Predictive Value of General Movement Assessment for Cerebral Palsy in Routine Clinical Practice

Background

Early identification of children at high risk of future neurodevelopmental disability is important for the initiation of appropriate therapy. In research settings, the assessment of fidgety movements (FMs) at 3 months supports a general movement assessment (GMA) as a strong predictor for subsequent motor development, but there are few studies from routine clinical settings.

Objective

The study objective was to examine the relationship between FMs and neurodevelopmental outcome by the age of 2 years in high-risk infants in a routine hospital clinical setting.

Design

This was a prospective study.

Methods

A GMA was performed in 87 high-risk infants at 3 months after term age. The infants were clinically assessed for cerebral palsy (CP) at 2 years. Sensitivity, specificity, likelihood ratios, and positive and negative predictive values were computed. The relative risk of motor problems by the age of 2 years, according to the GMA, was estimated.

Results

Of the infants with normal FMs, 93% (50/54) had normal development and none was diagnosed with CP, whereas 75% (12/16) with abnormal or sporadic FMs had normal development. In contrast, 53% (9/17) of those without FMs had CP. When the GMA was considered to be a test for CP and absent FMs were considered to be a positive test result, the sensitivity was 90% and the specificity was 90%. The likelihood ratios for positive and negative test results were 8.7 and 0.1, respectively. The negative predictive value was 99%, and the positive predictive value was 53%. The risk of motor problems by the age of 2 years increased linearly with the extent of pathological results on the GMA and was 10 times higher when FMs were absent at 3 months than when FMs were normal.

Limitations

The relatively small study sample was a study limitation.

Conclusions

When applied in a routine clinical setting, the GMA strongly predicted neurodevelopmental impairments at 2 years in high-risk infants.

Preterm children, especially those with extremely low birth weight, are at increased risk of cerebral palsy (CP).¹ Several other factors, such as intraventricular hemorrhage and periventricular leukomalacia, also increase the risk.² Early detection of CP is difficult, and the diagnosis is often not established until the child is about 1 year of age or, for mild cases, even later.^3,4 Because intervention programs may be most beneficial early in life, when the nervous system is most plastic,^5–9 it is important to identify infants who are at the greatest risk of later motor impairment as early as possible. In both infants at low risk of CP and those at high risk of CP, the observation of general movements (GMs)—specifically, whether so-called fidgety movements (FMs) are present or absent at 3 months after term age—has been found to be a good predictor for subsequent motor development.^6,10–16

General movements are spontaneous movements that arise during early fetal life and persist until 3 or 4 months after birth.¹⁷ General movements have different age-specific periods; the last phase (the fidgety period) is characterized by FMs. As described by Prechtl et al,⁶ normal FMs are an ongoing stream of circular movements described as small-amplitude, moderate-speed, and variable accelerations of the neck, trunk, and limbs in all directions. Fidgety movements can be observed when the child is awake, alert, and lying supine. They may be superimposed on other movements, such as reaching and grasping, leg lifts, and trunk rotation, or other movements may occur during pauses between FMs.¹⁷ Fidgety movements are present from 6 to 9 weeks until 16 to 20 weeks after birth, although they are most prevalent at 12 weeks. Initially and at the end of the fidgety period, FMs may appear as sporadic FMs, which are like FMs but are interspersed with long pauses.¹⁸ Impairment of GMs can take the form of abnormal FMs. These are movements that look like normal FMs but have moderately or greatly exaggerated amplitude, speed, and jerkiness.¹⁷ Finally, FMs may be totally absent.

Several studies^6,10–16 have reported that the absence of fidgety movements in the fidgety period is strongly related to the diagnosis of CP by the age of 2 years, with a sensitivity of about 90% in both infants at high risk of motor impairment and those at low risk of motor impairment. Infants with abnormal FMs are also at increased risk of CP, but the relationship is substantially weaker than that for infants who did not have FMs.^6,19–22 However, infants with abnormal FMs have a higher risk of milder neurological dysfunction than infants with normal FMs.^6,20,21

At the University Hospital North Norway, Tromsø, Norway, assessments of GMs have been done over the last decade as part of a clinical evaluation of infants at high risk of future neurological disability. The role of a general movement assessment (GMA) has been to ensure that infants at high risk of developmental deficits are identified so that early intervention can be initiated. Previously, 4 studies^23–26 described as being performed in a clinical setting focused on the clinical value of assessing general movements. The results of those studies indicated that the GMA is a reliable and valid predictor^23,26 and improves the ability to identify infants at risk²⁴ and that at least 2 trained therapists should evaluate GMs for an accurate result.²⁵ However, those studies were performed in a research environment with dedicated time and resources. In contrast, a routine hospital clinical setting is hectic, less predictable, and more diverse, with many different tasks that may influence the assessment of GMs. Therefore, there is still a need for more knowledge regarding the degree to which a GMA performed in a routine hospital clinical setting predicts subsequent motor development. The aim of this study was to examine the relationship between FMs (normal, absent, abnormal, sporadic) and neurodevelopmental outcome by the age of 2 years in high-risk infants when assessments are done in a routine hospital clinical setting.

Method

Study Population

The neonatal intensive care unit at the University Hospital North Norway serves the 2 northernmost counties in Norway (Troms and Finnmark); the region is sparsely populated, with 230,000 inhabitants,²⁷ and has approximately 3,000 deliveries per year.²⁸ All high-risk infants from this area who had been admitted to the University Hospital North Norway from November 3, 2002, to October 11, 2010, were included in the study (n=196). High risk was defined as a birth weight of less than 1,500 g, a gestational age of less than or equal to 32 weeks, or both or other risk factors for CP, such as hemorrhagic infarction, periventricular leukomalacia, infections, and asphyxia. Infants who had tumors, severe malformations, or syndromes (n=6) were not included. The data collected were primarily used locally for quality control, and a waiver of informed consent was granted by the local data protection official. This article is thus a publication of the results of an internal quality assessment procedure.

For the 196 children included in the study, the families of 19 children moved out of the counties, and 4 children died during follow-up. Thus, 173 infants were eligible for the 2-year follow-up. Unfortunately, 64 were not examined during the fidgety period (17 were examined too late and 47 were not examined at all). Moreover, for technical reasons, data from 22 children were lost. These 86 children were considered to be nonparticipants. Participants were the 87 children for whom complete data concerning FMs at 3 months of age and the diagnosis of CP or the development of functional disability at 2 years of age were available.

Baseline Registrations

The perinatal characteristics of the infants included gestational age, birth weight, and diagnosis in the neonatal period.

Assessment of GMs at 3 Months

As part of the follow-up program at the hospital, one 5- to 10-minute videotaped assessment of GMs was obtained at the beginning of the clinical assessment at 3 months after term age. General movements were analyzed according to complexity, variability, location, speed, and temporal organization.¹⁷ In line with recommendations,¹⁸ a small digital video recorder was used. The recorder was placed high (approximately 1.5 m) above the infant lying in the supine position on a mattress on the floor. There were no toys in the surroundings, the parents were watching from the side, and the infants were awake and alert.

The recorded GMs were rated as described by Prechtl et al.⁶ Two physical therapists conducted the ratings. All ratings were assigned by therapists certified by the Prechtl GM Trust at a basic or an advanced level, and throughout the study, at least 1 of the 2 therapists was certified at an advanced level. Over the years, 4 different physical therapists were involved.

The ratings were assigned whenever there was time, according to the physical therapist's clinical work, at noon, at the end of a workday, or between treatments of other children. Interruptions and distractions due to unscheduled patient care work could occur. One of the physical therapists had no knowledge about the infants' medical records. The quality of movements was rated as follows: normal FMs, sporadic FMs, abnormal FMs, and absent FMs. The 2 physical therapists scored the movements separately and compared their classifications afterward. In case of disagreement, the observations were repeated and the findings were discussed before a consensus was reached. The observation periods were 1.5 to 2 hours long. Fidgety movements were assessed at a mean of 12.7 weeks (SD=1.5, range=9–16, median=12.0) after term age. The median postpartum ages of the children with normal FMs, sporadic FMs, and abnormal FMs and children who did not have FMs were 12, 13, 12, and 12 weeks, respectively.

Assessment of Outcome by 2 Years After Preterm Birth

A pediatrician with extensive experience in the field of neurology and a physical therapist performed a systematic clinical assessment of the children at 2 years after term age. This assessment consisted of anamneses and observations of movement patterns, spontaneous functional skills, postural reactions, active/ passive movements, and muscle tonus—that is, an assessment focusing on the quality of movements and milestones in typical motor development.²⁹ A CP diagnosis was made by the pediatrician in accordance with the definition of Rosenbaum et al³⁰ and was based on the clinical examination and information from the hospital medical records. A child was defined as having mild motor impairment if he or she did not meet the standard expectation of functional activities for his or her age; the general impression of the movements could be slightly increased or decreased tone, slightly abnormal patterns of posture and movement, or delayed motor development (ie, milestones were delayed 5 or 6 months or more). Motor development was recognized as age appropriate when functional movements and activities were of high quality (fluent, adjusted, and complex) and milestones were achieved within the proper age range.

Data Analysis

Differences between participants and nonparticipants were tested with the chi-square test and the Student 2-sample t test. The relative risk and 95% confidence interval (CI) for CP or mild motor impairment by the age of 2 years, according to the different categories of FMs at the age of 3 months, were estimated, as was the P value for the linear trend in relative risk, according to 3 levels of FMs (normal, abnormal or sporadic, and absent). Sensitivity, specificity, likelihood ratio, positive predictive value, and negative predictive value were calculated. A 2-sided P value of less than .05 was considered to be statistically significant. The data were analyzed with the 19.0 version of IBM SPSS for Windows (IBM SPSS, Armonk, New York).

Results

The mean gestational age and birth weight of the 173 infants eligible for follow-up were 30.5 weeks and 1,422 g, respectively. The 87 children for whom data were available on both the examination of FMs at 3 months of age and the assessment of possible functional disability by the age of 2 years did not differ from the other children enrolled in the study, except for a smaller proportion of boys (P=.04) in the former group (Tab. 1). In particular, the prevalence of CP by the age of 2 years did not differ between participants and nonparticipants (P=.33).

Table 1

Selected Characteristics of Participants and Nonparticipants^a

Characteristic	Participants(n=87)	Nonparticipants(n=86)	P
Boys, % (no.)	47 (41)	63 (54)	.04
Birth weight, g, X̅ (SD)	1,528 (1,045)	1,314 (713)	.12
Gestational age, wk, X̅ (SD)	30.7 (5.0)	30.2 (4.0)	.45
% (n) at gestational age of:			.21
≤28 wk	38 (33)	33 (28)
28.1–36.9 wk	46 (40)	58 (50)
≥37 wk	16 (14)	9 (8)
Diagnosis (no. of children)b
Hemorrhagic infarction	6	6
Asphyxia	25	31
CNS infection	4	1
Respiratory problemsc	36	27
Dysmaturity	4	5
Periventricular leukomalacia	3	3
Epilepsy	1	0
Diagnosis of CP by 2 y, % (n)d	12 (10)	7 (6)	.33

^a

Participants were children for whom complete data concerning fidgety movements at 3 months of age and the diagnosis of CP or the development of functional disability at 2 years of age were available; nonparticipants were 64 infants who were not examined during the fidgety period and 22 infants for whom data were lost. CNS=central nervous system, CP=cerebral palsy.

^b

Other than low birth weight (<1,500 g), a gestational age of less than or equal to 32 weeks, or both.

^c

Including bronchopulmonary dysplasia and neonatal respiratory distress syndrome.

^d

Information on 3 nonparticipants was missing.

A total of 54 infants had normal FMs, 8 had abnormal FMs, 8 had sporadic FMs, and 17 did not have FMs (Tab. 2). Table 2 shows the distribution of children with normal development, mild motor impairment, and CP by 2 years corrected age according to the 4 categories of FMs. Of the 54 infants with normal FMs, none was diagnosed with CP and 93% (95% CI=82%, 98%) had normal development. Furthermore, of the 16 infants with abnormal or sporadic FMs, 1 was diagnosed with CP, 3 were diagnosed with a mild disability disorder, and the remaining 12 (75%; 95% CI=48%, 93%) had normal development. On the other hand, 9 of the 17 children without FMs (53%; 95% CI=28%, 77%) were diagnosed with CP by the age of 2 years, and only 4 had normal development.

Table 2

Outcome by 2 Years According to 4 Categories of Fidgety Movements

Fidgety Movements	No. of Children
	Total (n=87)	With Normal Development (n=66)	With the Following Diagnosis by 2 y:
			Mild Motor Impairment (n=11)	Cerebral Palsy (n=10)
Normal	54	50	4	0
Abnormal	8	6	1	1
Sporadic	8	6	2	0
Absent	17	4	4	9

The fidgety test at the age of 3 months may be considered to be a test for CP at the age of 2 years. When absent FMs were considered to be a positive test result (and normal, abnormal, or sporadic FMs were considered to be a negative test result), the sensitivity was 90% (95% CI=56%, 100%) and the specificity was 90% (95% CI=81%, 95%). The likelihood ratio for a positive test result was 8.7 (95% CI=4.4, 17.2), and that for a negative test result was 0.1 (95% CI=0, 0.7). The negative predictive value was high (99%), but the positive predictive value was relatively low (53%). Of the infants who did not have FMs, who had abnormal or sporadic FMs, and who had normal FMs, 76% (13/17), 25% (4/16), and 7% (4/54), respectively, had some kind of motor problems (CP or mild motor impairment) by the age of 2 years. Thus, abnormal or sporadic FMs are also of concern. When any finding but normal (normal FMs) was considered to be a positive test result, the sensitivity with regard to a CP diagnosis was 100% (95% CI=69%, 100%), the specificity was 70% (95% CI=59%, 80%), and the likelihood ratios for positive and negative test results were 3.3 (95% CI=2.4, 4.7) and 0 (95% CI=0, 1.1), respectively. The positive predictive value was only 30%, but the negative predictive value was 100%.

As shown in Table 3, the risk of having motor problems (the CP and mild motor impairment categories merged) by the age of 2 years was 10 (95% CI=3, 36) times higher when FMs were absent than when FMs were normal. The risk was 3.4 times higher when FMs were abnormal or sporadic than when FMs were normal, but it was not statistically significant (95% CI=0.8, 13.5). However, the P value for the linear trend over the 3 levels of FMs (normal, abnormal or sporadic, and absent) was highly significant, strongly suggesting that with increasingly pathological findings, the risk of CP or mild motor impairment increases.

Table 3

Relative Risk of Motor Problems (Cerebral Palsy or Mild Motor Impairment) by 2 Years

Fidgety Movements	No. of Children		Relative Risk (95% Confidence Interval)a
	Total	With Motor Problems
Normal	54	4	1
Abnormal or sporadic	16	4	3.4 (0.8, 13.5)
Absent	17	13	10.3 (3.0, 35.8)

^a

The Pvalue for a linear trend was <.001.

Discussion

The present study showed that none of the infants with normal FMs during the fidgety period developed CP and that 7% had mild motor impairment by the age of 2 years. A total of 76% of the infants who did not have FMs and 25% who had abnormal or sporadic FMs during the fidgety period had some kind of motor dysfunction (CP or mild motor impairment) at 2 years after term age. Furthermore, the risk of a clinical pathological outcome was 10 times higher in infants who did not have FMs at 3 months after term age than in infants who had normal FMs. The corresponding risk in infants with abnormal or sporadic FMs was 3.4 times higher. However, given the wide CIs for the point estimates, the risk of some kind of motor dysfunction for infants with abnormal or sporadic FMs was not statistically significantly higher than that for infants with normal FMs. The P value for a linear trend over the 3 levels of FMs (normal, abnormal or sporadic, and absent) was highly significant, however.

Our results are in accordance with those of several previous studies^6,10–16 showing that the absence of FMs is associated with a poor outcome and that normal FMs in high-risk infants are associated with a low risk of developing CP. In line with our study, 3 studies^6,20,21 also examined the relationship between abnormal FMs and neurological dysfunction less severe than CP. Romeo et al²⁰ showed that 88% of infants with abnormal FMs had mild disability by the age of 2 years and 12% had CP. Prechtl et al⁶ found that 44% of infants with abnormal FMs developed mental retardation, motor retardation, or both; 37% had CP; and 19% had normal development. Ferrari et al²¹ reported that 5% of infants with abnormal FMs developed mild motor impairment, 84.5% had CP, and 10.5% had normal development. Although those studies showed that abnormal FMs strongly predicted both motor retardation and severe CP, we found a somewhat weaker association between abnormal FMs and motor dysfunction. In the present study, fewer infants developed mild motor dysfunction (12.5%) and CP (12.5%) and as many as 75% of the infants had normal development. The discrepancies may be due, in part, to problems related to the present study was 2 years; more children with abnormal FMs may have developed problems at later ages. Hadders-Algra et al³¹ reported that abnormal FMs at 3 months were significantly related to minor neurological dysfunction or behavioral disorders at the ages of 4 to 9 years. In line with the findings from previous studies, the results of the present study indicate that infants with abnormal FMs should be monitored closely for years.

An interesting result of the present study is that sporadic FMs may be associated with an increased risk of mild motor impairment but the risk of CP is probably low. There were few infants with sporadic FMs in our study sample, and further studies with larger samples and long-term outcomes at toddler and school ages are needed. Finally, even though our study sample consisted of infants known to be at a relatively high risk of physical disability, none of those with normal FMs developed CP and 7% had mild motor impairment by the age of 2 years. We believe that this knowledge is important for health care personnel but is most important for reassuring parents that such infants are unlikely to develop CP.

The results of the present study indicate that assessing FMs in a clinical setting strongly predicts later outcome. Thus, a GMA has a clear clinical value. However, the use of a GMA in clinical practice will never enable clinicians to predict an infant's development with certainty. We agree with Spittle,³² who suggested that a GMA must be used as a tool supplementary to infants' medical history, physical examination, parents' concerns and, optionally, other standardized tests to identify infants most at risk for subsequent poor developmental outcomes. With this approach, infants at risk can be monitored and early physical therapy may be initiated.

A strength of the present study is that it provides information on the association between abnormal or sporadic FMs and outcome in high-risk infants at 2 years of age on the basis of data from an ordinary clinical setting. The fact that our study sample was diverse—because it included both preterm and full-term infants—may also be considered a strength. However, when the GMA is used as a tool in clinical practice, it is essential to assess children at the proper time for observing FMs. In the present study, 1 of 3 infants had to be excluded from the study population. The main reason was that the infants were not given an appointment at 3 months of age. Furthermore, the relatively high percentage that were lost to follow-up may be related to the special geographic and demographic conditions in northern Norway. Some of the infants included in the present study lived as far as 900 km from the hospital, and rough weather and difficult driving conditions during a long winter may have prevented families from attending scheduled follow-up appointments. Additionally, an infant or a caregiver may have been ill. In a review, Darsaklis et al¹² called attention to the fact that several studies did not include information about the infants who were omitted from the analysis. Therefore, it is not possible to determine how representative the study samples were. There are no reasons to believe that the infants who were included in the present study differed in a clinically significant way from those who were not included (Tab. 1).

A limitation of the present study may be that different physical therapists and different pediatricians participated, reflecting routine clinical practice. However, because the assessments of FMs and motor problems were independent, the relationships found in the present study most likely were attenuated. Moreover, it is plausible that the reliability of the measures used in the present study was lower than that in many other studies and may have diluted the associations between FMs and outcome.

The likelihood ratio for a positive test result was high, nearly 9 (95% CI=4, 17). From a statistical point of view, including more children would have been an advantage because the CIs were rather wide. For example, to reduce considerably the 95% CI for the sensitivity (90%; 95% CI=56%, 100%) of absent FMs as a positive test for CP by the age of 2 years, 100 children with CP would have to have been included in the analysis. In that scenario, the 95% CI would have been 82%, 95%. We reported the 95% CIs only for sensitivity and specificity, not for positive and negative predictive values, because those measures are dependent on the prevalence of the condition. Care should be exercised in comparing the positive and negative predictive values found in the present study with those found in other studies.

Another limitation of the present study may be that our study sample did not include all infants hospitalized during the study period; it included only high-risk infants. However, we believe that this group of infants constitutes the most likely candidates for assessment in routine clinical work. However, further studies, particularly those with larger groups of infants with abnormal and sporadic FMs, are needed.

A GMA is easy and quick to perform during the fidgety period because it usually takes just 5 to 10 minutes. The assessment is noninvasive, and most infants accept being left alone in a supine position for a few minutes. No editing of the video is necessary, and the entire procedure of video recording and classification takes approximately 15 to 30 minutes per assessment. Thus, in theory, the assessment of FMs is not time-consuming.

In clinical practice, however, the assessment of FMs may be more difficult. According to Prechtl et al,⁶ it is sufficient for the evaluation of GMs to be performed by one therapist certified by the Prechtl GM Trust. In the training courses and instructional videos, the examples are mostly typical and distinct. In real life, the FMs in some infants may be different and not as typical and therefore may be more difficult to classify. Moreover, rating can be challenging because of the assessment's high prediction of the child's neurological outcome and the subjective nature of scoring. Therefore, we have decided that all of the ratings in our hospital are to be assigned by 2 therapists, with discussion about disagreements to reach a consensus. To achieve higher levels of accurate scoring, others have implemented similar procedures.²⁵ However, finding time for 2 therapists to meet during hectic daily clinical practice may be challenging and may result in delayed scoring.

The guidelines¹⁷ for analyzing GM recordings were not followed strictly in the present study. According to these guidelines, the assessors should be alert and not tired, no more than 45 minutes should be spent rating sessions without taking a break, and a recalibration through watching a standard normal recording should take place when many recordings of abnormal GMs in a series are being viewed. However, the physical therapists in the present study were rarely able to follow these recommendations and performed assessments for periods longer than 45 minutes. Moreover, they sometimes had to assess recordings in between other tasks or at the end of the workday, when they may have been tired. In addition, they rarely found time for recalibration. Despite these weaknesses, the results show that the GMA is a valid tool in clinical practice.

In conclusion, the present study shows that the assessment of FMs is a valuable tool for detecting subsequent motor problems early in life when performed in a routine hospital clinical practice. The risk of developing motor problems by the age of 2 years increases linearly with the extent of pathological FM findings and is 10 times higher if FMs are absent by 3 months of age than if FMs are normal. Hospital routines must allow for the assessment of high-risk children at the proper time for observing FMs.

The local data protection official at University Hospital of North Norway approved the study.

Author notes

Dr Øberg provided concept/idea/research design. All authors provided writing and data analysis. Dr Øberg provided data collection and project management. Dr Jacobsen and Dr Jørgensen provided consultation (including review of manuscript before submission).