Psychometric Properties of the Turkish Version of the Abiloco: Factor Analysis

Introduction

Stroke is a very common neurological disease and ranks third among diseases that cause functional disability (WHO, 2020). Post-stroke disability is often chronic. Stroke patients experience difficulty in performing activities of daily living and have problems in basic mobility activities such as walking and transfers. Therefore, the treatment of the disease is planned to reduce these problems. Various assessment methods such as tests, questionnaires, and gait analysis are used to measure the level of disability (Dobkin, 2006; Lin et al., 2010; Pyo et al., 2020). Scales are simple, inexpensive, and easy to apply in clinics to evaluate any problem.

Questionnaires, a fast, practical, and inexpensive measurement tool, are frequently used to evaluate functional results. Recently, great emphasis has been placed on measurements that collect information directly from patients. Self-reported measurements that offer the possibility of self-assessment provide more realistic data that better reflect the clinical situation. The ABILOCO is a self-reporting scale developed by Gilles Caty in 2008 to evaluate mobility in stroke patients (Caty et al., 2008). The advantage of the ABILOCO is that it has been developed with a statistical model that allows ordinal scores to be converted into linear measures. Linear measurements give more reliable (inter-/intra-individual) and clinically significant results. The ABILOCO is a linear, unidimensional, invariant, valid, and reliable questionnaire.

The ABILOCO scale has been translated into Korean (Lee et al., 2013), Portuguese (Roberto Avelino et al., 2018), Benin (E. S. Sogbossi et al., 2014), and Persian (Didehdar & Kharazinejad, 2020). When a scale is translated, it should be adapted to the society in which it will be used, and the validity and reliability of the translation should be examined in the target society. The ABILOCO scale was developed using the Rasch model, and the versions for other languages were examined by that model as well. Investigating the scales with both the Rasch model and factor analysis increases reliability. The aim of the present study was to translate the ABILOCO scale into Turkish and adapt it to Turkish society using factor analysis and to investigate its reliability, validity, and construct validity.

Method

The ABILOCO is a self-reported measurement tool developed by Caty Gilles in 2008 (Caty et al., 2008). The scale consists of 13 items covering a wide range of movement activities. Patients rate their movement activities according to 3 difficulty levels, namely “impossible”, “difficult”, and “easy”. The overall score ranges from 0 to 26, with a higher score indicating more mobility and independence. Gilles et al. proved that the scale offered good psychometric properties and it is valid and reliable.

The study was approved by the XXX University Non-Interventional Clinical Studies Ethics Committee (#2019/09–43, Project No: GO19/351). The translation of the scale into Turkish and its cultural adaptation were carried out according to the steps suggested in the literature (Kalfoss, 2019; Wild et al., 2005). Permission was obtained from the developer of the scale.

The scale was translated into Turkish by a physiotherapist and a linguist with a good command of English. The translators convened and combined the translations into a single questionnaire. The Turkish scale obtained from the translation was translated back into English independently by two native English speakers fluent in Turkish. A single questionnaire was obtained again from the two English translations and the compatibility of the questionnaire with the original version was examined by a Turkish Language and Literature expert. The final version of the scale was examined in terms of intelligibility via a pilot study. The scale items were evaluated as “I understood”, “I partially understood”, and “I did not understand the question” by a total of 30 people, 15 with a stroke history and 15 healthy. As a result of the feedback, the expression “without bringing the feet together” was added in parentheses to the fourth item and “partial paralysis” was added in parentheses to the seventh and eighth items. In the pilot test, the participants did not indicate any difficulties in understanding the questions and no changes were required to the ABILOCO-Turkish.

Data Collection

The data were collected by the research physiotherapist between May 2019 and May 2021. The informed consent form and the ABILOCO scale forms were filled out by the participants. The demographic information and the Berg Balance Scale (BBS) and Timed Up & Go (TUG) test results were recorded by the research physiotherapist. The BBS evaluates balance and determines the risk of falling (Berg et al., 1989). It consists of 14 tasks for assessment. The participants were asked to complete each task. The TUG evaluates functional mobility and is valid and reliable in elderly and stroke patient groups (Podsiadlo & Richardson, 1991). The patient was asked to get up from a chair and timing started. After walking 3 m, they were asked to return from the designated point and sit on the chair again. The time elapsed was recorded in seconds. One week after data collection, the ABILOCO scale was re-administered to 40 volunteer participants in the same clinic.

Data Analysis

The data were analyzed using SPSS version 23 and SPSS AMOS version 21. The demographic data were assessed by descriptive statistical analyses. Test–retest and internal consistency analyses were performed to test the reliability of the ABILOCO scale. It was analyzed for stability using retest results from 40 participants. The internal consistency of the ABILOCO scale was evaluated by Cronbach’s alpha coefficient and item–total correlation, and test–retest reliability was evaluated by intraclass correlation coefficient (ICC). The acceptable value for the Cronbach’s alpha coefficient and ICC is .70 and above and for the item–total correlation is .30 and above (Nunnally, 1967).

In order to examine the content validity of the ABILOCO-Turkish, the content validity index was calculated (Polit & Beck, 2006). For the concurrent/criterion validity of the ABILOCO-Turkish, the TUG test (Podsiadlo & Richardson, 1991) and the BBS (Berg et al., 1989) were used. There was a close relationship between the ABILOCO-Turkish and the TUG test and BBS. The correlation of the tests applied simultaneously with the ABILOCO-Turkish was analyzed by Pearson’s correlation analysis.

Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were used to examine the basic structure of the Turkish version of the ABILOCO scale. Before the EFA, Bartlett’s sphericity (BS) test was used to test the factorability of the study sample and its suitability for factor analysis, and the Kaiser–Meyer–Olkin (KMO) test was used to decide on the adequacy of the sample (Watson, 2017). The KMO (KMO value ≥0.60) and BS tests (p < 0.05) were performed to determine the suitability and adequacy of the sample before the EFA. - ‘Principal Axis Factoring’ - Since it was reported that the original ABILOCO scale was unidimensional, the Turkish version was also examined in terms of its compliance with unidimensionality. The maximum likelihood estimation method was used to estimate the parameters in CFA. Chi-squared, degrees of freedom (df), χ²/df values, ​​and fit indices (incremental fit index (IFI), normed fit index (NFI), goodness of fit index (GFI), comparative fit index (CFI), and root mean square error of approximation (RMSEA)) were calculated for the scale structure (Schermelleh-Engel et al., 2003). The adequacy of the sample for analysis was determined by the Holter test. In addition, convergent validity was evaluated using the composite reliability (CR) coefficient.

Results

Participants' Characteristics

The sample of the study consisted of 140 participants with a mean age of 61.71 ± 11.78 years, 64.3% male and 35.7% female. While 16.4% of the participants were literate, 67.1% were primary school graduates, 12.9% were high school graduates, and 3.6% were university graduates. Moreover, 81.4% had had ischemic stroke and 18.6% hemorrhagic stroke. The left extremities were affected in 52.9% of the participants and the right extremities in 47.1% and 55% were not using any walking aid in their daily life. The mean MMST score of the sample was 26.64 and the mean time after stroke was 21.52 months (Table 1).

OPEN IN VIEWER

Table 1.

Participants' Characteristics

Characteristics	Value
Sex
Male	90 (64.3%)
Female	50 (35.7%)
Age (years)	61.71 ± 11.78 (20–85)
Bmi (kg/m2)	27.17 ± 3.49 (18.14–38.37)
Educational status
Literate	23 (16.4%)
Primary school	94 (67.1%)
High school	18 (12.9%)
University	5 (3.6%)
Stroke type
Hemorrhagic	26 (18.6%)
Ischemic	114 (81.4%)
Dominant side
Right	131 (93.6%)
Left	9 (6.4%)
Affected side
Right	66 (47.1%)
Left	74 (52.9%)
Use of walking aid
Yes	77 (55%)
No	63 (45%)
MMST	26.64 ± 1.31 (24–30)
Time after stroke (months)	21.52 ± 19.18 (2–120)

Reliability

The reliability coefficient of the ABILOCO scale was calculated as α = 0.96. Cronbach’s alpha reliability coefficients obtained by removing any of the items ranged from 0.957 to 0.966, while item–total score correlations ranged between 0.450 and 0.896 (Table 2). The item–total score correlations of the ABILOCO scale ranged from 0.464 to 0.899. The intraclass correlation coefficient (ICC: 0.989) obtained by the test–retest method showed that the scale is stable and invariant over time. The paired samples t-test showed that there was no significant difference between the first and second measurements (t: −1.727; p: 0.092). In addition, a high correlation was found between the first and second measurements by Pearson correlation analysis (r: 0.980; p < 0.001).

OPEN IN VIEWER

Table 2.

Item–Total Statistics

Item	Scale Mean if Item Deleted	Scale Variance if Item Deleted	Corrected Item–Total Correlation	Squared Multiple Correlation	Cronbach’s Alpha if Item Deleted
1	10.4576	61.508	.450	.464	.966
2	10.2034	57.326	.673	.627	.962
3	9.8253	52.978	.851	.784	.958
4	10.1610	55.768	.738	.728	.960
5	10.0110	54.648	.778	.690	.959
6	9.4253	53.687	.860	.879	.957
7	10.0467	55.139	.807	.780	.959
8	9.7753	52.944	.894	.853	.956
9	9.4610	53.787	.896	.899	.956
10	9.4110	53.559	.887	.887	.957
11	9.7538	53.261	.885	.821	.957
12	9.2753	54.324	.836	.887	.958
13	9.0967	56.019	.742	.766	.960

Corrected item–total correlations, which should be between 0.20 and 0.90, were between 0.450 and 0.896. The ABILOCO-Turkish total score showed a floor and ceiling effect of <15%. Hotelling’s T-squared test showed no response bias in the scale (p < 0.05).

Validity Analysis

For the validity analysis of the ABILOCO scale, content/scope validity, criterion validity, construct validity, and factor analysis were examined. Item-level content validity (I-CVI) for each item of the ABILOCO-Turkish was >0.78, the mean I-CVI was 0.97, and scale-level content validity (S-CVI) was 0.84 (Polit & Beck, 2006). In the concurrent validity analysis, a high negative correlation was observed between the TUG and the ABILOCO (r: −0.830, p < 0.001), while there was a strong positive correlation between the BBS and the ABILOCO (r: 0.919, p < 0.001).

Construct Validity

Exploratory: The KMO value was 0.942 and the BS value was significant (p < .001). These are excellent results and confirm that the scale is suitable for factor analysis (Tabachnick et al., 2007b). The ABILOCO-Turkish was a unidimensional measurement tool and all of the items were placed under one factor. While the factor is expected to explain at least 30% of the total variance in unifactor scales, in ABILOCO-Turkish this one factor explained 66.567% of the total variance. Factor loads of the items in the factors were 0.453–0.914.

Confirmatory: According to the results of the CFA, the factor loads of the items in the ABILOCO scale ranged from 0.36 to 0.95. The CFA results of the single factor scale structure are shown in Figure 4.1. The chi-squared value of the single factor scale structure was 284.952, with degrees of freedom of 61 (p = 0.000). χ²/SD was 4.671. The fit indices were as follows: IFI 0.896, NFI 0.871, GFI 0.708, CFI 0.895, and RMSEA 0.163. With the Holter test, it was concluded that the sample size was sufficient for analysis.

Discussion

The ABILOCO-Turkish was seen to be a valid and reliable scale that can be used as a self-assessment tool for patients who have had a stroke. The preliminary ABILOCO questionnaire included 43 items on locomotive activities. However, Caty et al. excluded 30 items in their study for various reasons, and the final ABILOCO-original consisted of 13 items. In our study, we also investigated 13 items. Sogbossi et al. (E. Sogbossi et al., n.d.) started with 36 items but deleted 21 of them. They showed that the 15-item ABILOCO-Benin is a one-dimensional scale with good model fit.

The steps recommended in the literature for the translation and cultural adaptation of the ABILOCO scale into Turkish were followed. All items in the scale were appropriate for and understandable by patients in Turkey. No items required removal or modification. Explanations were added in parentheses (“without bringing the feet together” and “partial paralysis”) to improve intelligibility.

Reliability

The homogeneity and stability of a scale are determined by internal consistency and test–retest method, respectively. These methods are frequently used in the literature (Kraemer, 1981), so they were employed for determining the reliability of the ABILOCO-Turkish. The Cronbach’s alpha value was 0.96 for the ABILOCO-Turkish. Item–total score correlations were examined and moderate and strong correlations were found (p < 0.00, ranged from 0.464 to 0.899), indicating that the scale had a high level of reliability. In other words, all items can adequately measure the structure related to mobility. Moreover, ICC, Pearson correlation, and paired samples t-test values were high, which verified the stability and invariance of the ABILOCO-Turkish and indicated a high level of reliability (Nunnally, 1967). ICC values between the first and second evaluations similar to our results have been shown in the literature (ICC = 0.77) (Caty et al., 2009) (ICC = .95; p < .001) (E. S. Sogbossi et al., 2014) (p < 0.05, ICC = 0.85) (Didehdar & Kharazinejad, 2020). In the ABILOCO-Benin validation study, acute stroke was retested within 3 days and chronic stroke within 2 weeks and showed a high correlation (ICCZ.95; p < .001) (E. S. Sogbossi et al., 2014). We re-tested our participants 1 week later so that their clinical status had not changed and the learning effect disappeared.

Validity

Content validity of the ABILOCO-Turkish scale was evaluated with CVI. The I-CVI and S-CVI values for the scale’s content validity were acceptable (Polit & Beck, 2006). The results obtained from the CVI showed that the items in the Turkish version of the scale were similar to those in the original scale (Caty et al., 2008).

The BBS, which was also used in previous studies, was used to test the concurrent validity of the ABILOCO-Turkish. In addition, the TUG test, which is accepted as the gold standard for assessing functional mobility, was used for concurrent validity analyses (Podsiadlo & Richardson, 1991). The ABILOCO-Turkish showed high correlation with the TUG test (Rho = −.881, p < 0.001) and the BBS (Rho = .903, p < 0.001). Various measurement tools such as the Korean Modified Barthel Index (K-MBI) (r = 0.51, p = 0.012), the Korean Modified Barthel Index Mobility (K-MBIM) (r = 0.53, p = 0.008), the Korean Berg Balance Test (K-BBS) (r = 0.57, p = 0.004), the Fugl-Meyer assessment, the 10-m walk test (r = .75 p < 0.001), the six-minute walk test (r = .76 p < 0.001), the Functional Walking Category (r = .81 p < 0.001), the Functional Ambulation Categories (r = .86 p < 0.001), the Functional Independence Measure (FIM) mobility (r = .87 p < 0.001), walking speed (r = .63 p < 0.0001), and the FIM walking ability (r = .81 p < 0.001) were used in studies examining the original and other versions of the ABILOCO (Avelino et al., 2018; Caty et al., 2008; Didehdar & Kharazinejad, 2020;Lee et al., 2013; E. S. Sogbossi et al., 2014).

Construct Validity

The original version of the ABILOCO scale developed with the Rasch model is unidimensional (Caty et al., 2008), as are the other versions (Avelino et al., 2018; Caty et al., 2008; Lee et al., 2013; E. S. Sogbossi et al., 2014). ABILOCO-Benin and ABILOCO-Brazil were also analyzed with the Rasch model, while ICC values were calculated for the Korean and Persian versions. Examining the questionnaires with various analytic methods increases their validity and reliability. For this purpose, the construct validity of the ABILOCO-Turkish was investigated using factor analyses in the present study. EFA was used to determine the construct validity of the ABILOCO-Turkish. As the Bartlett’s test result was statistically significant (p < 0.001) and the KMO value was close to 1 (0.942), achieving the perfect fit, factor analysis was appropriate. The EFA results of the ABILOCO-Turkish confirmed that the scale is unidimensional. This factor explains 66.56% of the total variance. The factor loads of the items in the factors vary between .453 and .914 (Table 2). The findings obtained as a result of EFA meet the criteria based on factor analysis (DeVellis, 2003). All items were correlated with each other and measured mobility. The ABILOCO-Benin (E. S. Sogbossi et al., 2014) and the ABILOCO-Brazil (Roberto Avelino et al., 2018) presented good internal consistency.

EFA defines only the sub-dimensions of the scale; it does not examine the scale structure. Therefore, the scale structure was examined by CFA. The fit indices of the ABILOCO-Turkish were examined and the χ²/df and CFI values ​​were within the acceptable range, while the GFI, NFI, and RMSEA values ​​were not. In addition, factor loadings calculated for each item by CFA were at an acceptable level (>0.32) and confirmed the unidimensionality of the scale (Figure 1). Examining the scales with methods such as both factor analysis and Rasch analysis increases their reliability. Such examples exist in the literature and are recommended. In the present study, we examined the ABILOCO-Turkish using factor analysis. It should also be examined with the Rasch model.

OPEN IN VIEWER

Figure 1.

ABILOCO-Turkish Confirmatory Factor Analysis

In the Brazilian version, a ceiling effect was observed in 21 individuals (15.4%) (Avelino et al., 2018). The ABILOCO-Turkish did not show ceiling or floor effects. Regarding the ABILOCO-Turkish, one person received a maximum score of 26 and no one received a minimum score of 0, suggesting that the scale was free from these effects.

Very satisfactory results were obtained in the present study in which the psychometric properties of the ABILOCO-Turkish were examined in stroke patients. This scale, which provides detailed clinical information about walking and mobility functions, is easy and quick to apply and is also available in Turkish and is suitable for the Turkish population.

A large sample size provides excellent results in validity and reliability studies. One of our limitations was that we performed CFA and EFA with the same sample group, since we could not achieve a sufficient sample size. In the literature, it is recommended to perform EFA and CFA in different sample groups. Moreover, the participants were evaluated in the university laboratory, i.e., they were able to get the laboratory and therefore had better mobility. Lastly, the validity and reliability of the ABILOCO scale were not investigated in stroke patients with cognitive impairment.

In conclusion, the 13-item ABILOCO, which evaluates mobility, is valid and reliable in Turkish and can be used safely by healthcare professionals. This self-report, easy-to-understand, and time-saving scale can be helpful in assessing mobility in the Turkish population.