The qualitative assessment of general movements in

preterm, term and young infants - review of the

methodology

Christa Einspieler”’ * , Heinz F.R. Prechtl”, Fabrizio Ferrarib,

Giovanni Cioni”, Arend F. Bosd
“Department of Physiology, Karl-Franzens-University, Graz, Austria
bDepartment of Paediatrics, Division of Neonatology, University Hospital, Modem, Italy
‘Institute of Developmental Neurology, Psychiatry and Educational Psychology, University of Pisa and
St&a Maris Foundation, Pisa, Italy
‘Department of Paediatrics, Division of Neonatology, Beatrix Children ‘s Hospital, University Hospital,
Groningen, The Netherkmds

Received 30 January 1997; accepted 27 August 1997

Abstract

We describe the state of the art of Prechtl’s method for the qualitative assessment of general
movements as a diagnostic tool for early detection of brain dysfunction. After discussing the
optimal technique for video recording general movements in preterm, term and young infants,
attention is focused on the proper analysis of this spontaneous motor pattern. Recently, a group
of active researchers in the field reached consensus on the various qualities of normal and
abnormal general movements. These definitions are reported here in full. Since it is a newly
introduced method careful investigation into its reliability is required. Various groups of
investigators have obtained data which demonstrate the robust character of the method
(interscorer agreement: 78-98%). Finally, we discuss the validity of this early assessment
method on the basis of the reports published so far. While the method’s sensitivity is similar in
all age groups studied (preterm, term, first month, second month, and third month age epochs),
and averages 94.5%, the specificity of the method is age-dependent. It is low during the early
ages, increases gradually and reaches 82 to 100% at 3 months post-term. This phenomenon is
explained by spontaneous recovery of early dysfunction. In contrast, consistent abnormalities
of general movements are linked to neurological deficits found at the 2 year follow-up.
0 1997 Elsevier Science Ireland Ltd.

*Corresponding author. Harmchgasse 21, A-8010 Gnu. Tel.: +43 316 3804266; fax: +43 316
3809630; e-mail: christa.einspieler@kfuaigraz.ac.at

0378-3782/97/$17.00 0 1997 Elsevier Science Ireland Ltd. All rights reserved

PII SO378-3782(97)00092-3
48 C. Einspieler et al. I Early Human Development 50 (1997) 47-60

Keywords: General movements; Spontaneous motility; Infant; Preterm infant; Neurological

assessment

1. Introduction

The young infant has a repertoire of distinct and spontaneous movement patterns
[7,14,23]. One set of these movement patterns is known as general movements
(GMs). GMs are complex, frequent and long lasting (for definition see Section 3.1.).
Changes in the normal quality of GMs are a reliable indicaror of brain dysfunction.
This holds true for the preterm and term ages as well as for the first 20 weeks
posttetm. Prechtl developed a method to assess the quality of GMs based on video
observation as a diagnostic tool for early detection of brain dysfunction [20).
The present paper summarises our experience regarding the optimal recording
procedure and describes the proper analysis of GMs. In addition, we provide the
definitions of the different GM qualities based on a recent consensus of opinion
amongst a group of active researchers in the field.
Since this method for assessing GMs consists of a judgement based on observation.
interobserver and test-retest reliability are crucial. Therefcre, we review previous
findings collected by various research groups and provide new data on a large study
on interscorer agreement. The existing results on the validity of GM assessment for
predicting neurological and developmental outcome at 2 years is also surveyed.

2. Video recording of general movements

The simplest way of assessing motor activity is by directly observing the

movements with the unaided eye. However, considerable improvement in the
assessment is achieved if the infant’s spontaneous movements are observed by
replaying a video recording. Undisturbed by other environmental impressions the
observer assessesthe video recording on the basis of his/her visual Gestalt perception
[20] of the movement quality. There is the added advantage of repeated playback.
even at different speeds, and of storing the recordings for documentation and future
reference.
For the assessment of GMs an adequate video recording is a prerequisite. We
recommend the observer to attend to the following details:

2.1. The infant’s position and behavioural state

Depending on its age the infant should lie in supine position in the incubator, the
infant’s bed or on a mattress on the floor. Recording infants lying on a table or baby
dressing table should be avoided. The for the infant’s safety necessary caretaker’s
presence will attract the infant’s attention and thus interfere with the infant’s
spontaneous activity.
 C. Einspieler et al. I Early Human Development 50 (1997) 47-60 49

If a preterm or term infant is lying in an incubator it could be filmed naked or

wearing a nappy only. During the postterm period infants should be dressed lightly
and comfortably, in for example a ‘body’. Usually, we leave the infant’s arms and
legs bare and avoid thick nappies. The room temperature should be comfortable,
fitting the infant’s age and clothing. If the ambient temperature is either too low or
too high it will affect the infant’s behavioural state and the movement quality.
Most important for the assessment of GM quality in infants older than about 36
weeks postmenstrual age is the correct behavioural state. Recordings should then
preferably be performed during active wakefulness (state 4, according to Prechtl
[17]). Younger preterm infants should be recorded when bouts of activity occur. For
the proper analysis of GMs it is contra-indicated to continue recording during
prolonged episodes of fussing and crying, during drowsiness and episodes of
hiccuping. Nor is it possible to judge the quality of GMs properly if the infant is
sucking on a dummy. Soothing the infant with a dummy results in the sucking
posture (flexed arms, fisting, and extended legs). In case of prolonged fussing and
crying the recording must be stopped. Only restart the recording after the infant has
been soothed. This may take some time. It might be necessary to repeat the recording
session on another day. However, it should be mentioned that infants suffering from
severe brain dysfunction may frequently cry after they start moving.
To obtain recordings of spontaneous motor activity it is a necessity to avoid every
possible interference by an observer (parents, examiner) and, in older babies, the
presence of any toy in the immediate surroundings. Watching a monitor outside the
observation room is a useful way of observing the baby without causing interference.
In this way, parents can be asked to soothe their baby if it starts crying because then
the recording must be interrupted anyway.
We advise against putting the infant in front of a mirror as sometimes used in
recording or treatment rooms. The mirror image, acting as a ‘twin’, destroys the
observer’s Gestalt perception. Older infants are immediately attracted by their ‘twin’,
touching the mirror and smiling towards it which, of course, interferes with true
spontaneous motor activity.

2.2. Position of the camera and duration of the recording

The best view of the baby is obtained by filming it mid-sagitally or laterally from
above. In any case, the observer must be able to see the infant’s face to make sure,
for example, that jerky movements are not due to crying. The camera should be
placed high above the infant to avoid irrelevant and uncalled-for exposure to the
camera of the infant’s private parts.
Particularly for infants older than 1 month, it is necessary to prevent the camera
from attracting the infant’s attention. This is not easy when using such a big, black
object with blinking lights. There are two possibilities to cope with this problem:
either leave the camera next to the infant’s cot until the baby is habituated to it or use
a single chip camera which is so small that the infant will be unaware of its presence.
Another solution mentioned by Geerdink and Hopkins [ 131, is to use a camera hidden
above a purpose-built bed surrounded by white curtains.
50 C. Einspieler et al. I Early Human Development 50 (1,797) 47-60

The duration of the recording depends on the age of the infant. In order to collect a
sufficient number of GMs for reliable judgement we usually record preterm and term
infants for about 1 h. Ten minutes of optimal recording are usually sufficient for older
infants.
We advise not to record GMs during the first 3 days after birth. Mainly during
these days, many physiological variables fluctuate more than later. There is also an
initial instability of behavioural states changing rapidly from quiet sleep to crying
which can interfere with a proper observation of GMs. It is therefore advisable to
avoid this period of instability if not otherwise indicated.
For the selection of GMs from the recording and the subsequentanalysis, a time
code signal superimposedon the tape (minutes and seconds)is most helpful.

2.3. Selection of general movements

Video recording and the analysis of GMs should be performed longitudinally. The
1 h recordings of the preterm and term period should be replayed at high speedand
three sufficiently long examples of GMs (preferably from the beginning, the middle
and the end of the recording) should be identified and copied onto a separatetape.
Including the beginning and the end of each GM is important for the assessment.
There are no other criteria for the selection of the particular GMs. The selection ot’
more than one GM is important. The necessary redundancy increasesthe correct
judgement. It should be noted that according to our experience. abnormal GMs arc
consistently abnormal during one recording.
In addition, three to 5 min of each recording of the awake and moving postterm
infant should be copied onto the new tape. This new tape documentsthe developmen-
tal courseof spontaneousmotor activity for each infant. We also advise adding a brief
recording of the outcome (for example at 2 years of age).
Thus, an optimal summary tape of the individual developmental course of GMI;
consists of (a) several recordings of the preterm period, preferably one for every
week; (b) one recording at term age; (c) one recording between 3 and 6 weeks; td) at
least one recording during the period when fidgety movements(seeSection 3.1) come
ro full expression (i.e. 9 to 15 weeks postterm age): two or three recordings durinp
this age are desirable, however.

3. Analysis of general movements

The analysis should be done from the summary tape. The examiner should never
assessfor longer than 45 min at a time without taking a break: he or sheshould watch
normal GMs in between the assessmentsessionsin order to ‘re-calibrate’ his or her
Gestalt perception. For this purpose a tape with the gold standard per age group
should be at hand. (We advise to collect a seriesof recordings of normal GMs in
infants at various ages’).

‘Information about a demonstration video can be requested from the tirst author.
 C. Einspieler et al. I Early Human Development 50 (1997) 47-60 51

3.1. Definitions of general movements

GMs can be observed in fetuses as young as 9 weeks postmenstrual age [24]. In

infants without neurological dysfunction, GMs continue in a similar pattern until
about the end of the second month postterm, which is then followed by a gradually
emerging new GM pattern.
Normal GMs are ‘gross movements, involving the whole body. They may last from
a few seconds to several minutes or longer. What is particular about them is the
variable sequence of arm, leg, neck and trunk movements. They wax and wane in
intensity, force and speed, and they have a gradual beginning and end. The majority
of sequences of extension and flexion movements of arms and legs is complex, with
superimposed rotations and often slight changes in direction of the movement. These
added components make the movements fluent and elegant and create the impression
of complexity and variability’ [20].
During term age and during the first postterm months, GMs are commonly referred
to as writhing movements [14,23]. GMs of a writhing pattern are characterised by
small to moderate amplitude and by slow to moderate speed. Fast and large extensor
movements may occasionally break through, particularly in the arms. Typically, such
movements are elliptical in form; this component creates the impression of a writhing
character of movement.
Types of abnormal GMs during preterm, term and early postterm age (first 2
months) are:

Poor repertoire of GMs: A sequence of the successive movement components is

monotonous and movements of the different body parts do not occur in the complex
way as seen in normal GMs [12,21].

Cramped-synchronised GMs: These appear rigid and lack the normal smooth and
fluent character; all limb and trunk muscles contract and relax almost simultaneously
[12,21].

Chaotic general movements: Movements of all limbs are of large amplitude and occur
in a chaotic order without any fluency nor smoothness. They consistently appear to be
abrupt [5].

At the age of 6 to 9 weeks postterm the form and character of GMs of normal
infants change from the writhing type to a fidgety pattern [14,23]. Fidgety Movements
(FMs) are circular movements of small amplitude and moderate speed and variable
acceleration of neck, trunk and limbs in all directions. They are continual in the
awake infant, except during focused attention. They may be concurrent with other
gross movements, such as kicking, wiggling-oscillating. [14] and swiping of the arms
1141 or ‘pleasure bursts’ (i.e. simultaneous kicking with the legs and downwards
swiping with the arms, frequent spontaneous smiling). FMs may be seen as early as 6
weeks postterm but usually occur around 9 weeks and are then present until 15 to at
most 20 weeks. This age range holds true for term as well as for preterm infants after
52 C. Einspieler et al. I Early Human Development 50 ‘1997) 47-60

correcting the age. Initially they occur as isolated events (Score: + ); they gradually
increase in frequency (Score: + + ) and then decrease once again (Score: f ) 12 I 1

FMs are judged as abnormal if they are:

(a) Absenr: FMs are never observed from ages 6 to ;!O weeks postterm. Other
movements can, however, be commonly observed [2 I].
(b) Abnormal: they look like normal FMs but their amplitude, speed and jerkiness
are moderately or greatly exaggerated [21].

3.2. Global judgement and individual developmental trajectories

In his paper ‘Gestalt Perception as a Source of Scientific Knowledge’ the Nobel

prize laureate Konrad Lorenz pointed out that “Gestalt perception is able to take into
account a greater number of individual details and more relationships between these
than in any rational calculation” (p 312) [16]. Gestalt perception is very vulnerable to
attention to detail but a powerful instrument in the ana1ysi.i of complex phenomena
[16]. Our visual Gestalt perception is a complex technique for a global judgement of
the quality of GMs [20]. Are certain GMs complex, fluent and variable, in other
words are they normal or does the absence or reduction of complexity and variability
indicate an abnormal quality? Thus, the first step in the judgement should be
concerned only with identifying normal ( ‘N’) versus ubnormal. Abnormality should
be specified in terms of ‘poor repertoire’ ( ‘PR’), ‘cramped synchronised’ i ‘CS ’ ).
‘chaotic GMs’ (‘Ch ‘), ‘abnormal FMs (‘AF ‘) or ‘absent FMs’ ( ‘F - ‘). respective-
ly.
As mentioned in Section 2.3., the assessment of the quality of GMs should be
repeated at different ages. The individual developmental trajectory obtained in this
way indicates the consistency or inconsistency of normal or abnormal findings. A
graphic display of a particular individual developmental trajectory is easily con-
structed (Fig. 1 and [8,12,22]). “Individuals with similar developmental trajectories
can then be grouped and of course, it is of considerable interest to discover whether
or not the outcome is similar in cases with similar developmental trajectories” [ 201.

3 * CS PR PR PR PR AF AF DR
4 * PR Ch Ch Ch PR PR PR AF AF
5 * PR PR CS CS CS CS CS F- F- F-
6 * cs cs cs cs cs CS CS CS F- F--.-A-LIP..-.

Fig. I. Examples of different individual developmental trajectories. t moment of birth; N, normal general
movements or normal outcome; PR, poor repertoire of general movemen-s; CS, cramped-synchronised
general movements; AF, abnormal fidgety movements; DR, developmental retardation; CP. cerebral palsy:
Ch, chaotic general movements; F-, absent fidgety movements.
 C. Einspieler et al. I Early Human Development 50 (1997) 47-60 53

We can now answer this question in the affirmative [3,5,21,22]. The close,
longitudinal surveillance of GM quality has an additional advantage: in affected cases
the developmental course of early and later symptoms of cerebral palsy can be fully
documented. This approach has provided new data on the natural history of
developing cerebral palsy [ 12,20-221.

3.3. Detailed analysis of general movements

Having carried out the global judgement, it can be worthwhile to look at different
aspects and components of GMs, especially if they are abnormal. Ferrari et al.
attempted such an analysis for preterm infants. They used a list of criteria including
amplitude, speed, movement character, sequence of GMs, range in space, onset and
offset of GMs and kinds of finger movements [12] (this list is also included in the
paper by Ferrari et al., this Issue). In addition, a semi-quantification of the GM quality
was achieved by applying Prechtl’s optimality concept [18]. A score of 2 is given for
every optimal aspect (e.g. ‘variable speed’) while the non-optimal aspects (e.g. a
‘predominately slow speed’) score a 1. Thus, the higher the score the more optimal
the movements and the lower the score the more reduced the optimality of GM
quality is. Interestingly, in none of the cases more than one or two of the separate
aspects were scored as optimal if the previous, independently obtained, global
judgement was abnormal.
Einspieler [lo] adapted this semi-quantitative scoring system for older infants
(3-12 weeks, and 13-26 weeks postterm age) by calculating a ‘motor optima&y
score’. Such a score can be easily used for further statistical treatment, e.g. by
comparing the quality of GMs with data obtained from polysomnographical record-
ings [lO,ll] or from biochemical analysis. An obvious limitation, however, is that
this detailed analysis does not allow a re-synthesis from the description of the details
to the total picture. Too much Gestalt is lost with this semi-quantitative approach
which focuses on details [20]. Hence, it should be stressed that the detailed procedure
cannot replace the global judgement and must always be carried out as a second step
following global judgement.

4. Reliability and validity

4.1. Interscorer agreement and test-retest reliability

The first study to report interobserver reliability for the classification of the
different spontaneous movement patterns in preterm and term infants revealed an
overall kappa 193 of 0.91 [7]. Recently, this result was replicated [3] and resulted in a
kappa of 0.92.
The first results on interscorer agreement of GM assessment concerning the
judgement normal versus abnormal GM quality, were published in 1990 [20]. Ten
observers with varying experience in the field of paediatric neurology but without
specific training with this specific tool, assessed a test tape with GMs of 20 preterm
54 C. Einspieler et al. I Early Human Development 50 (1997) 47-60

infants during their preterm and term ages (ten with normal GMs and ten with
abnormal GMs, presented in random order). The average agreement was 90% with a
range of 75 to 100%. Another study [l], again on preterm infants (22 cases), reported
an interpersonal consensus between eight observers ranging from 67% to 99% with
an average of 78%. Yet another study on 35 preterm infants reached an interobservel
agreement of 87% (preterm age) and 93% (early postterm age) between two observer\
] 131. Most recently, three different studies on different groups of infants born preteml
have replicated the high values for interscorer agreement between two observers for
the global judgement [3-51 (Table 1). The largest study on inter- and intraobserver
agreement so far dealt with 15 small-for-gestational-age and 15 appropriate-for-
gestational-age newborn infants between their 3rd and 9th postnatal day [ 151. The
main focus of the study concerned different aspects of GMs such as ‘amplitude of
arms compared to legs’, ‘speed of arms compared to legs’, etc., thus destroying the
visual Gestalt perception of GMs. However, one item among several detailed
judgements was ‘global judgement of the movement: normal vs. abnormal’. The
inter- and intraobserver agreement of this global judgement for chance turned out to
be high (group kappa = 0.84, kappa values for intra-observer agreement: 1.00, 0.77
and 0.69).
All studies mentioned above dealt with the assessment of GMs during preterm and
term age. So far, only one study dealt with older infants, but used a detailed analysis
(motor optima&y score, see Section 3.3.). In that study [ 1O] test-retest reliability was
checked: the analysis of 20 GM recordings was repeated after a time-interval of 2
years. The same scores were obtained in 17 infants (85%) whereas the remaining
scores differed only by one point each.
During the last few years it has become increasingly evident that FMs especially
have a high prognostic value [21]. Hence, we studied interscorer agreement using
examples of FMs in order to compare the interscorer agreement between GMs of

Table 1
Interscorer agreement for the assessment of GM quality during different age periods in nine studies on a
total number of 207 infants. All together 80 observers participated in these studies. A11 studies except one
(I 0] refer to global judgement of normal versus abnormal quality of GM!;.
-.
37 weeks 38-42 weeks 43-47 weeks 2 48 weeks
PMA PMA PMA. PMA

Prechtl, 1990 [20] 90%

v.Kranen-Mastenbroek Kappa = 0.84
et al., 1992 [I51
Geerdink and Hopkins,
1993 [13] 87% 93%
Albers and Jorch, 1994 [I ] 78%
Einspieler, 1994 [lo] Test- retest: 85%*
Bos et al., 1997 [3] 96%
Bos et al., 1997 [4] 98%
Bos et al., 1997 [5] Kappa = 0.92
Present investigation 88% 88% 84%

PMA, postmenstrual age; * for detailed analysis

 C. Einspieler et al. I Early Human Development 50 (1997) 47-60 55

different age groups. An additional question was whether there is a difference in

recognising different types of abnormal GMs.
Forty-six observers, divided into three groups (N = 11, 7 and 28), were given three
different test tapes after 2 days of GM assessment (global judgement) training. The
observers (N = 7) who had to judge the cases on tape 2 had attended a GM
assessment teaching course a few years earlier. The observers were all unaware of the
clinical history of the infants. In addition, all three tapes were again judged by one of
us (H. P.), to whom the clinical history of the infants was unknown and tapes 1 and 2
were judged by two of us (F.F. and G.C.) who knew the clinical history. These
judgements were taken as reference when we compared the interscorer agreement for
different types of GM abnormalities. It should be mentioned that the interobserver
agreement between the reference observers was 100%. Tape 1 contained eleven cases,
tape 2 ten cases and tape 3 nine cases; in total there were 51 judgements of 30 cases.
However, the cases on the three tapes were of varying difficulty from a judgement
point of view and the quality of the recordings was not the same in all three tapes
either. Half the cases were more or less clear-cut, whereas the remainder were not
easy to judge even for very experienced observers. Especially tape 2 contained a
number of these ‘grey’ cases.
We obtained an overall agreement of 87% and the score was almost the same for
each tape (Table 2). This was an unexpected and surprising result considering the fact
that the quality of the recordings differed, the grade of difficulty varied and the
observers had had different amounts of training. The interscorer agreement for
normal GMs averaged 89%. The lowest score (82%) was obtained for the five cases
with normal GMs on tape 2 because three of these were borderline cases and proved
very difficult to judge. The overall agreement for abnormal GMs was 85%. When we
look into the different types of abnormality, the interscorer agreement was highest for
cramped-synchronised GMs (Mean: 89%; Range: 56 to 100%; nine cases) and

Table 2
Interscorer agreement (mean and range) of 51 observers of GMs of 30 infants on three different video tapes
Tape 1 Tape 2 Tape 3 Total
Observers 11 7 28 46
+ Reference observer +2 +2 +1 +5

Normal GMs 96% 82% 90% 89%

(92-100%) (67-100%) (86-93%) (67-100%)
Cases 4 5 3 12

Abnormal GMs 82% 87% 87% 85%

(62-100%) (56-100%) (62-100%) (56-100%)
Cases 1 5 6 18

Overall Agreement 87% 85% 88% 87%

Cases (62-100%) (56-1008) (62-100%) (56-100%)
11 10 9 30
GMs, general movements.
56 C. Einspieler et al. I Early Human Development 50 (1997) 47-60

sufficiently high for judging GMs with poor repertoire (.Mean: 83%; Range: 62 to
93%; five cases). It seems to be more difficult to recognise abnormal FMs (Mean:
78%; Range: 62 to 100%). One has to be aware, however, that this result is based on
four cases only. Observer’s comments on those cases which resulted in an inter-
observer agreement of less than 75% referred to the sequences being too short and a
non-optimal recording procedure in some cases. Thus, poor quality recordings clearly
decrease the possibility for accurate judgement.
The relationship between interscorer agreement and the age of the observed infants
is given in Table 1. Our investigation confirmed previous reliability data obtained for
preterm infants (Mean: 88%; Range: 62 to 100%; 13 cases) as well as for the early
postterm period (Mean: 88%; Range: 56 to 100%; 8 cases). Moreover, the inter-scorer
agreement for the assessment of GMs during the age period when FMs are present.
was also sufficiently high (Mean: 84%; Range: 62 to 100%; nine cases).
In conclusion, this survey provides ample evidence for the robust nature of the
qualitative assessment of GMs and it can now be trusted as a reliable method.

4.2. Sensitivity and specijkity

When introducing a new assessment technique the following aspects must be

considered carefully: how accurate is the test when it comes to detecting disease
positives (i.e. neurological deficits), and how accurate is the test at excluding disease
negatives, i.e. those who do not suffer neurological deficits. The conventional indices
employed to determine this, are sensitivity (number of disease positives and test
positives/number of disease positives) and specificity (number of disease negatives
and test negatives/number of disease negatives) [2].
Table 3 provides an overview of these predictive values obtained in several studies
[6,8,12,13,21,22]. The sensitivity of GM assessment was extremely high (Median:
94.5%; Range: 60 to 100%) regardless of the infant’s age at which the method had
been used. Only one study revealed a sensitivity of 60% for the GM assessment
during preterm age [13]. This study is not comparable with the others insofar as the
outcome was measured at 1 year of age and disease positive was defined as ‘poor
outcome according to neurological examination’. All the other studies measured the
outcome at the end of the second year and cerebral palsy or developmental retardation
(developmental scores below two standard deviations) or both were taken as ‘disease
positives’.
Specificity was low during the early ages (46 to 65%) due to the consistent number
of infants with abnormal GMs (mainly poor repertoire) at this early age who
normalised before or at ‘fidgety age’ (third month postterm) and who had a normal
outcome. However, with increasing age specificity increased, revealing a relatively
high value (Median: 85%; Range: 82 to 100%) during the third month ( ~48 weeks
postmenstrual age) when normal FMs are present in normal infants. which is
indicative of a normal outcome.
The largest study so far, in which 130 infants participated [21] stressed the
importance of the quality of FMs. Ninety-six percent of the infants with normal FMs
had a normal neurological outcome, whereas abnormal quality or total absence of
 C. Einspieler et al. I Early Human Development 50 (1997) 47-60 51

Table 3
Specificity and sensitivity obtained for GM assessmentduring different age periods in six different studies.
The comments indicate: number of infants, children’s age at outcome and impairment at outcome
5 37 weeks 38-42 weeks 43-47 weeks 2 48 weeks Comments
PMA PMA PMA PMA
Ferrari et al., 29 preterm infants.
1990 [12] Outcome: 2 years.
Sensitivity 100% Cerebral palsy
Specificity 59%

Geerdink and 35 preterm SGA inf.

Hopkins, 1993 [13] Outcome: 1 year.
Sensitivity 60% 80% 100% Poor outcome
Specificity 58% 61% 100% according to
neuro-logical
examination

Prechtl et al., 26 asphyxiated

1993 [22] fullterm infants
Sensitivity 100% 85% Outcome: 2 years.
Specificity 46% 82% Cerebral palsy.

Cioni et al., 66 preterm infants.

1997 [6] Outcome: 2 years.
Sensitivity 91% 100% 100% 100% Cerebral palsy or
Specificity 58% 65% 74% 85% developmental
retardation or both.

Cioni et al., 58 fullterm infants.

1997 [8] Outcome: 2 years.
Sensitivity 94% 94% 94% Cerebral palsy or
Specificity 59% 86% 83% developmental
retardation or both.

Prechtl et al., 130 pre- and

1997 [21] fullterm infants.
Sensitivity 95% Outcome: 2 years.
Specificity 96% Cerebral palsy or
developmental
retardation or both.
PMA, postmenstrual age; SGA, small for gestational age; inf, infant.

FMs was followed by neurological abnormalities in 95% of the cases (cerebral palsy
in 82% and developmental retardation and minor neurological signs in 13%). A
recent study on 19 preterm, small-for-gestational-age infants confirmed the high
correlation between the quality of FMs and neurological outcome (P < 0.006) [5].
Another study replicated these results [3]. They found that in 15 very low birth
weight infants at-risk for severe chronic lung disease, the development of cerebral
palsy was related to the presence of cramped synchronised movements near term
58 C. Einspieler et al. I Early Human Development 50 (1997) 47-60

(P < 0.02) and the absence of fidgety movements at the age of 3 months postterm
(P < 0.05) [3].
When we compared the results of this new technique for the assessment of GMs to
those of traditional neurological examination, sensitivity and specificity of GM
assessment with respect to outcome were consistently superior to those of the
neurological examination, in both preterm [6] and term infants [8].
Most striking was the fact that the predictive values on the basis of the assessment
of GMs during the third months of life, were superior to predictions based on brain
ultrasound images [6,21]. In the study on 130 infants mentioned above 1211, the
specificity (83%) and sensitivity (80%) of predictions based on the ultrasound
assessment were clearly lower than the specificity (96%) and sensitivity (95%) 01
predictions based on the assessment of FMs.

5. Discussion

Often there is an urgent need for early recognition of those infants who require
intervention and treatment, particularly if they belong to a high risk group and to
distinguish them from those who are not in need of special attention. Equally
important is the need to provide parents with a realistic prognosis regarding their
infant’s development.
A quick, non-invasive, even non-intrusive and cheap method with high reliability
and high validity is, therefore, most desirable and much in demand for early
assessment of neurological deviations which lead to cerebral palsy and developmental
deficits later on. Such a method is now available.
Provided the quality of the video recording is optimal (as described in this paper).
the assessment of the quality of normal and abnormal GMs is highly reliable. Most
people can make perfect use of their visual Gestalt-perception which was already
recognised by Lorenz as an important scientific method [ 161. A respectable number of
reliability studies provides sound proof of the robust nature of Prechtl’s assessment
technique.
The validity of this new assessment technique is just as high as the reliability. The
most important aspect of the validity of GM assessment lies in the introduction of
‘developmental trajectories’ by Prechtl [20]. Only if abnormal findings, such as
cramped-synchronised or poor repertoire GMs, ate consistently found over time, then
the prediction for major later deficits becomes increasingly solid. The well known
clinical phenomenon of spontaneous normalisation of abnormal neurological findings.
particularly around the third month, is the reason why these early abnormal tindings
cannot be interpreted as false positives. They are real, albeit transient abnorxnaI&ies
and disappear at the time of the major neural transformation at 3 months [ 191. This
developmental process explains why specificity is relatively low until the third month.
If the abnormalities persist until this age, the chances of spontaneous remission
become small.
Great care was given to the proper classification of the abnormal qualities of GMs.
There is always the problem of a sufficient differentiation between types o!
 C. Einspieler et al. / Early Human Development 50 (1997) 47-60 59

phenomena and the remaining of sufficient reliability of the assessment. If splitting is

taken too far, reliability decreases rapidly. The optimal solution between these two
risks is only reached by a sufficiently large empirical basis which we have now
obtained. Modifications of our method have been suggested but these did not provide
any gain in accuracy, reliability, or validity.

6. Notation

GM general movement
FM fidgety movement, i.e. general movement of fidgety quality

Acknowledgements

The authors greatly acknowledge Professor Peter F. Wolff (Harvard University,

Boston, USA), Dr. Simone Albers (University Miinster, Germany), Dr. Heidi Engele
(University Graz, Austria), and Dr. Dieter Sontheimer (University Heidelberg,
Germany) for their invaluable support in connection with reaching consensus of
opinion on the definitions of fidgety movements. In addition, we thank Dr. Mijna
Hadders-Algra (University Groningen, the Netherlands), for providing a test tape
(tape 3) which was used for the reliability test. Our gratitude goes to Dr. Titia
Brantsma-van Wulfften Palthe for correcting the English.

References

[I] Albers S, Jorch G. Prognostic significance of spontaneous motility in very immature preterm infant
under intensive care treatment. Biol Neonate 1994;66:182-7.
[2] Bland M. An introduction to medical statistics. Oxford Medical Publications, 1996:273.
[3] Bos AF, Martijn A, van Asperen RM, Hadders-Algra M, Okken A, Prechtl HFR. Qualitative
assessment of general movements in high risk preterm infants with chronic lung disease requiring
dexamethasone therapy. J Ped, 1997.
[4] Bos AF, van Asperen RM, de Leeuw DM, Prechtl HFR. The influence of septicaemia on spontaneous
motility in preterm infants. Early Hum Dev 1997;50:61-70.
[5] Bos AF, van Loon AJ, Hadders-Algra M, Martijn A, Okken A, Prechtl HFR. Spontaneous motility in
preterm, small for gestational age infants. II Qualitative aspects Early Hum Dev 1997;50:131-47.
161 Cioni G, Ferrari F, Einspieler C, Paolicelli PB, Barbani MT, Prechtl HFR. Comparison between
observation of spontaneous movements and neurological examination in preterm infants. J Pediatr
1997;130:704-11.
[7] Cioni G, Prechtl HFR. Preterrn and early postterm motor behaviour in low-risk premature infants.
Early Hum Dev 1990;23:159-93.
[8] Cioni G, Prechtl HFR, Ferrari F, Paolicelli PB, Einspieler C, Roversi MF. Which better predicts later
outcome in fullterm infants: quality of general movements or neurological examination?. Early Hum
Dev 1997;50:71-85.
[9] Cohen J. A coefficient of agreement for nominal scales. Educ Psycho1 Meas 1960;20:37-46.
[lo] Einspieler C. Abnormal spontaneous movements in infants with repeated sleep apnoaes. Early Hum
Dev 1994;36:31-49.
60 C. Einspieler et al. I Early Human Development 50 (1997) 47-60

[Ill Einspieler C, Prechtl HFR, van Eykem L, de Roos B. Observation of movements during sleep m
ALTE and apnoeic infants. Early Hum Dev 1994;40:39-50.
[12] Ferrari F, Cioni G, Prechtl HFR. Qualitative changes of general movements in preterm infants with
brain lesions. Early Hum Dev 1990;23:193-233.
[13] Geerdink JJ, Hopkins B. Qualitative changes in general movements and their prognostic value trl
preterm infants. Eur J Pediatr 1993;152:362-7.
I141 Hopkins B, Prechtl HFR. A qualitative approach to the development of movements during earl!
infancy. In: Prechtl HFR, editor. Continuity of Neural Functions from Prenatal IO Posmatal Lttc
Oxford: Blackwell Scientific Publications, Oxford. Clin Dev Med 1984;94: 179- I97
[IS] van Kranen-Mastenbroek V, van Oostenbtugge R, Palmans L ct al. Inter- and intra-observer
agreement in the assessment of the quality of spontaneous movements in the newborn. Brain I)e\
1992;14:289-93.
116) Lorenz K. Gestalt perception as a source of scientific knowledge. Ergbsh translation from a German
paper in 1959. In: Lorenz K, editor. Studies in Animal and Human Behavmur. vol II London
Methuen, 1971:281-322.
[17] Prechtl HFR. The behavioural state of the newborn (a review). BraIn Res 1974:76: 185-2 12.
[18] Prechtl HFR. The optimality concept. Editorial. Early Hum Dev 1990;4:201-S.
[ 191 Prechtl HFR. Continuity of neural functions from prenatal to pcstnatal life. Oxford: Blackwell
Scientific Publications, 1984. Clin Dev Med 1984;94:255.
[20] Prechtl HFR. Qualitative changes of spontaneous movements in fetus and preterm infants are .I
marker of neurological dysfunction. Early Hum Dev 1990;23:15 l-9.
1211 Prechtl HFR, Einspieler C, Cioni G, Bos AF, Ferrari F, Sontheimer D. An early marker tot
neurological deficits after perinatal brain lesions. Lancet 1997;349: 136 1-3.
1221 Prechtl HFR, Ferrari F, Cioni G. Predictive value of general movements in asphyxiated fullterm
infants. Early Hum Dev 1993;35:91-120.
[23] Prechtl HFR, Hopkins B. Developmental transformations of spontaneous movements in early infancy
Early Hum Dev 1986;4:233-8.
[24] de Vries JIP, Visser GHA, Prechtl HFR. The emergence of fetal bthaviour. 1. Qualitative aspect\
Early Hum Dev 1982;7:301-22.