Permission for translating the FMA to the Korean language from the original form of the FMA was obtained from the University of Gothenburg, which possesses the copyright of the FMA. Forward and reverse translations were performed by experienced translators, and evaluation, revision, and cognitive debriefing of the translated product were conducted by a translation committee comprising eight experts including physiatrists, physical therapists, and occupational therapists. Forward translation was performed separately by two translators fluent in both English and Korean after receiving an explanation regarding the questionnaires of the FMA. The translation committee then created K-FMA version 1.0 with the aim of resolving any inconsistencies between the two versions.

Another bilingual translator who was fluent in both English and Korean conducted reverse translation of K-FMA version 1.0 into English [21]. The translation committee compared the forward and reverse translations to examine whether K-FMA version 1.0 and the original version of the FMA were consistent. In consistencies were resolved by the translation committee, and K-FMA version 2.0 was created. Cognitive debriefing was performed on K-FMA version 2.0 to identify other conceptual issues that could cause confusion, and the final version of the K-FMA was established by resolving these issues (Appendix 1).

Throughout the translation process, medical terms were selected from Medical Terminology, which was published by the Korean Medical Association and the KARM. In addition, external advisers from the KARM also consulted on version 1.0 and version 2.0.

A total of 34 stroke survivors who were admitted to a rehabilitation hospital were recruited between November 2019 and April 2020. The inclusion criteria were as follows: (1) first diagnosis of haemorrhagic or ischaemic stroke, (2) age of >19 years, (3) Brunnstrom stage of upper or lower extremity from 2 to 4, and (4) sufficient cognitive function to allow understanding and following the instruction of the researcher. Participants with history of psychiatric or neurological diagnosis, severe aphasia, decreased level of consciousness, disorders affecting movement other than stroke such as motor neurone disease, Parkinson’s disease, polyneuropathy, or mononeuropathy; and uncontrolled comorbidities such as cancer and severe organ failure were excluded. Among the 34 participants who were recruited, three were transferred to other hospitals for further treatment during the course of the study; thus, the data of 31 participants were finally analysed in this study to verify the reliability and validity of the K-FMA. This study was approved by the Institutional Review Boards of the National Rehabilitation Center (NRC-2019-03-017), and all participants provided written informed consent prior to enrolment.

A prospective evaluation was conducted to verify the intra-rater reliability, inter-rater reliability, and concurrent validity of the final version of the K-FMA. According to a test-retest method for the analysis of reliability, two experienced physical therapists, evaluator 1 and evaluator 2, measured all participants with the K-FMA, and repeated the measurement after 1 week. The two evaluators performed evaluations on the same day. For intra-rater reliability, the first and second assessments of each evaluator were compared, and for inter-rater reliability, the first K-FMA results of each evaluator were compared. For robust comparisons, the second K-FMA results of each evaluator were also compared. Concurrent validity analysis was carried out using Pearson correlation test of the mean value of K-FMA and other evaluation tools for patients with stroke in clinical areas, the Motricity Index (MI), MAS, and Berg Balance Scale (BBS).

The original FMA consists of four domains: motor, sensory, joint range of motion (ROM), and joint pain [5]. Each evaluation item is graded on a 3-point ordinal scale from 0 to 2, with a higher score representing less impairment [5]. In the motor function domain, the upper extremity (UE) motor function evaluation consists of 33 items with scores ranging from 0 to 66, and the lower extremity (LE) motor function evaluation consists of 17 items with scores ranging from 0 to 34. The domain of sensory function includes four items evaluating the light touch sensation of the UE and LE and eight items assessing the proprioception of the UE and LE. The sensory function score ranges from 0 to 24. The domains of joint ROM and joint pain are composed of 12 items for UE and 10 items for LE, and the scores range from 0 to 44 in both domains.

The MI is an index used to measure limb strength, consisting of three items each in the UE (pinch grip, elbow flexion, and shoulder abduction) and LE (ankle dorsiflexion, knee extension, and hip flexion) [22]. Scores in the MI are calculated based on the sum of the scores of the three items in the UE and LE each, and range from 0 (complete paresis) to 100 (normal strength) [22].

The MAS is an assessment tool for evaluating the functional capabilities of patients with stroke, which consists of eight motor items: supine to side lying onto intact side, supine to sitting over the side of the bed, balanced sitting, sitting to standing, walking, upper arm function, hand movement, and advanced hand activities [12]. Each motor item is rated on a 7-point scale, ranging from 0 to 6, with higher scores representing lesser impairment. The total MAS score ranges from 0 to 48 [12].

The BBS is a quantitative evaluation tool consisting of 14 items measuring static and dynamic balance while sitting, standing, or shifting weight [23,24]. Each item is graded on a 5-point scale from 0 to 4, and 4 points represent independent completion of the item. The total score is a sum of the scores obtained for each item, out of an overall score of 56 [23].

The intra-rater reliability and inter-rater reliability of the K-FMA were evaluated with the intraclass correlation coefficient (ICC) with values between 0.74 and 1.0 representing excellent reliability, between 0.60 and 0.74, good reliability; between 0.40 and 0.59, fair reliability; and below 0.4, poor reliability [25]. The concurrent validity of the K-FMA was tested using Pearson’s correlation test and compared with the MI, MAS, and BBS scores. A correlation between the K-FMA and other evaluation tools with a correlation coefficient of 0.7 or higher was regarded as highly positive [26]. A value of p<0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics version 20.0 for Windows (IBM SPSS, Armonk, NY, USA).

Subjects’ baseline characteristics including sex, age, and duration after stroke onset were collected. The demographic characteristics and results of other evaluation tools of the participants are presented in Table 1.

The intra-rater reliabilities are show in Table 2. All domains had ICC >0.74, except passive ROM, in which variation was found for the LE, and joint pain. All domains were statistically significant at p<0.05. According to the results of intra-rater reliability of evaluator 1, the total motor score recorded the highest ICC (0.961), and the passive ROM of the LE showed the lowest ICC (0.596). Similarly, as per the intra-rater reliability results of evaluator 2, the motor function of the UE had the highest ICC (0.977) and the passive ROM of the LE had the lowest ICC (0.512).

The inter-rater reliability results between evaluators 1 and 2 are provided in Table 3. The assessment was divided into two parts: part 1 was the first-test comparison and part 2 was the retest comparison. The analysis found all ICCs >0.74, except the passive ROM for LE and joint pain for the UE, and all these results were statistically significant at p<0.05. Tactile sensitivity for the UE had one of the highest ICCs (0.995) in both parts 1 and 2. Joint pain for the LE had the highest ICCs in both parts 1 and 2 (0.851 and 0.997, respectively). The LE passive ROM had the lowest ICC in both parts 1 and 2 (0.782 and 0.498, respectively).

Results of the concurrent validity analysis of the K-FMA with MI, MAS, and BBS with Pearson correlation coefficient are provided in Table 4.

The total score (sum of all four domains) was significantly correlated with the MI, which had the highest correlation of all evaluation tools examined (r=0.788). It was also significantly correlated with the MAS, with a comparatively lower correlation coefficient (r=0.496). The total K-FMA score was not significantly correlated with the BBS score (p=0.22), unlike the other tools. The total K-FMA scores for the UE and LE showed similar results. The total score for the motor domain (sum of the UE and LE motor scores) was also significantly correlated with the MI, which had the highest correlation coefficient (r=0.891) among all evaluation tools. The total score for the K-FMA motor domain was also significantly correlated with the MAS score (r=0.613). The motor-domain subscores for the UE and LE showed similar significant correlations. None of the K-FMA motor domains showed significant correlations with the BBS score (p=0.22), similar to the K-FMA total score.

The total score and all domains of the K-FMA showed significant intra- and inter-rater reliabilities, with good to excellent reliability in most domains of the K-FMA except joint ROM for LE. In the intra-rater reliability analysis, the sum of the motor functions of the UE and LE in evaluations by evaluator 1 and the motor function of UE in the evaluation by evaluator 2 showed the highest degree of reliability, whereas passive ROM of the LE had the lowest reliability in the evaluations of both evaluators. Regarding inter-rater reliability, the tactile sensitivity of the UE and joint pain of the LE represented the highest reliability, while the passive ROM of the LE had the lowest reliability in both test and retest.

Similar to previous findings, the motor domains of the K-FMA (total motor, UE motor, and LE motor scores) showed excellent reproducibility in intra- and inter-rater reliabilities [7-10]. The motor domain of the FMA is often used alone for evaluating motor function because of its superiority and consideration of the time required [6].

The sensory domain of the K-FMA showed excellent intra-rater and inter-rater reproducibility, and the joint pain domain showed good to excellent reliability, although these two domains rely on the subjective appeal of the participants, which can result in loss of objectivity [6].

Conversely, the ICC for the domain of joint ROM for LE was low in both the intra-rater and inter-rater analyses. Since the intra-rater reliability was low, the inter-rater reliability was also low. The result of the intra-rater reliability analysis might be influenced by inconstant spasticity and change in the ROM after interventions. In particular, changes in ROM according to the pre- and post-physical therapy status might significantly impact the test results, although there were no changes in other functions. Additionally, there was no unified training programme on standardisation of the evaluation of LE joint ROM for evaluators before this study began. In a previous study on the Brazilian version of the FMA, the ICC value of the domain for joint ROM for LE improved significantly from 0.50 (p>0.05) to 0.90 (p<0.001) in the inter-rater reliability analysis after the evaluators adapted to the improved and detailed manual attached with illustrating photographs [7]. After the adaptation of the standardised manual, the reliability of the sensory function domain was also enhanced. Another study reported excellent intra-rater and inter-rater reliabilities of the FMA with a standardised manual and rater training programme [10]. A study suggested the use of a standardised manual for FMA providing detailed instructions on the test process, grading, patient guidelines, and equipment [11]. Hence, we also translated the original manual for the FMA into Korean with standardisation of the measurement instructions to increase the reliability of the K-FMA. Similar to the process for the translation of the FMA, with consent from the original author for usage of the original manual, the manual for the K-FMA was prepared through forward translation, revision, reverse translation, and cognitive debriefing. Images of postures for evaluation were created and added to the manual (Appendix 2). We expect this manual to be widely used to improve the reliability of the K-FMA.

The MI and MAS scores were statistically significantly correlated with the K-FMA score when analysed with Pearson correlation test, while the BBS score was not. This result was expected, considering that the MI and MAS are evaluation tools for motor function. Specifically, the MI showed a highly positive correlation with the K-FMA score, but the MAS showed a moderately positive correlation, probably because there were no evaluation items for the trunk muscles or balance in the K-FMA, unlike in the MAS, which has items to evaluate the trunk muscles and balance, such as supine to sitting over the side of the bed or balanced sitting, whereas the MI consists of items for evaluating muscles of the extremities. Moreover, the FMA and MI both evaluate patient impairment, unlike the MAS, which evaluates activities such as gait and movements of the extremities. Similar to the modified Rivermead Mobility Index, the FMA also showed a statistically significant correlation with the tools evaluating activities such as the Modified Barthel Index [27,28]. Accordingly, the FMA is considered useful for evaluating both motor function and the mobility of patients with stroke.

Although the BBS, which is a widely used evaluation tool for measuring balance, was previously reported to have moderate to good correlation with the motor function of the FMA, the results of our study showed no statistically significant correlation between the K-FMA and BBS [29]. In another study, the FMA did not show a statistically significant correlation with the evaluation tool regarding balance ability, including static balance, dynamic balance, and weight distribution asymmetry indices, which was consistent with the results of our study [30]. It is considered that classifying the FMA and BBS together as similar types of assessment tools for motor function would be difficult, since the K-FMA mainly focuses on motor and sensory function evaluation and does not include balance evaluation items. The FMA is especially valuable for evaluating recovery of motor impairment after stroke, but other domains are considered inadequate, as they do not correspond to the intended aim of the FMA [6].

There are some limitations to this study. First, it was difficult to generalise the results of this study because only participants with Brunnstrom stages 2–4 were recruited, whereas those with Brunnstrom stages 1, 5, and 6 were not. However, considering that patients in Brunnstrom stages 2–4 have the greatest variability, which affects the reliability and complicates the evaluation, the high reliability in our study implies that the K-FMA can be used in the clinical field. Second, raters were not trained to use a standardised manual for the K-FMA and evaluated patients based on their own standards, which could have adversely affected the objectivity of the evaluation. Improvement of reliability would be expected with the application of a standardised manual for the K-FMA, as in previous studies [7,10]. Nevertheless, the K-FMA was verified as an evaluation tool with high reliability and validity for patients with stroke and can be used effectively in the domestic rehabilitation field. We also produced a manual of K-FMA containing a training video and graphic illustration in each evaluation process as a standardised measurement method. Additional studies are being conducted on the effectiveness of training using this manual, and the manual is expected to play a vital role as a standardised training guideline.

In conclusion, the FMA and the manual for the FMA were translated into Korean through standardised translation and several verification processes, and this newly translated K-FMA was verified as a tool for sensorimotor evaluation in patients with stroke with a high level of reliability and validity.

The K-FMA is expected to be used as a standardised tool to replace the original FMA in Korea with high reliability and validity. Additionally, the newly developed manual for the K-FMA is anticipated to minimise errors that can occur in the evaluation and help improve the reliability of motor function evaluation in patients with stroke.