The Assisting Hand Assessment: current evidence of validity, reliability, and responsiveness to change

The Assisting Hand Assessment: current evidence of validity, reliability, and responsiveness to change Lena Krumlinde-Sundholm* PhD Reg OT, Neuropediatric Research Unit; Marie Holmefur MSc Reg OT PhD Student, Neuropediatric Research Unit, Department of Women and Child Health; Anders Kottorp PhD Reg OT, Division of Occupational Therapy, Department of Neurotec; Ann-Christin Eliasson PhD Reg OT, Neuropediatric Research Unit, Department of Women and Child Health, Karolinska Institutet, Stockholm, Sweden. *Correspondence to first author at Neuropediatric Research Unit, Department of Women and Child Health, Karolinska Institutet, Stockholm, Sweden. E-mail: Lena.Krumlinde.Sundholm@ki.se The Assisting Hand Assessment (AHA) provides a new perspective of hand function evaluation relevant for children with unilateral upper limb disabilities. It measures how effectively the involved hand is actually used for bimanual activity, which, for these children, might be the most important aspect of their hand function. The aim of this paper is to report the conceptual framework and the evidence for validity, reliability, and responsiveness to change for the measures. Previously, the AHA has been evaluated for children aged 18 months to 5 years and excellent inter- and intrarater reliability was demonstrated. This paper reports further evidence of construct validity and reliability for the AHA measures involving an extended age range of children with hemiplegic cerebral palsy or obstetric brachial plexus palsy from 18 months to 12 years of age (mean age 4y 11mo [SD 2y 9mo] range 18mo 12y 8mo). A Rasch measurement model was used to analyze 409 assessments from 303 children (170 males, 133 females). The analysis generated a scale demonstrating large capacity to reliably separate and spread personal ability measures, indicating sensitivity to change and a hierarchy of the items ranging them from easy to hard. Aspects of item fit, relationship between age and ability measures, and development of assisting hand function are discussed. See end of paper for list of abbreviations. When hand function impairment leads to intervention, the overall aim is generally to increase the functional use of the hands for everyday task performance. Most activities in daily life normally involve the collaborative use of two hands together. However, instruments evaluating hand functioning commonly entail testing one hand at a time. Furthermore, they evaluate maximum capability in contrast to performance, i.e. what the person can do rather than what he/she actually does do. In bimanual performance, the two hands have different roles: the dominant hand manipulates objects to a higher extent and is also quicker than the non-dominant hand, which more often has a holding and stabilizing role. 1,2 Given a unilateral deficit, the difference between these roles becomes even more pronounced. It is natural for people with unilateral hand impairment from early age to use the well-functioning hand as a first choice whenever this is possible. Even though some activities can be performed one-handed, and compensatory strategies may be functional, the use of two hands together is often essential for successful task accomplishment. 3 The affected hand/arm can be described as a helper or an assisting hand rather than a non-dominant hand. 4 Characteristics that specify a useful assisting hand are, to our knowledge, not described elsewhere. The Assisting Hand Assessment (AHA) is a recently developed test intended for use with children who have a unilateral upper limb dysfunction, in particular children with hemiplegic cerebral palsy (CP) or obstetric brachial plexus palsy (OBPP). 4 The core concept of the AHA is based on recognition of the different roles the two hands play and the fact that an assisting hand does not necessarily need to be as quick and as manipulative as a dominant hand to effectively assist in bimanual task performance. The AHA describes and measures how effectively people with a unilateral dysfunction actually use the affected hand/arm with the well-functioning hand to perform tasks requiring bimanual performance. The AHA is based on observations of actions performed in relevant activities and is meant to reflect the person s usual performance, not best capacity. Furthermore, the AHA is not intended to explain underlying reasons for limitations of performance on impairment level, e.g. spasticity, muscle strength, or sensory functions. Nor are results of the AHA norm referenced in a population of typically developed children because such children have two well-functioning hands, no assisting hand, and, thus, have another way of using both hands together. The AHA is developed for a targeted group of people with a unilateral upper limb impairment. The development of AHA was guided by models from modern test theory. 5 Applications of such models (e.g. Rasch analysis and item response theory) are recommended 6,7 and are currently used in paediatric rehabilitation services to an increasing extent to validate clinical assessments (e.g. Pediatric Evaluation of Disability Inventory [PEDI], 8 The Gross Motor Function Measure [GMFM-66], 9,10 ABILHAND-Kids 11 ). The AHA is administered in two steps. First, a semi-structured video-recorded play session lasting 10 to 15 minutes is conducted in which specific toys from the AHA test kit requiring bimanual handling are used. Second, the scoring is performed by a review of the video on 22 items with a 4-point criterion-referenced rating scale. The administration procedures and criteria for each scale step on each item are defined in a manual. 12 The items describe different types of objectrelated actions of the assisting hand under the following headings: general use, arm use, grasp and release, fine motor Developmental Medicine & Child Neurology 2007, 49: 259 264 259

adjustments, and coordination and pace. The general meaning of the rating scale categories is: 4=effective, 3=somewhat effective, 2=ineffective, and 1=does not do. Specific criteria describing behaviours within the categories for each item are defined in the manual. 12 The child s most commonly used performance is scored in the AHA and, thereby, described by the scale. EVIDENCE OF THE PSYCHOMETRIC PROPERTIES OF AHA TO DATE The first steps of the development and validation of AHA were presented in 2003. 4 Sixty assessments of children aged between 18 months and 5 years were analyzed by means of a Rasch measurement analysis. This study supported evidence of construct validity and reliability of the measures. Inter- and intrarater reliability of AHA measures was evaluated by Holmefur et al. 13 Two different designs were used for interrater trials, one with two raters evaluating the same 18 children and one with 20 raters assessing the same eight children. Two language versions of the test were used in the 20- rater design, Swedish and English, and were found to be equally reliable. The interrater intercorrelation coefficient (ICC) for sum scores were high: 0.98 (two-rater design) and 0.97 (20-rater design) respectively. Moreover, the intrarater ICC was found to be high, 0.99, when 20 raters each scored one session twice. In this article, the standard error of measurement (SEM) was calculated for inter- and intrarater conditions. These numbers can be applied to individual results for score interpretation in clinical practice and indicate the size of the error due to rater in the unit of raw scores. The SEM was 1.2 (intra-) and 1.5 (inter-), which gave error intervals of SD 2.4 and SD 3 raw scores respectively, which should be considered in outcome analysis. To find tests and questionnaires evaluating hand functioning that were useful for children with upper limb reduction deficiencies (ULRD), Buffart et al. 14 reviewed the literature. The AHA was one of two tests they found to be relevant out of 14 revised tests. They concluded that the tests need further studies concerning psychometric properties for this patient group and have started that evaluation. The AHA scale s responsiveness to change has been shown in a study in which Eliasson et al. 15 used the AHA as the outcome measure in evaluating the effects of a modified model of constraint-induced movement therapy. This study involved 41 children with hemiplegia (aged 18mo 4y). The results revealed significant differences between treatment and comparison groups, and also reported the amount of change in individuals and on individual AHA items. It was demonstrated that almost Table I: Demographics of children who were assessed using Assisting Hand Assessments Total 18 60mo 61 152mo n (%) n (%) n (%) Assessments 409 262 (64) 147 (36) Hemiplegia 362 (88) 231 (88) 131 (90) OBPP 47 (12) 32 (12) 16 (10) Males 228 (56) 136 (52) 92 (63) Right side affected 229 (56) 156 (60) 73 (50) OBPP, obstetric brachial plexus palsy. all of the items in the AHA detected changes. The items changing the most were arm-related items, such as moves upper arm, reaches and stabilizes by weight or support. As the AHA has only recently been developed, other results have not yet been published. As described above, recent reports on the AHA indicate that aspects of validity, reliability, and responsiveness to change are promising for use with children with unilateral deficits in the age range of 18 months to 5 years. The purpose of this paper was to examine whether the AHA measures demonstrate validity and reliability concerning a larger sample of children with hemiplegic CP or OBPP for an expanded age range of 18 months to 12 years. Method SAMPLE Scores from 409 AHA play sessions of children in the age range of 18 months to 12 years (mean age 4y 11mo [SD 33mo]) were analysed. Table I shows details of demographic information. Sixty-four per cent of the assessments were from children aged 5 years. The majority of the assessments (88%) involved children who had hemiplegic CP and the others had OBPP. In total, 303 children were included and among them 48 were represented by two to five assessments performed at least 2 months apart. All videos were scored by one of two experienced raters. The children represent a clinical convenience sample of patients attending paediatric services from different outpatient settings. Informed consent was obtained from all parents and ethical approval for this study was granted by the Ethics Research Committee of Karolinska Hospital in Stockholm. ANALYZES An application of the Rasch measurement model was implemented to evaluate aspects of validity and reliability of the AHA. The analysis is based on a simplistic model that uses probability estimations to order items and participants concurrently along a continuum, by using the following formula for a test with dichotomous items: Log (P ni /(1 - P ni )) = B n - D i, where P ni is the probability that the person n passes the item i, B n is the person s ability measure, and D i is the item difficulty calibration. 5,16 Thus, the Rasch measurement model orders the items on a scale from easiest to hardest, as well as individuals in terms of their ability on the trait being measured based on the actual responses on items and individuals. Both item calibrations and person ability measures are placed on a common measurement line with the interval unit (logits) log-odds-probability unit. The logarithmic transformation of the raw data commonly stretched out the logit measures in the low and the high ends of the scale (S-curved). This wider spreading reflects that changes in the ends of a continuum are often larger than seen from the change of raw scores and give a fairer representation of changes of abilities in different locations of the scale. For an overview of this reasoning and the Rasch Model in general, see Bond and Fox. 5 The Rasch analysis additionally provides each measure and each calibration with information about its standard error (SE), indicating the precision of the measure. Goodness-of-fit statistics for each item and each participant are used to evaluate the degree of fit between the actual patterns of responses from the AHA and the model assertions. This underlying assertion 260 Developmental Medicine & Child Neurology 2007, 49: 259 264

is that easier items are likely to be easier for all persons than more difficult items. In a corresponding way, the person-fit to the construct is evaluated from the assertion that more able people are more likely to pass any item than less able people are. A person will misfit when he or she obtains an unexpected low score on an easy item or an unexpectedly high score on a difficult item. Acceptable levels of goodness-of-fit for clinical observations are described by Wright and Linacre 17 and were in this study determined by analysis of infit mean square (MnSq) residual values of 1.4 in combination with standardized z-values of <2 for both items and participants. If 95% of the items in the AHA demonstrate acceptable goodness-of-fit, they provide evidence of unidimensionality and internal scale validity. Similarly, if 95% of the children s responses demonstrate acceptable goodness-of-fit, they support evidence of person response validity. These criteria have commonly been used earlier in the development of performance evaluations. 18 20 To evaluate the capability of the AHA scale to spread the children according to their ability measures along the ability continuum, the person separation index was used. This estimation is based on the SD in the sample and the individual measurement error. The person separation index defines the statistically distinct number of ability strata for the scale and indicates its sensitivity to detect differences among person ability measures. 5,21 The data from 409 AHAs were entered into the Rasch measurement model computer software program Facets 3.11 (Copyright 1987 1998, John M Linacre), and analyzed according to a partial credit model. The partial credit model was chosen because it could not be assumed that the scale steps were evenly spread across all items and rating scale categories. 9 Results The children demonstrated a wide range of abilities to use the affected arm as reflected by the range of the AHA measures (Table II). The corresponding raw score range was 23 to 87 points on the full scale of 22 to 88 points. Thus, no ceiling or floor effect was detected in this sample. The person separation reliability estimate (based on the same concept as Cronbach s alpha) 5 was high (0.97/0.96;Table II) and the AHA was found to effectively discriminate between children with different ability in seven distinct strata, indicated by a separation index of 5.42. The range of the person s ability measures was satisfactorily covered by the range of the item s difficulty measures (Table II), especially when the full range of the 4-point rating scale was considered. The difficulty of the test items was well matched to the ability of the person and was found to be appropriate for the targeted group of children, which was further confirmed by the closeness between the mean values of the items and persons respectively (Table II). The results of the Rasch analysis of the 22 items are shown in Table III. Two items (9%), moves forearm and moves upper arm, did not demonstrate acceptable levels of fit to the model. Since this was slightly higher than expected, a stepwise deletion of these two items was analyzed. This resulted in a 20- item scale on which 95% of the items demonstrated an acceptable level of goodness-of-fit (Table IV) and unidimensionality was confirmed. The overall rate of unexpected responses was low for both conditions, less than 5% (Table II). One consequence of the reduction of items was a decrease in the person separation index from 5.42 to 5.20, indicating a somewhat lower ability of the 20-item assessment to differentiate between participants (Table II). Since this decrease had a minimal impact on levels of strata identified, 22 it was considered marginal. More than 97% of the person measures demonstrated acceptable goodness-of-fit to the model expectations (Table II). There was no systematic pattern among those participants who did not meet the model expectations, according to sex, age, involved side, or diagnosis. To investigate further to what extent a reduction of items from 22 to 20 impacts on individual person measures, the ability measures generated from each of the versions were compared and evaluated in relation to the individual SE of each participant, as generated from the Rasch analysis. Only 16 (4%) of the person ability measures changed more than 2 SE. Thus, the choice of scale used had only a marginal impact Table II: A comparison of scale characteristics of 22- and 20- item versions of Assisting Hand Assessment 22-item scale 20-item scale Items Item measure range, logits 4.42 to 4.97 4.91 to 5.36 Item measure mean (SD), logits 0.0 (2.19) 0.0 (2.47) Mean SE, logits 0.13 0.13 Item fit, % 91 95 Persons Person measure range, logits 7.29 to 8.58 7.64 to 8.96 Person measure mean (SD), logits 0.16 (3.38) 0.10 (3.76) Person separation 5.42 5.20 Mean SE, logits 0.59 0.66 Reliability 0.97 0.96 Person fit, % 97.6 97.4 Unexpected responses, % 4.6 4.5 SE, standard error. AHA ability measure, logits 8 6 4 2 0 2 4 6 8 10 10 20 40 60 80 100 120 140 160 Age, mo Figure 1: Relationship between age (mo) and ability measure (logits) for 409 Assisting Hand Assessments (AHA) of children with hemiplegic cerebral palsy or obstetric brachial plexus palsy. Assisting Hand Assessment, Evidence of Psychometric Properties Lena Krumlinde-Sundholm et al. 261

on the participants ability measures. To review the relationship between the children s ability measures and age, a scatter plot was produced (Fig. 1). The ability measures for this sample were distributed over the whole age span with no clear association between the variables. The use of the Rasch model for data analysis has been vital to the development of the AHA as the Rasch method creates the hierarchical ordering of items as well as people along an equal-interval measure line. Through this ordering, the difficulty of the items was established (Tables III and IV). That is, the response patterns of the children involved in this study showed which of the items were easier and which were more difficult. This item hierarchy was found to correspond well with clinical experience. Table III: Item measures and fit statistics for the Assisting Hand Assessment 22-item scale. Items demonstrating misfit in bold Measure SE Infit Outfit Item hierarchy MnSq Z MnSq Z 4.42 0.15 0.7 2 0.6 0 Puts down 3.14 0.12 0.9 1 0.7 0 Manipulates 1.87 0.15 1.7 6 2.3 5 Moves forearm 1.81 0.13 0.8 2 0.7 1 Grasps 1.42 0.11 0.8 3 0.6 1 Readjusts grip 1.36 0.13 0.7 4 0.6 2 Varies grasp 1.18 0.12 1.3 3 1.2 1 Reaches 0.82 0.13 0.7 5 0.6 2 Calibrates 0.79 0.13 0.9 1 1.1 0 Chooses AH when closest to object 0.73 0.14 0.6 6 0.5 5 Flow in bimanual performance 0.68 0.12 0.7 4 0.7 2 Releases 0.55 0.13 0.6 6 0.6 2 Stabilizes by grip 0.06 0.12 0.7 4 0.6 3 Orients objects 0.38 0.11 1.1 1 1.2 1 Moves fingers 0.52 0.14 1.7 8 1.8 7 Moves upper arm 0.71 0.13 0.8 2 0.7 2 Initiates use 0.95 0.12 0.8 3 0.8 2 Coordinates 1.7 0.12 1.3 3 1.3 2 Changes strategies 2.33 0.12 1.3 3 1.6 4 Proceeds 3.01 0.11 1.0 0 0.7 0 Stabilizes by weight or support 4.24 0.13 1.0 0 0.8 0 Holds 4.97 0.20 0.6 2 0.2 0 Approaches objects SE, standard error; MnSq, mean square; AH, assisting hand. Table IV: Item measures and fit statistics for the Assisting Hand Assessment 20-item scale. Item demonstrating misfit in bold Measure SE Infit Outfit Item hierarchy MnSq Z MnSq Z 4.91 0.15 0.8 1 0.8 0 Puts down 3.58 0.13 0.9 1 0.7 0 Manipulates 2.05 0.13 0.9 1 0.8 1 Grasps 1.60 0.11 0.8 2 0.7 1 Readjusts grip 1.56 0.13 0.7 3 0.7 1 Varies grasp 1.38 0.14 1.6 6 1.5 2 Reaches 0.96 0.13 0.7 4 0.6 2 Calibrates 0.92 0.13 1.0 0 1.3 1 Chooses AH 0.86 0.15 0.6 5 0.5 4 Flow in bimanual performance 0.81 0.12 0.8 3 0.7 1 Releases 0.66 0.13 0.6 5 0.6 2 Stabilizes by grip 0.13 0.12 0.8 3 0.7 2 Orients objects 0.36 0.12 1.2 3 1.3 3 Moves fingers 0.70 0.13 0.9 1 0.8 1 Initiates use 0.97 0.12 0.8 3 0.8 1 Coordinates 1.80 0.12 1.3 4 1.4 2 Changes strategies 2.50 0.13 1.3 4 1.8 4 Proceeds 3.18 0.12 1.0 0 0.7 0 Stabilizes by weight or support 4.53 0.13 1.0 0 0.9 0 Holds 5.36 0.21 0.6 2 0.2 0 Approaches objects SE, standard error; MnSq, mean square; AH, assisting hand. 262 Developmental Medicine & Child Neurology 2007, 49: 259 264

Discussion The results of the Rasch analysis concerning children in the extended age range up to 12 years correspond well with the results previously reported for children up to 5 years of age. 4 These findings, together with the outcome evaluation study by Eliasson et al. 15 and the inter and intrarater reliability evaluation, 13 indicate that the AHA can be used to reliably measure individuals demonstrating a wide range of abilities, and that the AHA measures can be responsive to changes. The AHA measures were able to separate the individuals in this sample and spread them along the ability continuum to a high degree, as seen by the person separation reliability estimate and the number of statistically distinct difficulty strata. The clinical sample involved in this analysis seemed to represent well the varied ability spectra of children with hemiplegia or OBPP. The cohort contained individuals who hardly used the arm/hand at all to individuals who used the assisting hand/arm with minimal limitation involving only a discrete quality decrease in the smoothness of in-hand manipulation. None of the participants received a maximum or minimum sum score. Thus, the ceiling and floor effect was avoided. ITEM AND PERSON-FIT Although unidimensionality was demonstrated for 95% of the responses in the 20-item version of the AHA, and only for 91% of the responses in the 22-item version, a decision to omit the two items not demonstrating acceptable levels of goodness-offit is not straightforward. 16,22 It is reasonable to view the moves upper arm and moves forearm items as having different constructs than the other items as these two have a range of movement relation that is different to the other items. However, the misfit of these items can also be explained by the different response patterns that can be expected from the two diagnostic groups included in the analysis. The children with OBPP typically have a more severely involved shoulder and forearm than hand involvement. 23 Children with hemiplegia more commonly display a pattern where poorer shoulder and forearm movements relate to an overall lower performance. Some of them may even demonstrate the opposite pattern, with rather effective movements in shoulder and forearm and more severely affected hand function. 24 This can explain the reasons for misfit. The two items demonstrating misfit are seen as clinically important items that can be directly attributed to intervention, e.g. with botulinum toxin A or surgery, and, therefore, can be expected to change. Furthermore, the arm-related items were the ones that demonstrated the biggest changes in the intervention study by Eliasson et al. 15 Removing these items from the scale would then possibly decrease the scales responsiveness to change, which is not desirable. The use of the 22- item or the 20-item item scales did not change the person s measures notably, and participant separation index was slightly higher for the 22 items. This implies that the 22-item scale could still be used to obtain valid measures and further points out that removing the moves upper arm and moves forearm items from the scale needs to be carefully considered. As all items seem to provide clinically useful information, all 22 items are kept in the scale for further scrutiny. 16,22 As stated by Buffart at al., 14 before using the AHA for children with ULRD, the scale needs further evaluation for this group of children. Item and person-fit should be evaluated as well as the difficulty hierarchies using the Rasch measurement model. Also, for other groups of patients with unilateral arm/hand deficiencies, the use of the AHA should be evaluated before being used. Future studies must also involve a differential item functioning analysis for all specific diagnostic groups, to detect whether there are different response patterns among those groups and the impact this may have on the item fit and the estimations generated. DEVELOPMENT OF ASSISTING HAND FUNCTION Hand function in children without disability typically develops dramatically with age. Aspects of accuracy, speed, grip strength, grip regulation, and ability to use the two hands simultaneously but for complementary actions, increase considerably over the age span 18 months to 12 years. 25 27 In this light, the wide span of ability measures spread over all ages is an interesting finding (Fig. 1). It suggests that age, per se, may not be the most important factor for children s ability to use the affected arm/hand as an effective assisting hand. A few items in the AHA scale have criteria, according to which development norms are expected to influence the score. These are Manipulates, which describes in-hand manipulation skills that have been shown to develop with age, 26 and Calibrates, which is also age-dependent. 25 The Changes strategies item is also expected to be scored higher with increased cognitive development. However, the low association between age and AHA measures indicates that the item difficulty hierarchy, generated by the Rasch analysis, reflects steps of increasing ability rather than agedependent development. From this perspective, the hierarchical ordering of items deserves greater attention. This hierarchy may prove to be useful to define goals for hand function treatment adjusted to the child s ability level. It may also serve as a first step towards the development of a theory pertaining to the development of assisting hand function in children with unilateral impairments. Thus, the ordering of the items from easy to difficult reflects an increasing ability of assisting hand function. However, longitudinal studies are needed to learn more about individuals assisting hand development. The framework of the ICF 28 can be used to classify an instrument into what domain of functioning it is situated. For the AHA, the activity domain is used, and the instrument has a clear focus on performance as opposed to capacity. For example, it is not the person s best capacity to grasp, release, or manipulate objects that is requested; rather, it is how effectively these different actions are performed spontaneously while playing that is assessed. Play is used as the basis for the observation and toys are presented playfully and without instructions as to which hand to grasp or hold with. This makes the test attractive to children and promotes a typical performance. Conclusions The current study demonstrated evidence that AHA is a valid instrument for use with children with hemiplegic CP or OBPP between the ages of 18 months and 12 years. Inter- and intrarater reliability was earlier shown to be excellent. 13 The AHA scale demonstrated a considerable ability to separate and distribute person ability measures, indicating sensitivity to change. Responsiveness to change was also confirmed in an intervention outcome study. 15 Although validation of an instrument is an ongoing process and more studies are needed to acquire a deeper knowledge, the current evidence of sound psychometric properties for the measures is positive and indicates that the AHA can be useful for both clinical practice and research. Assisting Hand Assessment, Evidence of Psychometric Properties Lena Krumlinde-Sundholm et al. 263

Accepted for publication 8th November 2006. Acknowledgements This study was supported by grants from the Center for Health Care Sciences, Swedish Research Council and Health Care Sciences Postgraduate School. References 1. Fagard J. (1990) The development of bimanual coordination. In: C Bard, M Fleury, L Hay, editors. Development of Eye-hand Coordination Across the Life Span. Columbia: University of South Carolina Press. p263 282 2. Kimmerle M, Mainwaring L, Borenstein M. (2003) The functional repertoire of the hand and its application to assessment. Am J Occup Ther 57: 489 498. 3. Skold A, Josephsson S, Eliasson AC. (2004) Performing bimanual activities: the experiences of young persons with hemiplegic cerebral palsy. Am J Occup Ther 58: 416 425. 4. Krumlinde-Sundholm L. (2003) Development of the Assisting Hand Assessment: a Rasch-built measure intended for children with unilateral upper limb impairments. Scan J Occup Ther 10: 16 26. 5. Bond TG, Fox CM. (2001) Applying the Rasch Model. Fundamental Measurement in the Human Sciences. Mahawa: Lawrence Erlbaum. 6. Velozo CA, Kielhofner G, Lai JS. (1999) The use of Rasch Analysis to produce scale-free measurement of functional ability. Am J Occup Ther 53: 83 90. 7. Fisher AG. (1993) The assessement of IADL motor skills: an application of many faceted Rasch analysis. Am J OccupTher 47: 319 329. 8. Haley SM, Coster WJ, Ludlow LH, Haltiwanger JT, Andrellos PJ. (1992) Pediatric Evaluation of Disability Inventory (PEDI), Development, Standardization and Administration Manual. 1.0 ed. Boston: PEDI Research Group, New England Medical Centre Hospital. 9. Avery LM, Russell DJ, Raina PS, Walter SD, Rosenbaum PL. (2003) Rasch analysis of the Gross Motor Function Measure: validating the assumptions of the Rasch model to create an interval-level measure. Arch Phys Med Rehabil 84: 697 705. 10. Russell DJ, Avery LM, Rosenbaum PL, Raina PS, Walter SD, Palisano RJ. (2000) Improved scaling of the Gross motor Function Measure for children with cerebral palsy: evidence of reliablility and validity. Phys Ther 80: 873 885. 11. Arnould C, Penta M, Renders A, Thonnard JL. (2004) ABILHAND-Kids: a measure of manual ability in children with cerebral palsy. Neurology 63: 1045 1052. 12. Krumlinde-Sundholm L, Holmefur M, Eliasson A-C. (2006). Manual: Assisting Hand Assessment (English Research Version 4.3). Stockholm: Karolinska Institutet. 13. Holmefur M, Krumlinde-Sundholm L, Eliasson AC. Interrater and intrarater reliability of the Assisting Hand Assessment. Am J Occup Ther. (Forthcoming) 14. Buffart LM, Roebroeck ME, Pesch-Batenburg JM, Janssen WG, Stam HJ. (2006) Assessment of arm/hand functioning in children with a congenital transverse or longitudinal reduction deficiency of the upper limb. Disabil Rehabil 28: 85 95. 15. Eliasson A-C, Krumlinde-Sundholm L, Shaw K, Wang C. (2005) Effects of constraint-induced movement therapy in young children with hemiplegic cerebral palsy: an adapted model. Dev Med Child Neurol 47: 266 275. 16. Wright BD, Stone MH. (1979) Best Test Design, Rasch measurement. Chicago: Mesa Press. 17. Wright BD, Linacre JM.(1994) Reasonable mean-square fit values. Rasch Meas Trans 8: 370. 18. Fisher AG. (1993) The assessement of IADL motor skills: an application of many faceted Rasch analysis. Am J Occup Ther 47: 319 329. 19. Atchison BT, Fisher AG, Bryze K. (1998) Rater reliability and internal scale and person response validity of the School Assessment of Motor and Process Skills. Am J Occup Ther 52: 843 850. 20. Tham K, Bernspång B, Fisher AG. (1999) Development of the assessment of awareness of disability. Scan J Occup Ther 6: 184 190. 21. Fisher WP Jr. (1992) Reliability Statistics. Rasch Meas Trans 6: 238. 22. Bohlig M, Fisher WP Jr, Masters GN, Bond T. (1998) Content validity, construct validity, and misfitting items. Rasch Meas Trans 12: 607. 23. Strömbeck C, Krumlinde-Sundholm L, Forssberg H. (2000) Functional outcome at 5 years in children with obstetrical brachial plexus palsy with and without microsurgical reconstruction. Dev Med Child Neurol 42: 148 157. 24. Uvebrant P. (2001) Clinical presentation and neurology. In: Neville B, Goodman R, editors. Congenital Hemiplegia. London: Mac Keith Press. p 53 64. 25. Forssberg H, Eliasson AC, Kinoshita H, Johansson RS, Westling G. (1991) Development of human precision grip I: basic coordination of forces. Exp Brain Res 85: 451 457. 26. Case-Smith J. (2006) Hand skill development in the context of infants play: birth to two years. In: Henderson A, Pehoski C, editors. Hand Function in the Child: Foundations for Remediation. 2nd edn. St Louis: Mosby. p117 141. 27. Pehoski C. (2006) Object manipulation in infants and children. In: Henderson A, Pehoski C, editors. Hand Function in the Child: Foundations for Remediation. 2nd edn. St Louis: Mosby. p143 160. 28. World Health Organisation (2001) International Classification of Functioning, Disability and Health. Geneva: World Health Organization. List of abbreviations AHA Assisting Hand Assessment ICC Intercorrelation coeffecient ICF International Classification of Functioning, Disability and Health OBPP Obstetric brachial plexus palsy 264 Developmental Medicine & Child Neurology 2007, 49: 259 264