained.

  • Patient consent: Obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.