Cerebral palsy in children: Movements and postures during early infancy, dependent on preterm vs. full term birth

Abstract

Background

A deviant motor behaviour at age 3 to 5 months is predictive of cerebral palsy (CP). Particular features of the early motor repertoire even proved predictive of the degree of functional limitations as classified on the Gross Motor Function Classification System (GMFCS) in children with CP, born preterm.

Aims

We aimed to determine whether an association between the early motor repertoire and the GMFCS also holds true for children born at term.

Study design

Longitudinal study.

Subjects

79 infants (60 boys and 19 girls; 47 infants born at term; video recorded for the assessment of movements and posture at age 9 to 20 weeks postterm age) who developed CP.

Outcome measures

The GMFCS was applied at age 2 to 5 years.

Results

Motor optimality at age 3 to 5 months showed a significant correlation with functional mobility and activity limitation as classified on the GMFCS at age 2 to 5 years in both children born at term (Spearman rho = − 0.66, p < 0.001) and born preterm (rho = − 0.37, p < 0.05). Infants born preterm were more likely to show normal movement patterns than infants born at term. A normal posture and an abnormal, jerky (yet not monotonous) movement character resulted in better levels of function and mobility. With the exception of one, none of the infants showed fidgety movements. A cramped-synchronised movement character, repetitive opening and closing of the mouth, and abnormal finger postures characterised children who would show a poor self-mobility later.

Conclusions

Assessing the quality of motor performance at 9 to 20 weeks postterm age (irrespective of the gestational age) improves our ability to predict later functional limitations in children with CP.

1. Introduction

Cerebral palsy (CP) is a well-recognized neurodevelopmental condition which manifests early in childhood, usually before 18 months of age [1]. A comprehensive meta-analysis revealed that the prevalence of CP decreases with increasing gestational age: 14.6% of affected children were born extremely preterm (below 28 weeks' gestational age), 6.2% between 28 and 31 weeks' gestation, 0.4% to 0.7% were born moderately preterm (i.e. at 32 to 36 weeks' gestation), and 0.1% at term [2]. In children born preterm, spastic CP is predominant, whereas in children born at term dyskinetic and ataxic forms prevail [2]. Bilateral spastic CP is more widespread than unilateral spastic CP both in infants born preterm (median prevalence: 73% vs. 21%) and born at term (median prevalence 48.5% vs. 36.5%) [2]. Figures show a similar prevalence of CP in China as in Europe or the USA [3,4].

Apart from the low gestational age and low birth weight, perinatal asphyxia, white matter disease, severe intraventricular haemorrhage, cerebral infarction and deep grey matter lesions were identified as predictors of CP [4–6]. Abnormal spontaneous ‘general movements’ (GMs) were found to be among the most reliable markers for CP [7–17]. The assessment of GMs is based on visual Gestalt perception in preterm infants and term infants aged up to 5 months. Normally, GMs comprise the entire body and manifest themselves in variable sequences of arm, leg, neck and trunk movements. They come and go gradually, varying in intensity and speed. Rotations and frequent slight variations of the direction of motion make them look complex, though smooth [13]. GMs occur in age-specific patterns. During the postterm age of 3 to 5 months, they are described as ‘fidgety movements’, i.e. small movements of the neck, trunk and limbs in all directions and of variable acceleration [7]. Virtually all infants develop normally if such fidgety GMs are present and normal, even if their brain ultrasound findings indicate a disposition to later neurological deficits. Conversely, almost all infants with no fidgety movements develop neurological deficits, even if their ultrasound indicates no significant risk [7,13]. A recent review of 15 studies on the predictive value of fidgety movements reported a sensitivity > 91% and a specificity > 81% [18]. The most comprehensive study so far, in which more than 900 children participated, yielded a sensitivity of 98% and a specificity of 94% [19].

The motor repertoire of infants aged 3 to 5 months consists not only of fidgety movements but also of other movements and postural patterns [20]. Detailed studies have shown that several qualitative and quantitative aspects of the motor patterns are predictive of the level of self-mobility of children with CP [16], as well as of minor neurological dysfunctions [21–23], and lower intelligence at school age [24]. Bruggink and colleagues [16] found a non-optimal motor behaviour at 11 to 17 weeks to be indicative of poor self-mobility in 37 children with CP (9 unilateral spastic CP and 28 bilateral spastic CP). Apart from the absence of fidgety movements, more severe functional limitations in 6- to 12-year-old children with CP were mainly associated with a cramped movement character, an age-inadequate motor repertoire, and monotonous kicking [16]. Interestingly enough, the same study demonstrated that a so-called flat posture (i.e. all limbs were mainly lying on the surface; antigravity movements and flexions in hips and knees were rare) was associated with a less severe functional limitation [16]. However, the sample of 37 children only included preterm infants born before 34 weeks' gestation, which raised the question of whether such a detailed analysis of movements and postures in infants (aged 3 to 5 months) could also be predictive of the functional mobility of children with CP who were born at term. We therefore did the same assessment both in term-born and in preterm-born children with CP. The aims of our study were (1) to describe movements and postures in 9- to 20-week-old infants who later developed CP; (2) to analyse to what extent the motor performance at 9 to 20 weeks' postterm age was related to (a) the anatomical (uni- vs. bilateral) distribution of CP and (b) the functional mobility and activity limitation at 2 to 5 years of age as classified on the Gross Motor Function Classification System (GMFCS) [25]; and (3) to analyse whether such associations are different in term-born children with CP and preterm-born children with CP.

2. Methods

2.1. Participants

The study comprised 79 children, 60 boys (76%) and 19 girls (24%), who had been admitted to the Department of Rehabilitation at the Children's Hospital of the Fudan University, Shanghai, PR China, between September 2003 and May 2009, and developed cerebral palsy by the age of 2. The reasons for admission were the following: (a) high-risk for neurodevelopmental problems due to pre- and perinatal histories; (b) abnormal findings at paediatric examinations; or (c) parental concerns. The inclusion criterion for this study was that their motor performance was videoed around their 4th month postterm age.

The infants' gestational ages at birth ranged from 28 to 42 weeks (median = 38 weeks), with a birth weight range of 880 to 4500 g (median = 2962 g). Thirty-two infants were born preterm (40.5%). Twelve of the infants born preterm had a very low birth weight (< 1500 g) and one had an extremely low birth weight (< 1000 g). Apart from gestational age and birth weight, full-term infants (n = 47; 59.5%) differed from preterms in the following aspects: periventricular leucomalacia was more common in preterm infants (p < 0.05), whereas hypoxic ischaemic encephalopathy and neonatal seizures were more common in fullterm infants (p < 0.05; Table 1). Other clinical characteristics are presented in Table 1.

Table 1.

Clinical characteristics and neurodevelopmental outcome of 79 participants according to preterm or term birth.

	Children born preterm n = 32 (40.5%)	Children born at term n = 47 (59.5%)
Male:female	23:9 (72%:28%)	37:10 (79%:21%)	n.s.
Gestational age in weeks	Median = 32 (P25–P75 = 30–34)	Median = 39 (P25–P75 = 38–40)	p < 0.001a
Birth weight in grams	Mean = 1548 (SD = 549)	Mean = 3200 (SD = 423)	p < 0.001b
Periventricular leucomalacia	n = 12 (38%)	n = 8 (17%)	p < 0.05c
Intracranial haemorrhage	n = 15 (47%)	n = 14 (30%)	n.s.
HIE	n = 1 (3%)	n = 11 (23%)	p < 0.01c
Brain malformation	–	n = 6 (13%)
Normal brain imaged	n = 1 (3%)	n = 5 (11%)
Congenital heart disease	n = 3 (9%)	n = 2 (4%)	n.s.
MODS	–	n = 2 (4%)
Purulent meningitis	–	n = 1 (2%)
Septicaemia	n = 4 (12%)	n = 1 (2%)	n.s.
Hyperbilirubinaemia	n = 3 (9%)	n = 3 (6%)	n.s.
Hypoglycaemia	n = 2 (6%)	n = 2 (4%)	n.s.
IRDS	n = 2 (6%)	n = 1 (2%)	n.s.
Severe apnoeas	n = 5 (16%)	n = 4 (8%)	n.s.
Neonatal seizurese	n = 2 (6%)	n = 11 (23%)	p < 0.05c
Outcome
Unilateral spastic CP	–	n = 5
Bilateral spastic CP	n = 32	n = 41f
Dyskinetic CP	–	n = 1
Epilepsy	n = 4 (12%)	n = 11 (23%)	n.s.
Severe visual impairment	n = 2 (6%)	n = 14 (30%)	p = 0.01c

Abbreviations: CP = cerebral palsy; HIE = hypoxic ischaemic encephalopathy; IRDS = idiopathic respiratory distress syndrome; MODS = multiple organ dysfunction syndrome.

^aIndependent sample median test.

^bT-test for equality of means.

^cPearson Chi-Square test.

^dBrain imaging was performed by means of magnetic resonance (n = 14), cranial ultrasound or computer tomography; four infants did not receive brain imaging.

^eThree of these 13 children developed epilepsy.

^fOne child died during the second year of life.

All parents gave written informed consent. The Ethical Review Board of the Fudan University Shanghai approved the study.

2.2. Recording and evaluation of movements and postures in early infancy

Seven-minute video recordings were made of the spontaneous motility of each infant at a median age of 14 weeks (range = 9 to 20 weeks). The recordings were made during periods of active wakefulness between feedings, with the infant partly dressed, lying in supine position [26]. The video recordings were evaluated by two raters (HY and CE) according to Prechtl's method of GM assessment [27]. Scorer CE was not familiar with the participants' clinical histories, the anatomical distribution of their respective CP, or their GMFCS levels. In case of disagreement, the raters re-evaluated the recording of the infant in question until consensus was reached on a final score.

Fidgety movements and the concurrent motor and postural repertoires were assessed independently in separate runs of the video recordings. Using the score sheet for the assessment of motor repertoire at 3 to 5 months (Einspieler et al. [27]; page 26), we defined a motor optimality score with a maximum value of 28 (for the best possible performance) and a minimum value of 5. The score sheet comprises the following five sub-categories: (1) fidgety movements, (2) age-adequacy of motor repertoire, (3) quality of movement patterns other than fidgety movements, (4) posture, and (5) overall quality of the motor repertoire [16,22–24,27–29]. Fjørtoft and colleagues found a very high inter-observer reliability in the assessment of the motor optimality score with an intra-class correlation coefficient ranging between 0.80 and 0.94 [29].

2.3. Functional assessment of the neurological findings

All children were regularly examined clinically, neurologically, and psychologically, up to the age of 5 years. Diagnosis of CP was based on the definition proposed recently [1,30]. Five children (6%) developed unilateral spastic CP. Bilateral spastic CP was diagnosed in 73 children (93%), whereby the arms were less affected in 23 children than the legs. One child with bilateral spastic CP died during the second year of life. The remaining infant (1%) developed dyskinetic CP, further differentiated into choreoathetotic CP. Epilepsy (two or more afebrile, non-neonatal seizures) was diagnosed in 15 children (19%); severe visual impairment was found in 16 children (20%); profound hearing loss was found in 7 children (9%); data on cognitive development were not available.

Apart from the child who died, 13 families could not be traced after the age of 1.5 years and therefore could not be assessed by means of the GMFCS [25]. Although the majority of these children (n = 11) were born at term, their distributions of birth weight and gestational age did not statistically differ from participants who were eventually evaluated by means of GMFCS (Mann–Whitney-U test; p = 0.29 and p = 0.11, respectively).

At 2 to 5 years of age, 65 children (49 boys and 16 girls) were scored and classified by two rehabilitation doctors (HY, SW) according to the GMFCS. The GMFCS is a standardised method of classifying the gross motor function of children with CP. It is based on a 5-level classification system: the higher the level, the more severe the CP. The classification consists in determining which of the 5 levels best corresponds to the child's abilities and limitations in gross motor function in home and community settings [25].

2.4. Statistical analysis

Statistical analysis was performed using SPSS package for Windows, version 18.0 (SPSS Inc., Chicago, IL). Pearson Chi-Square test was used to evaluate associations between nominal data. In order to put the medians of the non-normally distributed continuous data (e.g. gestational age, birth weight, optimality scores) in relation to the nominal data (e.g. gender, categories of motor behaviour), we applied the Mann–Whitney-U test and the Kruskal–Wallis test. In order to assess the relative strength of the association between variables, we computed the following correlation coefficients: Cramer-V coefficient was applied when at least one of the two variables was nominal (e.g. gender and quality of movement patterns); Kendall-Tau-c coefficient was applied when both variables were ordinal (e.g. quality of movement patterns and GMFCS level). To assess the relation between two continuous variables (e.g. gestational age and motor optimality score), we applied Pearson product‐moment correlation coefficient. We used the Spearman rank order correlation coefficient when one score was ordinal and the other continuous (e.g. GMFCS level and motor optimality score). Throughout the analyses, p < 0.05 (two-tailed) was considered to be statistically significant.

3. Results

3.1. The motor performance at early infancy

Only one infant had normal fidgety movements; the other infants (78; 99%) did not display fidgety movements and their fidgety movements were classified as absent (Table 2). None of the infants displayed age-adequate movement repertoires. The repertoire was scored as reduced in 10 infants (13%) and as not adequate to age in the remaining 69 infants (87%).

Table 2.

Abnormal movement and postural patterns in 79 participants at 9 to 20 weeks postterm age. Patterns that occurred in fewer than three infants are not reported on here.

Abnormal pattern	n (%)
Movements
•Absence of fidgety movements	78 (99%)
•Long lasting repetitive swipes	3 (4%)
•Circular arm movement	9 (11%)
•Asymmetrical segmental movements	5 (6%)
•Prolonged, monotonous kicking	14 (18%)
•Burst of excessive ‘excitement’ though without joyful facial expression	8 (10%)
•Lack of leg movements	13 (16%)
•Monotonous side-to-side movements of the head	21 (27%)
•Repetitive opening and closing of the mouth	23 (29%)
•Repetitive tongue protrusion	16 (20%)
Posture
•Inability to keep the head in midline	50 (63%)
•Asymmetrical body posture	12 (15%)
•Trunk and limbs lying flat on the surface	13 (16%)
•Persistent asymmetric tonic neck response	26 (33%)
•Extended arms	20 (25%)
•Extended legs	19 (24%)
•Hyperextension of the neck and trunk	9 (11%)
•Fisting	28 (35%)
•Synchronised opening and closing of the fingers	15 (19%)
•Finger spreading	13 (16%)

The quality of the various movement patterns (other than fidgety movements) was scored as predominantly normal in 11 infants (14%) and predominantly abnormal in 52 infants (66%); 16 infants (20%) showed an equal number of normal and abnormal movements. Among the normal movement patterns were swipes, wiggling–oscillating arm movements, hand-to-mouth movements, kicking, and legs lifted. Abnormal movement patterns are listed in Table 2.

Posture was rated as predominantly normal in 15 infants (19%) and predominantly abnormal in 53 infants (67%); 11 infants (14%) showed an equal number of normal and abnormal postures. Infants with predominantly normal postural patterns were able to hold their head in midline and showed a symmetrical body posture as well as variable finger postures; persistent asymmetric tonic neck response was absent. The occurrence of normal postural items was associated with the occurrence of normal motor patterns (Pearson r = 0.36; p < 0.01). The majority of infants, however, had abnormal postures (Table 2); up to 7 items (median = 3) related to the posture were scored as abnormal. The occurrence of abnormal postures was not related to the occurrence of abnormal motor patterns (Pearson r = 0.13; p = 0.26; n.s.).

In none of the infants was the overall movement character rated as normal. Fifty-eight infants (73%) had a monotonous, stiff, jerky or tremulous movement character, and in 21 infants (27%) the movement character was scored as cramped-synchronised.

The median motor optimality score was 6 (inter-quartiles: P25 = 6; P75 = 9; range = 5–20).

Neither the motor optimality score nor the details of the motor behaviour were different between infants who were lost to follow-up (n = 14 including the infant who died) and those who participated in the outcome evaluation (Pearson Chi-Square; Mann–Whitney-U test; p-values ranged from 0.27 to 0.64; n.s.).

3.2. Functional mobility and activity limitation at 2 to 5 years of age

GMFCS level I was diagnosed in 18 children (28%; four with unilateral spastic CP, 14 with bilateral spastic CP), level II was diagnosed in nine children (14%; one with unilateral spastic CP, 8 with bilateral spastic CP), level III was diagnosed in eight children (12%; all with bilateral spastic CP), level IV was diagnosed in 15 children (23%; 14 with bilateral spastic CP, one with dyskinetic CP), and level V was diagnosed in 15 children (23%; all with bilateral spastic CP). Children born at term had a significantly poorer self-mobility than children born preterm (Kendall-Tau-c = 0.37; p < 0.01; Table 3).

Table 3.

Functional mobility and activity limitation according to the Gross Motor Function Classification System [25] of 65 participants with respect to their preterm or term birth.

	Children born preterm n = 29 (45%)	Children born at term n = 36 (55%)
Level I	n = 11 (38%)	n = 7 (19%)	Kendall-Tau-c = 0.37 p < 0.01
Level II	n = 4 (14%)	n = 5 (14%)
Level III	n = 7 (24%)	n = 1 (3%)
Level IV	n = 4 (14%)	n = 11 (31%)
Level V	n = 3 (10%)	n = 12 (33%)

3.3. Association between the motor performance in early infancy and the anatomical distribution of CP

The one child with normal fidgety movements developed unilateral CP. Normal and abnormal movement patterns other than fidgety movements (at 9 to 20 weeks postterm age) were equally distributed between children with later unilateral CP and those who developed bilateral CP (Pearson Chi-Square; p-values ranged from 0.33 to 0.91; n.s.). An asymmetry of segmental movements was found in three infants who later developed unilateral CP (60%) and in two infants with bilateral CP (3%; Pearson Chi-Square; p < 0.001). Twelve out of 13 infants with no leg movements at 9 to 20 weeks postterm age developed bilateral CP; the remaining infant developed dyskinetic CP.

Children who later developed unilateral CP and those who developed bilateral CP had neither differed in their abnormal postures nor in their overall movement character (Pearson Chi-Square; p-values ranged from 0.33 to 0.98; n.s.); the same goes for the motor optimality score (Mann–Whitney-U test, p = 0.34; n.s.).

3.4. Association between the motor performance in early infancy and the functional mobility and activity limitation at 2 to 5 years of age

A Spearman correlation coefficient of rho = − 0.56 (p < 0.001) between the motor optimality score and the GMFCS level indicated that a higher motor optimality score at 9 to 20 weeks of age was associated with a lower GMFCS score indicating a better level of self-mobility.

The one infant with normal fidgety movements was classified as GMFSC level I. There was a significant association between the age-adequacy of the motor repertoire at 9 to 20 weeks postterm age and the GMFCS level later on (Kendall-Tau-c = − 0.22; p < 0.001); i.e. the more age-adequate the motor repertoire during infancy, the lower (i.e. better) the GMFCS levels.

The association between the quality of movements at 9 to 20 weeks postterm age and the GMFCS levels was − 0.52 (Spearman correlation coefficient; p < 0.001): seven out of ten children (70%) with predominantly normal movement patterns (other than fidgety movements) were rated as GMFCS level I. Plus, the more abnormal movement patterns were observed at 9 to 20 weeks, the higher (i.e. worse) the GMFCS level would be later (Spearman correlation coefficient rho = 0.40; p = 0.001). Hardly any normal movement patterns were found in infants who were later rated as GMFCS level IV or V. With the exception of one, all abnormal movement patterns were equally distributed between the children who were later rated as GMFCS levels I and II (i.e. better levels of self mobility) as opposed to levels III to V, which represent poor self-mobility. Only repetitive opening and closing of the mouth was found significantly more often in children who would show a poor self-mobility (Pearson Chi-Square; p < 0.01).

The association between the postural patterns found at 9 to 20  weeks postterm age and the children's later GMFCS levels was − 0.19 (Kendall-Tau-c; p < 0.05). Nine out of 13 children (69%) with predominantly normal posture were classified with GMFCS level I or II. The occurrence of a persistent asymmetric tonic neck response at 9 to 20 weeks postterm age was not associated with later GMFCS levels (Pearson Chi-Square; p = 0.61; n.s.). However, finger spreading and synchronised opening and closing of the fingers occurred more often in infants with later GMFCS levels IV and V (Pearson Chi-Square; p < 0.01).

The association between a cramped-synchronised movement character and the later GMFCS levels was 0.41 (Kendall-Tau-c; p < 0.001): 15 out of 18 infants (83%) with cramped-synchronised GMs had later GMFCS level IV or V. A monotonous but not cramped-synchronised movement character was also associated with GMFCS levels IV and V (Pearson chi-square; p < 0.01). Tremulous movements or slow or fast movements were not associated with GMFCS levels. Interestingly, jerky movements were observed significantly more often in infants who were later classified with GMFCS level I or II (10/15 infants; 67%; Pearson Chi-Square; p < 0.05).

3.5. Difference between children born preterm and children born at term

The total sample consisted of 32 participants born preterm and 47 born at term (Table 1). GMFCS data were available of 29 children born preterm and 36 children born at term (Table 3). Children born at term had higher (i.e. worse) GMFCS levels than children born preterm (Kendall-Tau-c = 0.37; p < 0.01). Apart from the number of normal movements, the movements and postures of preterm infants during their early infancy did not differ significantly from those of infants born at term. Infants born preterm were more likely to show normal movement patterns than infants born at term (Table 4; Pearson Chi-Square; p < 0.05). Normal wiggling–oscillating arm movements occurred solely in infants born preterm, whereas abnormal, monotonous and long-lasting wiggling–oscillating arm movements occurred solely in infants born at term (Pearson Chi-Square; p < 0.05). Though statistically significant, their rate of occurrence (6% in each group) is negligible.

Table 4.

Movements and postures in 79 participants at 9 to 20 weeks postterm age, according to preterm or term birth.

	Children born preterm n = 32 (40.5%)	Children born at term n = 47 (59.5%)
Fidgety movements normal:absent	0:32	1:46	n.s.
Movement repertoire adequate:reduced:inadequate	0:4:28	0:6:41	n.s.
Quality of movement patternsa N > A:N = A:N < Ab	6:8:18	5:8:34	n.s.
Number of normal movement patterns	Median = 1 P25 = 0; P75 = 2	Median = 0 P25 = 0; P75 = 1	p < 0.05c
Number of abnormal movement patterns	Median = 2 P25 = 1; P75 = 2	Median = 2 P25 = 1; P75 = 3	n.s.
Quality of postural patterns N > A:N = A:N < Ab	5:4:23	10:7:30	n.s.
Number of normal postural patterns	Median = 2 P25 = 1; P75 = 3	Median = 2 P25 = 1; P75 = 2	n.s.
Number of abnormal postural patterns	Median = 3 P25 = 2; P75 = 4	Median = 3 P25 = 2; P75 = 4	n.s.
Movement character N:A but not CS:CS	0:25:7	0:33:14	n.s.
Motor optimality score	Median = 7 P25 = 6; P75 = 9 Range = 5–22	Median = 6 P25 = 5; P75 = 9 Range = 5–26	n.s.

Abbreviations: A = abnormal; CS = cramped-synchronised; N = normal.

^ai.e. other than fidgety movements.

^bN > A (predominantly normal); N = A (equal number of normal and abnormal movement patterns); N < A (predominantly abnormal).

^cMann–Whitney-U test.

The motor optimality score was significantly associated with the GMFCS levels in both children born preterm and born at term (Table 5). The particular components of the motor optimality score, however, were only significantly associated with the GMFCS classification of children born at term (Table 5).

Table 5.

Association between movements and postures at 9 to 20 weeks postterm age and GMFCS levels at 2 to 5 years of age in 65 children, according to preterm or term birth.

	Children born preterm n = 29 (45%)	Children born at term n = 36 (55%)
Movement repertoire × GMFCS (Cramer-V)	− 0.24 p = 0.22; n.s.	− 0.61 p < 0.01
Quality of movement patterns × GMFCS (Kendall-Tau-c)	− 0.24 p = 0.11; n.s.	− 0.52 p < 0.001
Quality of posture × GMFCS (Kendall-Tau-c)	− 0.10 p = 0.49; n.s.	− 0.28 p < 0.05
Movement character × GMFCS (Cramer-V)	− 0.23 p = 0.28; n.s.	− 0.54 p < 0.05
Motor optimality score × GMFCS (Spearman rho)	− 0.37 p < 0.05	− 0.66 p < 0.001

Abbreviation: GMFCS = Gross Motor Function Classification System [25].

4. Discussion

Himmelmann et al. [31] reported that GMFCS level I was positively correlated with increasing gestational age. By contrast to this finding, our study revealed that children born at term had a significantly lower GMFCS level than children born preterm (Table 3). This was ascribed to the relatively high rate of occurrence of hypoxic–ischaemic encephalopathy (HIE), since central grey matter damage, the hallmark of HIE, leads to severe CP [31–33]. Likewise, each component of the motor optimality score (assessed at 9 to 20 weeks postterm age) correlated with the GMFCS levels of children born at term but not of children born preterm (Table 5). The motor optimality score, however, was significantly associated with the level of functional mobility in both children born at term and born preterm. In accordance with a previous study on preterm infants [16], a higher motor optimality score was significantly related to a better level on the GMFCS, notably in infants born at term.

4.1. Fidgety movements and CP

Only one of the 79 children with CP had normal fidgety movements. This boy, born fullterm with an intraventricular haemorrhage developed mild unilateral spastic CP (GMFCS level I). In the study by Bruggink et al. [16] three out of 38 infants had fidgety movements, and all three of them were at GMFCS level I or II at school age. A recent study compared Prechtl's assessment of GMs with the GM assessment devised by Hadders-Algra, and likewise reported on five children who had some fidgety activity but nonetheless developed CP, albeit in its mild form [17].

Safe for one infant, none of the infants of our study (99%) had any kind of fidgety movements whatsoever, which once again shows the high predictive power of absent fidgety movements for CP [7–9,11,13–19,27,28,33]. The fact that fidgety movements are absent in all subtypes of CP indicates that it takes intact corticospinal fibres and a normal output of the basal ganglia and cerebellum to generate typical fidgety movements [7,9,15]. Magnetic resonance studies have demonstrated that in very preterm infants white matter lesions [34] and a reduced cerebellar diameter [35] were associated with absent fidgety movements. In infants born at term the severity of the injury to the central grey matter correlated with absent fidgety movements [33].

4.2. Age-adequacy of the motor repertoire during early infancy, and later CP

In none of the participants was the movement repertoire age-adequate (9 to 20 weeks). Arm and leg movements towards the midline and antigravity movements were absent in the majority of infants. Similar observations have been published before [9,36]. Those infants who had almost no identifiable movement patterns were later rated as GMFCS levels IV and V. Previous studies also demonstrated that an age-inadequate motor repertoire was associated with later severe neurological impairment, including CP [16,28].

4.3. Eventual CP despite normal movement patterns in infants aged 9–20 weeks

In spite of absent fidgety movements, some children showed normal movement patterns in early infancy. In one child, no less than six normal movement patterns could be observed including normal swipes and wiggling–oscillating arm movements, hand-to-mouth movements, kicking, and leg lifting. Also social smiles and visual scanning were observed, although rarely. Infants born preterm were more likely to have normal movement patterns than infants born at term (Table 4). This might be related to their better GMFCS levels at ages 2 to 5 (Table 3).

4.4. Association between abnormal movement patterns in early infancy and later CP

The majority of the participants not only had a lack of fidgety movements but other abnormal movement patterns as well. A higher number of abnormal movement patterns during 9 to 20 weeks postterm age were related to GMFCS levels IV and V. Monotonous side-to-side movements of the head were frequent (Table 2), usually accompanied by an asymmetric tonic neck response. Only few infants showed long-lasting repetitive swipes or monotonous and long-lasting kicking; some infants showed a combination of swipes and kicking in what one could call bursts of excitement but showed no facial expression of excitement or pleasure. Yet again, other infants did not move their legs at all (Table 2). A lack of leg movements in 3-month-old infants who later developed CP was also described by Pizzardi et al. [37], although during vertical suspension. Pizzardi et al. emphasised that an abnormal movement quality (assessed with the Hammersmith Neurological Examination for Infants) had the highest predictive value for the occurrence of CP when assessed at 3 months of age [37].

Einspieler et al. [9] identified the co-occurrence of circular arm movements and absent fidgety movements as an early marker for dyskinetic CP. This observation was confirmed by the one participant of the present study who developed choreoathetotic CP. However, there were also eight participants in the present study who developed bilateral spastic CP in spite of circular arm movements. The previous study by Einspieler et al. [9] also comprised 12 children with spastic CPs, one of whom had shown circular arm movements during the third month of life. Similar observations were made in infants with hypoxic–ischaemic encephalopathy [38].

Abnormalities of facial movements included repetitive opening and closing of the mouth and repetitive tongue protrusion. Especially repetitive opening and closing of the mouth was more frequent in infants who would later be rated as GMFCS levels IV and V.

Only five participants developed unilateral CP, yet three of them (60%) showed a reduction of segmental movements in the arm and leg that were later paretic. Such an asymmetry was also found in two participants with later bilateral CP (3%). Segmental movements are isolated movements of the wrist, fingers, ankles and toes, which were identified as an early marker of later unilateral CP if their frequency of occurrence is not balanced on both sides [8,11]. Cioni et al. [8] reported that such an asymmetry could also be observed in infants with a normal neurological development, though very rarely.

4.5. Association between posture in early infancy and later CP

Fifteen infants (19%) had a predominantly normal, age-adequate posture, which was associated with the occurrence of normal movements. Their head was in midline, the body symmetrical, the asymmetric tonic neck response was not persistent and variable finger postures could be observed. Most of these children had GMFCS level I or II at the age of 2 to 5. The majority of the participants, however, showed abnormal postural patterns. Sixty-three percent could not keep their head in midline. Similar observations were made in preterm infants with unilateral intraparenchymal echodensity [8].

Touwen and Hadders-Algra [39] identified hyperextension of the neck and trunk as an early marker for CP. We found such a posture in merely 11% of our participants. Other infants could not keep their body in a symmetrical position; in some infants the trunk and limbs were flat on the surface. While these observations are related to a lack of antigravity postures, some infants stiffly stretched their arms or legs upwards and often sustained in this posture.

A persistent asymmetric tonic neck response was observed in 33% of the participants but was not associated with later GMFCS levels. The same percentage was reported in a recent study by Hamer et al. [17]. Einspieler et al. [9] observed this posture in half of the infants who later developed spastic CP.

Fisting was more frequent, but we also observed finger spreading and synchronised opening and closing of the fingers. The latter two patterns were associated with GMFCS levels IV and V, whereas fisting was not. This is in line with the findings of Konishi and Prechtl [40], who reported that children with later spastic CP had shown all kinds of finger movements in early infancy other than tight fisting.

4.6. Association between cramped-synchronised movements during early infancy and CP

In none of the infants was the overall movement character rated as normal. Seventy-three percent had a monotonous, stiff, jerky or tremulous movement character, while the movement character of 27% of the infants was rated as cramped-synchronised. The study by Hamer et al. [17] revealed a similar percentage, although it only included ten participants. Other studies have yielded a clear correlation between cramped-synchronised GMs at 3 to 4 months and eventual CP [10,28,33]. Bruggink et al. [16] also found that a cramped-synchronised movement character was associated with severe functional limitations in 6- to 12-year old children with CP. Our results confirm this finding.

Interestingly, jerky (yet not monotonous) movements were observed significantly more often in infants who later were classified with GMFCS level I or II. Similarly, Groen et al. [41] reported that the presence of jerky movements at 3 to 4 months was not associated with a less favourable outcome at school age, whereas stiff movements were.

4.7. Limitations of our study

Although our findings are based on a large sample of infants developing CP, the proportion of participants with unilateral CP (6%) is rather small. Therefore, we are far from drawing general conclusions about the early motor behaviour in infants eventually developing unilateral CP. Furthermore, our participants are a sample of convenience: inclusion criteria were that their motor performance was videoed around their 4th month postterm age and that they had developed CP. They were recruited from a single medical centre; their parents consulted the department for rehabilitation for various reasons; the majority of them could be considered as high-risk for maldevelopment. An extrapolation of the findings to the general population is not feasible but was also not the aim of the study.

5. Conclusions

Assessing the quality of motor performance at age 3 to 5 months considerably improves our chances not only of establishing the risk of an infant to develop cerebral palsy but also of predicting the later level of functional ability on the GMFCS. Apart from the absence of fidgety movements, a cramped-synchronised movement character, repetitive opening and closing of the mouth as well as abnormal finger postures were found more often in infants who were later rated as GMFCS levels IV and V. By contrast, a normal posture, absent fidgety movements, and an abnormally jerky but neither monotonous nor cramped-synchronised overall movement character were identified in infants who were later classified with GMFCS level I or II.

Detailed assessment of movement and postural patterns at 3 to 5 months of age is indeed the most time-consuming and it requires a lot of expertise, but then it enables a more precise identification of infants with a high risk of CP. Apart from establishing abnormal movements it is also important to evaluate which motor patterns are still carried out normally, especially in view of early intervention, as the focus of treatment has shifted from decreasing impairments to promoting normal functions.

Conflict of interest

None to declare.