Main

Preterm infants frequently show a hyperextended posture of neck and trunk, a phenomenon that is also called "transient dystonia"(1,2). The hyperextended posture can be a precursor of cerebral palsy, but usually this is not the case(1,3). Recent studies have shown that a hyperextended posture interferes with the development of arm and hand function(2,4), social interaction and communication(5), and cognitive development(6). It has also been suggested that the presence of a hyperextended posture during infancy is related to motor and learning problems at later age(1,7–8). This suggestion is supported by the high prevalence figures of hyperextension during infancy and those of learning disorders and coordination and balance problems at school age and adolescence in preterm children(9–13).

Little is known about the underlying neural mechanisms of the hyperextension. It has been suggested that disturbances in muscle tone regulation, mainly affecting the axial muscles, are responsible for the abnormal posture in the preterm infants(2–3,14). Not only is knowledge on the mechanisms of hyperextension scarce, we also know relatively little about the development of postural control in preterm children.

Hadders-Algra et al.(15) recently studied with the help of so-called perturbation experiments the development of postural control in preterm children with and without lesions of the PWM between 1½ and 4½ y of age. They found that the basic organization of the postural adjustments in response to sudden destabilizations was usually intact. This means that perturbations inducing a forward sway of the body elicited a primary response in the dorsal muscles, and those inducing a backward sway of the body evoked a primary response in the ventral muscles ("direction-specificity"). But the postural adjustments of the preterm children were not entirely normal, they were characterized by two types of dysfunction. In infants with a PWM-lesion, a limited repertoire of postural responses was observed. In preterms without a PWM-lesion a change in the ability to modulate the postural responses occurred. The preterm children showed a higher sensitivity to the velocity of the perturbation than the full-term control children, and they lacked the capacity to modulate EMG-amplitude with respect to initial sitting position. This was interpreted as a shift from the normally developing control that is guided by feedforward processes based on prior experience, to a form of postural control that is dominated by feedback mechanisms.

The development of postural adjustments during voluntary movements has not yet been studied in preterm infants. Recently, we studied the development of postural adjustments accompanying reaching movements in healthy full-term infants between 3 and 18 mo(16). The first successful reaching movements, present at 4-5 mo, are accompanied by postural adjustments with a large variation and a specific organization. From the age onward, the postural adjustments are direction specific (activity in the dorsal trunk muscles precedes ventral trunk muscle activity), are characterized by a co-activation of the neck muscles, have a top-down recruitment order (the neck muscles are activated before the trunk muscles), and are dependent on the position of the infant (lying supine and varying sitting positions). Around the age of 6 mo, postural activity during reaching is temporarily low. This transient period of little postural activity probably serves as a transition phase in postural development, as after the transition the ability emerges to scale postural adjustments with respect to arm movement velocity and the amount of forward or backward rotation of the pelvis.17 Between 12 and 15 mo of age another transition takes place. From 15 mo onward children develop anticipatory postural control, as reflected by the consistent activation of one of the postural muscles before the arm muscle that initiates the reaching movement.17 In addition, children are, from this age onward, able to adjust postural activity to an increased task-load. The latter was demonstrated by the finding that the infants could adapt the output of the postural muscles to a situation in which the reaching arm was loaded with a small, weighty bracelet.17 The developmental changes in postural activity were related to the development of standing and walking skills, but not to the development of sitting abilities.17

Our purpose was to investigate the development of postural adjustments accompanying reaching movements in preterm infants. Therefore, we longitudinally studied 12 preterm infants without cerebral palsy between the (corrected) ages of 4 and 18 mo. During each session we recorded surface EMGs of arm, neck, trunk, and leg muscles while the infant was placed in various positions. We tested the following hypotheses: Reaching movements in preterm infants are accompanied by postural adjustments which 1) are direction specific, 2) are recruited in a top-down order, 3) show a reduced amount of co-activation in the neck muscles, 4) lack features of anticipatory postural control, and 5) cannot be scaled with respect to task-specific parameters, such as arm movement velocity and sitting position. In addition, we hypothesized that postural dysfunctions, if present, would be related to gestational age at birth, lesions observed on neonatal sonographic brain scans, signs of hyperextension of neck and trunk, and neurologic outcome at 18 mo of age.

METHODS

Subjects. Twelve preterm infants (five boys and seven girls) participated in the study. They were born between 26 and 32 wk (median value: 30 wk) PMA and admitted to the Beatrix Children's Hospital in Groningen. Birth weight ranged from 850 to 1910 g (median value: 1230 g) and was appropriated for gestational age in all but one of the infants. The infant who was small-for-gestational age at birth caught up and continued her growth from 6 mo of age along the 25th percentile. Additional clinical details are provided in Table 1. None of the infants developed cerebral palsy.

Table 1 Clinical details of preterm infants
Full size table

The infants were assessed longitudinally at the corrected ages of 4, 5, 6, 8, 10, 12, 15, and 18 mo. Seven infants entered the study at 4 mo (infants A, B, C, G, H, I, and J), four at 5 mo (D, E, F, and K), and one infant at 6 mo (infant L). Due to the different ages of study-entry, to parental holidays and to technical problems during some recordings the number of recordings at each age varied between 7 and 10 (median value: 8.5). Ten full-term infants (6 boys and 4 girls) with a normal neurological development served as controls. They were born between 37 and 43 wk PMA (median value: 40 wk). Their birth weights were appropriate for gestational age and ranged from 2870 to 3850 g (median value: 3265 g). The study was approved by the Medical Ethics Committee of the University Hospital Groningen. All parents gave informed consent.

Procedure. We tested the infants in various positions: lying supine in an infant bed (LY), sitting semi-reclined at an angle of about 45° in a so-called maxi-cosi infant seat (SR), upright sitting in an infant chair (US), and from the moment the infants were able to sit without help "long-leg" sitting (legs semi-flexed) without support on a flat surface (LL). Reaching was elicited by presenting small, attractive toys in the midline at arm length distance. The toys were presented when the infants were in a quiet and alert behavioral state. From 10 mo onward task-load effects were studied by adding small weights (56 g) in the form of a bracelet around both wrists of the infant in the upright sitting position (LOAD). When the infant became fussy or tired, the session was shortened. This resulted in a smaller number of trials or in the absence of data in a particular condition. On average, the recording session lasted about 45 min.

Surface EMGs were recorded with the help of a software program for long-lasting recordings (POLY, Inspector Research Systems, Amsterdam, The Netherlands; sampling rate 1000 Hz). Disposable bipolar electrodes (Twente Medical Systems) were placed over the DE, PM, BB, TB, NF (sternocleidomastoideus), NE (at the C7 level), RA, LE (at the L4 level), the RF, and the HAM muscles at the right side of the body. Simultaneously, the whole session was recorded on video. The video registrations and the EMG recordings were time-coupled. After each session body length and body weight were recorded and a standardized neurologic examination was carried out based on the method of Touwen(18) by an examiner who was unaware of the EMG data. During this examination special attention was paid to postural function, such as the presence of hyperextension of neck and trunk, and the development of sitting, standing, and walking abilities. The neurological assessment at 18 mo resulted in a classification of either normal or MND, the latter denoting the presence of minor neurological dysfunction that did not result in an overtly disabling condition.

Serial brain ultrasound scans were in general made with a high-resolution 7.5 MHz transducer. The infants were scanned during the first months after birth at varying intervals, ranging from 3 d until 2 wk. Periventricular hemorrhages and leukomalacia in the preterm infants were classified according to the methods of Levene et al.(19) and De Vries et al.(20), respectively.

Data analysis. The video recordings served two goals. First, they were used to select those reaching movements that were appropriate for further analysis (e.g. a reaching movement on the side of the EMG electrodes). Second, an observer scored the onset of the reaching movements, arm movement velocity (slow, moderate, or fast), the position of the pelvis at movement onset (anteflexion, neutral, or retroflexion), and the absence or presence of trunk rotation. The video data were stored in a computer program (TAPETIME) and subsequently merged with the EMG data.

For the EMG analysis we used a computer algorithm, which was developed for the detection of phasic muscle activity (see for detailsRef. 21). The EMG analysis resulted in four parameters: 1) the frequencies of postural muscle activation and the frequencies of so-called complete postural response patterns (patterns consisting of the in concert activation of NE+NF+LE+RA, NE+NF+LE, or NE+LE+RA; seeRef. 17). The frequencies were calculated by dividing the number of trials during which a specific muscle or pattern was activated by the total number of trials in a specific condition, 2) the pattern variation index, which was obtained by dividing the number of different response patterns by the total number of trials in a specific condition, 3) the median latencies of activation of the postural muscles with respect to the activation of the first arm muscle (usually the DE), and 4) the median amplitudes of postural muscle activity for each infant at all ages and in each position. Data were only included when at least three trials per position were available.

Nonparametric tests were used to analyze differences in the EMG data (relative frequencies, latency, and amplitude values) between the various conditions (Wilcoxon), between the preterms and full-terms (Mann-Whitney), and between the ages (Mann-Whitney and Kruskall-Wallis). The Pearson correlation was used to evaluate the correlation between arm movement velocity and sitting position on the one hand and EMG data on the other hand. The various relations between clinically observed neuromotor behavior and the postural muscle activity parameters were controlled for age by using a partial correlation analysis. Throughout the analyses, differences with a p < 0.05 and correlations with a p < 0.01 were considered statistically significant.

RESULTS

Motor development. Motor development tended to be delayed in the preterm infants compared with the full-term infants, but the differences did not reach levels of significance For instance, at the age of 4 mo three of nine preterm infants could grasp successfully, whereas seven of nine full-terms could (Fisher, p = 0.07, NS). Similarly, at 8 mo of age 4 of 12 of the preterms could sit independently, compared with 7 of 10 full-terms (Fisher, p = 0.08, NS). Reaching movements in the full-term infants were accompanied by trunk rotation from 8 mo onward. In the preterm group only 6 of the 12 infants showed such a trunk rotation (Table 1; Fisher, p = 0.01), especially those infants without PVH (Fisher, p = 0.04).

Hyperextension of neck and trunk was present in 6 of the 12 preterm infants during the early phases of the study. In two of them the hyperextension lasted until 8 mo and in the other four until 10 mo of age (Table 1). The preterms with hyperextension tended to develop their motor milestones somewhat later than the preterms without hyperextension, but the differences between the small groups did not reach statistical significance. Hyperextension of neck and trunk was not related to the presence of brain lesions on the neonatal ultrasound scans and to the outcome at 18 mo (Table 1). At 18 mo five preterm infants showed minor neurologic dysfunction, whereas none of the full-terms did (Fisher, p = 0.03).

The development of postural adjustments. We focus on the EMG data in the upright sitting position. The postural adjustments accompanying successful reaching movements and those during reaching attempts did not differ significantly. Therefore these data were pooled.

The reaching movements of the preterm infants were accompanied by a remarkably high, but variable amount of postural activity at all ages (Fig. 1). The degree of variation in postural muscle activity, reflected by the pattern variation index, was similar for preterm and full-term infants. But throughout the age period studied the activation rates of the postural muscles in the preterms were higher than those in the full-term infants. As with the full-term infants, the preterms most frequently activated NE, NF, and LE (median frequencies: in preterms 80-100% and full-terms 60-80%; p < 0.05 for NE and, from 10 mo onward, for LE), and less often RA, HAM, and RF (preterms 30-50%, full-terms 0-20%; p < 0.05 for RA from 4 to 8 mo of age). The high rate of postural activity in preterms was also reflected by a high rate of occurrence of the complete response patterns. From 5 mo onward, the frequency of complete response patterns was significantly higher in the preterms than in the full-terms (Fig. 2; p < 0.05). This high rate of postural activity in the preterm infants was associated with an absence of the normally occurring transient decline in the occurrence of the complete response patterns around 6 mo of age. Moreover, due to the persistently high rates of complete response patterns significant relationships with the development of standing and walking abilities could not be found.

Figure 1
figure 1

Postural activity during reaching movements in the upright sitting position. Typical examples of postural activity accompanying a reaching movement during upright sitting of a full-term infant (left panel) and of preterm infant E (right panel) at the age of 12 mo. EMG activity of the following muscles is displayed: DE = deltoid, NF = neck flexor, NE = neck extensor, RA = rectus abdominis, LE = lumbar extensor, RF = rectus femoris, and HAM = hamstrings. Electrocardiac activity is present in RA and LE. The solid vertical line denotes the onset of the arm movement as scored on the video, and the dotted vertical lines mark the window in which phasic EMG bursts are detected. The horizontal bars delineate the presence of significant EMG burst as defined by the computer algorithm. Note the presence of reciprocal activity in the neck muscles and the absence of direction specific trunk muscle activation in the EMG pattern of the preterm infant.

Full size image
Figure 2
figure 2

Developmental changes in the frequencies of "complete" postural response patterns. The development of "complete" postural response patterns (NE+NF+LE+RA, NE+NF+LE, and NE+LE+RA) during reaching movements in the upright sitting position. The filled bars represent the group ranges of the frequencies in the preterm infants, the open bars represent the group ranges of the full-terms. The median group values are displayed by the horizontal lines. A significant decrease was found in the amount of complete postural response patterns in the full-term infants between 4 and 6 mo of age (p < 0.05; Fisher), which was followed by a significant increase until 18 mo (p < 0.05; Kruskall Wallis). Asterisks indicate significant differences between the two groups of infants: *p < 0.05, **p < 0.01 (Mann-Whitney).

Full size image

In full-term infants postural activity varied with testing position, for instance the neck muscles were predominantly used in the semi-reclined position. Such a position dependency was absent in the preterm infants. In addition, visual inspection of the EMGs suggested that the preterm infants exhibited a higher level of tonic background activity in NE and LE than the full-term infants did. This did not result in significantly higher rates of extensor inhibition than those observed in the full-term infants.

The analysis of the latencies focused on the neck and trunk muscles, as they were most frequently activated during the reaching movements. At all ages and in both groups of infants, the postural activation latencies showed a large variation (individual values ranging between - 188 and 2366 ms). The preterm infants tended to activate NF later, and RA earlier than the full-terms did, but the differences only occasionally reached statistical significance (NF at 10, 15, and 18 mo). Closer inspection of the temporal organization revealed, however, four differences between the two groups (Fig. 3). The first difference concerned the coordination of neck muscle activity. Full-term infants consistently co-activated the NE and NF muscles (median latencies for both muscles around 200 ms), but such a co-activation was only present in preterm infants younger than 8 mo of age. From 10 mo onward preterm infants activated the NE substantially earlier than the NF (median latencies of NE around 200 ms and of NF around 600 ms). The second difference was observed in the activation latencies of the trunk muscles. From early age onward the full-term infants activated LE and RA in a direction- specific way (LE on average 400 ms before RA). In preterm infants this form of trunk muscle coordination developed first at 12 mo of age, but the latency differences between LE and RA remained smaller than in the full-term infants (around 250 ms). The third difference was found in the development of the top-down recruitment order. A top-down recruitment of the postural muscles was present in full-term infants from 4 mo onward, but in the preterms only from 15 mo on. The fourth difference concerned the effect of position on the latency values. In full-term infants of at least 10 mo of age, the LE muscle was activated earlier during long-leg sitting than during sitting upright in the infant chair. Such a position dependency of LE-latency was absent in the preterm infants. Remarkably, preterm and full-term infants did not differ in the development of anticipatory postural activity. Both groups developed this ability at 15 mo of age.

Figure 3
figure 3

Latencies of neck and trunk muscle activity. Group data on the latencies of the neck and trunk muscle activation during upright sitting at 10 and 18 mo in the preterm infants (filled bars) and the full-term infants (open bars). The vertical line (t = 0) denotes the onset of EMG activity in DE. The data are presented by the ranges (horizontal bars) and the median values (vertical lines) at the various ages. Asterisks denote statistically significant differences between the activation latencies of the two groups: *p < 0.05 (Mann-Whitney). Additionally, significant differences were present at both ages in the full-term infants between the latencies of LE and RA (p < 0.05; Fisher) and between NE and LE (p < 0.05; Fisher at 10 mo, Wilcoxon at 18 mo).

Full size image

Unlike the full-term infants, the preterms did not develop the ability to modulate their postural output with respect to arm movement velocity, the pelvis position at the onset of the reaching movement, or the LOAD condition.

The postural dysfunctions found in the preterm infants did not show a consistent relationship with the gestational age at birth, birth weight, body weight at the age of assessment, the presence of neonatal complications, the presence or absence of trunk rotation and hyperextension of neck and trunk, the lesions found on the neonatal ultrasound scans of the brain (Table 1), and the neurological outcome at 18 mo.

DISCUSSION

Our study demonstrated that the organization of postural adjustments accompanying reaching movements in relatively healthy preterm infants differs from that in term infants. Preterm infants showed an excess of postural activity. Moreover, their postural activity showed temporal disorganization and was not adapted to task-specific conditions.

Neurophysiological considerations. Increased levels of muscle activity during the development of postural adjustments in preterm infants has not been previously reported. The finding is, however, in agreement with the EMG findings during general movements (GM) of preterm infants. Healthy preterms often show mildly abnormal GM during the first months after birth and this is associated with an increased risk for the development of minor neurological dysfunction, attention problems, and aggressive behavior at school age(22). The EMG patterns of mildly abnormal GM are characterized by excessive activity, which is reflected by increased durations and amplitudes of the phasic EMG bursts(23). Hadders-Algra and Groothuis(22) hypothesized that these EMG findings point to dysfunctions in the monoaminergic (dopaminergic and serotonergic) systems. Possibly, such dysfunctions result from minor hypoxic insults to the vulnerable preterm brain, as recent animal experiments indicated that chronic or repetitive mild hypoxic insults may induce long-term changes in the striatal dopaminergic system(24,25). Theoretically, high EMG activity might point to the activation of a high proportion of fast twitch muscle fibers, but as surface EMG primarily reflects motoneuronal recruitment, this explanation is less likely(26). Others(27) have speculated that changes in muscle fiber composition might underlie the clinically observed postural dysfunctions of preterm infants, but up until now evidence pointing in that direction has been lacking.

The high levels of postural activity were associated with an inability to scale the postural adjustments with respect to task-specific conditions, such as initial sitting position and arm movement velocity. In full-term infants such a modulatory capacity emerges around the age of 8-9 mo(17,28), indicating that from this age onward healthy full-terms are able to use feedforward processes, based on information from prior experience, in the organization of postural responses. Hadders-Algra et al.(15) showed that preterm birth is associated with a reduced capacity to modulate postural adjustments. The latter study demonstrated that preterm children, with and without brain lesions, cannot modulate postural responses during platform perturbations with respect to the initial sitting position. In addition, the preterm children showed an increased sensitivity to the rather crude stimulus of platform velocity. On the basis of these findings the authors argued that postural control in preterm infants shifts from a feedforward control based on prior experience, to a form of postural control that is dominated by feedback mechanisms. Our study suggests that the altered modulatory capacity of the preterm children is not necessarily based on a deficit in feedforward processing, as the preterm infants did develop normal anticipatory postural control. It seems more likely that the altered modulatory capacity can be due to a failure to adequately learn from prior experience, or to put it in other words, to a deficiency in the formation of an appropriate sensorimotor memory. It is conceivable that inadequately organized sensorimotor integration might play an important role in the later development of clumsiness in preterm children, such as has been claimed for older children by Ayres(29).

The temporal aspects of the postural adjustments in preterm infants were markedly disorganized. A similar temporal disorganization has been reported during mildly abnormal GM(23), but it was not found in the postural patterns in response to platform perturbations(15). Possibly, the temporal disorganization is only present in self-generated motor activity and not during reactive motor control.

The postural dysfunctions of the preterm infants were not clearly related to the clinical data. The absence of such a relationship may be due to the small sizes of the subgroups. This might be the explanation for the absence of a relation between the presence of hyperextension or the outcome at 18 mo on the one hand, and the postural data on the other hand(3). However, the absence of a relationship between the findings on the ultrasound brain scans and the postural data, is in agreement with the finding of postural dysfunction in preterms without brain lesions(15). But it is possible that the previously mentioned minor insults to the striatum escape the sonographic view.

Clinical considerations. The presence of postural dysfunction in relatively healthy preterm infants might be due to minor repetitive hypoxic insults of the preterm brain. This indicates that postural development might improve when the care of preterm infants is further optimized, for example, by means of accurate Po2 monitoring or by mimicking the intra-uterine environment by means of nest-like body support(30).

Our study demonstrates that relatively healthy preterm infants mainly show a deficit in the ability to fine-tune postural activity to task-specific constraints. Possibly, this disability is related to the continuous presence of high levels of postural activity, which might hamper the adequate processing of sensory information. Therefore, it is conceivable that a combination of relaxation and balance exercises could enhance proper postural development in preterm infants from about 5-6 mo corrected age onward. In turn, this might result in less clumsy behavior at later ages.