Is it possible to predict the infant's neurodevelopmental outcome at 14 months of age by means of a single preterm assessment of General Movements?

BACKGROUND: It continues to be a challenge for clinicians to identify preterm infants likely to experience subsequent neurodevelopmental deficits. The Test of Infant Motor Performance (TIMP) and the assessment of spontaneous general movements (GMs) are the only reliable diagnostic and predictive tools for the functionality of the developing nervous system, if applied before term. AIM: To determine to what extent singular preterm assessments of motor performance can predict the neurodevelopmental outcome in 14-month olds.
METHODS: Thirty-seven preterm infants born <34weeks gestational age were recruited for the study at the NICU of the São Lucas University Hospital, Porto Alegre, RS, Brazil. At 34weeks, their GMs were assessed; and the Test of Infant Motor Performance (TIMP) was applied. A prospective design was used to examine (A) the association between the GM assessment and the TIMP; and (B) the relation between GMs or the TIMP and the developmental status at 14months, assessed by means of Alberta Infant Motor Scales (AIMS) and the Pediatric Evaluation of Disability Inventory (PEDI).
RESULTS: Nineteen infants (41%) had abnormal GMs; only one scored within the TIMP average range. Hence, GMs and TIMP were not related. Children with cramped-synchronized GMs at 34weeks preterm had a lower AIMS centile rank than those with poor repertoire or normal GMs. There was a marginal association between cramped-synchronized GMs and a lower PEDI mobility score.
CONCLUSIONS: A single preterm GM assessment is only fairly to moderately associated with the 14-month motor development. The TIMP is not suitable as a complementary assessment tool at such a young age.

Introduction

Preterm infants are at a greater risk for developmental deficits than their term peers [1-3]. Thanks to advanced care and technology, the number of infants who survive in spite of a very low birth weight keeps increasing, but along with this trend developmental morbidity has gone up [4,5]. Motor impairments as well as severe visual and hearing impairments are clearly associated with preterm birth; they are found in one out of three infants with a birth weight that is extremely low [5]. There is increasing evidence that the most common area of maldevelopment is the cognitive domain [6,7]. Compared to classmates, 40% of 6-year-old children born with an extremely low birth weight have moderately to severely decreased IQs, and a further 30% have mild cognitive impairments [8]. Preterm infants are also at greater risk for long-term deficits in their attentional, perceptual-motor, and visuo-spatial abilities [1,9].

Early identification of infants who are likely to experience neurodevelopmental deficits continues to be a challenge for clinicians. Only few methods can be applied for a functional assessment of a preterm's nervous system; one is to assess the quality of the so-called General Movements (GMs) [10]. Its non-intrusiveness makes this method an important diagnostic tool in the assessment of fragile, physiologically unstable preterm infants.

During the last 15 years considerable research has been devoted to GMs, whose quality has proved to be most indicative of the functional integrity of the young nervous system, with a specificity of 82–99%, a sensitivity of 95–100%, a negative likelihood ratio of 0.05, and a positive likelihood ratio > 51 [11]. In its predictive power, the GM assessment is superior to cranial ultrasound [12], or neurological examination [13], and equivalent to MRI (white matter assessment) [14].

Several studies found that abnormal GMs from preterm age onwards up to 3–4 months post term were associated not only with a high risk for cerebral palsy [12], genetic disorders [15], minor neurological deficits [16-18], and behavioral problems [19], but also lower intelligence [7,20].

Focusing on the preterm age alone, the sensitivity of consistently abnormal GMs reaches 98–100%, whereas specificity is low, as a large number of preterm infants normalize by term or shortly afterward [11,21].

In a review of the assessment methods currently used in Brazil for an early identification of developmental problems, Santos and colleagues recommended the assessment of GMs and the Test of Infant Motor Performance (TIMP) due to their high sensitivities and specificities [22]. Recently, a systematic review on neuromotor assessments of preterm infants also revealed that the TIMP and the assessment of GMs are the only tools appropriate for use before term [23]. Both methods focus on spontaneous movements. The GM assessment is solely based on global pattern recognition of qualitatively different motor patterns [24], whereas the TIMP concentrates on the organization of posture in response to demands for movements placed on infants by caregivers in naturalistic interactions [25]. The sensitivity of the TIMP is between 45% and 92%, whereas the specificity reaches 68–78% [26].

To our knowledge, both assessments are rarely applied in Brazil [22,27]. So far, only the study by Garcia and associates from São Paulo confirmed that normal preterm GMs were associated with a normal neurological outcome, whereas abnormal GMs were only predictive when assessed at an age of 3 to 4 months post term [27].

The assessment of normal or abnormal GMs, which is a matter of minutes for an experienced observer, is usually done off-line from video [24]. Producing the very assessment tape, however, can be rather time-consuming, for the mere recording of GMs depends on the behavioral state of the infant; 30- to 60-minute recordings have to be previewed so as to select some GMs for further analysis; the montage of a final assessment tape requires technical experience [28]. Not every neonatal unit can afford to carry out the serial videotaped observations of GMs, required by the standards of the method. As for ourselves, due to internal technical and organizational restrictions we were able to apply the GM assessment only once during the later preterm period.

The aims of our study were (1) to compare the GM assessment with the applied TIMP, which was carried out on the same day; (2) to analyze to what extent the quality of the GMs and the TIMP scores were associated with the motor and social performances around one year of age; and (3) to analyze to what extent the GMs and the TIMP results were related to the infants' neonatal characteristics.

Methods

The study was conducted at the Neonatal Intensive Care Unit of the São Lucas University Hospital, which belongs to the Catholic University of Rio Grande do Sul (PUCRS), and at the Ceneffi Clinic in Porto Alegre, RS, Brazil. The study protocol had been approved beforehand by the Ethics Committee of the PUCRG; an informed consent was signed by the parents.

Participants

Between June 2006 and April 2007, 169 preterm infants born at 33 weeks or younger were admitted to the Neonatal Intensive Care Unit of the Hospital São Lucas, PUCRG. From this population, a convenience sample of 37 infants (17 boys and 20 girls) was recruited. Apart from the need for timely notification of the principal investigator (S.A.M., who is not a member of the hospital staff), the inclusion criterion was the authorization of the medical staff. None of the infants had a diagnosis of metabolic disorder, congenital, chromosomal or cardiac abnormality; and none of the infants received sedative drugs at the age of assessment.

The infants' gestational ages ranged from 24 to 33 weeks (median = 31 weeks); their birth weights ranged from 640 to 2050 g (median = 1455 g). Twenty-one infants had a very low birth weight (VLBW; <1500 g); six infants had an extremely low birth weight (ELBW; <1000 g). Two infants were small for their gestational ages, and two infants had bronchopulmonary dysplasia (oxygen > 28 days). Table 1 shows further characteristics of the study group.

Table 1

Clinical characteristics and risk factors (median and inter-quartiles) according to the quality of GMs at 34 weeks postmenstrual age.

	Normal GMs n = 18 (49%)	Poor repertoire GMs n = 12 (32%)	Cramped-synchronized GMs n = 7 (19%)	p-Value < 0.05
Boys	6 (33%)	8 (67%)	3 (43%)	n.s.
Gestational age (weeks)	32 (29–33)	30.5 (28–32)	29 (27–31)	n.s.
Birth weight (grams)	1637.50 (1149–1924)	1397.50 (977.50–1654)	1190 (960–1470)	n.s.
Apgar score at 1 min	8 (7–8)	8 (7–9)	5 (3–8)	p < 0.05
Apgar score at 5 min	9 (8–9)	9 (8–9)	8 (7–9)	n.s.
Ventilated (number of days)	0 (0–2)	1 (0–6)	1 (0–10)	n.s.
Abnormal brain ultrasound	1 (6%)	2 (17%)	3 (43%)	n.s.
	IVH grade 1	PVL grade	IVH grade 1
		3 IVH grade 1	IVH grade 1
			1 IVH grade 1

PVL = periventricular leucomalacia, graded according to de Vries et al. [33].

IVH = intraventricular hemorrhage graded according to Volpe [34].

The assessment of motor behavior at preterm age

Recording and analysis of GMs

One-hour digital video recordings were made of the spontaneous motility of each infant at the postmenstrual age of 34 completed weeks (with a tolerance of + 4 days). In the incubator the infants, who wore diapers, lay in supine position. They were either in active wakefulness or in active sleep. Whenever they fussed or cried for a longer period the recording was interrupted for a few hours (or even up to 1–2 days). The videos were edited according to the standards of GM assessment [28]. Five-minute video clips with at least three GMs were analyzed by two trained observers (S.A.M., C.E.), the latter of who – an expert in GM assessment – was not familiar with the infants' clinical histories and the results of other assessments. When the evaluators disagreed on particular results of the GM assessment, they re-evaluated the recording of the infant in question. In all cases consensus was reached on a final score.

Normally, the GMs of a preterm or term infant comprise the entire body and manifest themselves in a variable sequence of arm, leg, neck and trunk movements. They appear and cease gradually, varying in intensity and speed. Rotations and frequent slight variations of the direction of motion make them look complex but smooth [10,24].

GMs are categorized as normal or abnormal. Abnormal GMs are classified into (1) “poor repertoire GMs”, whereby the sequence of movement components is monotonous; the amplitude, speed, and intensity lack the normal variability; (2) “cramped-synchronized GMs”, which appear rigid as they lack the usual smoothness and fluent character; the limb and trunk muscles contract almost simultaneously and relax almost simultaneously; and (3) “chaotic GMs”, which appear jerky and abrupt due to their large amplitude and high speed [24].

TIMP

The TIMP was applied by the principal investigator (S.A.M., who had been trained in the matter) on the same day of the GM recording. The TIMP is a functional motor scale that can be used for infants between 34 weeks postmenstrual age and 4 months postterm age [25]. It measures spontaneous movements as well as infants' motor responses relevant to their daily environmental interaction such as head orientation, body alignment, antigravity leg movements, and responses to auditory and visual stimulation. Each pattern has its own scale, the score varying from 1 to 6. The pattern scores sum up to a total raw score, yielding the categories such as “average” (+ 1 to − 0.5 standard deviation (SD) from age mean), “low average” (− 0.5 to − 1 SD below age mean), “below average” (− 1 to − 2 SD below age mean), and “far below average” (>−2 SD below age mean). The normative sample was 990 infants from USA at risk of poor neurological outcome.

Motor performance and social function early in the second year of life

Five children (3 boys and 2 girls) could not be traced and were lost for follow-up. Thirty-two children (14 boys and 18 girls) participated in the outcome assessment at a median age (corrected for preterm birth) of 14 months (inter-quartiles = 13–16 months; range = 12–18 months). We applied the Alberta Infant Motor Scale (AIMS) and the Pediatric Evaluation of Disability Inventory (PEDI).

AIMS

The AIMS is an observational scale devised to measure gross motor development as related to weight-bearing, posture and antigravity movement in infants up to the stage of independent walking [29]. A total raw score was calculated and converted into an age-related centile rank, based on a sample of more than 2200 children from Alberta, Canada. A score equal or below the fifth centile classifies the motor development of the child as suspicious/abnormal.

PEDI

The PEDI is a tool to document the development and capability of functional skills in children in terms of mobility, self care, and social function [30]. A normative score < 30 indicates developmental delay.

The PEDI was validated for Brazil [31]; the AIMS was adapted according to the specific cultural parameters [22]. Apart from the normative values for North America, we also used the raw scores for statistical computation. AIMS and PEDI were performed by the principal investigator (S.A.M.), trained in the standardized administration and scoring of both assessments.

Statistics

Statistical analysis was performed by means of an SPSS package for Windows, version 16.0. (SPSS Inc, Chicago, IL).

In order to compare the medians of non-normally distributed continuous data, like e.g., gestational age, birth weight or the scores of the outcome assessments in relation to the GMs or TIMP, we applied the Kruskal–Wallis test. In order to assess whether independent samples had yielded equal values, we applied the Mann–Whitney U test (e.g. boys' vs. girls' test scores); to assess the relative strength of the association between variables, we computed the following correlation coefficients: Cramer's V coefficient was applied when at least one out of two variables was nominal (e.g. GMs and AIMS categories); Kendall's Tau-c coefficient was used when both variables were ordinal (e.g. GMs and TIMP categories). To assess the relationships between two continuous test scores, we applied the Pearson's product moment correlation (e.g. AIMS centile rank and PEDI). We used the Spearman rank order correlation when one score was ordinal and the second score was continuous (e.g. GMs and PEDI). The relative strength of the association between variables was evaluated along Portney's guidelines: values between 0.25 and 0.50 showed an association of fair strength; a value of 0.50 to 0.75 indicate a moderate association; values above 0.75 showed a very good to excellent association [32].

Throughout the analyses p < 0.05 (two-tailed) was considered to be statistically significant.

Results

Eighteen infants (49%) had normal GMs; twelve infants (32%) scored with a poor repertoire of GMs; and seven infants (19%) showed cramped-synchronized GMs (Table 1). No chaotic GMs were observed. Although five children were lost for follow-up, the GM patterns were similarly distributed: 47% normal; 37.5% poor repertoire GMs; and 15% cramped-synchronized GMs. The distribution of postnatal age was the same across all categories of GMs (Kruskal–Wallis test, n.s.).

The association between GMs and TIMP

Only one infant scored within the average (although low average) range of the TIMP and had normal GMs. Eleven infants scored below average, 24 infants far below average (Table 2). The results of the TIMP were not related to the GM quality (Kendall's tau-c = 0.13, n.s.). The correlation between GMs and the TIMP raw score was 0.29 (Spearman rank order correlation; n.s.).

Table 2

The quality of GMs in relation to the TIMP at 34 weeks postmenstrual age.

TIMP	Normal GMsa	Poor repertoire of GMs	Cramped-synchronized GMs	p-Value
Low average	1	0	0
Below average	7	1	3
Far below average	9	11	4	n.s.

TIMP was not applied in case of one infant due to medical instability.

Motor performance at preterm age and motor and social functions at 14 months

Eight children scored below or at the 5th AIMS centile, which is a suspect/abnormal value. Six children scored below a PEDI normative score of 30, which indicates developmental delay.

The AIMS centile rank correlated with the PEDI mobility score (Pearson correlation coefficient r = 0.63, p < 0.001) and the PEDI social function (r = 0.53, p < 0.01) but not with the PEDI self care score (r = 0.30, n.s.).

GMs and AIMS

Children with cramped-synchronized GMs at 34 weeks postmenstrual age had a significantly lower AIMS centile rank than those with poor repertoire or normal GMs (Kruskal–Wallis test, p < 0.05; Table 3). Four out of five children with cramped-synchronized GMs scored below the 5th AIMS centile; three of the twelve children with poor repertoire GMs had a suspect/abnormal score; one of the 15 children with normal GMs also scored abnormally on the AIMS. The strength of association between the quality of GMs and the categories of the AIMS was moderate (Cramer's V = 0.58; p < 0.01; Table 4).

Table 3

The quality of GMs at 34 weeks postmenstrual age; and motor performance, self care and social function (tested by means of AIMS and PEDI; median and inter-quartiles) at a median age of 14 months post term.

	Normal GMs n = 15 (47%)	Poor repertoire GMs n = 12 (37.5%)	Cramped-synchronized GMs n = 5 (15.5%)	p-Value
Age at outcome assessment (months; median, IQ)	15 (13–16)	14 (12.5–16)	14 (11–16.5)	n.s.
AIMS centile ranks	90 (50–90)	37.50 (5–90)	3 (3–47.50)	p < 0.05
PEDI mobility	43 (37–47)	38 (29–46)	31.5 (15–37)	p = 0.06, n.s.
PEDI self care	53 (48–56.5)	47 (37–53)	47 (23–56)	n.s.
PEDI social function	58 (51–61.5)	56 (34.5–61)	49 (24–54)	n.s.

Table 4

Strength of association (Spearman rank order coefficient and Cramer's V coefficient) between assessments at 34 weeks postmenstrual age and 14 months (range 8–18 months) corrected age.

	AIMS centile rank	AIMS categories	PEDI mobility	PEDI self care	PEDI social function
GM assessment	r = 0.43	Cramer's V = 0.58	r = 0.41	r = 0.22	r = 0.30
	p = 0.01	p < 0.01	p < 0.05	n.s.	n.s.
TIMP	r = 0.06	Cramer's V = 0.06	r = 0.12	r = 0.20	r = 0.22
	n.s.	n.s.	n.s.	n.s.	n.s.

GMs and PEDI

Table 2 shows that cramped-synchronized GMs at 34 weeks postmenstrual age were marginally associated with a lower PEDI mobility score. The strength of association between GMs and the PEDI mobility scale was fair (Table 4). The quality of GMs was not associated with the PEDI self care or social function scales (Tables 3, 4).

TIMP and outcome measurements

The results of the TIMP did not correlate with the outcome measurements at 14 months (Table 4).

Motor behavior at preterm age and neonatal characteristics

Apart from a low Apgar Score at 1 min (which was associated with cramped-synchronized GMs), there was no relation between the quality of GMs and neonatal data (Table 1); nor was there an association between the results obtained by the TIMP and neonatal data.

All the test scores were equally distributed in infants with normal and abnormal brain ultrasound findings (Mann–Whitney U test, n.s.).

Discussion

Our study revealed that the GM quality measured at a single assessment during preterm age was only fairly to moderately associated (correlation coefficients between 0.41 and 0.58; Table 4) with the functional motor outcome at a median age of 14 months. Although our aim was to examine whether it was possible to predict the infants' neurodevelopmental outcome by means of a single GM assessment, we should keep in mind that the high predictive value of GM assessment is based on developmental trajectories of GMs. A single recording does not reveal whether cramped-synchronized GMs are only transient or are indeed present for several weeks, which would be a specific marker for spastic cerebral palsy [12,35]. The predictivity of poor repertoire GMs is low [36]; it is therefore all the more recommendable for affected children to have regular GM assessments carried out until 3 to 5 months postterm age [21,24,28]. A recent review on neuromotor assessments for preterm infants has stressed that no single assessment has yet managed to reliably identify normal or atypical motor development in a given group of infants [23].

In our study, the TIMP was not an optimal tool to complement the GM assessment at preterm age. All but one infant scored below the age-specific average in the TIMP, although 59% of them had normal GMs. One explanation for such a discrepancy could be that the sample on which the normative data of the TIMP scores were based was very small — 17 preterm infants aged 32 to 34 weeks [37]. Moreover, the validity data of the TIMP for prediction of AIMS scores were scarce if the TIMP was applied during the first weeks of life [26]. Our study also revealed no association between the TIMP and the outcome measurements (Table 4). A recent study showed that, even if applied at term age, the TIMP did not correlate with the outcome measures at 1 year of age, among which was the AIMS [38].

With a value of 41%, the percentage of abnormal GMs in our study (including both poor repertoire and cramped-synchronized GMs) seems to be rather high. In a recent study by de Vries and Bos [39] preterm infants often showed abnormal GMs during the first 2 weeks of life; on postnatal day 10, 66% of their study group (N = 38; selected for ELBW) still had abnormal GMs. De Vries and Bos argued that such a high percentage of abnormal GMs could be explained by a number of factors that influenced early brain function in a preterm infant [39]. In our study, however, the GM quality was neither related to postnatal age nor to lower birth weight (Table 1). The strength of association between GM quality and TIMP was low (Kendall's tau-c = 0.13), but almost identical with the results recently published by a Canadian group (GMs and TIMP assessed at term age: Kendall's tau-c = 0.12) [40]. Cioni et al. reported that there was a 73% accordance between the quality of GMs during preterm age and the results of the Dubowitz neurological examination [13]. We agree with Snider and co-workers [40] that future studies should focus on the correlation between TIMP and neurological examination in order to investigate whether GM assessment and TIMP represent different constructs.

Most of the work groups that applied the GM assessment waited for at least 24 months to assess the neurodevelopmental outcome [11-13], only a few groups assessed the outcome at 1 year [14,27,40]. The study by Snider and associates revealed a significant relation between GMs and the walking ability measured on the AIMS standing subscale: none of the children with cramped-synchronized GMs at term age was able to walk a year later [40]. Garcia et al. found a significant and positive relationship between GM quality and the neurological outcome if the GMs were evaluated at term or postterm age, but not if they were assessed preterm. The study's outcome assessment was based on a neurological examination and on the Denver developmental screening test, carried out at 12 months. At preterm age the negative predictive value was 100%, but the positive predictive value was only 36% [27]. Spittle et al. examined the motor development in 86 preterm infants (likewise by means of AIMS) at 12 months corrected age. The Spearman rank order coefficient between the GM quality and AIMS was 0.31, which is slightly lower than in our study, although here the GMs were assessed at 1 month post term. In the same study, the MRI findings (white matter assessment) correlated with the AIMS even to a lesser extent than they did with GM quality (r = 0.27) [14].

There is a limitation to our study in that we did not cover the full range of brain ultrasound findings. Furthermore, our results are based on data from a small group of children from a single medical center. The infants had not been carefully selected but were a sample of convenience. All assessments were done by the same person; neither TIMP nor the outcome assessments were double scored. These limitations raise the question to what extent the findings can be generalized to a broader population of children born preterm.

Apart from these limitations, our results confirm that early intervention should not only address preterm infants with typical risk factors but should also include infants with cramped-synchronized GMs. However, we should be extremely cautious when informing parents about their infant's prognosis. The assessment of GM quality can only be part of additional neurological assessment, most notably brain ultrasound, MRI or cerebral function monitoring. Ultimately, we must always bear in mind that preterm infants have an increased risk for motor, cognitive and behavioral difficulties that only become apparent at school age, but are not necessarily identifiable at 14 months of age.