The Action Research Arm Test (ARAT) is an observational activity-based measure used to determine the quality of upper limb function that was developed by Lyle (1981) as a modification of the Upper Extremity Function Test. The test consists of 19 items encompassing 4 subtests (grasp, grip, pinch, and gross arm movement) where the subject’s performance on each item is rated along a 4-point scale (0 to 3) where (0) equates cannot perform any part of the test, (1) can partially perform the test, (2) can complete the test but took abnormally long or had great difficulty, and (3) can perform the test normally. A maximum score of 57 is considered for each extremity tested and indicates the absence of any dysfunction (Koh et al., 2006). The test has a hierarchical design in which the subject attempts the most difficult item of each subtest first. If subjects correctly complete the first, most difficult item, they are positively credited for all items of the subtest without having to be tested any further (McDonnell, 2008). Thus not all items need to be tested. If the patient fails the first most difficult item the second item tested is the easiest, if he or she receives a score of 0 for the easiest item (i.e., fails) the assessor moves onto the next subtest and so on until all subtests have been attempted (McDonnell, 2008). While the test requires substantial set up as well as a fair amount of equipment, once set up it can be completed in about 10 minutes.
A study by Nordin, Alt-Murphy, and Danielsson (2014) using 2 raters with 35 subjects ≤ 22 months post-stroke found that the median ARAT total score for raters A and B was 37 (range; 3 to 54) and 38 (range; 3 to 56). Intrarater reliability for total score was r = 0.94 for each of the raters. At the subtest level, a satisfactory level of agreement was also found within each rater for all 4 subtests, ranging from 0.91 to 0.94 for the subtests grasp and pinch, and 0.83 to 1.0 for the subtests grip and gross movement. Inter-rater reliability for the ARAT total score was r = 0.91 and 0.99, respectively while inter-rater agreement at the subtest level ranged from 0.81 to 0.91 for grasp and pinch, and from 0.86 to 1.0 for grip and gross movement. An earlier study by Platz et al. (2005) found a significant correlation (0.92) between the ARAT, the Fugl-Meyer Motor Assessment upper limb section, and the Box and Block Test. The same study also showed that the ARAT had moderate to significant correlations with the Hemispheric Stroke Scale (0.66), suggesting that it too is measuring the same phenomenon as other established assessments.
No special training is required for administration and once set up the assessment can be completed in < 10 minutes for most patient populations. The ARAT has also been shown to have good responsiveness evidenced by its ability to detect clinically relevant changes in motor function during the acute phase following stroke as well as in patients with chronic conditions (McDonnell, 2008). Another advantage of the ARAT is its unique hierarchical format and scoring design where once an item is completed successfully it is assumed that all easier items can also be completed, which can shorten testing time without sacrificing a valid score. Conversely, if the easiest item is failed it can be assumed that any attempt at a harder item will also result in failure (McDonnell, 2008).
Koh et al. (2006) examined the construct validity of the current 19-item ARAT and found that although it closely resembled the original design and displayed a consistent hierarchical order relative to continuum of motor function the pinch ball bearing with 3rd finger and thumb item was inconsistent with other items in its subtest and its removal should be considered. The poor fit of this item as well as others were also discussed in a study by van der Lee, Roorda, Beckerman, Lankhorst, and Bouter (2002) in which they argued that although overall the assessment displayed a good uni-dimensional hierarchy with a scalability coefficient of 0.79 (range = 0.69 to 0.86) the removal of four items would allow even higher values of the scalability coefficient H.
Testing procedures, scoring, and materials have been standardized and outlined in an article by Yozbatiran, Der-Yeghiaian, and Cramer (2008), where the full notation is under permissions. Conversely, a testing kit can be purchased from a number of therapy supply retailers. The complete list of items needed are a chair without armrests, a table, various-sized wooden blocks, a cricket-type ball, a sharpening stone, alloy tubes, a washer and bolt, 2 glasses, marbles, ball bearings, 2 planks for placing the alloy tubes, 1 plank to place the washer, 2 tobacco tin lids, and 1 37-cm high shelf (Yozbatiran et al., 2008). During assessment, the subject is seated upright and each of the 19 tasks are performed unilaterally until completion of the task or until reaching a time limit defined as 60 seconds (Yozbatiran et al., 2008) The subject starts with the least affected arm first where quality of movement is scrutinized.
• Pinch ball bearing with 3rd finger and thumb
• Pinch marble with 3rd finger and thumb
• Pinch ball bearing with 2nd finger and thumb
• Pour water glass to glass
• Grasp block (10 cm3)
• Pinch ball bearing with 1st finger and thumb
• Pinch marble with 2nd finger and thumb
• Pinch marble with 1st finger and thumb
• Grasp block
• Grip washer over bolt
Adapted from Koh, C. L., Hsueh, I. P., Wang, W. C., Sheu, C. F., Yu, T. Y., Wang, C. H., & Hsieh, C. L. (2006). Validation of the Action Research Arm Test using item response theory in patients after stroke. Journal of Rehabilitation Medicine, 38, 378.