Identifying premature infant at high and low risk for motor delays using motor performance testing and MRS Patricia Coker-Bolt, PhD OTR/L, FAOTA1; Kathryn Hope2, MPH; Viswanathan Ramakrishnan, PhD3; Truman Brown, PhD4; Denise Mulvihill, MD1, and Dorothea Jenkins, MD2

1Division of Occupational Therapy, 2Department of Pediatrics, 3Public Health Sciences, 4Radiology, Medical University of South Carolina, Charleston, SC

Background:

• Premature infants with normal cranial ultrasounds remain at risk for later

motor delays, which go undetected in early infancy.2

• Of the 12-16% of children with developmental delay, only half will be

identified by the time they enter kindergarten.1

• The Test of Infant Motor Performance (TIMP) is the current gold

standard infant motor assessment, but rarely used by pediatricians

during well-child visits due to lack of time and special training required.

• A short, standardized screening test administered to infants in the first

months of life would target early intervention to those most at risk.3

• In our previous work, Rasch partial credit model was used to analyze

and select 10 TIMP items with strongest correlation to motor ability at 12

months. New scales were then developed for more sensitive scoring of

these 10 items, comprising a novel screening tool, the Specific Test of

Early Infant Motor performance (STEP) (Figure 1).

Specific aims: Evaluate the robustness of a new, shortened, screening

assessment (STEP) through factor analysis of 10 motor items, and test the

STEP against current validated infant motor skills assessments and

outcome measures of development.

Design: A secondary analysis of an existing cohort of 22 preterm infants

(24-35 weeks gestation), with video recorded motor tests at 12 weeks and

12 months corrected gestational age (CGA). (Table 1)

Measures of Outcome:

• STEP scored from existing video recordings of TIMP at 12 weeks CGA.

• Test of Infant Motor Performance (TIMP) at term and 12 weeks CGA.

• Bayley III Scales of Infant and Toddler Development (Bayley III) at

12 months CGA. Bayley subscale scores ≤8 = low/below average 4,5.

• MRS: Siemens 3T: Single voxel [15 x 15 x 15 mm] in basal ganglia (BG)

and frontal lobe white matter (WM) [echo PRESS sequence128 avg,

TR=2000ms, TE 30ms, 270ms].

• Data Analysis: Pearson’s correlational coefficient was used to relate

TIMP, Bayley and STEP scores. Exploratory factor analysis (EFA) was

used to identify latent constructs of STEP. Logistic regression, using EFA

scores, was performed to predict dichotomized Bayley outcome.

Sensitivity and specificity of EFA STEP scores were evaluated using

predictive outcomes versus actual outcomes. A drop-one-predict

evaluation was used to assess external validity.

ABSTRACT

DESIGN/ METHODS

RESULTS

Table 1. Patient Demographics Total Infants (N=22)

Sex 10 Females, 12 Males

Race 11 Caucasians, 11 African Americans

GA at Birth (mean ± SD) 29.6 ± 2.9 weeks

Birth weight (mean ± SD) 1259 ± 634 grams

Intracranial lesions 4 IVH (3 grade I, 1 grade II), 0 infarcts/PVL

Risk (TIMP at 12-weeks CGA) 9 High; 13 Low

Bayley Motor Outcome (N=19) 8 Average; 11 Low/Below Average

Figure 1. Representative schematic of STEP scale for “Prone Extension”

Head Control Upper

Extremity

Lower

Extremity

Pull to sit 0.89540 -0.23615 -0.06632

Prone extension 0.89496 -0.15936 -0.21346

Supine with no vision 0.84610 0.20399 -0.16812

Supine with vision 0.76003 -0.30853 0.16172

Standing 0.69165 -0.43411 0.23230

Supine sitting 0.76553 0.36989 0.00095

Grasp 0.24003 0.72577 -0.29594

Kicking 0.38154 0.48327 0.76373

Rolling with leg 0.65745 0.11327 -0.17037

% variance explained 68% 20% 12%

Table 2. Latent constructs of STEP using Exploratory Factor Analysis (EFA)

• STEP correlated with TIMP scores at

term (r=0.062, p=0.003), 12 weeks CGA

(r=0.79, p<0.001), as well as the change

in TIMP scores from term to 12 weeks

(r=0.46, p=0.035), indicating it might be a

short, specific surrogate measure of a

much longer motor test. (Fig. 2)

Figure 2. Correlations between the STEP and TIMP

• STEP correlated better with Bayley outcomes than the TIMP (Table 3,

Figure 3)

Bayley: Gross

Motor

Fine

Motor

Total Motor

Score

STEP Scale:

R value 0.68519 0.47682 0.68634

p-value 0.0012 0.039 0.0012

TIMP Scale:

R value 0.474 n/s

0.437

p-value 0.04 0.061

Table 3 & Figure 3. Correlations of STEP and TIMP scales at 12-weeks

CGA to Bayley scores at 12-months

• 1 item (rolling with arm) was unable to accurately be scored by viewing

existing video tapes, therefore the STEP score was constructed from 9

items for the EFA analysis.

• A pattern of relationships exists between STEP items that evaluate

head, arm, and leg movements (Table 2).

• STEP items related to head movements are likely influencing the overall

score (Table 2).

DISCUSSION

CONCLUSIONS

REFERENCES 1.Mackrides PS & Ryherd SJ. Screen for developmental delay. American Family Physician 2011;

84:544-560.

2.Donahue, P.K., Graham, E.M. Earlier markers for cerebral palsy and clinical research in

premature infants. Journal of Perinatology. 2007; 27: 259-261

3.Coker-Bolt, P., Woodbury, M., Perkel, J., Moreau, N., Hope, K., Brown, R., Ramakrishnan, R.

Mulvihill, D., & Jenkins, D. (2012). Identifying premature infant at high and low risk for motor

delays using motor performance testing and MRS. Journal of Pediatric Rehabilitation Medicine,

2014 (in press).

4.Spittle AJ, et al. Does the Bayley II Motor scale at 2 years predict motor outcome at 4 years in

very preterm children? Developmental Medicine and Child Neurology. 2013; 55:448-452.

5.Vohr BR, et al. Are outcomes of extremely preterm infants improving? Impact of Bayley

assessment on outcomes J Peds. 2012; 161:222-8.

6.Bosanquet, M., Copeland, L., Ware, R., & Boyd, R. (2013). A systematic review of tests to

predict cerebral palsy in young children. Developmental Medicine & Child Neurology, 55,418-

426.

7.Barbosa, V. M., Campbell, S. K., Smith, E., &Berbaum, M. (2005).Comparison of test of infant

motor performance (TIMP) item responses among children with cerebral palsy, developmental

delay, and typical development. American Journal of Occupational Therapy, 59, 446-456.

• The STEP is quick and easy to administer and score and requires

minimal training of health care professionals.

• The 10-item STEP at 12-weeks showed a much stronger correlation than

the concurrent full 42-item TIMP with the Bayley at 12-months.

• Early motor skills should reflect biochemical evidence of subtle brain

injury. Preliminary analysis of the STEP scales showed correlations to

key brain metabolites.

• The STEP demonstrated good psychometric properties and predictive

abilities for infant developmental outcomes on the Bayley at 12

months.

• The motor items of STEP were grouped into three meaningful latent

constructs which show a pattern of relationships between motor

movements which evaluate head, arm, and leg movements.

• Early head movements, patterns of standing, kicking, and visual

responses may be sensitive to discriminate differences between typical

and at-risk infants.

• Quality of movement is a marker of the integrity of an infant’s nervous

system and could be a predictor of developmental disability and provide

valuable information on how at-risk infants interact with people and

objects in their environment.3,7

Table 5. Comparison of common infant tests and STEP:

Sensitivity and Specificity6

Assessment Sensitivity Specificity

STEP 0.75 0.82

TIMP 0.86 0.68

Cranial ultrasound 0.70 0.69

Neurologic Exam 0.88 0.87

• Ability of the STEP to predict Bayley Motor outcomes , using EFA scores,

in the current cohort: Sensitivity =1.0 and Specificity= 0.91

• Ability to prospectively identify at risk infants (External Validity):

Sensitivity=0.75 and Specificity=0.82 (Table 5). Funded by MUSC’s SCTR Discovery & Early Career Grant FY13

UL1 TR000062

• Though our numbers are small, the strength of the associations are

encouraging and provide strong pilot data and proof of concept for

potential validity of the STEP assessment.

• This is a preliminary analysis of the psychometric properties of the STEP.

Validation of the STEP as a clinically useful infant motor assessment

requires further analysis.

• Future work: currently enrolling and testing infants using a revised STEP

for adequate subject to variable ratio needed to complete a more through

factor analysis of the psychometric properties of the STEP.

EVALUATION OF ROBUSTNESS OF THE STEP

CONCURRENT AND EXTERNAL VALIDITY OF THE STEP

• MRS metabolite ratios mI/Cr,

mI/NAA, and NAA/Cho

showed the most significant

associations with the 12 week

TIMP and STEP scores

(Table 4)

Metabolites STEP

(p=)

TIMP

(p=)

WM mI/Cr 0.008 0.007

WM mI/NAA 0.016 0.026

WM NAA/Cho 0.006 0.0197

Table 4. Motor test associations with common MRS metabolites in WM