and reliability System (GMFCS; Palisano et al. 1997). The GMFCS describes
gross motor function in terms of self-initiated movements
with emphasis on function in sitting and walking. This classi-
fication is based on the concept of disability (World Health
Ann-Christin Eliasson* PhD OT; Organization 1980) and of functional limitation (National
Lena Krumlinde-Sundholm PhD OT, Department of Woman Institutes of Health 1993). The GMFCS has been widely adopt-
and Child Health, Karolinska Institute, Stockholm; ed internationally, suggesting that it has filled a gap in func-
Birgit Rösblad PhD PT, Department of Community Medicine tional classification (Morris and Bartlett 2004). No similar
and Rehabilitation, University of Umeå, Umeå; classification of hand function has been available, but one is
Eva Beckung PhD PT, Queen Silvia Children’s Hospital, needed to broaden the functional perspective of CP beyond
Göteborg; gross motor issues.
Marianne Arner PhD MD, Hand Unit, Department of Classification is the process of grouping data, persons, or
Orthopaedics, Lund University Hospital, Lund; objects into classes according to common characteristics,
Ann-Marie Öhrvall MSc OT, Department of Habilitation, thereby reducing the number of data elements. The usefulness
Stockholm County Council, Stockholm, Sweden; of a classification depends on how understandable and clear
Peter Rosenbaum MD FRCP, CanChild Centre for Childhood the descriptions are, and how meaningful different levels are.
Disability Research, McMaster University, Hamilton, The most common way to classify hand function in CP has
Ontario, Canada. been to use terms like ‘mild’, ‘moderate’, and ‘severe’ impair-
ment (Claeys et al. 1983). Although some other classifica-
*Correspondence to first author at Neuropaediatric tions are available (House et al. 1981, Zancolli and Zancolli
Research Unit Q2:07, Astrid Lindgren Children’s Hospital, 1981, Krägeloh-Mann et al. 1993, Beckung and Hagberg 2002),
SE-171 76 Stockholm, Sweden. all of them classify aspects of grasping rather than functional
E-mail: ann-christin.eliasson@kbh.ki.se performance and none has been tested for reliability.
This proposed new classification, the Manual Ability
Classification System (MACS), is designed to classify how chil-
dren with CP use their hands when handling objects in daily
activities. The focus is on manual ability, as defined in the
The Manual Ability Classification System (MACS) has been International Classification of Functioning, Disability and
developed to classify how children with cerebral palsy (CP) use Health (ICF; World Health Organization 2001). It has its start-
their hands when handling objects in daily activities. The ing point in upper limb function but is also influenced by envi-
classification is designed to reflect the child’s typical manual ronmental, personal, and contextual factors. MACS follows
performance, not the child’s maximal capacity. It classifies the Penta and coworkers’ definition of manual ability as ‘the capac-
collaborative use of both hands together. Validation was based ity to manage daily activities that require the use of the upper
on the experience within an expert group, a review of the limbs, whatever the strategies involved, which can be observed
literature, and thorough analysis of children across a spectrum from activity performance in the person’s everyday context’
of function. Discussions continued until consensus was reached, (Penta et al. 2001). MACS is not designed to classify best capaci-
first about the constructs, then about the content of the five ty and does not mean to distinguish different capacities
levels. Parents and therapists were interviewed about the between hands. MACS reports the collaboration of both hands
content and the description of levels. Reliability was tested together: it is not an assessment of each hand separately. It
between pairs of therapists for 168 children (70 females, 98 looks for the children’s manual ability to ‘handle object(s) in
males; with hemiplegia [n=52], diplegia [n=70], tetraplegia daily life’. By that we mean those activities that are relevant and
[n=19], ataxia [n=6], dyskinesia [n=19], and unspecified CP age appropriate for the child including, for example, eating,
[n=2]) between 4 and 18 years and between 25 parents and dressing, playing, writing, etc.; these are distinct from
their children’s therapists. The results demonstrated that MACS advanced skilled activities that require special training for per-
has good validity and reliability. The intraclass correlation formance, such as playing a musical instrument. Neither does it
coefficient between therapists was 0.97 (95% confidence classify activities closely related to academic skills in school.
interval 0.96–0.98), and between parents and therapist was 0.96 The emphasis of MACS is on handling objects in an individual’s
(0.89–0.98), indicating excellent agreement. personal space, the space immediately close to one’s body, as
Statistics
For analysis of interrater reliability a one-way random effects Table II: Types of cerebral palsy (CP) by Manual Ability
model (referred to as intraclass correlation coefficient [ICC] Classification System (MACS) rating by professionals
1) was used, which, according to Shrout and Fleiss (1979), is
suitable when different raters are assessing the children using Type of CP MACS levels
I II III IV V Total %
average measures. The ICC result is equivalent to a weighted
kappa (Fleiss and Cohen 1973). Rater agreement, the coeffi- Hemiplegia 10 38 4 – – 52 31
cient of total agreement between therapists (number of chil- Diplegia 12 23 18 15 2 70 42
dren with total agreement divided by total number of children; Tetraplegia/quadriplegia – – 1 9 9 19 12
Polit and Hungler 1994) was calculated. For descriptive analy- Ataxia 1 2 2 1 – 6 4
sis, the result from the first rater of MACS was used. For cor- Dyskinetic CP – – 3 4 12 19 12
relation between MACS and GMFCS, Spearman’s rank corre- Total 23 63 28 29 23 166a –
lation was used. % 14 38 17 17 14 – –
aCP type was unavailable for two children.
Results
VALIDATION OF THE CLASSIFICATION
All parents and therapists responded positively to the ques-
tion about whether MACS was built on a useful construct for Table III: Numbers of children 4 to 18 years in each Manual
describing the child’s hand function. They also found the dif- Ability Classification System level as rated by two therapistsa
ferences between the five levels meaningful and easy to under-
Therapist 2
stand. Parents emphasized the advantage of describing what
I II III IV V Total
the children actually can do instead of focusing on their limi-
tations. The distinction between capacity and performance Therapist 1
did not cause confusion for the families. Several therapists I 19 4 – – – 23
also emphasized that MACS was giving a structure to their II 2 56 6 – – 64
thinking, and that it was important to have one classification III – 5 22 1 – 28
for all kinds of CP. IV – 1 3 22 3 29
V – – – 1 23 24
The rationale for choosing a particular level instead of
Total 21 66 31 24 26 168
others was most often clear for both parents and therapists.
Although parents and therapists mainly agreed about the lev- aIntraclass correlation coefficient=0.97 (95% CI 0.96–0.98).
els there were some disagreements (reported below). These Numbers in bold represent total agreement.
Manual Ability Classification System for Children with CP Ann-Christin Eliasson et al. 551
between 8 and 12 years. The total agreement was 72%. There DISTRIBUTION OF MACS LEVELS AMONG SUBTYPES OF CP
was disagreement about one level for seven children, five of Children with all different subtypes of CP were included in
whom were rated on a higher level by the parents and two by the study (Table II, Fig. 1). The children with hemiplegia
the therapists. The disagreement was distributed across all were mainly seen in Level II but also in Levels I and III.
levels of MACS. Children with diplegia were found in Levels I to IV, whereas
children with dyskinetic CP and tetra- or quadriplegia were
apparent in Levels IV to V.
Table IV: Distribution of intraclass correlation coefficient
RELATION BETWEEN MACS AND GMFCS
(ICC) for different age groups
The correlation between GMFCS and MACS was 0.79 (p<0.05;
Age (y) n ICC (95% CI) Table V), with complete agreement in 49%. A large variation
occured at MACS Level II, where 40% of the children were
4–7 45 0.98 (0.97–0.99)
scored and where the agreement between GMFCS and MACS
8–12 90 0.96 (0.94–0.98)
was just 35%. Differences were also observed in the other levels
13–18 33 0.98 (0.96–0.99)
All: 4–18 168 0.97 (0.96–0.98) (Fig. 2).
100 100
90 90
80 80
70 70
60 60
Per cent
Per cent
50 50
40 40
30 30
20 20
10 10
0 0
I II III IV V I II III IV V
MACS levels MACS levels
Figure 1: Distribution of children (per cent of collected Figure 2: Distribution of children between levels of Manual
material) according to subtypes of cerebral paresis (CP). Ability Classification System (MACS) and Gross Motor
MACS, Manual Ability Classification System; dys, dystonia; Function Classification System (GMFCS).
tetra, tetraplegia; quadri, quadriplegia.
Manual Ability Classification System for Children with CP Ann-Christin Eliasson et al. 553
Bax M, Goldstein M, Rosenbaum P, Leviton A, Paneth N. (2005) Morris C, Bartlett D. (2004). Gross Motor Function Classification
Proposed definition and classification of cerebral palsy, April System: impact and utility. Dev Med Child Neurol 46: 60–65.
2005. Dev Med Child Neurol 47: 571–576. Mutch L, Alberman E, Hagberg B, Kodama K, Velickovic Perat M.
Beckung E, Hagberg G. (2002) Neuroimpairments, activity (1992) Cerebral palsy epidemiology: where are we now and
limitations, and participation restrictions in children with where are we going? Dev Med Child Neurol 34: 547–551.
cerebral palsy. Dev Med Child Neurol 44: 309–316. National Institutes of Health. (1993) Research Plans for Center for
Claeys V, Deonna TH, Chrzanowski R. (1983) Congenital hemiparesis: Medical Rehabilitation Research (NCMRR). National Institutes
the spectrum of lesions. Helv Paediatr Acta 38: 439–455. of Health publication no 93-3509. Bethesda, MD: National
Exner CE, Henderson A. (1995) Cognition and motor skills. In: Institutes of Health.
Hendersson A, Pehoski C, editors. Hand Function in the Child. Öhrvall AM. (2005) MACS – a classification of manual ability for
Foundation for remediation. New York: Mosby. p 93–110. children with cerebral palsy. A study of validity and reliability.
Fink A, Kosekoff J, Chassin M, Brook RH. (1984) Consensus (Thesis). Karolinska Institute, Sweden.
methods: characteristics and guidelines for use. Am J Public Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B.
Health 74: 979–983. (1997) Development and reliability of a system to classify gross
Fleiss JL, Cohen J. (1973) The equivalence of weighted kappa and motor function in children with cerebral palsy. Dev Med Child
the intraclass correlation coefficient as measures of reliability. Neurol 39: 214–223.
Educ Psychol Meas 33: 613–619. Penta M, Tesio L, Arnould C, Zancan A, Thonnard J-L. (2001) The
Gorter JW, Rosenbaum PL, Hanna SE, Palisano RJ, Bartlett DJ, ABILHAND questionnaire as a measure of manual ability in
Russell DJ, Walter SD, Raina P, Galuppi BE, Wood E. (2004) Limb chronic stroke patients. Rasch-based validation and relationship
distribution, motor impairment, and functional classification of to upper limb impairment. Stroke 32: 1627–1634.
cerebral palsy. Dev Med Child Neurol 46: 461–467. Polit D, Hungler B. (1995) Nursing Research. Principles and
Hagberg B, Hagberg G. (1993) The origins of cerebral palsy. In: Methods. 5th edn. Philadelphia: Lippincott.
David TJ, editor. Recent Advances in Paediatrics XI. Edinburgh Shrout PE, Fleiss JL. (1979) Intra-class correlations: uses in
and London: Churchill Livingstone. assessing rater reliability. Psychol Bull 86: 420–428.
House JH, Gwathmey FW, Fidler MO. (1981) A dynamic approach to Wood E, Rosenbaum P. (2000) The Gross Motor Function
the thumb-in palm deformity in cerebral palsy. J Bone Joint Surg Classification System for cerebral palsy: a study of reliability and
Am 63: 216–225. stability over time. Dev Med Child Neurol 42: 292–296.
Ingram TTS. (1964) Paediatric Aspects of Cerebral Palsy. London: World Health Organization. (1980) International Classification of
Livingstone. Impairments, Disabilities and Handicaps. Geneva: World
Krägeloh-Mann I, Hagberg G, Meisner C, Schelp B, Edebol Eeg- Health Organization.
Olofsson K, Selbmann HK, Hagberg B, Michaelis R. (1993) World Health Organization. (2001) International Classification of
Bilateral spastic cerebral palsy – a comparative study between Functioning, Disability and Health. Geneva: World Health
south-west Germany and western Sweden. I: Clinical patterns Organization.
and disabilities. Dev Med Child Neurol 35: 1037–1047. Zancolli EA, Zancolli ER. (1981) Surgical management of the hemiplegic
Kvale S. (1996) Interviews. An Introduction to Qualitative spastic hand in cerebral palsy. Surg Clin North Am 61: 395–406.
Research Interviewing. California: Sage Publications.
Appendix I: The Manual Ability Classfication System (MACS) leaflet (which can be downloaded in several languages from
www.macs.nu)