Assessing bimanual performance in young children with hemiplegic cerebral palsy: a systematic review
Abstract
Aim This systematic review investigated the availability of assessment tools to evaluate bimanual skills in young children (≤ 3y) with hemiplegic cerebral palsy. Evidence for validity, reliability, and clinical utility of the identified instruments was sought.
Method Ten electronic databases and grey literature were searched (earliest 1806) to February 2009 to identify articles that met criteria related to the child’s age and diagnosis, and included a discrete bimanual assessment. Further searches for evidence of psychometric properties of each identified assessment were undertaken. For quality criteria, we evaluated the internal validity of each study and appraised the validity and reliability of identified assessments.
Results From 1435 papers retrieved, 15 were eligible for inclusion, and 11 assessments of bimanual performance were identified. Ten assessments had inadequate evidence for reliability and validity. Only the Assisting Hand Assessment had evidence for reliability and validity for its intended purposes.
Interpretation Reliability and validity are fundamental considerations when developing or selecting assessments. Additional considerations for assessing young children include the following: (1) standardized administration and scoring; (2) items that capture the complexity of bimanual behaviour; and (3) protocols and items targeted within a small age range to meet specific developmental needs. A validated assessment of younger infants is needed to support early intervention during this period of rapid brain development.
LIST OF ABBREVIATIONS
-
- AHA
-
- Assisting Hand Assessment
-
- CFUS
-
- Caregiver Functional Use Survey
The importance of upper limb assessment for children with hemiplegic cerebral palsy (CP) is now well recognized.1,2 An upper limb evaluation might be conducted for a variety of purposes, such as to guide treatment planning, to measure the effectiveness of an intervention, or to compare and discriminate between individuals. Clinicians and researchers use assessment as a basis for making decisions and drawing conclusions.3 As children with hemiplegic CP primarily have one well-functioning side of their body and one ‘more affected’ side, evaluation of the upper limb often targets the more affected limb using unimanual assessments. Yet clinical experience shows that children with hemiplegic CP rarely use their impaired hand for unimanual tasks. This hand is typically used when they need it, i.e. during bimanual task performance. Bimanual actions are more complicated than unimanual actions as the movements of both arms and hands must be coordinated temporally and spatially to complete a task or achieve a desired goal.4 With many everyday tasks requiring cooperative use of both hands, poor bimanual performance is often the greatest functional impairment for children with hemiplegia. Careful evaluation of bimanual ability should be an integral component of an upper limb assessment.
This systematic review investigated what bimanual assessments or outcome measures are available for use with children aged 3 years or younger with hemiplegic CP. Early childhood was targeted as it is a time of great change and potential for acquisition of motor skills, as well as a time when motor behaviours associated with the impairment can start to emerge.5 Careful evaluation of bimanual abilities and their progression could provide valuable information for clinicians and researchers about how these skills are affected by the evolving impairment. Evaluation could also assist in determining whether the infants’ skills are different to those of older children whose motor patterns have become established. Carefully collected information about bimanual performance could then be used to guide and/or evaluate the effectiveness of intervention programmes.
In addition to identifying which bimanual assessment tools were being used with young children with hemiplegic CP, this review aimed to investigate evidence for the validity and reliability of each assessment as well as their clinical utility. Assessment tools used in clinical practice and for research purposes should have sound psychometric properties, initially reported by test developers and further evaluated in subsequent research. Criteria to ensure that an assessment tool is administered in a consistent manner, evidence that it is appropriate to a particular population or purpose, and that the results can be interpreted confidently, are essential properties.6 In addition to sound psychometric evidence, the assessment should also be considered from a clinical perspective. Aspects such as the availability of an instrument, its ease of use, and the relevance and meaningfulness of the information that it provides will help determine whether it can be realistically incorporated into clinical practice.7,8 The specific aim of the systematic review was to investigate the availability of assessment tools for evaluating bimanual skills in young children with hemiplegic CP and to assess the validity, reliability, and clinical utility of the identified instruments.
Method
Search strategy
The search strategy was conducted in two steps. Step one involved identifying assessment tools; in step two, a second search aimed to find evidence for validity, reliability, and clinical utility for each identified instrument.
Step one: identification of outcome measures
We comprehensively searched the following electronic databases: Medline (1950–February 2009), Medline In-Process, CINAHL (1982–February 2009), Embase (1980–February 2009), PsychINFO (1806–February 2009), ERIC, OTseeker, PEDro, and the full Cochrane Library Issue 1, 2009. Relevant terms and synonyms were identified under four key concepts by scoping appropriate medical subject headings (MeSH terms) from Medline and CINAHL and by identifying keywords in key papers. The concepts and examples of the search terms were the following: (1) diagnosis (e.g. CP, hemiplegia, movement disorder); (2) assessment tool (e.g. general terms such as outcome assessment, disability evaluation, or by using specific outcome measures such as the Assisting Hand Assessment [AHA]) or interventions that might have been evaluated using a bimanual assessment (e.g. constraint-induced movement therapy, botulinum toxin); (3) upper limb (e.g. upper extremity, bimanual skills, hand function); and (4) age (e.g. child, infant, toddler).
Additional articles were identified through a search of the grey literature (trial registries in Australia9 and America10 and theses online11), by searching the reference lists of retrieved articles, by searching for additional papers published by significant authors, and by citation tracking of key papers using Google Scholar.
The full references of the identified papers were exported to a reference management system (EndNote, version X.0.2) and duplicates were removed. Two reviewers (SG and KD) independently assessed the papers by title and abstract for inclusion and exclusion using criteria described below. Consensus for the inclusion or exclusion of relevant articles was reached by discussion.
Articles were included if they included the following: (1) children who were 3 years of age or younger; (2) children who had a diagnosis of unilateral or hemiplegic CP, or who had been identified as having a unilateral movement disorder as a result of an injury to the central nervous system; and (3) either a dedicated assessment of bimanual skills or an assessment that included discrete evaluation of bimanual skills as a component of the assessment.
Articles were excluded if they were (1) abstracts from conference papers or letters to the editor, because these did not supply enough information about the outcome measure or its uses; (2) papers written in languages other than English owing to the lack of resources available for translation; and (3) journal articles that used upper limb assessments considered unsuitable. Assessments were considered unsuitable for the following reasons: (1) not valid for young children (e.g. Shriners Hospital for Children Upper Extremity Evaluation, ABILHAND-Kids); (2) they were unimanual instruments (e.g. Quality of Upper Extremity Skills Test, Melbourne Assessment of Unilateral Upper Limb Function); (3) if there was a mix of both bimanual and unimanual items, or if it was ambiguous as to whether the task was bimanual or unimanual and there was no discrete bimanual score (e.g. Pediatric Motor Activity Log, Toddler Actual Arm Use Test, Modified House Functional Classification12); (4) if there were no objective criteria that enabled evaluation of bimanual skills (e.g. descriptive evaluation); (5) if they indirectly evaluated bimanual skills such as activities of daily living assessments or measures of individualized outcome performance (e.g. Goal Attainment Scale, Canadian Occupational Performance Measure).
Step two: evidence for validity, reliability, and clinical utility
After identifying the bimanual assessments, a second search was conducted for additional papers providing evidence for validity, reliability, and clinical utility. Each assessment was searched as text and then combined with MeSH terms or keywords for validity, reliability, or clinical utility. The four major databases (Medline, Embase, CINAHL, and PsychINFO) were searched using the same search dates as previously reported, and a search of the Health and Psychological Instruments database (1985 to February 2009) was also undertaken. Articles were selected for inclusion from this second search if they (1) evaluated the psychometric properties or clinical utility of an instrument; (2) included infants who had a diagnosis of hemiplegic CP; and (3) included children who were 3 years of age or younger. Articles were excluded if they were abstracts or conference presentations, or papers written in languages other than English. In addition to the article search, the test manual (if available) of each assessment was used to clarify details of the tool or provide evidence of its psychometric properties.
Data extraction
A data extraction sheet was specifically developed for this review, which summarized information from published papers, manuals, and correspondence with authors13,14 into four sections. The first section contained general information about a published paper that identified a bimanual assessment, such as the type of study, its objective, and the name of the bimanual assessment. The second section contained information about the assessment itself, such as its purpose, age range, items used to score bimanual behaviours, and scoring system. The third section detailed evidence supporting validity and reliability of the assessment. The fourth section detailed evidence for the clinical utility of the assessment such as format and ease of administration, instructions, examiner qualifications, and costs involved.
Quality assessment
A two-tiered approach was used to evaluate the quality of the retrieved evidence. The first level rated the methodological quality of each individual article that provided evidence to support the reliability or validity of an assessment tool. The second level of evaluation examined the aggregated body of information, i.e. it considered the combined results from the individual studies to evaluate the quality of the assessment tool. The quality of evidence for clinical utility was not considered in this review, as each individual clinical situation would have different requirements of an assessment and its utility. Instead, descriptive information was extracted using guidelines adapted from the Outcomes Measure Rating Form.15
Level one: rating the quality of the published evidence from individual studies
An important aspect of systematically evaluating literature is consideration of the quality of individual studies that provide evidence for the question under review.16 In this review, studies directly evaluating aspects of reliability or validity of an outcome measure were considered as the most important or primary source of evidence. This evidence might have been located within published manuals or journal articles. The quality, or internal validity, of findings from these studies was evaluated using two rating scales: one investigating biases of design and conduct of reliability studies and one of validity studies. These scales were devised using guidelines from studies evaluating similar aspects,17,18 from articles providing methods for critiquing the quality of studies,16,19 and from general principles about reliability and validity.3,20 The reliability rating scale contained 12 criteria to rate the design and method of the study, whereas the validity rating scale contained four to six criteria depending on the type of validity evaluated. The two rating scales and their criteria can be found in Appendices SI and SII (supporting information published online).
It was anticipated that there might be little primary evidence for many of the outcome measures. Therefore, a secondary level of evidence, that of studies using a bimanual assessment for children with hemiplegic CP within the desired age range, was also considered. The following types of study could be considered secondary evidence to support psychometric properties of an assessment tool. First, rigorously conducted randomized controlled trials in which the objectives included changes in bimanual performance were considered as evidence for responsiveness of the assessment tool if there was a significant difference in outcomes between an experimental and a no-treatment or sham control group. The PEDro scale was used to evaluate the quality of the randomized controlled trials.21,22 For this review, we considered randomized controlled trials to be rigorous if they achieved a PEDro ranking of 6/10 or greater. Second, rigorously conducted randomized controlled trials were considered as evidence for test–retest stability of the outcome measure if there was no change in the no-treatment group and a significant difference between the groups. Third, we considered a study with at least two baseline measurement points during which there was no change in the outcome measure or any other variable that might also contribute secondary evidence for test–retest stability. At best, secondary evidence was considered to provide only a low level of support for the assessment tool.
Level two: rating the aggregated evidence for each outcome measure
All information gained about a bimanual assessment from both primary and secondary sources and manuals was used to evaluate the aggregated evidence. Although there are currently no accepted quality criteria against which to rate the psychometric properties of health measurement scales,23 this review adapted criteria from Terwee et al.24 who identified eight critical measurement properties of health status questionnaires. These were content validity, internal consistency, criterion validity, construct validity, reproducibility (reliability and agreement), responsiveness, floor and ceiling effects, and interpretability. Terwee et al. provided specific criteria against which to measure these properties. In this review, the criteria of Terwee et al. were extended to include criteria suitable for assessments developed using Rasch analysis, although differences in approach and wording in modern and classical test theory on central concepts made placement of the criteria difficult.25 Although these extended criteria might not be complete, their inclusion enabled assessment tools that were developed using the Rasch measurement models to be evaluated. These extended criteria can be found in Appendix SIII (supporting information published online).
Quality criteria were applied independently by two of the researchers (SG, CI) both to the individual studies supporting the assessment tool and the individual assessments identified. Disagreement was resolved by discussion until consensus was reached. As four different scales were used to rate the evidence for the assessment tools, the quality of the evidence was summarized descriptively as being poor if it scored positively in 0 to 33% of the criteria in the appropriate rating scale, moderate if it scored positively in 34 to 66% of the criteria, or high if it scored positively in 67 to 100% of the criteria.
Results
Database search
The initial search of electronic databases produced 1431 articles after duplicates were removed. Searches of the grey literature, citation tracking, and reference lists produced a further four articles. After applying the inclusion/exclusion criteria to these papers, 15 papers, which included 11 different bimanual assessments, remained for inclusion in this systematic review. Of these published papers, three were studies whose aim was to evaluate aspects of validity or reliability.26–28 The remaining 12 papers were either intervention trials or longitudinal studies that had used a bimanual assessment tool to evaluate children with hemiplegic CP within the desired age range.13,14,29–38 A second search for evidence of validity, reliability, and clinical utility of the identified assessments for the specified diagnosis and age range found no further study.
In this review, the assessments will be identified by a known name or acronym (e.g. House, AHA) or identified by the first author of the primary study using this assessment. Table I provides an overview of the 11 bimanual assessments identified from the searches, and includes the primary reference for each assessment. The AHA had a test administration manual available to clinicians who have participated in certification courses39 in addition to journal publications. Guidelines for administration and scoring of the Caregiver Functional Use Survey (CFUS) were provided by the authors.13 Satila et al. report that an unpublished manual is available in Finnish for the assessment identified in this review as Satila.14
| Outcome measure | Purpose(s) | Target population | Age range | Scales/items/tasks | Format of outcome measure | Method of scoring |
|---|---|---|---|---|---|---|
| Assisting Hand Assessment39 | Descriptive, evaluative | Children with hemiplegic or unilateral CP/obstetric brachial plexus palsy | 18mo–12y | One scaleTwenty-two items evaluating various behavioural aspects of bimanual performance | Videotaped recording of 10–15min semi-structured play session using specific test objects later scored by therapist | Twenty-two bimanual items each rated on a four-point rating scale of quality (1–4) when completing bimanual activities. Three final scores are possible: sum of scores (22–88); percentage score (0–100); conversion by Rasch analysis to logits (available to certified raters from test developers) |
| Caregivers Functional Use Survey (CFUS)13 | Evaluative | Children with hemiplegic CP | 3y 6mo–15y 6mo | Two scales: bimanual and unimanual tasks10 tasks in each scale | Parental observation during specific everyday tasks | Parents rate 10 bimanual and 10 unimanual tasks on six-point Likert scale (0–5) designed to measure ‘how well’ and ‘how much’ the ‘affected’ arm is used. Scores averaged for each set of tasks and combined score calculated |
| House Functional Classification System (House)44 | Evaluative | Children with CP | 2–20y | No specific items/tasks | Observation of functional status by patient, parents, therapist, or physician | Nine-point rating scale (0–8) evaluating how ‘functionally’ the impaired hand is used during everyday activity. Levels include non-use, use as a passive assist, active assist, and spontaneous use |
| Goldner et al.33 | Evaluative | Children and adults with CP | 3–34y | One scaleNo specific items or tasks | Observation of ‘functional’ skills of impaired hand by parents, therapist, or physician | Functional result graded on four-point scale (P1–P4) based on active movement and ability of impaired hand to be involved in unimanual and bimanual actions and everyday tasks (such as dressing) |
| Satila et al.14 | Evaluative | Children with CP | 3y 4mo–17y | One scaleSeven bimanual tasks | Videotaped recording of child performing specific bimanual tasks later scored by physician/therapist | Different standardized grips scored while completing 10 bimanual items/tasks. Grips rated on a four-point scale ranging from cannot grasp to grasps using normal grip and completes task |
| Pagliano et al.34 | Evaluative | Congenital hemiplegia | 2–17y | One scaleNon-specified tasks chosen according to age | Observation of bimanual skills during spontaneous play and everyday activities | Overall spontaneous use and support functions of ‘impaired’ hand scored on four-point scale ranging from 0 (the impaired limb is not used) to 3 (the impaired hand cooperates with holding and manipulation using a varied repertoire of patterns) |
| Fedrizzi et al.29 | Evaluative, descriptive | Children with hemiplegic CP | 2y 7mo–12y 9mo | One scale.Four structured bimanual tasks dependent on age at assessment | Video recording of children completing the four tasks | Overall quality of bimanual manipulation scored on a four-point scale based on variability and stereotypy of movement patterns ranging from 0 (no use of impaired limb) to 3 (cooperation of the impaired limb holding and manipulation using a varied repertoire of patterns) |
| Crocker et al.35 | Evaluative | Child with spastic hemiplegia | 2y | One bimanual item: two-handed use (Six other behaviours that may or may not be completed bimanually) | Videotaped 15min session of free play with variety of toys | One bimanual action scored for frequency of observed behaviour during the play session |
| Dickerson and Brown36 | Evaluative | Child with hemiplegia | 2y | One bimanual task (Eight other unimanual tasks) | Catching rolling ball | One bimanual task (ball catching using both hands) scored for frequency and five-point Likert scale evaluating quality of upper extremity movement of ‘affected’ arm/hand during this task |
| Smelt37 | Evaluative | Child with spastic hemiplegia | 17mo | One item: bimanual play | Use of ‘affected’ hand during play with a ball | One bimanual item scored on a five-point scale used to evaluate ability to perform the play activity. |
| Fergus et al.38 | Evaluative | Child with hemiparesis | 13mo | Two bimanual items (Six other unimanual actions/behaviours) | Videotaped 30min session | One bimanual action (transfers) scored for success and smoothness. Percentage of successful and smooth transfers calculated out of total number of transfers performed |
Overview of identified bimanual assessments
Each assessment had been developed or used as an evaluative tool, i.e. they were used to measure the magnitude of change over a period or after an intervention.40 Evidence for use as a descriptive assessment (a measure that describes the difference among individuals within groups) was also found for the AHA and Fedrizzi.
One assessment was developed to evaluate infants from 18 months of age (the AHA). Two assessments, developed for use with individuals in a case study (Fergus) and a single subject design study (Smelt), evaluated infants at 13 and 17 months respectively. The youngest age for the remaining assessments was typically 2 or 3 years of age. The age range of children for whom assessments were designed was often large, with seven of the 11 assessments spanning more than 10 years.
Although each assessment evaluated bimanual skills, the format varied. Some assessments relied on real-time observations made in natural settings (CFUS, House, Goldner, Pagliano, Fergus). Only the CFUS specified the activities to be observed. Four assessments used video observation of specific bimanual tasks or play (AHA, Satila, Fedrizzi, Crocker). Three assessments (Dickerson, Smelt, Fergus) evaluated only one specific activity or skill (ball catching, transferring an object from hand to hand). The scoring systems were also variable, ranging from simple four-point rating scales or frequency counts of a single observed behaviour, to more complex scoring systems covering several behaviours.
Quality assessment
Results from application of the criteria rating the internal validity of primary evidence (i.e. studies evaluating aspects of validity and reliability) are presented in Table II. Three papers provided primary evidence for one assessment, the AHA. All three papers were of a high methodological quality (reliability 10/12; validity 3 to 3½/4). Enough information was reported in a longitudinal study29 to provide primary evidence for the reliability of Fedrizzi, i.e. the authors described a pilot study conducted to evaluate the interrater agreement of the tool. However, the methodological quality of this evidence was assessed to be of a moderate level (5/12). None of the remaining studies met the criteria considered to be secondary evidence for a bimanual assessment tool. The identified papers and their reasons for exclusion are included in Table III.
| Identified studies/assessment tool | Study design | Quality of reliability study methods | Quality of validity study methods |
|---|---|---|---|
| Krumlinde-Sundholm and Eliasson:26 Assisting Hand Assessment (AHA) | Validity study | Content validity: 3½/4Construct validity: 3/4 | |
| Krumlinde-Sundholm et al.:28 AHA | Validity study | Construct validity: 3/4Content validity: 3/4 | |
| Holmefur et al.:27 AHA | Reliability study | 10/12 | |
| Fedrizzi et al.:29 Fedrizzi | A pilot reliability study (reported in a longitudinal trial) | 5/12 |
| Identified studies/assessment tool(s) | Study design | Reason for exclusion |
|---|---|---|
| Gordon et al.:13 Assisting Hand Assessment (AHA)/Caregiver Functional Use Survey | Intervention trial: RCT | PEDro score <6/10 and lacked repeated baseline measures |
| Eliasson et al.:30 AHA | Intervention trial: case-controlled clinical trial | Not RCT and lacked repeated baseline measure |
| Wallen et al.:31 AHA | Intervention trial: prospective pre-post feasibility study | Not RCT and lacked repeated baseline measure |
| Satila et al.:14 House/Satila | Intervention trial: retrospective case–controlled trial | Not RCT and lacked repeated baseline measure |
| Roth et al.:32 House | Intervention trial: retrospective case series | Not RCT and lacked repeated baseline measure |
| Goldner et al.:33 Goldner | Intervention trial: retrospective longitudinal case series | Not RCT and lacked repeated baseline measure |
| Pagliano et al.:34 Pagliano | Longitudinal study | Not RCT and lacked repeated baseline measure |
| Crocker et al.:35 Crocker | Intervention trial: SSD | Not RCT and unstable baseline measurements |
| Dickerson and Brown:36 Dickerson | Intervention trial: SSD | Not RCT and lacked repeated baseline measurements of bimanual measure |
| Smelt 1989:37 Smelt | Intervention trial: SSD | Not RCT and unstable baseline measurements |
| Fergus et al.:38 Fergus | Intervention trial: single case study | Not RCT and lacked repeated baseline measurements |
- RCT, randomized controlled trial; SSD, single subject design.
Aggregated evidence for the measurement properties of the 11 bimanual assessments is presented in Table SI (supporting information published online). No data were available for nine assessments (CFUS, House, Satila, Crocker, Dickerson, Goldner, Pagliano, Smelt, and Fergus). One assessment, Fedrizzi, could be scored against one criterion, reliability, for which it was rated as questionable owing to the doubtful methodology as reported within the longitudinal study. One assessment, the AHA, could be scored on six of the nine measurement criteria and achieved positive ratings in all six. Criterion-related validity could not be evaluated as there is currently no criterion standard assessment against which to compare.
The clinical utility of each bimanual assessment, evaluated using the guidelines by Law,15 is summarized in Table IV. The administration formats of the assessments ranged from naturalistic observation to videotaped recordings of specific bimanual task performance. Three assessments had manuals (AHA, CFUS, Satila) and a further three provided enough detail in the published studies to replicate the assessment (Fedrizzi, Dickerson, Smelt). However, five assessments did not provide enough information to ensure each user would administer them in the same way. Most assessments appeared easy to administer, score, and interpret, except for the AHA. This was the only assessment that required users to purchase an assessment kit and undertake training to ensure reliable administration and scoring.
| Assessment | Format of administration | Procedures | Ease of administration | Assessor training | Cost | |
|---|---|---|---|---|---|---|
| Outcome measure | Training | |||||
| Assisting Hand Assessment | Video recording of bimanual task performance10–15min session | Manual available with trainingStandardized assessment | Administration: easyScoring: more complexInterpretation: more complex | Required | Expensive | Expensive |
| Caregivers Functional Use Survey | Interview with parent | Manual available from authorStandardized assessment | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| House Functional Classification System | Naturalistic observations by physician, therapist, patient, or parent | No manual available. Difficult to replicate assessment from studies | Administration: DTRScoring: easyInterpretation: more complex | Not required | Inexpensive | Not required |
| Goldner | Naturalistic observations | No manual availableDifficult to replicate assessment from studies | Administration: DTRScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Satila | Video recording of bimanual task performanceApproximately 20–30min | Finnish manual available from author, otherwise difficult to replicate from studies | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Pagliano | Naturalistic observations | No manual availableDifficult to replicate from study | Administration: DTRScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Fedrizzi | Video recording of bimanual task performance15–20min | No manual availableInstructions provided in study enough to replicate assessment | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Crocker | Video recording of 15min free play sessionNaturalistic observations | No manual availableDifficult to replicate from study | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Dickerson | Video recording of bimanual task performance15min session | No manual availableInstructions provided in study enough to replicate assessment | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Smelt | Bimanual task performanceApproximately 5–10min | No manual availableInstructions provided in study enough to replicate assessment | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
| Fergus | Video recording of 30min play sessionNaturalistic observations | No manual availableDifficult to replicate from study | Administration: easyScoring: easyInterpretation: easy | Not addressed | Inexpensive | Not addressed |
- Adapted from Law 2004.15 Administration, scoring and interpretation scored as (1) easy, (2) more complex. When rating administration, set-up time and time to complete were considered easy if under 30min. When rating scoring and interpretation, time taken to score and interpret results and ease/difficulty with which results could be interpreted for the average service provider were considered. Cost: expensive, total more than Aus$1,000; moderate, Aus$500–$1,000; inexpensive, less than Aus$500. DTR, difficult to rate as not enough information provided.
Discussion
Fifteen published papers were identified in this review as evaluating or using one of 11 bimanual assessments for young children with hemiplegic CP. Three papers provided high-quality evidence for validity or reliability, which we considered as primary evidence, for one of the assessments (AHA), and one paper provided moderate-quality evidence for the reliability of another assessment (Fedrizzi). The remaining 11 papers were intervention trials or longitudinal studies that had used a bimanual assessment tool. Evidence provided by these papers might have been considered as low-level or secondary evidence for two measurement properties (responsiveness and test–retest reliability) if they met defined criteria. However, no paper met the criteria and these secondary sources did not contribute to evidence for any of the bimanual assessment tool identified in this review.
Assessment tools are designed, and should be chosen, for a specific purpose and population.40 The primary purpose of all 11 assessments identified in this review was to evaluate change over time or in response to an intervention. Evidence for content, construct validity, and responsiveness as well as good test reliability (intrarater, interrater, and test–retest) are important properties for evaluative instruments.41 When there is evidence supporting these measurement properties, clinicians and researchers have an authoritative basis for explaining and supporting their findings and recommendations.6 The AHA is the only bimanual assessment that has used systematic test-development procedures to provide evidence supporting most of these critical aspects. In addition, its development using Rasch modelling has resulted in an interval-level scale that truly makes it an instrument that allows us to measure the difference both within individuals and between individuals or groups of individuals.25 Only three of the 12 intervention or longitudinal studies used this instrument.13,30,31 Of the remaining 10 bimanual assessments identified, there was no evidence for any measurement property. Results reporting on changes in bimanual skills from studies using these 10 assessments should, therefore, be interpreted with caution.
The development of the AHA using the Rasch measurement model has provided strong evidence that the items in the assessment both cover and are specific to the construct of bimanual skills. The AHA has established inter- and intrarater reliability for children with hemiplegic CP as young as 18 months.27 Training requirements and use of standardized equipment and procedures, although adding to the cost of the assessment, ensure that it is administered and the results interpreted in a valid and reliable manner. Clinicians and researchers can have some confidence when reporting assessment findings that these are the true abilities of the individual and when a difference in performance is identified that it is likely to represent real change. However, what has not yet been evaluated in the AHA is the smallest detectable change determined from a test–retest evaluation, and the minimal important change. Such information can contribute to information about the tool’s responsiveness and would strengthen the AHA’s use as an outcome measure.25
Reviewing the reported bimanual assessments used with young children with hemiplegic CP also highlighted several critical issues that were related to the age of the children and the construct of bimanual performance. First, using standardized procedures and protocols might be a critical factor when developing a reliable outcome measure for young children as this might help reduce systematic and random error variance resulting from variable and inconsistent motor responses. Bimanual assessments for very young children should have standard and consistent administration procedures.
The items included in the assessment are also important. Using a single, simple task (such as catching a rolling ball or transferring objects), although easy for young children to complete, would appear to have limited face validity in terms of measuring the complex construct of bimanual performance. When assessing bimanual skills, many of the assessments aimed to use naturalistic play behaviours or real-life observations to evaluate a child’s abilities. Using these methods acknowledges that it is important to assess bimanual performance, or what the child actually does, rather than the capacity of the affected limb, to understand bimanual ability.
The age range of the children for whom the test is valid also needs to be considered in relation to test items. When assessments aim to evaluate children across a wide age span, it complicates the selection of developmentally appropriate activities, especially for children at the younger end of the age range. For two of the assessments in this review (CFUS, Satila), young children would have difficulty completing some of the stipulated bimanual tasks, again compromising the validity of the content. For example, Satila used drawing a circle with the help of a glass and then cutting it out. This is a complex task for young children, and therefore poor performance owing to impairment cannot be separated from a developmentally appropriate lower level of skill.
The youngest children assessed were 13 months (Fergus), 17 months (Smelt), and 18 months old (AHA). The remaining assessments did not evaluate children until 2 or even 3 years of age, nearing the end of the age range for children in this review. No extensive assessment evaluated infants younger than 18 months. Reasons for this, although not stated by the developers of the measures, might be that younger infants would require different assessment protocols and activities owing to their limited understanding and skill level. Occupational interests and abilities will also be very different in younger children. Additionally, continuous neurobiological and functional changes during early infancy increase the need for age-specific assessments in this period.42 Assessment at this young age might be critical because, as well as being a time when movement skills develop rapidly, the motor impairment associated with abnormal pathophysiology is also emerging in children under 2 years of age.5 Advances in neuroimaging indicate that different types of neural reorganization might influence these developing motor patterns.43 It is important when studying the mechanisms of neural plasticity of the developing brain also to scrutinize carefully any associated functional abilities. An assessment tool that could track these changes in bimanual abilities over time, that could evaluate how these abilities are affected by the emerging motor impairment, and that examined whether the bimanual ability of younger infants with hemiplegia differed from older children with established motor patterns, could provide crucial information to clinicians and researchers when planning interventions or evaluating intervention effectiveness. As the AHA has already established content and construct validity and reliability in children from 18 months, future research should assess whether this instrument could be further developed to include younger infants.
Although aiming to be thorough, some limitations might have affected the results of this systematic review and, hence, the interpretation of the findings. Restricting the search to papers published in English might have reduced the number of assessments found. The use of quality criteria to rate both the internal validity of each individual study providing evidence as well as criteria against which to measure the collective or aggregated evidence for each assessment was made difficult by the lack of standardized critical appraisal tools addressing these areas. It required some tools to be adapted or extended and other tools to be formulated. These new and adapted criteria have not been formally appraised for their own internal validity.
Conclusions
Eleven different assessment tools were found during this systematic review, indicating that clinicians and researchers consider the evaluation of bimanual performance, not just unimanual capacity, to be an important component of upper limb assessment for young children with hemiplegic CP. However, the measurement properties of the identified assessments also needed to be considered. This review found that only one assessment, the AHA, had evidence supporting six of nine critical measurement properties. The remaining assessment tools had poor or no evidence to support them. Clinicians and researchers should endeavour to assess bimanual performance using instruments with proven validity and reliability. Without this evidence, both clinical information and research results gained from the assessment tools should be interpreted with caution.
Special consideration might need to be given when assessing bimanual skills of very young children with hemiplegic CP. Because infancy is a time of rapid physical and psychological change, assessments need to be designed specifically for this age range. Instruments should take into account the likely variability of responses, the rapidly changing fine motor and adaptive skill levels, and the more limited cognitive understanding of younger children. By developing bimanual assessments for young infants with hemiplegic CP, crucial insights into the development of these abilities and the role of neural plasticity might become evident. However, these younger infants need to be assessed using valid and reliable measurement tools. The first author is currently exploring the potential for the AHA, an assessment with established psychometric properties, to be extended to this younger age range.
Acknowledgements
Public Health (Dora Lush) NHMRC Postgraduate fellowship (AS); NHMRC Postdoctoral Training Fellowship (PA), NHMRC Career Development Fellowship Project Grant (RB): National Health and Medical Research Council (NHMRC) 284512. A previous version has been published in The Cochrane Library, Issue 2, April 2007.
