Abstract
The BAM was shown to be a valid, reliable measure for people with chronic stroke. It can be used to measure bimanual functioning, which may help people return to prestroke hand roles.
Temporal and spatial coordination in the performance of bimanual skills is important for functional independence. Although upper extremity (UE) sensorimotor deficits are often described in unilateral terms (Hunter & Crome, 2002; Nakayama et al., 1994), challenges also exist in bimanual control because of the spatial and temporal dependencies between the two hands (Cauraugh et al., 2010; McCombe Waller & Whitall, 2004). In addition to bimanual coordination impairments, the loss of dexterity of the hemiparetic hand during bimanual tasks can lead to role reversals of one or both hands (Haaland et al., 2012; McCombe Waller & Whitall, 2004). More specifically, the dominant hand can switch from a predominantly manipulative role toward a supportive function, or the nondominant hand may lose its supportive role in favor of using external objects for stabilization.
It is critical to recognize that bimanual and unimanual tasks engage different neural mechanisms (McCombe Waller et al., 2008), and recovery of bimanual function cannot be completely captured by unimanual outcomes (Kang & Cauraugh, 2014). Although bilateral UE training has been investigated and integrated into the clinical practice of occupational therapy, the focus has primarily been on the transfer effect, leading to improved unilateral movements (Cauraugh et al., 2009; Stoykov et al., 2009) but no improvement in bimanual outcomes (Smith & Staines, 2006; van Delden et al., 2012). Moreover, the resulting effectiveness of available bimanual rehabilitation approaches have been inconsistent and even contradictory, which may be due to the use of unimanual outcomes (Cauraugh et al., 2010; McCombe Waller & Whitall, 2008). There is a need to identify both people who may benefit from bimanual training and which specific coordination deficits need to be addressed so that functional outcomes can be maximized.
Most clinical outcome scales have omitted an assessment of bimanual function. The few measures of bimanual function that are available either do not consider, or inadequately consider, prestroke hand dominance, or they lack specificity to a population of adults with stroke (Barreca et al., 2004; Desrosiers et al., 1993; Krumlinde‐Sundholm et al., 2007; Penta et al., 2001; Smith, 1973). As a consequence, there is a gap in stroke rehabilitation by which bimanual functions can be assessed and tracked over time with consideration of the residual capacity of prestroke hand dominance. Without recognizing and monitoring changes in bimanual control, clients may be inaccurately labeled as having reached a therapeutic plateau, thereby leading to prematurely terminated rehabilitation services. Our objective in this study was to develop and test the reliability and validity of a low-cost, easy-to-use instrument, the Bimanual Assessment Measure (BAM), to use with people with chronic stroke (see the Supplemental Appendix, available online with this article at https://research.aota.org/ajot).
Method
Development and testing of the BAM was conducted in two phases: (1) test construction and assessment of face validity and (2) reliability and known-groups validity testing. Approval was obtained from the local institutional review board, and all participants provided informed consent before taking part.
Phase 1: Test Construction and Face Validity
Principles of motor control, motor learning, rehabilitation, and stakeholder input were used to design the BAM. During Phase 1, test construction evolved without formal statistical analysis. The method and results of this phase are presented in chronological order.
Item Selection
Bimanual tasks were first identified on the basis of their importance for community independence. Requests were emailed to stroke participants from previous studies conducted in our department who were >6 mo poststroke and who agreed to be recontacted. Three people with stroke responded to the email and were interviewed using a series of 12 questions, each with subquestions, regarding the way they conduct specific bimanual tasks. The questions differed depending on whether the dominant (e.g., “How do you put toothpaste on a toothbrush?”) or nondominant (e.g., “How do you hold a cup steady while pouring yourself a drink?”) arm was affected. The interviews provided insight into everyday bimanual tasks that were perceived to be both important and challenging. Variations in bimanual task requirements were determined to ensure that different neuromotor control requirements were represented in the measure (e.g., symmetrical vs. asymmetrical movement), and key bimanual resources (Woytowicz et al., 2016) in body structure and function (e.g., fine motor vs. gross motor) were matched to each task (Table 1). This process resulted in six tasks: (1) using a knife and fork, (2) taking a lid off of a plastic container, (3) holding a tray while removing a cup, (4) opening a water bottle (changed to opening a pill bottle in a later iteration), (5) catching a falling object, and (6) zipping up a zipper.
Final List of BAM Tasks and Bimanual Components of Motor Control
Note. BAM = Bimanual Assessment Measure.
Instructions
The test instructions were designed to account for the prestroke roles of the paretic and nonparetic hands given that dominant and nondominant hands differentially contribute to interlimb coordination (Sainburg, 2005). Therefore, the BAM starts with questions regarding hand dominance and the affected side. Participants were then initially asked to perform the tasks without prompts for hand roles.
Scoring
Three scoring criteria were chosen: (1) spontaneous role of the affected hand, (2) timing, and (3) quality of movement. Spontaneity was scored on three ordinal levels (affected hand used in the prestroke role as manipulator, used as stabilizer, or not used). Timing was scored on three levels (task completed in <15 s, 16–30 s, or >30 s, when indicated). The timing cutoffs were established by determining the time in which a healthy older adult completed the task (lower limit) and the time in which participants with stroke completed the task (upper limit). Movement quality was scored on the following six ordinal levels: (1) task completed with both hands without external assistance; (2) task completed with both hands and external assistance for setup only; (3) task completed with both hands and external assistance or excessive trunk movement during task execution; (4) task partially completed with both hands, with or without external assistance; (5) task completed without any use of the affected hand; and (6) task completed partially, without any use of the affected hand.
A scoring method that accounted for prior hand dominance was derived for the three scoring criteria. The score for the first attempt, without instructions for hand role, is referred to as BAM(A). For tasks in which the affected hand is spontaneously used differently than the prestroke role (i.e., using the dominant affected hand as a stabilizer) or not at all, a second attempt at the task is prompted with instructions to use the affected hand in the prestroke manipulator role or as a stabilizer. This second attempt was recorded as BAM(B) or BAM(C), respectively. Thus, BAM(A) reflects spontaneous use, and BAM(B) and BAM(C) reflect functional capacity for participants who used the hand in a role different from that prestroke. Given that people may not perform all BAM tasks for a given scoring variant, the raw score is converted to a 0-to-100 performance scale relative to the number of tasks performed. Higher scores indicate a higher level of function.
Face Validity
Additional stakeholder input was sought on the initial six tasks and the instructions and scoring and was obtained through a focus group of five occupational therapists (recruited from a local inpatient rehabilitation hospital via email) and individual meetings with five volunteers with stroke. The inclusion criterion for occupational therapists was at least 4 yr of clinical experience working with clients with stroke. People with stroke were recruited via emails sent to previous study participants who were >6 mo poststroke and who had provided permission to be recontacted. For people with stroke, the inclusion criteria were as follows: age 18 to 80 yr, unilateral stroke >6 mo before study participation, and visible voluntary contraction of at least one finger of the affected hand. Exclusion criteria were as follows: brainstem or cerebellar lesion or severe cognitive deficits that would interfere with testing (e.g., inability to follow two-step commands).
Focus group and interview questions were different for occupational therapists and patients with stroke. During the meeting with occupational therapists, questions were focused on whether task items were representative of therapeutically relevant activities involving coordinated bimanual function, suggestions for alternate or additional tasks, clarity of instructions, scoring criteria, and feasibility. During the meetings with patients, questions were focused on which items were challenging or easy, clarity of instructions, and suggestions for alternate or additional tasks. On the basis of this clinician and patient input, the following six tasks were added to the original six tasks: (1) unbuckling a clip belt; (2) lifting a large, heavy object with back support (removed in later iteration); (3) lifting a large, heavy object without back support; (4) writing; (5) clipping a chip bag closed; and (6) closing a large plastic jar.
The revised 12-item BAM was then developed as a mobile web application (Qualtrics online survey software). The app was sent to 40 occupational therapists, recruited via word of mouth across the United States, who had at least 4 yr of clinical experience working with patients with stroke. Survey questions were focused on task clarity and appropriateness, terminology, starting position, instructions, criteria for task completion, and whether the pictures enhanced written instructions. Questions were rated on a Likert scale that ranged from 1 (strongly disagree) to 4 (strongly agree.)
Feedback from occupational therapists (n = 40) via the online survey was analyzed qualitatively and is presented in Table 2 as mean percentiles across all tasks. The occupational therapists had a mean of 10.35 yr (SD = 8.04) of clinical experience. The 26 participants who reported their state of residence were spread across the United States: Maryland (n = 10), Wisconsin (n = 8), Arizona (n = 2), Georgia (n = 1), Indiana (n = 1), Illinois (n = 1), Nebraska (n = 1), New Jersey (n = 1), and New York (n = 1). There was mostly positive support for the list of tasks (see Table 2) and scoring criteria. More than 91% of responses were rated as “agree” or “strongly agree” for all survey categories.
Survey Ratings From 40 Occupational Therapists on BAM Tasks 1–11
Note. All table values are percentages. BAM = Bimanual Assessment Measure.
Phase 2: Psychometric Evaluation
Phase 2 consisted of psychometric evaluation of the BAM(A) for internal consistency, interrater reliability, intrarater reliability, test–retest reliability, and known-groups validity.
Participants
Participants included occupational therapists with >1 yr of experience and patients with and without stroke who were ages 18 to 80 yr. Inclusion criteria for patients with stroke were as follows: unilateral lesion >6 mo before study participation and visible voluntary contraction of at least one finger of the affected hand. Exclusion criteria included brainstem or cerebellar lesion or severe cognitive deficits that would interfere with testing. Participants in the age-matched control group were within 5 yr of age of an included participant with stroke. Participants were excluded if they demonstrated other neurological or musculoskeletal deficits that could affect study participation.
Procedure
To determine test–retest, intrarater, and interrater reliability, participants with chronic stroke (n = 25) completed the BAM during two in-person visits (<2 wk apart). The BAM was administered by the same study team member (Brian P. Johnson), and each visit was videotaped. Afterward, the occupational therapists (n = 5) individually scored BAM(A) participant videos during two visits (<2 wk apart).
To assess known-groups validity, the BAM was administered to a group of people with stroke and a group of age-matched control participants during a single study visit (n = 25). These participants were recruited via word of mouth and digital advertisements.
Statistics
We used IBM SPSS Statistics (Version 25) for all statistical analyses. Internal consistency between scores on the BAM(A) items was determined via interitem correlations using Cronbach’s α. Interrater and intrarater reliability were analyzed via two-way mixed-effects model intraclass correlation coefficients (ICCs). Test–retest reliability was assessed using Pearson’s correlation between the BAM(A) scores during Visit 1 and those during Visit 2. Known-groups validity was determined by assessing total BAM(A) score differences between the participants with stroke and age-matched controls using an independent t test. The level of significance was set at α = .05.
Results
Overall, 24 people with chronic stroke and 23 age-matched controls completed the BAM, with 5 occupational therapists scoring the recorded videos of BAM tasks (Table 3). As indicated by the average Wolf Motor Function Test score of 43/66 with a standard deviation of 26, the functional ability of patients with stroke in this study ranged from low to high functioning on the basis of a midpoint cutoff score of 33 (Wolf et al., 2001). The occupational therapists had an average of 6.40 yr (SD = 3.36) of clinical experience.
Participant Characteristics
Note. Dashes indicate not applicable. F = female; L = left; M = male; R = right; UE = upper extremity; WMFT = Wolf Motor Function Test.
Internal Consistency
One task (lifting a large, heavy object without back support) was removed because of a high intertask correlation (r = .99) with another task (lifting a large, heavy object with back support), indicating redundancy. This item removal resulted in 11 total tasks in the final version of the BAM (see Table 1). Internal consistency for the final 11 tasks was found to be excellent (Cronbach’s α = .97). Intertask correlations ranged from .54 to .89 (Table 4).
BAM Intertask Correlations
Note. BAM = Bimanual Assessment Measure.
Reliability
Interrater reliability of the BAM(A) was excellent during the first (ICC = .95) and second (ICC = .99) visit and was excellent for the mean of all raters’ intrarater reliability (ICC = .95) and excellent for four of the five individual rater scores (ICCs = .98, 1.0, .99, .78, and .99 for Raters 1–5, respectively). Test–retest reliability was excellent (r = .94, p < .0001).
Validity
Participants with stroke had significantly lower scores on the BAM(A) (M = 61.63%, SD = 31.42%) compared with age-matched controls (M = 95.30%, SD = 3.69%), t(45) = –5.21, p < .0001).
Discussion
A stroke affects control of contralesional and, to a lesser extent, ipsilesional UEs (Johnson & Westlake, 2020), as well as bimanual control (McCombe Waller & Whitall, 2004). Bimanual control is needed for most activities of daily living, such that bimanual function is important to assess and treat in people with stroke undergoing physical rehabilitation. Depending on whether the dominant or nondominant UE is more affected by the stroke, a person may experience role reversals of their hands (i.e., the dominant hand switches from being used in a manipulative role to a supportive role). It is therefore clinically important to evaluate bimanual function with consideration of both pre- and poststroke hand roles.
In this study, we chronicled the development and initial psychometric testing of the BAM, a relatively quick measure of bimanual function in people with chronic stroke. Although measures of bimanual functioning exist (Barreca et al., 2004; Desrosiers et al., 1993; Krumlinde‐Sundholm et al., 2007; Penta et al., 2001; Smith, 1973), the BAM is different in that it includes a range of neuromotor control requirements in addition to preferred-versus-prestroke hand roles. In Phase 1 of this study, we designed the BAM to incorporate suggestions made by people with a history of stroke and occupational therapists, thereby establishing face validity. In Phase 2, excellent internal consistency, interrater, intrarater, and test–retest reliability, as well as known-groups validity, were established.
During Phase 1, we drew on our previously defined framework of bimanual control to narrow down the list of items that had been identified during patient and clinician focus groups (Woytowicz et al., 2016). Within this framework, we defined symmetrical (each hand performs the same activity) and asymmetrical (each hand performs a different activity) movements that are defined by spatial, temporal, and force characteristics. As part of the BAM, we identified functional tasks using the same terminology defined in our previous framework: hands symmetric in phase (mirror-image spatial and temporal movements with equal forces), asymmetric complementary with a common goal (spatial and temporal movements are complementary with distinct stabilizing and manipulating forces), and asymmetric independent with an uncommon goal (spatial, temporal, and force demands of each hand are independent; see Table 1). Although the majority of laboratory-based investigations of bilateral arm and hand function after stroke have focused on symmetrical tasks (McCombe Waller & Whitall, 2008), we recognize that the challenge in measuring asymmetric functional tasks is a lack of measurement tools. The BAM is a novel tool that may address this gap.
Both occupational therapists and participants with stroke agreed that the BAM appears to measure what it intends to measure (i.e., bimanual function), thereby establishing face validity. By gaining and integrating the viewpoints of people who will be directly involved in the use of the BAM (i.e., occupational therapists and people with stroke) at several points throughout development, the BAM task items and scoring were identified as clinically relevant, meaningful, and feasible for use in clinical settings. Obtaining such stakeholder input also increases the likelihood that clinicians will be willing and able to use the BAM and that people with stroke may be more inclined to find meaning in the results of the assessment.
During Phase 2, the BAM was found to have high internal consistency and excellent interrater, test–retest, and intrarater reliability, as well as known-groups validity. After removing one of the two task items that demonstrated a correlation of .99 (indicating redundancy in measurement), its internal consistency was found to be excellent, with correlations between individual task items ranging from .54 to .89. These values suggest that the properties of the test are consistent and are similar to what has been reported for other unilateral UE measures after stroke, such as the Wolf Motor Function Test (Morris et al., 2001), the Action Research Arm Test (Nijland et al., 2010), and the Fugl-Meyer Assessment (Lin et al., 2004). The excellent reliability scores also indicate consistent test properties such that day-to-day variations in test scores measured within the same client, by the same clinician, or among different clinicians administering the assessment are minimal. It is likely that the task photos and detailed instructions, including starting position, task execution, and completion of each task, contributed to psychometric stability. Finally, the BAM was found to demonstrate good known-groups validity, meaning that the BAM score can reasonably differentiate bimanual function in people with and without sensorimotor deficits leading to impaired bimanual function after stroke.
Limitations
A limitation of our study is that the selected tasks do not necessarily capture meaningful tasks across different cultures and occupations. Our intent in creating the BAM was to include clinically meaningful tasks that each captured different bimanual neuromotor control requirements. Therefore, we consulted with diverse groups of people with stroke and occupational therapists to generate an initial list of meaningful and challenging bimanual tasks. Of these tasks, we chose ones that would each capture distinct neuromotor control requirements (e.g., symmetrical forces and timing, asymmetrical forces and symmetrical timing, asymmetrical forces and timing, time-critical reflexive movement), as described in Table 1. A future version of this measure could become more meaningful in different cultures and occupations by establishing a list of alternative tasks for each neuromotor control category. The rubric for the BAM is already standard, so it can be easily applied to the alternate tasks.
Two additional limitations of our study are (1) that the reliability testing was conducted using video recordings rather than real-time assessments and (2) the relatively small convenience sample. Although the video recording method ensures that assessments of the same patient can be assessed by multiple therapists without patient fatigue, as done elsewhere (Kopp et al., 1997; Morris et al., 2001), future studies should consider an assessment of real-time reliability. Our sample size for the reliability testing, although small, is on par with what has been used for initial psychometric testing of other UE tests (Desrosiers et al., 1993; Morris et al., 2001). Nevertheless, future reliability and validity studies should include a larger sample size to ensure that the current results can be replicated.
Future Directions
As a future goal of this research, we aim to develop a mobile app–based version of the BAM and conduct further validation testing of it. As part of the app format, we will provide users with narrative interpretations of test scores and treatment plan suggestions. The clinician may then effectively use these outcomes to define and validate a bimanual rehabilitation protocol. Future research will also need to establish other psychometric properties of the BAM, such as discriminant and convergent validity, minimal detectable change, and minimal clinically important difference. Moreover, as stated earlier, future versions of the BAM may include an expanded list of alternate tasks that add to the cultural and occupational meaningfulness of this assessment tool.
Implications for Occupational Therapy Practice
As a clinical assessment tool, the implications of the BAM for occupational therapy practice are twofold. It can be used to identify (1) deficits in bimanual control that are above and beyond unimanual deficits and (2) the capacity of a person with stroke to regain their prestroke hand dominance. Allowing a person to perform a task with their preferred hand roles and then requesting that they reverse hand roles to prestroke dominance allows therapists and clients to identify a possible mismatch between what is preferred versus what is possible. Such knowledge can be used to guide therapeutic decision making as to whether a focus on prestroke hand roles or compensatory training in preferred poststroke hand roles is more beneficial. In either case, the BAM will be able to provide an objective outcome regarding impairments in bimanual function.
Conclusion
Using a multiphase process, this study has established that the BAM is clinically relevant, internally consistent, and psychometrically reliable. It can therefore be used to measure bimanual functioning in people with chronic stroke.
Supplemental Material
Supplementary material for Development and Psychometric Testing of the Bimanual Assessment Measure for People With Chronic Stroke
Supplementary material, sj-pdf-1-aot-10.5014_ajot.2022.048995.pdf for Development and Psychometric Testing of the Bimanual Assessment Measure for People With Chronic Stroke by Brian P. Johnson, Jill Whitall, Sandy McCombe Waller and Kelly P. Westlake in The American Journal of Occupational Therapy
Footnotes
Acknowledgments
This research was partially funded by a Qualtrics Behavioral Research Award to Brian P. Johnson. We thank Nicole Shires and Cindy Uruburo for assisting in the early literature review and item selection and Gregory Hancock for methodological consultation.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
