Motor Optimality Scores are significantly lower in a population of high-risk infants than in infants born moderate-late preterm

Abstract

Background:

The Motor Optimality Score-Revised (MOS-R) is a detailed scoring of the General Movement Assessment (GMA), measuring the spontaneous behaviors of infants. Infants born moderate-late preterm are not traditionally followed in high-risk clinics, but have increased risk of neurodevelopmental disability.

Aims:

Compare MOS-R at 3 months corrected age (CA) in high-risk (HR; very preterm or abnormal neuroimaging) infants to infants born moderate-late preterm (MLP).

Study design:

In this prospective cohort study, parents of enrolled infants created video recordings using an app at 3 months CA. Videos were scored with the General Movement Assessment (GMA) and MOS-R. MOS-R scores were divided into “higher-risk” (≤19) and “lower-risk” (≥20).

Subjects:

181 infants born MLP or categorized as HR.

Results:

Among enrolled infants, 68 (38 %) were in the MLP group, and 113 infants were in the HR group. The HR group had 3.8 increased odds of having an aberrant GMA score compared to the MLP group (p < 0.01, 95 % CI 1.38–10.52). The HR group had significantly lower MOS-R scores (mean 20) than the MLP group (mean 24; p < 0.001; 95%CI 3.3–7.3). The HR group had 11.2 increased odds of having a higher-risk MOS-R score (95%CI 2.5–47.6, p < 0.001) than MLP group. Infants were most likely to have a lower MOS-R score if they had any of the following: VP shunt placement, periventricular leukomalacia, or bronchopulmonary dysplasia.

Conclusions:

Aberrant GMA and higher-risk MOS-R scores were more common in infants at high-risk, reflecting history of brain lesions and younger gestational age at birth.

1. Introduction

The incidence of adverse neurodevelopmental outcomes increases with each week of decreasing gestation in preterm birth [1]. Children who are born at earlier gestational ages and those born at term with brain lesions, carry the highest risk of adverse neurodevelopmental outcomes [1], [2] and are traditionally followed in high-risk (HR) follow-up clinics after leaving the neonatal intensive care unit (NICU). However, the majority of children born preterm are born moderate-late preterm (MLP; 32 to 36 weeks gestation) [3]. Infants born MLP also have increased risk of adverse neurodevelopmental outcome compared to those born full-term without brain lesion [4], but are not often included in HR developmental follow-up clinics. Consequently, infants born MLP receive less targeted developmental screening than those infants deemed HR.

A short-term neurodevelopmental assessment that can be done remotely, through parent-collected video recording of infants moving spontaneously, is the General Movement Assessment (GMA) [5]. The GMA is now more widely utilized in high-risk follow-up clinics at 3–5 months of corrected age [6] because it is the most sensitive clinical tool for the prediction of cerebral palsy [7]. The presence of “fidgety movements” (FM), defined as small, multiplanar movements of the hips, knees, wrists, shoulders and neck, (which are spontaneously generated in the awake and calm infant) [8] is a reassuring developmental pattern, while the absence of FM is 95 % sensitive for the diagnosis of cerebral palsy(CP) [9]. The Motor Optimality Score-Revised (MOS-R) [8], a revision of the original Motor Optimality Score [5], classifies FM and, additionally, systematically assesses other behaviors in the infant’s motor repertoire including, posture, movement patterns, and movement quality [8]. The MOS-R can be scored from the same video recording as the GMA and has been associated with functional outcomes in CP [8] and other non-CP neurodevelopmental outcomes including cognitive, language, and non-CP motor delays in toddlerhood [10], [11] and school-age children [12], [13].

The use of a smartphone app allows parents to film an infant’s spontaneous movements with a short video and has been used successfully to collect GMA and MOS-R data in a large sample of infants born extremely low birth weight (ELBW) [14]. This parent-collected data was used to establish a relationship between MOS-R scores and adverse neurodevelopmental outcomes at two years of age [10], [11]. Parent-recorded data using a smartphone app provides parents an alternative to traditional in-person developmental screening and may be useful in a population of MLP infants who are not routinely followed. The purpose of our study was to determine if the MOS-R scores in HR infants were different from infants who were born MLP, using parent-collected video recordings.

2. Material and methods

2.1. Participants

Infants were recruited prospectively from four neonatal intensive care units (NICU) in Chicago between August 2019 and May 2022. Recruited infants required either a NICU hospitalization for MLP birth (32–36 weeks gestation), for HR of cerebral palsy including very preterm birth (<32 weeks) or other neurological risk factors (abnormal head ultrasound or MRI findings). Demographic characteristics and medical histories were extracted from the medical records for each participant. Infants were then stratified by group into the MLP group (born with no other risk factors), or into the HR group of infants who were either <32 weeks gestational age or had an abnormal head ultrasound or an abnormal brain MRI. Informed parental consent was obtained for each infant, and ethics approval for the study was granted by the institutional review board at each participating university hospital.

2.2. Data collection

Parents of enrolled infants created GMA video recordings using the Baby Moves smartphone application (app) [15] between 12- and 16-weeks corrected age. After enrolling in the study, the Baby Moves app notified participating families to take a video of the infant at 12- and 14-weeks corrected age. The app provided families with detailed instructions to record and upload a 3-minute video of the child while actively moving. Infants were filmed in a quiet, alert state (without a pacifier) in the supine position. Videos were then uploaded directly to a research database (REDCap 11.1.21; Vanderbilt University) to be scored by assessors certified in the Prechtl General Movement Assessment.

2.3. General Movement Assessment

A team of trained assessors with advanced certification in the GMA, including training in the use of MOS-R, assessed infants between 12 and 16 weeks (and 6 days) CGA, according to Prechtl methodology [5]. As we have done previously [16], we classified FMs as either normal or aberrant. FMs were classified as aberrant if abnormal (exaggerated with respect to speed and amplitude) [17], sporadic (lasting ≤3 s) [18], or absent. At least two raters scored each video together. If a disagreement in scoring occurred between the two raters, a third rater served as a tiebreaker. Raters were blinded to the infant’s clinical histories at time of assessment.

2.4. Motor Optimality Score for 3-to 5-month-old infants – Revised and Motor Repertoire Score

We next scored the videos with the MOS-R [8]. In addition to scoring FM, the MOS-R assesses movement behavior and assigns points in the following four domains: 1) ratio of normal to atypical movement patterns (observed movement patterns), 2) number of pre-specified age adequate movement patterns (age-adequate motor repertoire), 3) ratio of normal to atypical body, neck, and finger postures (observed postural patterns), and 4) whether movements are normally smooth and fluent or have one or more atypical and well-described characteristics, including jerky, monotonous, tremulous, or stiff movements (movement character). The sum of these scores ranges between 5 and 28 [8]. There are four categories of optimality derived from the MOS-R score: optimal (25–28), mildly reduced (20–24), moderately reduced (9–19), and severely reduced (5–8) [8]. We created a binary score for the MOS-R called “higher-risk” (score 5–19) or “lower-risk” (score 20–28). If families submitted more than one video for review, the video with the higher MOS-R score was used for analysis. To understand how infant characteristics were related to the group of non-FM behaviors captured in the MOS-R, we created a Motor Repertoire score by subtracting the fidgety movement score (up to 12 points) from the MOS-R, as we have done previously [19], where the score ranged from 4 to 16.

2.5. Statistical analyses

Statistical analyses were conducted using RStudio, version 1.4.1717 (2009–2021 RStudio, PBC). To characterize the sample, descriptive statistics were used to portray medical and demographic characteristics of infants whose families did and did not submit video data. We first tested for medical and demographic differences among children who did and did not submit video data. Next, we tested for differences in GMA, MOS-R, and Motor Repertoire Scores between HR and MLP groups. We compared mean differences using the t-test, or Wilcoxen rank sum test when data were skewed. Differences among categorical groups were compared with Fisher’s exact test and Chi-Squared tests. When Chi-Squared tests were significant for differences among groups (race, Table 1; GMA and MOS-R categories, Table 2), a post-hoc analysis was performed examining residuals and employing the Bonferroni correction for multiple comparisons. Finally, to determine if there were associations between either the MOS-R and Motor Repertoire Scores and infant medical and/or demographic characteristics of the sample, we used two multivariate linear regression models. Univariate linear regression models were first used to determine if medical and/or demographic characteristics were significantly associated with MOS-R or Motor Repertoire Scores. Characteristics which were significantly correlated were then entered into a multivariate linear regression model. To determine if characteristics in the models were collinear, we performed the Variance Inflation Factor test for each model. A best subsets regression was then performed for each model to determine which model had the best goodness of fit. A p value of <0.05 was considered statistically significant.

Table 1.

Medical and demographic characteristics among 244 infants enrolled in the study.

	Did not send videos	Sent videos	p-Value
	(n = 62)	(n = 182)
English is family’s primary language, n (%)	57 (95)	153 (85)	0.14
Maternal level of education, n (%)			0.27
High school graduate or less	13 (21)	26 (14)
Some college or associate’s degree	13 (21)	30 (17)
Bachelor’s or advanced degree	20 (32)	84 (46)
Unknown	15 (24)	42 (23)
Race/ethnicity, n (%)			<0.02a
Asian	0	3 (2)
Black	26 (40)	66 (36)
Hispanic	7 (11)	25 (14)
White	13 (21)	74 (41)
Multiracial	3 (5)	5 (2)
Unknown	12 (20)	9 (5)b
Male, n (%)	36 (58)	86 (47)	0.11
Singleton gestation, n (%)	42 (68)	157 (86)	<0.01a
IVH, n (%)	8 (13)	31 (17)	0.45
Ventriculoperitoneal shunt, n (%)	1 (17)	7 (4)	0.34
Periventricular leukomalacia, n (%)	1 (17)	6 (3)	0.47
Bronchopulmonary dysplasia (on oxygen support >36 wk), n (%)	13 (21)	56 (31)	0.14
Necrotizing enterocolitis (with laparotomy or drain) n (%)	2 (3)	15 (8)	0.25
Treated retinopathy of prematurity (injections or laser), n (%)	1 (2)	14 (8)	0.01a
Birthweight, mean ± SD (grams)	1655 ± 809	1652 ± 942	0.98
Birthweight range (grams)	560–3620	430–4460
Gestational age, mean ± SD (weeks)	30 ± 4	30 ± 5	0.58
Gestational age range (weeks)	23–40	22–39

IVH: Interventricular hemorrhage; unknown: Parents chose “prefer not to say” or data on race was unavailable.

^aSignificant finding.

^bSignificant finding on pos-hoc analysis.

Table 2.

General Movement Assessment, Motor Optimality Score Revised and Motor Repertoire Scores among 182 infants, categorized by risk.

	Moderate-late preterm group	High-risk group	P value
General Movement Assessment (GMA)
GMA category
Empty Cell	n = 68	n = 114	0.02a
Normal, n (%)	64 (94)	87 (77)b
Sporadic, n (%)	3 (4)	11 (9)
Abnormal, n (%)	0 (0)	2 (2)
Absent, n (%)	1 (1)	14 (11)
Aberrant fidgety, n (%)	4 (6)	27 (24)	<0.01a
Motor Optimality Score-Revised (MOS-R)
	N = 66	N = 112
MOS-R Score, mean ± SD	24 ± 3	20 ± 7	<0.001a
MOS-R Score, median (range, IQR)	24 (12–28, 24–26)	23 (4–28, 19–24)
Optimal MOS-R (25–28), n (%)	28 (42)	18 (16)b	<0.001a
Mildly reduced MOS-R (20–24), n (%)	36 (55)	65 (59)b
Moderately reduced MOS-R (9–19), n (%)	2 (3)	19 (16)
Severely reduced (5–8), n (%)	0 (0)	10 (9)
Higher-risk MOS-R score (5–19), n (%)	2 (3)	29 (25)	<0.001a
Motor Repertoire Score
Motor Repertoire score, mean ± SD	13 ± 2	10 ± 3	<0.001a
Motor Repertoire score, median (range, IQR)	12 (8–16, 12–14)	11 (3–16, 9–12)

Moderate-late preterm infants: born 32–36 weeks gestation with no risk factors; High-risk group: infants born <32 weeks gestational age and/or with abnormal head ultrasound or brain MRI; Aberrant fidgety: combined categories of sporadic, abnormal, and absent, MOS-R: Motor Optimality Score-Revised.

^aSignificant finding.

^bSignificant finding on pos-hoc analysis.

3. Results

3.1. Participants

A total of 243 infants were enrolled in the study, however, 62 families of enrolled infants (26 %) did not submit videos due to attrition or technical difficulties with the Baby Moves app. The infants whose families did not submit videos were more likely to have race categorized as unknown (p < 0.02), were more often from twin gestations (p < 0.01), and were less likely to have treated retinopathy of prematurity (p = 0.01) compared to infants whose videos were submitted (Table 1). There were no other significant differences in medical and demographic characteristics among infants whose families did and did not submit videos for analysis (Table 1).

3.2. General Movement Assessment

Among the 181 infants whose videos were submitted for analysis, 68 infants (38 %) were categorized as MLP. Infants in the MLP group were more likely to have normal fidgety movements than infants in the HR group (Table 2). Among the 68 infants in the MLP group, one infant (1 %) had absent fidgety movements as compared to the HR group where 13/113 (11 %) had absent fidgety movement. Infants in the HR group had 4.78 increased odds (95 % CI, 1.59–14.29; p = 0.003) of having aberrant fidgety movements as compared to those in the MLP group.

3.3. Motor Optimality Score-Revised and Motor Repertoire Scores

Four infants whose videos were analyzed with the GMA (2 %) were unable to maintain a calm state for the required 2–3 min needed to score the MOS-R (and derived Motor Repertoire Score). Accordingly, 178 infant videos were scored with the MOS-R and the Motor Repertoire Score. Of the infants scored with MOS-R, 66 infants (37 %) were categorized as MLP. Infants in the MLP group had significantly higher scores on the MOS-R and Motor Repertoire Scores (p < 0.001, Table 2) as compared with the infants in the HR group. Infants in the high-risk group were less likely to have an optimal MOS-R score and had a higher proportion of infants who had a mildly reduced MOS-R score compared to infants in the MLP group (p < 0.001, Table 2). In fact, infants in the HR group had 11.2 increased odds of having a high-risk MOS-R (score ≤19; Table 2) than infants in the MLP group (95 % CI, 2.6–48.6; p < 0.001). Interestingly, the majority of infants in the MLP group (55 %) had a mildly reduced MOS-R score (Table 2).

3.4. Predictors of Motor Optimality Score-Revised and Motor Repertoire Scores

To understand which infant characteristics predicted the Motor Optimality Score-Revised, we ran a series of univariate models; the following characteristics were significantly associated with the MOS-R score: gestational age, history of periventricular leukomalacia (PVL), history of ventriculoperitoneal (VP) shunt placement, history of bronchopulmonary dysplasia (BPD) and history of treated retinopathy of prematurity (ROP). After entering these characteristics in a multivariate model, gestational age and treated ROP were no longer significantly associated with MOS-R score. There were several factors that could independently account for a decrease in the 28-point MOS-R score (Table 3). They included having a history of VP shunt placement (10 points, p < 0.001), a history of PVL (6 points, p = 0.007), and history of BPD (3 points, p = 0.020). In addition, children in our study had 11.0 increased odds of having aberrant FMs if they had a history of PVL (95%CI 1.9–63.3, p = 0.03).

Table 3.

Independent predictors of MOS-R and Motor Repertoire scores among 178 infants.

MOS-R (maximum score of 28 points)
Predictor	Score change	Standard error	P value
Gestational age (per week)	0	0.1	0.74
History of PVL	−6	2	0.007a
History of ventriculoperitoneal shunt placement	−10	2	<0.001a
History of BPD	−3	1	0.02a
History of treated ROP	−2	1	0.16
Motor Repertoire score (maximum score of 16 points)
Gestational age (per week)	0.1	0.1	0.09
History of BPD	−2	0.78	0.04a
History of VP shunt placement	−4	1	0.006a

^aSignificant finding; PVL: periventricular leukomalacia; BPD: bronchopulmonary dysplasia; ROP: retinopathy of prematurity; VP shunt: ventriculoperitoneal shunt.

In the Motor Repertoire Score model, the following three predictors were significantly associated with Motor Repertoire Score in univariate analysis: gestational age, history of BPD, history of VP shunt placement. As in our previous model, gestational age was no longer significant in a multivariate model, however, having a history of BPD was independently associated with a two-point (out of 16 possible) decrease in Motor Repertoire Score. Infants with a history of VP shunt placement had a four-point decrease in Motor Repertoire Score after accounting for gestational age and history of BPD (Table 3).

4. Discussion

In this prospective cohort study, we found that the MOS-R, scored from parent-recorded videos, were lower in children born very preterm and/or with abnormal neuroimaging findings, as compared to children born MLP (with no other risk factors). Children from the HR group were more likely to have aberrant fidgety movements and lower scores on the additional items on the MOS-R, which we have named Motor Repertoire Score. Histories of VP shunt placement, PVL, and BPD were factors responsible for lowered MOS-R scores and histories of VP shunt placement and/or BPD were factors responsible for reduced Motor Repertoire Scores.

While the assessment of FM using the GMA is highly sensitive to the development of CP [7], the original MOS and MOS-R are often more predictive of other non-CP neurodevelopmental adversity [10], [11], [12], [20], [21]. Previous studies have described MOS-R scores in populations of infants born VLBW [22], ELBW [10], [13], [14], or at term with hyperbilirubinemia [23], but to the best of our knowledge, MOS-R scores in a cohort of infants born MLP have not previously been reported. The mean MOS-R score of 24 seen in our infants born MLP, was the same as a large cohort of healthy term born infants [10] and 2 points lower than another cohort of term-age infants [12]. The percentage of infants with aberrant FM was higher in our group of infants born MLP (6 %, n = 4/68) than those in the full-term cohorts (2.2 %, n = 4/180 [12]; 2.7 %, n = 5/184 [10]), and may reflect future increased incidence of neurodevelopmental delays seen in infants born MLP compared to those born full-term [4]. Furthermore, over half of our sample of infants born MLP (55 %) had mildly reduced MOS-R scores (20–24), which may also indicate an increased chance of experiencing future neurodevelopmental challenges. Long-term follow-up is needed to understand how the MOS-R score at 3–5 months is related to outcome in toddlerhood and school age in populations of children born MLP.

The MOS-R scores in our HR group was significantly lower than the group of infants born MLP. While we have not yet described the MOS-R scores in relationship to later neurodevelopmental outcomes of the infants, the range of scores in our sample likely presages the future performance, stratified by risk [1], [24]. In fact, higher MOS-R scores have been associated with improved cognitive, language, and motor outcomes in toddlers [11] and school-age children [12], [13] with a history of very low or extremely low birth weight.

Previously we found that the Motor Repertoire Score, derived by removing the FM score from the MOS, was able to better account for the differences seen in long-term cognitive, language and motor outcomes at age two [19], than the presence of FM alone. This finding was recently replicated for two-year motor and cognitive outcomes in a cohort of infants born extremely preterm [21]. In our current study we found that the Motor Repertoire Score was also significantly lower in our high-risk population as compared to those infants born MLP and may suggest that these non-FM behaviors reflect the higher burden of non-CP neurodevelopmental adversity seen in high-risk populations compared to infants born MLP [4].

Additional medical complexity and required procedures early in life have been confirmed in our study to have an impact on movement and, likely, long-term outcomes. Children with history of a VP shunt placement had lowered scores on both the MOS-R and the Motor Repertoire Score. Indeed, infants with a history of VP shunt placement have both elevated risk of cerebral palsy and other non-CP neurodevelopmental adversity [25] as compared to infants who were born at term without complications, suggesting that the requirement of VP shunt placement affects early motor behavior and long-term neurodevelopmental outcome.

Unsurprisingly, children with history of PVL had lower MOS-R scores, by approximately 7 points, and were more likely to have aberrant FMs. Both aberrant FMs and history of PVL have been linked to the development of CP [7], [26]. School-aged children with PVL and motor impairment had injury in the descending corticospinal motor pathways, decreased functional connectivity among brain regions and reduced cortical volumes (as measured by quantitative MRI [27]). Our findings suggest that early motor behavior in infancy was also affected by the presence of PVL and may reflect the onset of neurologic structural and functional differences seen later in childhood. However, the overall number of children in our study with VP shunt placement (n = 7) and a history of PVL (n = 6) are low, representing only a small percentage of our overall sample.

Having a history of BPD was also associated with lower performance on both the MOS-R and the Motor Repertoire Score. In a separate prospective cohort of very preterm infants, we previously found an association with a history of BPD and lower Motor Repertoire Score [19], again suggesting that the presence of BPD may inhibit early motor behavior and predict atypical neurodevelopment. In fact, BPD has been independently associated with increased risk of the development of cerebral palsy, developmental delays [28] (including motor, language, cognitive and executive functioning abilities [29]) throughout childhood, and poor academic achievement [30]. The underlying hypoxia and chronic inflammation associated with BPD provides a mechanistic explanation for delayed brain development and subsequent neurodevelopmental impairment [31] and may also be reflected in decreased performance on the MOS-R and Motor Repertoire Score seen here in early infancy.

Our study was strengthened by the diverse sample which varied by race, maternal level of education, and geographic location within a large urban metropolis. The use of the Baby Moves smartphone app allowed us to collect high-quality data from many different neighborhoods and towns in the Chicago region. Parent-recorded videos were easily scorable and the high-quality instructions provided in the app, allowed parents to obtain standardized data. Most of our data collection occurred during the COVID-19 pandemic during a period of time where travel or visits to medical centers were often limited. The use of the Baby Moves app provided families with an alternative to in-person assessment. While our study is limited to the initial motor behaviors at 3 months, others have found that use of parent-recorded videos was feasible [14] and performance of motor behavior at 3 months was related to two year neurodevelopmental outcomes in high-risk infants born ELBW [10], [11].

Our study was limited when technical difficulties arose as we established the initial use of the app in the United States and, as a result of these difficulties, some families were unable to participate. While those who did participate were similar in most demographics, we were less likely to know the race of those who did not send videos and had more families send videos of children from singleton births than those of twin gestation. Our dataset was also limited to the 3-month time period; therefore, we are not able to draw conclusions about how the early behavior seen in our cohort relates to later outcome. We plan to continue tracking the developmental trajectories of this cohort to learn how the GMA, MOS-R and Motor Repertoire Scores relate to long-term neurodevelopmental outcomes.

While specific neurodevelopmental outcomes of our cohort have not yet been established, the overall MOS-R scores appear to mirror the well-established trajectory of risk-stratification in at-risk infants. Because the parent-collected data was feasible in our diverse population and was convenient for families, it might be useful as an alternative developmental screening for infants born MLP. Providing an expedient screening for families who otherwise do not receive it may be beneficial for infants at lower risk of developmental delays.

5. Conclusion

MOS-R scores were significantly lower in a high-risk cohort of infants when compared to a lower risk cohort of infants born MLP. Infants at high risk had increased likelihood of having aberrant FM and also lower scores on the remaining behaviors captured in the MOS-R (which we defined as Motor Repertoire Score). Understanding the outcomes of this population and MOS-R cutoff scores in future work will help us to determine clinically meaningful criteria for targeted intervention. Because over half of our infants born MLP experienced mildly reduced MOS-R scores, screening of infants born MLP may be warranted. Long-term follow-up is required to understand the correlates of early spontaneous motor behavior to the developmental trajectories of infants with various risk factors.

Abbreviations:

MOS-R

Motor Optimality Score Revised

GMA

General Movement Assessment

FM

fidgety movements

HR

high-risk

MLP

moderate-late preterm