Validity and Reliability of the Korean-Translated Version of the International Cooperative Ataxia Rating Scale in Cerebellar Ataxia
Article information
Abstract
Objective
The International Cooperative Ataxia Rating Scale (ICARS) is a semiquantitative clinical scale for ataxia that is widely used in numerous countries. The purpose of this study was to investigate the validity and reliability of the Korean-translated version of the ICARS.
Methods
Eighty-eight patients who presented with cerebellar ataxia were enrolled. We investigated the construct validity using exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). We also investigated the internal consistency using Cronbach’s α and intrarater and interrater reliability using intraclass correlation coefficients.
Results
The Korean-translated ICARS showed satisfactory construct validity using EFA and CFA. It also revealed good interrater and intrarater reliability and showed acceptable internal consistency. However, subscale 4 for assessing oculomotor disorder showed moderate internal consistency.
Conclusion
This is the first report to investigate the validity and reliability of the Korean-translated ICARS. Our results showed excellent construct and convergent validity. The reliability is also acceptable.
Ataxia is a term for impairment in the coordination of movement and presents as a disorder involving complex multiple functional movements. Due to various clinical presentations, it is difficult to objectively measure the severity of ataxia. The International Cooperative Ataxia Rating Scale (ICARS) was first developed in 1996 and has been widely used for semiquantitative measurement of ataxia [1]. The ICARS is composed of 19 items and 4 subscales for assessing gait and ability to stand, motor function, speech, and eye movements. The ICARS was originally written in English and required translation and validation for use in countries where other languages are spoken [2]. However, there has been no report on the validity and reliability of the Koreantranslated version of the ICARS. As the prevalence of patients with ataxia in Korea has been increasing, it is necessary to validate the Korean-translated version of ICARS.
MATERIALS & METHODS
Translation process
The translation and back translation methods were applied for translation to Korean [3]. The group of translators was composed of two Korean neurologists, one of whom had lived in an English-speaking country for 3 years. The group of back translators was composed of a Korean neurologist and a nonmedical native speaker of English. After backtranslation, all translators reviewed and compared the translated piece to the original version, and any errors were corrected with consideration of Korean cultural differences. The final translated version was supervised by the Korea–US education and cultural exchange association called KoAm Education Alliance (Supplementary Material in the online-only Data Supplement).
Participants
This study was a multicenter cross-sectional study. The inclusion criteria were patients aged 20–80 years who presented with cerebellar ataxia. Exclusion criteria included the presence of 1) other neurological and orthopedic symptoms that affect movement and gait.; 2) severe fall risk, deeming patients unable to undergo ICARS; and 3) fluctuating symptoms within one month.
All participants performed the Korean translated version of the ICARS, Timed Up and Go (TUG) test and the Korean Tinetti mobility test (TMT) [4,5]. Informed consent was obtained from all participants, and this study was approved by the Institutional Review Board (IRB number: HP2021-05-04).
Statistical methods
Validity
For construct and convergent validity, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were performed. Before performing EFA, the Kaiser–Mayer–Olkin and Bartlett’s tests were used for sampling adequacy. To assist in interpreting the factors, varimax orthogonal rotation was used. Before constructing a common factor model, initial eigenvalues from a screen plot were used as guidelines for deciding the number of factors. Item loading was used with absolute values greater than 0.4 to describe the factors.
For CFA, five model fit indices and their criteria were used to examine the goodness-of-fit. We evaluated the minimum chisquare/degrees of freedom (CMIN/DF), Tucker–Lewis index (TLI), comparative fit index (CFI), standardized root mean square (SRMR), and root mean square error of approximation (RMSEA) as model fit indices. CMIN/DF values below 3, SRMR values below 1.0, RMSEA values below 0.8, and CFI and TLI values above 0.9 were considered to indicate acceptable model adjustment [6]. For concurrent validity, Spearman’s rank correlation coefficient with the TMT, TUG and disease duration was calculated.
Reliability
Cronbach’s alpha coefficient was used to calculate internal consistency. A value above 0.7 was considered good, and above 0.6 was considered moderate [7]. The intraclass correlation coefficient (ICC) was used for intrarater and interrater reliability, and above 0.75 was considered good [8]. For interrater reliability, the test–retest method was used. The two raters facilitated the ICARS at the same time, and one of them performed the ICARS again in the same set of patients within 4 weeks.
RESULTS
Participants
We enrolled 88 patients in 6 different movement clinics in tertiary hospitals in this study. All subjects were Korean, and the mean age was 61.25 years; 56.8% of the participants were men. Causative diseases included multiple system atrophy (53.4%), spinocerebellar ataxia (18.2%), idiopathic cerebellar ataxia (17%), postinfectious cerebellar ataxia (2.3%), and others (15.7%).
Construct and convergent validity
Table 1 shows the factor loading of each item derived from EFA. Subscale 2 has three factors, and a total of six factors were extracted in the ICARS from EFA. The factor loading of each item showed a high correlation with other items that belonged to the same subscale. However, Item 14 under subscale 2 showed a relatively low correlation with other items in the same subscale (0.5), but it showed a high correlation with items in subscale 3. For CFA, all the fit indices were satisfactory: CMIN/DF = 1.519, TLI = 0.908, CFI = 0.924, SRMR = 0.694, and RMSEA = 0.77.
Factor lording in all items in exploratory factor analysis
Criterion validity
Concurrent validity was used for the criterion validity. The total score of the ICARS was significantly correlated with the scores of TMT (r = -0.695, p < 0.01) and TUG (r = -0.308, p < 0.01), indicating high concurrent validity of the total score of ICARS. Subscales 1, 2, and 3 were significantly correlated with TMT (r = -0.820, p < 0.01; r = -0.428, p < 0.01; r = -0.361, p < 0.01, respectively). However, only subscale 1 had a significant positive correlation with TUG (r = 0.621, p < 0.01). The total ICARS score showed a significant correlation with disease duration (r = 0.251, p = 0.02). Subscale 3 was significantly correlated with disease duration (r = 0.280, p < 0.01).
Reliability
Table 2 shows the results of the evaluation of the reliability. Cronbach’s alpha coefficient for the total ICARS score was 0.907, indicating sufficient reliability. Cronbach’s alpha coefficients of subscales 1, 2, and 3 were acceptable (0.916, 0.837, 0.838, respectively); however, that of subscale 4 showed a relatively lower value (0.654). The interrater ICC of the total ICARS score was 0.96, and the intrarater ICC of the total ICARS score was 0.98. Subscale 1 had the highest reliability, while subscale 4 showed the lowest reliability despite an acceptable range.
Crohnbach’s α coefficient for internal consistency and ICC for reliability
DISCUSSION
In our results, the Korean-translated version of the ICARS showed acceptable validity and reliability. There have been two commonly used scales for patients with ataxia, including the Scale for the Assessment and Rating of Ataxia (SARA) and ICARS. Only one report on SARA has been translated into the Korean language and has been validated in stroke patients [9]. There are still no clinical scales applicable for Korean-speaking individuals for cerebellar ataxia except in cases of stroke.
After applying EFA to investigate the factorial structure, the number of factors was 6, with eigenvalues greater than 1 even if the ICARS was composed of 4 subscales. The validation study of English ICARS in SCA revealed that 4 factors had eigenvalues greater than 1 [10]. However, some previous studies have often reported more than 4 factors in factor analysis. The validation study of ICARS in focal cerebellar lesions revealed 5 extracted factors with eigenvalues greater than 1 [11]. In another study, the Turkish version of the ICARS in multiple sclerosis also extracted 5 factors in principal component analyses [12]. Subscale 2 is an assessment of kinetic function that measures movement of the upper and lower extremities. Therefore, factor loadings are grouped according to the movement of body parts, including the lower extremities, arm, and fingers.
Previous studies on the convergent validity of the ICARS have shown inconsistent results [13,14]. In our study, Item 14, which measures the drawing skill of the Archimedes loop, showed relatively low association with other items under subscale 2 in EFA. Similar results regarding Item 14 as a hindrance factor for validity have been reported in previous studies [11]. We postulate several limitations of Item 14 that might cause this result. First, the size of the Archimedes loop can affect the drawing performance; however, there are no exact descriptions of the size of the Archimedes loop in ICARS. Second, a Korean word for “hypermetric swerve” was not often used; therefore, this may be unfamiliar to Korean-speaking individuals.
The English version of the ICARS has been proven to have excellent reliability [15]. Cronbach’s alpha coefficient and ICC for assessing the reliability of total ICARS scores revealed acceptable internal reliability in our results. However, Cronbach’s alpha coefficient and the ICC score of subscale 4 were lower than those of the other subscales despite a moderate range of internal consistency. These findings are also often reported in other reports, suggesting an inherent limitation of ICARS itself and not the translation process [10,13]. It is difficult to quantitatively estimate eye movement by observation because subtle differences in ocular dysfunction can be influenced by the subjective judgment of the observer.
This study has several limitations. First, we only analyzed correlations with scales of gait for concurrent validity. Overall assessment of daily function, such as the modified Barthel index, could not be measured for concurrent validity. Additionally, we could not evaluate other ataxia scales, such as SARA, because there are no validated Korean-translated versions of these scales. Second, our data collection was limited to the clinical symptoms of cerebellar ataxia in patients and not specific diseases.
In conclusion, the Korean-translated version of the ICARS showed excellent validity and reliability for cerebellar ataxia. Although some hindrance factors were identified, these results have often been reported as limitations of the ICARS itself. Our study makes it possible to apply the ICARS to Korean-speaking patients with ataxia in clinical and research settings.
Supplementary Materials
The online-only Data Supplement is available with this article at https:// doi.org/10.14802/jmd.22137.
Acknowledgements
We would like to show our gratitude to Dr. Ji Min Choi and Soo Hyun Lim for their comments that greatly improved our manuscript.
Notes
Conflicts of Interest
The authors have no financial conflicts of interest.
Funding Statement
This study was supported by Corestem Inc.
Author contributions
Conceptualization: Hee-Tae Kim, Joong-Seok Kim. Data curation: Chang- Whan Ryu, Jinse Park, Yoon-Sang Oh. Formal analysis: Wooyoung Jang, Engseok Oh. Funding acquisition: Seong-Beom Koh, Hee-Tae Kim. Investigation: Jee-Young Lee. Methodology: Jinse Park, Hyungyoung Hwang. Project administration: Jin-Young Ahn, Jinse Park. Resources: Jin Whan Cho. Supervision: Jae H. Park. Validation: Jinyoung Youn. Writing—original draft: Jinse Park. Writing—review & editing: Hee-Tae Kim.