Use of the Alberta Infant Motor Scales in Screening for Motor Abnormality

The Alberta Infant Motor Scales (AIMS) is a tool for assessing motor development. The AIMS has high sensitivity for ruling out motor abnormality and this study recommends the use of the 5th percentile as a cutoff score for ruling in motor abnormality.

This study is level 4 evidence due to its non-independent reference standard and non-blinded design.

Citation/s: Campos D, Santos DC, Goncalves VM, Goto MM, Arias AV, Brianeze AC, Campos TM, Mello BB. Agreement between scales for screening and diagnosis of motor development at 6 months. J Pediatr (Rio J). 2006; 82(6):470-4.

Lead Author's name and fax: Vanda Maria Gimenes Goncalves.

Three-part Clinical Question: What is the most appropriate cutoff score on the Alberta Infant Motor Scales (AIMS) to identify motor development abnormalities in 6 month old infants?

Search Terms: A broad, sensitive defined search on PubMed Clinical Queries in the category Diagnosis using the terms "Alberta Infant Motor Scales" brought up 5 citations. The title and a quick peek at the abstract on the cited article seemed right on target for my question.

The Study: The Study Patients: The sample population in this study were 43 infants, age 6 months (+/- 7 days) who were part of a cohort that had been enrolled as neonates all being born single-fetus pregnancies, at gestational ages of 37 to 41 week, in the standard maternity ward, requiring no special care, were discharged from the maternity unit 2 days after birth, residents of the metropolitan area of Campinas, Brazil. Infants were excluded if they had low birth weight, exhibited genetic syndromes, malformations, or congenital infections. Demographic data on this sample showed a healthy population with: birth weight (grams) mean 3065.93, SD 370.26, gestational age (weeks) 39.9, SD 1.03, 1-minute Apgar (1-10 scale) mean 8.18, SD 1.72, 5-minute Apgar mean 9.45, SD 0.55. The spectrum of healthy infants aged 6 months seemed appropriate. The study made comparison with a reference (gold) standard for each subject, but non-independently and non-blinded. Target disorder and Gold Standard: Target disorder was "motor abnormality". The "Gold Standard" was an index score less than 85 (or greater than one standard deviation below the mean) on the Bayley Scales of Infant Development II (BSID-II). The BSID-II is a developmental diagnostic tool that assesses children aged 1 to 42 months and requires about 40 minutes to administer. It is considered among the best scales available in the area of child development assessment, providing reliable, valid and precise results. Diagnostic test: The Alberta Infant Motor Scale (AIMS) is an assessment instrument that evaluates gross motor function from birth to 18 months. This instrument classifies infants against a development curve between the 5th and 90th percentiles, but does not establish a range of percentiles as cut-off points for further diagnostic assessment. Assessment with the AIMS is a process of scoring observed behaviors of an illustrated sequence of postural control development in the positions of prone, supine, sitting, and standing. Administration of the AIMS takes approximately 15 minutes and has excellent psychometric properties of test-retest and interobserver reliability and validity.

The Evidence: Note: Alied 0.5 to the counts in all four cells of the observed tables below.

Target Disorder: Motor Abnormality
Test: AIMS <=5% / Present / Absent
Test Result / Num / Prop / Num / Prop / Likelihood Ratios
Positive / 6.5 / a / 8.5 / b / 4.15 / 2.22 to7.77
Negative / 0.5 / c / 29.5 / d / 0.09 / 0.01 to 1.34
Sensitivity: / 93% / 74 to 100
Specificity: / 78% / 64 to 91
Prevalence: / 16% / 5 to 26
Positive Predictive Value: / 43% / 18 to 68
Negative Predictive Value: / 98% / 94 to 100
Target Disorder: Motor Abnormality
Test: AIMS <=10% / Present / Absent
Test Result / Num / Prop / Num / Prop / Likelihood Ratios
Positive / 6.5 / a / 19.5 / b / 1.81 / 1.25 to 2.62
Negative / 0.5 / c / 18.5 / d / 0.15 / 0.01 to 2.16
Sensitivity: / 93% / 74 to 100
Specificity: / 49% / 33 to 65
Prevalence: / 16% / 5 to 26
Positive Predictive Value: / 25% / 5 to 42
Negative Predictive Value: / 97% / 90 to 100

Comments: Are the results valid? A significant threat to the validity of this study is the lack of blinding. Assessments of the infants in this sample were made simultaneously by an examiner and two observers with scores recorded by consensus among these three team members. Although the diagnostic test and the gold standard test were both administered for each subject, they were administered by the same team, and at the same time. The examiners knew the score on both the diagnostic test and the gold standard, and although there is no description of which assessment was administered first, the absence of blinding allowed for bias. In regard to the sample spectrum, I am not too concerned about the omission of the gender characteristics. Neither am I bothered about the possible cultural differences between the Brazilian infants and my American infants, after all, the AIMS normative values are based on Canadian infants. But my concern is that the study's healthy-at-birth group of infants had no description of their characteristics at 6 months, which was at the time of the testing. The study provides no information about the health status of the subjects between birth and age 6 months. Since the intent of the study was to ascertain agreement between the screening and diagnostic tool and define the best cutoff point for the screening, the lack of description of the current medical status of the population does not further question the validity of the study.

What are the results? The study reports by two cutoff point (5th and 10th percentile scores on the AIMS) the frequency of infants identified as true positive and negatives and as false positives and negatives, and the degree of agreement between the AIMS and BSID-II. The sensitivity of the AIMS at both cutoff points is high 93% (range 74 to100). Applying the SnNout concept, this indicates that the AIMS is very good at ruling out motor abnormality when the test is negative (or the score is at or above the 10th percentile score). The specificity of the AIMS at the 10th percentile cutoff is 49%, with a broad confidence interval 33-65%, which indicates that at this cut point that the AIMS is not very good at ruling in a motor abnormality. In aliition, given the pre-test likelihood of 16% and the small positive likelihood ratios at 1.81 (range 1.25 to 2.62) a positive test does not provide a large change or convincing post-test likelihood (even though the PLR is significant as the range does not include the 1.0 null point). But the specificity of the AIMS at the 5th percentile cutoff is 78%, still with a broad confidence interval 64-91%, indicating that the 5% cutoff point is better than the 10% cut-point at ruling in a motor abnormality given a low (positive) test score. Looking at it another way, the lower cutoff score reduces the number of false positives, thus reducing the likelihood of inappropriately identifying an infant which may lead to unnecessary evaluation or bring unnecessary concern. The PLR at this lower cutoff point is 4.15 (range 2.22 to 7.77), and though still small is up one level per Jaeschke's taxonomy, as compared to the 10% cut-point PLR, and may generate small yet important changes in probability. Aliitionally the NLRs for both cut-points are not significant since the sample results include the null value, 1.0. In regard to the degree of agreement between the AIMS and the BSID-II, the authors report accuracy and kappa indices for the two cut points. But since they were all together, observing and scoring each infant after agreement had been reached, the biases introduced would influence these tests.

How can I apply the results to patient care? Since the purpose of my health promotion program is to promote parent-infant interaction to facilitate normal sensorimotor development, screening to identify infants as typically developing is important. The AIMS is an appropriate tool because of its high sensitivity making it appropriate for use in screening. The AIMS is already available in my practice, without aliitional cost, which is important in my small private business. After reviewing the sections on psychometric properties and norm-referencing of the AIMS in Motor Assessment of the Developing Infant by Piper MC and Darrah J. WB Saunders (1994), I have enough confidence to over-ride my concerns of the validity of the study, and I will plan to use the AIMS, with the 5% cutoff score for motor screening of the babies enrolling in Massage and Movement class. This study confirmed the sensitivity of the AIMS as a screening tool and statistically defined the specificity value of using the 5% cutoff score to ascertain ruling in motor abnormality.

Appraised by: Dawn Stanuley, November 17, 2007 Email: