Baby Observational Selective Control AppRaisal (BabyOSCAR): Convergent and discriminant validity and reliability in infants with and without spastic cerebral palsy
This original article is commented by Heathcock on pages 1406–1407 of this issue.
Abstract
Aim
To describe the development of an observational measure of spontaneous independent joint motion in infants with spastic cerebral palsy (CP), the Baby Observational Selective Control AppRaisal (BabyOSCAR), and to test its convergent validity and reliability.
Method
A retrospective sample of 75 infants (45 with spastic CP and 30 without CP) at 3 months of age were scored with the BabyOSCAR and compared with diagnosis of spastic CP, limbs affected, and Gross Motor Function Classification level at 2 years of age or later for convergent validity using t-tests, Kruskal–Wallis tests, and Spearman's rank correlation coefficients. BabyOSCAR interrater and test–retest reliability was also evaluated using intraclass correlation coefficients.
Results
Infants with spastic CP had significantly lower BabyOSCAR scores than children without CP (p < 0.001) and scores were significantly correlated with Gross Motor Function Classification System levels (p < 0.001). Children with unilateral CP had significantly higher asymmetry scores than children with bilateral CP or no CP (p < 0.01). Interrater and test–retest reliabilities were good to excellent.
Interpretation
Reductions in independent joint control measured in infancy are a hallmark of eventual diagnosis of spastic CP, and influence gross motor function later in childhood (with or without a diagnosis of CP).
What this paper adds
- Early brain injury causing spastic cerebral palsy results in fewer independent joint movements in infants.
- Baby Observational Selective Control AppRaisal (BabyOSCAR) score at 3 months depends on limbs affected by early brain injury.
- BabyOSCAR scores at 3 months correlate with Gross Motor Function Classification System level at ≥2 years.
- BabyOSCAR has excellent interrater reliability.
- BabyOSCAR, scored with a 1-minute video recording, has good to excellent test–retest reliability.
What this paper adds
- Early brain injury causing spastic cerebral palsy results in fewer independent joint movements in infants.
- Baby Observational Selective Control AppRaisal (BabyOSCAR) score at 3 months depends on limbs affected by early brain injury.
- BabyOSCAR scores at 3 months correlate with Gross Motor Function Classification System level at ≥2 years.
- BabyOSCAR has excellent interrater reliability.
- BabyOSCAR, scored with a 1-minute video recording, has good to excellent test–retest reliability.
Selective motor control at 3 months of age can be quantified by observing spontaneous movements in supine for 1 minute using the BabyOSCAR. The BabyOSCAR score represents the number of joints across the body that demonstrated independent movement during the minute of observation. This score is strongly correlated with gross motor function (GMFCS) at 2 or more years of age.
This original article is commented by Heathcock on pages 1406–1407 of this issue.
Abbreviations
-
- BabyOSCAR
-
- Baby Observational Selective Control AppRaisal
-
- SMC
-
- selective motor control
One of the primary challenges to movement function in children and adults with spastic cerebral palsy (CP) includes the lack of selective voluntary motor control.1 Selective motor control (SMC), otherwise known as independent joint control, is defined as the ability to move one joint in isolation without the concurrent movement of other joints.2 When done in response to a specific instruction where intention is known, it can be defined as selective voluntary motor control.3 SMC affects the function of individuals with CP more than spasticity, strength, or range of motion deficits,1 but has not often been measured in young children and infants.
Selective voluntary motor control has been studied with engineering techniques2 as well as clinical assessments in cohorts of children and adults who are able to follow directions.4, 5 These studies reveal that having greater ability to selectively activate muscles is strongly related to independence in mobility both in gross motor4 and in manual ability5 tasks.
Despite the quantification of these behaviors in childhood, consensus on when SMC emerges is lacking. It is hypothesized that the maturity of the corticospinal system is responsible for the ability to completely differentiate volitional movements.6 During the first weeks of an infant's life, the corticospinal system is not yet myelinated and control of the movements and behaviors is probably more influenced by subcortical structures.7 As the brain develops, a cortical influence on behavior increases and a new functional repertoire of social, cognitive, language, and motor performance emerges.8 One change coinciding with this neurodevelopmental shift in function is the endogenously generated appearance of fidgety movements, defined as small multiplanar joint motions.9 The absence of fidgety movements at 3 months corrected age is predictive both of spastic and of dyskinetic CP diagnoses,9 indicating that spontaneous motor behavior reflects the integrity of the nervous system. Therefore, the observation of self-generated SMC at 3 months of age may provide specific insights into the integrity of the corticospinal system and the development of spastic CP. Currently, no tool exists to measure SMC clinically in infants.
In this paper we introduce the Baby Observational Selective Control AppRaisal (BabyOSCAR), which we created to provide a systematic approach for evaluating SMC in infants. BabyOSCAR quantifies an infant's capacity to perform isolated movements in joints of the arms and legs, identifying which joints the infant can and cannot move in isolation. We counted the number of joints in which individual movements were observed in infants at 10- to 16-weeks post-term age and hypothesized that having a higher number of joints where individual movements were seen (higher BabyOSCAR score) would be related to better gross motor function in childhood in a cohort of infants with and without spastic CP. Our objective was to evaluate and describe the reliability and convergent validity of the BabyOSCAR tool.
METHOD
Tool development
We began by reviewing current tools, the Selective Control Assessment of the Lower Extremity4 and the Test of Arm Selective Control,5 that measure selective voluntary motor control in older children and adults who are able to follow verbal directions. We next compiled a list of movements that could be observed in the extremities while in a supine position. A score sheet (Figure 1) was created where there was one point possible for each observed isolated movement, and a manual of definitions and procedures was developed (Appendix S1). An isolated movement was defined as one that was done in the absence of movement at other joints in the same limb, and without the same movement occurring simultaneously on the opposite side of the body (i.e. mirroring). In total there are 18 items for the upper extremities (arm score) and 14 items for the lower extremities (leg score), for a total of 32 items (16 from each side of the body). Scores were determined in total score (sum of all observed movements), by limb (right arm, right leg, left arm, or left leg), and using an asymmetry value calculated by subtracting the left-side score from the right-side score for arm, leg, and total scores.
BabyOSCAR scoresheet. The scoresheet is used to record observations during the 1-minute video clip of an infant moving spontaneously. Further instructions are available in the manual (Appendix S1).
Abbreviation: BabyOSCAR, Baby Observational Selective Control AppRaisal.
Scoring
One-minute video recordings of the spontaneous movements of alert and calm infants aged 10 to 16 weeks corrected age in supine position were obtained. During the video recording all limbs, toes, and fingers could be visualized, pacifiers were removed, and the infant was not crying or irritable. The angle of the video was ideally overhead with the head on the top part of the screen; however, any video where all limbs were easily visualized was used. Ratings were completed by experienced pediatric physical therapists and physical therapy students who were trained using the manual and who discussed and reviewed videos to establish initial reliability. The raters were divided into pairs and scored together in real time. One rater pair blinded to infant outcome scored all videos using the BabyOSCAR. A second rater pair blinded to the infant outcome scored a subset of videos for interrater reliability. An additional rater scored two separate 1-minute clips from the same session to evaluate test–retest reliability.
Participants
Infants with and without spastic CP who had previously participated in research studies using the General Movement Assessment where they had at least one video recording of spontaneous movements between 10 weeks and 16 weeks corrected age were retrospectively identified for inclusion in this study. Videos were included in analysis only if there were data on the 2-year outcome of the infants (presence of spastic CP diagnosis, side of diagnosis, and Gross Motor Function Classification System [GMFCS] level), and parents had provided permission for their use. CP diagnosis and GMFCS level assignments were made by trained physicians at 2 years or later corrected age. Infants were excluded from the study if they had a known orthopedic condition that prevented joints from moving independently or were diagnosed with either dyskinetic CP or a mixed dyskinetic and spastic CP type. Infant data were collected with institutional review board approval from four sites conducting longitudinal research on infant development and general movements: (1) Northwestern Memorial Hospital, Chicago, IL, USA, (2) Lurie Children's Hospital, Chicago, IL, USA, (3) University of Chicago Comer Children's Hospital, Chicago, IL, USA, and (4) Beatrix Children's Hospital, Groningen, the Netherlands.
Convergent validity
For a person with CP, the GMFCS classifies the level of self-initiated mobility on the basis of their capabilities in sitting, walking, and wheeled mobility.10, 11 Gross motor function has been previously demonstrated to be related to selective voluntary control in children,4 and therefore was used to evaluate the convergent validity of BabyOSCAR scores (arm, leg, total, asymmetry). Although not a true classification, we assigned a GMFCS score of 0 for those participants without CP for the purposes of correlation analysis (Table 1). A Shapiro–Wilk test was used to examine normality of data, and relationships between tests were examined using Spearman's rank correlation coefficient (ρ), secondary to the non-parametric nature of the data.
| Property | Operational definition |
|---|---|
| Convergent validity | Strength of the correlation between two tests of the same or similar constructs (selective motor control and gross motor function).33 We used Spearman's rank correlation coefficient. |
| Known-groups validity | General evidence in support of construct validity. Known-groups validity is the ability of a tool to predict how certain groups are expected to score (those with and without a CP diagnosis; those with different types of CP).33 We used t-tests and Kruskal–Wallis tests. |
| Interrater reliability | Similarity in scores between two raters evaluating the same video segment.33 This was quantified using ICC. ICC reliability was interpreted using general guidelines for reliability coefficients, with values <0.40 indicating poor agreement, 0.40–0.75 indicating good agreement, and >0.75 indicating excellent agreement.12 |
| Test–retest reliability | Consistency of measurement of selective control at different administrations (different sections of video, rated on different days to reduce bias).33 This was evaluated using ICC. ICC reliability was interpreted using general guidelines for reliability coefficients, with values <0.40 indicating poor agreement, 0.40–0.75 indicating good agreement, and >0.75 indicating excellent agreement.12 |
- Abbreviations: CP, cerebral palsy; ICC, intraclass correlation coefficient.
Known-groups validity
To evaluate two types of known-group validity (Table 1), we first completed a t-test on scores between infants with and without a future diagnosis of CP. Furthermore, we evaluated the impact of limb involvement (bilateral CP, left-side unilateral CP, right-side unilateral CP, no CP diagnosis) on the BabyOSCAR scores (arm, leg, total, asymmetry) using Kruskal–Wallis tests. Post-hoc comparisons were Bonferroni corrected.
Relationship between gestational age and filming age with BabyOSCAR
To understand whether there was a relationship between either gestational age or corrected age of infant at time of BabyOSCAR assessment, we first used a Mann–Whitney U test to evaluate the differences between infants born preterm and at term in the group without a CP outcome on the BabyOSCAR scores (arm, leg, total, and asymmetry). Next, we used Spearman's rank correlation coefficient to look at the relationship between (1) gestational age and (2) the age at time of BabyOSCAR assessment for all infants in the study.
Interrater reliability and test–retest reliability
A subset of videos (n = 30: 15 without CP, 10 with bilateral CP, 5 with unilateral CP) were scored by two different rater pairs for interrater reliability (Table 1). To understand whether a 1-minute video clip was sufficient to capture an infant's performance, we assessed test–retest reliability (Table 1) using a subset of randomly selected infant videos (n = 12: 7 with CP, 5 without CP) that lasted at least 2 minutes (in the same session). These videos were trimmed into two different 1-minute segments and were scored on the BabyOSCAR by the same rater, but on different days with at least 1 week between scoring instances. Both interrater and test–retest reliability were assessed using intraclass correlation coefficients (ICCs). Two-way random effects models and corresponding 95% confidence intervals (CIs) were fitted for arm, leg, and total scores. ICC reliability was interpreted using general guidelines for reliability coefficients, with values less than 0.40 indicating poor agreement, 0.40 to 0.75 indicating good agreement, and greater than 0.75 indicating excellent agreement.12
Statistical analysis
SPSS version 28 (IBM Corp., Armonk, NY, USA) was used to complete all statistical analyses described above. A p-value of less than 0.05 was considered to be significant, except in the cases that were Bonferroni corrected.
RESULTS
Table 2 shows the sample distribution of 75 infants analyzed. Among these infants, 30 did not have CP (40%) and 45 had spastic CP. Of the 45 infants with spastic CP, 16 had unilateral CP (8 with right unilateral CP, 8 with left) and GMFCS levels ranged from I to V (Table 2). In the General Movements Assessment, 43 of the infants (43 with CP, 0 without CP) had absent fidgety movements.
| Mean gestational age, weeks (SD) | Mean age at BabyOSCAR, weeks (SD) | Mean BabyOSCAR total score (SD) | BabyOSCAR total score range | Mean absolute asymmetry (SD) | n | |
|---|---|---|---|---|---|---|
| Participants without CP (n = 30) | ||||||
| Preterm (<36 weeks) | 32.6 (2.64) | 13.3 (1.11) | 26.9 (2.62) | 24–31 | 1.1 (1.53) | 15 |
| Term | 38.7 (1.28) | 12.2 (3.15) | 26.9 (2.08) | 23–30 | 1.1 (0.74) | 15 |
| Participants with left unilateral CP (n = 8) | ||||||
| GMFCS level I | 34.2 (6.79) | 11.5 (1.38) | 17.8 (3.43) | 14–22 | 1.8 (0.75) | 6 |
| GMFCS level II | 25.5 (2.12) | 13 (0) | 16.5 (2.12) | 15–18 | 9.5 (2.12) | 2 |
| Participants with right unilateral CP (n = 8) | ||||||
| GMFCS level I | 34.8 (6.38) | 12.2 (1.3) | 15.4 (4.28) | 9–21 | 8.2 (3.70) | 5 |
| GMFCS level II | 26.5 (3.54) | 11.5 (0.71) | 11 (4.24) | 8–14 | 7.0 (1.41) | 2 |
| GMFCS level IV | 24 (0) | 10 (0) | 13 (0) | 13 | 11.0 (0) | 1 |
| Participants with bilateral CP (n = 29) | ||||||
| GMFCS level I | 25.33 (2.52) | 13.0 (1.0) | 17.3 (1.53) | 16–19 | 1.3 (0.58) | 3 |
| GMFCS level II | 23 (0) | 13 (0) | 17 (0) | 17 | 1.0 (0) | 1 |
| GMFCS level III | 27 (3.39) | 12.2 (1.48) | 11.2 (8.35) | 6–26 | 0.8 (1.30) | 5 |
| GMFCS level IV | 26.4 (2.45) | 13.0 (1.94) | 6.22 (4.05) | 1–14 | 2.0 (1.41) | 9 |
| GMFCS level V | 29.6 (6.06) | 13 (1.34) | 3.5 (2.88) | 0–9 | 1.7 (1.27) | 11 |
- Abbreviations: BabyOSCAR, Baby Observational Selective Control AppRaisal; CP, cerebral palsy; GMFCS, Gross Motor Function Classification System; SD, standard deviation.
Convergent validity between BabyOSCAR score and GMFCS level at ≥2 years
There was a strong correlation between GMFCS level at 2 years of age or later and total score (ρ = −0.891, p < 0.001; Figure 2a), leg score (ρ = −0.842, p < 0.001; Figure 2b), and arm score (ρ = −0.798, p < 0.001; Figure 2c). The Kruskal–Wallis results indicate the main effect of GMFCS was significant for total score, leg score, and arm score, but none of the asymmetry scores (Table 3 and Figure 2). Post-hoc analysis revealed significant correlations between infants without CP and those with CP in each GMFCS level.
Boxplots of BabyOSCAR scores by GMFCS level. GMFCS was assessed at 2 or more years of age and compared with the BabyOSCAR score from a video taken at 10 to 16 weeks corrected age. There was a strong correlation between GMFCS level at 2 years of age or later and (a) total score (ρ = −0.891, p < 0.001), (b) leg score (ρ = −0.842, p < 0.001), and (c) arm score (ρ = −0.798, p < 0.001).
Abbreviations: BabyOSCAR, Baby Observational Selective Control AppRaisal; CP, cerebral palsy; GMFCS, Gross Motor Function Classification System.
| Main effect H (p for GMFCS) | Significant pairwise comparisons for GMFCS level (p) | Main effect H, for spastic CP diagnosis (p) | Significant pairwise comparisons for CP diagnosis (p) | Main effect H, for spastic CP distribution (p) | Significant pairwise comparisons for CP distribution (p) | |
|---|---|---|---|---|---|---|
| Total score | 62.54 (<0.001) |
V < no CP (<0.001) IV < no CP (<0.001) III < no CP (0.018) I < no CP (0.007) V < I (0.041) |
57.161 (<0.001) |
Bi < no CP (<0.001) uniL < no CP (0.038) uniR < no CP (0.003) |
56.776 (<0.001) |
Bi < no CP (<0.001) Uni < no CP (<0.001) Bi–Uni (0.071)a |
| Leg score | 56.671 (<0.001) |
V < no CP (<0.001) IV < no CP (<0.001) III < no CP (0.014) I < no CP (0.017) II–no CP (0.072)a |
56.262 (<0.001) |
Bi < uniL (0.033) Bi < no CP (<0.001) uniR < no CP (0.007) |
55.378 (<0.001) |
Bi < Uni (0.015) Bi < no CP (<0.001) Uni < no CP (0.002) |
| Arm score | 52.785 (<0.001) |
V < I (0.022) V < no CP (<0.001) IV < no CP (<0.001) III < no CP (0.040) I < no CP (0.046) |
43.951 (<0.001) |
Bi < no CP (<0.001) uniR < no CP (0.012) uniL < no CP (0.018) |
43.942 (<0.001) |
Bi < no CP (<0.001) Uni < no CP (<0.001) |
| Right leg | 51.230 (<0.001) |
IV < no CP (<0.001) V < no CP (<0.001) III < no CP (0.045) I < no CP (0.002) |
55.693 (<0.001) |
uniR < no CP (<0.001) Bi < uniL (<0.001) Bi < no CP (<0.001) |
48.616 (<0.001) |
Bi < no CP (<0.001) Uni < no CP (0.001) |
| Left leg | 51.840 (<0.001) |
V < I (0.014) V < no CP (<0.001) IV < no CP (<0.001) IV–I (0.064)a III < no CP (0.031) II < no CP (0.042) |
49.706 (<0.001) |
Bi < uniR (0.015) Bi < no CP (<0.001) |
49.203 (<0.001) |
Bi < Uni (0.003) Bi < no CP (<0.001) Uni < no CP (0.030) |
| Right arm | 45.019 (<0.001) |
V < no CP (<0.001) IV < no CP (<0.001) III–no CP (0.087)a I < no CP (<0.001) |
42.944 (<0.001) |
uniR < no CP (<0.001) Bi < no CP (<0.001) |
39.217 (<0.001) |
Bi < no CP (<0.001) Uni < no CP (<0.001) |
| Left arm | 43.378 (<0.001) |
V < I (0.005) V < no CP (<0.001) IV < no CP (0.002) III–no CP (0.074)a |
40.198 (<0.001) |
Bi < uniR (0.007) Bi < no CP (<0.001) uniL < no CP (0.007) |
35.931 (<0.001) |
Bi < Uni (0.035) Bi < no CP (<0.001) Uni < no CP (0.036) |
| Asymmetry | 20.435 (<0.001) | No CP < II (0.020) | 25.467 (<0.001) |
uniL–no CP (0.051)a no CP < uniR (<0.001) Bi < uniR (0.001) |
22.667 (<0.001) |
Bi < Uni (0.002) no CP < Uni (<0.001) |
| Leg asymmetry | 19.295 (0.002) |
III < I (0.009) III < II (0.036) |
18.492 (<0.001) |
Bi < uniR (<0.001) no CP < uniR (0.005) |
16.591 (<0.001) |
Bi < Uni (<0.001) no CP < Uni (0.004) |
| Arm asymmetry | 18.244 (0.003) |
III–II (0.064)a No CP < II (0.006) |
18.976 (<0.001) |
no CP < uniR (<0.001) Bi < uniR (0.011) no CP < uniL (0.079)a |
17.522 (<0.001) |
Bi < Uni (<0.001) no CP < Uni (0.007) |
- Main effect H and p-value for corrected pairwise comparisons.
- Abbreviations: BabyOSCAR, Baby Observational Selective Control AppRaisal; Bi, bilateral cerebral palsy; CP, cerebral palsy; GMFCS, Gross Motor Function Classification System; Uni, unilateral cerebral palsy; uniL, left unilateral cerebral palsy; uniR, right unilateral cerebral palsy.
- a Trend towards significant value.
Known-groups validity of BabyOSCAR score with CP diagnosis
Children with spastic CP had significantly fewer independent joint movements than children without a CP diagnosis, as measured by BabyOSCAR (total t = 7.39, leg t = 6.87, arm t = 5.04, all p < 0.001). There was also a significant main effect of CP diagnosis on all outcome measures investigated (Table 3 and Figure 3). Children with unilateral CP had significantly higher asymmetry scores than those without unilateral CP (bilateral and no CP), and children with right unilateral CP had significantly higher asymmetry scores than all other children tested. Additionally, all children who did not have unilateral CP (no CP and bilateral CP) had total asymmetry scores of less than 5.
(a) Main effect of limb distribution on BabyOSCAR total score. (b) Main effect of limb distribution on BabyOSCAR asymmetry score. Those with right unilateral CP had significantly greater asymmetry than those in the three other groups (p < 0.05).
Abbreviations: BabyOSCAR, Baby Observational Selective Control AppRaisal; CP, cerebral palsy.
Relationship between gestational age and filming age
There was no effect of preterm birth in the group without a CP diagnosis at 2 years of age in the BabyOSCAR scores (total p = 0.436, leg p = 0.806, arm p = 0.267). There was a relationship between gestational age and BabyOSCAR scores, where a younger gestational age was correlated with a lower BabyOSCAR score (total ρ = 0.492, p < 0.001; leg ρ = 0.563, p < 0.001; and arm ρ = 0.316, p = 0.006). There were no significant findings related to the age at the time of filming (all p > 0.05).
Interrater and test–retest reliability
Interrater reliability was excellent for total score (ICC = 0.88, 95% CI 0.74–0.94, p < 0.001) and leg score (ICC = 0.95, 95% CI 0.81–0.98, p < 0.001), and good for arm score (ICC = 0.54, 95% CI 0.04–0.78, p = 0.02). Test–retest reliability among the 12 reviewed cases was excellent for the total BabyOSCAR score (ICC = 0.87, 95% CI 0.55–0.96, p < 0.001) and leg score (ICC = 0.95, 95% CI 0.81–0.98, p < 0.001), and good for the arm score (ICC = 0.59, 95% CI −0.41 to 0.88, p = 0.07). There were no notable differences in ICC scores among children with and without CP.
DISCUSSION
We found a strong relationship between observations of individual joint movements at 10- to 16-weeks corrected age and (1) having a diagnosis of spastic CP and (2) the gross motor functional abilities of children with spastic CP at 2 years of age, demonstrating strong overall convergent and discriminant validity for the BabyOSCAR.
Although both GMFCS level and CP type are related to selective control, previous work has shown the clinical and scientific use of quantifying SMC specifically.13 When comparing groups of children with varying gross motor abilities and CP types, we consistently found that the BabyOSCAR arm, leg, and total scores were lower in children in all GMFCS levels and types of spastic CP than in those without CP. Furthermore, limb scores (right arm, right leg, left arm, left leg) also supported the type and topography of CP distribution, as the involved limbs often had significantly lower BabyOSCAR scores, showing early evidence of differences in these limbs.
Interestingly, infants in this study were not performing instructed or elicited movements. These self-generated ongoing movement patterns seem to demonstrate connectivity of the corticospinal tract to individual muscle motor units in a robust manner even before the onset of volitional, goal-directed movement required for most later tools to evaluate selective control.5, 14 The potential for this to serve as a non-invasive biomarker of corticospinal integrity presents significant advantages and augmentative information with tools such as magnetic resonance imaging, electroencephalography, electromyography, and others that have a higher cost and analysis burden than the 1-minute video recording used by BabyOSCAR. It further supports the clinical value of observing spontaneous movement during the fidgety period and the value of saving videos in an appropriate manner for retrospective review.
When we designed and validated this tool, our target population included children with spastic CP and excluded those with dyskinetic or mixed spastic and dyskinetic type CP; our rationale was that difficulties in selective voluntary motor control are a feature of individuals with spastic CP, probably a manifestation of injury to the corticospinal motor pathways. The corticospinal connections are responsible for discreet and fine motor control15 and are shown to demonstrate functional connectivity during the first few months of life.16, 17 Before this, subcortical and bulbospinal projections are likely to have a stronger influence on movement. These brainstem projections synapse across multiple levels of the spinal cord,15 resulting in grouped movements. When there is an injury to the brain involving cortical projections, it manifests in the character of movement overall.9 Our findings indicate that individual joint movements are also affected in a way that is strongly related to future gross motor capacity. Because our test was designed specifically to evaluate SMC in children with a high chance of developing spastic CP, it should not be used as a substitute for early detection of CP as we do not know how the test would perform in children with dyskinetic or mixed-type CP. We therefore advocate the use of currently recommended early detection tools that predict all types of CP in infants under 5 months or age (i.e. Prechtl General Movement Assessment,9 Hammersmith Infant Neurological Assessment18), including dyskinetic and mixed types of CP. On the other hand, the BabyOSCAR could be used after a child has received a General Movement Assessment indicating absent fidgety movements, a low Hammersmith Infant Neurological Assessment score, or has a history of a brain injury affecting the corticospinal pathways. In these children the BabyOSCAR may provide more information about CP distribution and joints that have less independent movement, with implications for potential targeted treatments that focus on those joints. For example, those with reduced selective control of the lower extremities may benefit from a targeted kicking intervention.19, 20
The BabyOSCAR scores revealed evidence of asymmetry as early as 10 weeks of age. Specifically, children with unilateral CP had significantly greater asymmetry scores than those without unilateral CP (no CP and bilateral CP), significantly different in children with right unilateral CP (p < 0.001), and trending towards significance in children with left unilateral CP (p = 0.051). The deviation in behavior that we measured with BabyOSCAR is consistent with previous studies that show structural connectivity changes between lesioned and non-lesioned hemispheres21 and behavioral motor changes that begin to deviate at 2 to 3 months of age in children with unilateral CP.22, 23 Interestingly, we found a difference between right and left unilateral CP where those with right-sided weakness had more robust asymmetry findings. Previous studies of laterality indicate that older children with left unilateral CP were better responders to constraint-induced movement therapy,24 had better communication skills,25, 26 and better motor planning27 than children with right unilateral CP. How the early deviation in asymmetry findings measured by BabyOSCAR at 3 months relates to these findings is still unknown and should be studied longitudinally and in a larger sample. In either case, an asymmetry score greater than 5 should be an indication for exploring evidence-based interventions such as constraint-induced movement therapy.28
The ability of outcome measures to improve decision-making in clinical research relies heavily on their psychometric strength. Our rating groups had good to excellent reliability in scoring the BabyOSCAR, although the arm scoring was found to have a lower interrater and test–retest reliability than leg scoring. Although not specifically measured, our rating teams reported more anecdotal difficulty in rating the arms, which may be related to having more joint motions and degrees of freedom to score. Our findings confirm that while 1 minute of active movement is sufficient to understand the infant's capacity to independently isolate joint motion throughout the body, it may be that a longer observation of arm movement may improve a total score. This can be more formally studied in future cohorts, along with evaluating the consistency of scores for the same infant on different days. Currently, the strong relationship found between a single minute of video and long-term abilities is encouraging and well suited to clinical implementation. We used a combination of individual raters and rating pairs, both of which were feasible and acceptable to experienced therapists and students.
Our study was limited in that we had a smaller number of children classified in GMFCS levels II and III than other GMFCS levels, making it harder to understand discriminant validity among individual GMFCS levels. Future directions will include studying a larger sample of children longitudinally and comparing the BabyOSCAR scores with current measures of selective voluntary motor control in older children such as the Selective Control Assessment of the Lower Extremity or Test of Arm Selective Control.
CONCLUSION
Early identification of children with CP can contribute to early CP-specific intervention to improve outcomes and prevent secondary problems associated with CP later in life.28 While existing tools can help to identify infants early,29 the BabyOSCAR can be used concurrently with these tools in the population of infants with the highest chance of developing spastic CP. We have demonstrated strong validity and reliability for BabyOSCAR as a tool for assessing SMC in young infants. A 1-minute video of an infant in supine is very accessible to collect in several ways, including parents recording at home and sharing using app platforms,30, 31 and can be collected concurrently with other detection tools such as the General Movement Assessment. Making observations with the BabyOSCAR framework in mind may aid in future prognosis and other clinical applications.32
ACKNOWLEDGEMENTS
We thank the children and their families for their participation. Research reported in this publication was supported, in part, by the National Institutes of Health's National Center for Advancing Translational Sciences, grant number UL1TR001422. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The first author was supported by the National Center for Advancing Translational Sciences, grant number KL2TR001424.
FUNDING INFORMATION
National Institutes of Health's National Center for Advancing Translational Sciences, grant number UL1TR001422; National Center for Advancing Translational Sciences, grant number KL2TR001424.
CONFLICT OF INTEREST STATEMENT
Authors AFB and CP are members of the General Movements Trust Speaker's Bureau.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
