Movement Assessment Battery for Children (Movement-ABC)
Relative age effects in the Movement Assessment Battery for Children-2: age banding and scoring errors
Purpose
To determine and discuss error caused by age-grouped scoring of relative motor function using the MABC-2. Due to the MABC-2 using 6 month wide scoring bands for children 3 and 4 years old and 1 year wide bands for children 5 and older, children who have just had their birthday may appear to score lower than children about to turn a year older. This is referred to as an ‘age effect’ and it causes error in scoring.
Study population
N= 278 (mean (SD) age: 5 years, 0 months (9.6 months); 142 female) recruited from the Hamilton and Ontario areas from 2010 to 2014.
Methods
Data was pooled from two validation studies, Veldhuizen et al. 2015b, and Cairney et al. 2015. The participants for the these two studies were recruited from the Hamilton and Ontario areas from 2010 to 2014. Inclusion criteria included literacy and english speaking guardians or parents. A coefficient for relative age was calculated, which shows the difference between the standard MABC-2 score and the age adjusted score across one year. Estimates of the rates of misclassification of recommended thresholds were calculated from the regression analysis of standard score and relative age. Misclassification rates were calculated for the recommended cutoff points at the fifth and the 16th percentile (used to identify children with probable motor impairment and ‘at risk’ children respectively)
Outcome measures
Misclassification rates of the MABC-2.
NA
Intervention
NA
Results
There was a significant relative age effect, with older children within the same age band receiving higher standard scores. The standard score varied by 2.76 points on the MABC-2 per year of relative age. The rate of misclassification was found to be 9-23% depending on the child’s actual age.
Strengths
This study identifies and characterizes a major scoring flaw in the MABC-2 that could have implications for care that is available to children for whom this outcome measure is used. Solutions to mitigate the scoring errors are suggested in this study. The MABC-2 is one of the most widely used outcome measures for children with motor coordination impairment. Results were highly significant.
Limitations
The sample size was relatively small and all from a single geographical area, so generalizability cannot be assumed. The only age range tested was between 4 and 6 years old. The error misclassification rate may be different for other age ranges tested on the MABC-2. Error rates will be highest in the most rapidly developing age ranges.
Conclusion
There are significant and avoidable age effects in the MABC-2 test that make children closest to the age band limits prone to misclassification. Relatively younger 5 year old children are more likely (7 times more likely) to be identified, or misidentified as having motor control deficits than older 5 year old children. Misclassification rates are estimated to be between 9 and 23%. Implications are that as many as 23% of children functioning below normal ranges would be identified as normal, which could make justification for health services more difficult. Other methods of calculating scores exist that could mitigate these errors.
Reference
Veldhuizen, S., Rivard, L., & Cairney, J. (2017). Relative age effects in the Movement Assessment Battery for Children-2: age banding and scoring errors. Child: Care, Health and Development, 43(5), 752-757. doi:10.1111/cch.12459
2 responses to “Movement Assessment Battery for Children (Movement-ABC)”
Leave a Reply
You must be logged in to post a comment.
Devan good article review!
My question is given the effect of miscalculating motor control deficits for younger children, what is one of the solutions the article offer to mitigate this miscalculation? Would adjusting the scoring ages in months for children of 5 -6 year of age allow for less disparity?
Great question! The authors recommend several approaches to mitigating the error rate. The first is computerized scoring to produce ‘developmental ages’ from raw scores, and/or produce growth curve charts. The Lambda–Mu–Sigma method, for example, produces a z-score that is more representative of the child’s exact age. These methods are not easily implemented in a clinical setting, however, as you suggested, using more narrow age banding such as scoring ages by month would also reduce the error rate.
To make these calculations clinically useful they would likely need to be incorporated into an electronic medical record that could convert the scores based on the child’s exact age.