Assessment of the Validity and Reliability of the Turkish Version of the Modified Tampa Scale for Kinesiophobia in Patients With Shoulder Pain

Introduction

Shoulder pain is a prevalent musculoskeletal condition affecting individuals across all age groups, with common etiologies including impingement syndrome, bicipital tendinitis, and rotator cuff tears, while less frequent causes include referred pain conditions, osteoarthritis, thoracic outlet syndrome, and brachial neuritis [1]. Epidemiological studies report a global incidence ranging from 0.7% to 6.2%, with higher prevalence observed among women and individuals of higher socioeconomic status [2]. Patients with shoulder pain often report exacerbation of symptoms during mechanical activity, particularly during abduction or repetitive movements [3]. Consequently, many patients may adopt avoidance behaviors, believing that limiting movement will reduce pain intensity [4]. Such avoidance behaviors, when persistent, can contribute to disuse of the affected limb and the development of joint stiffness [5].

Fear of movement-related pain is a broad construct describing an emotional-cognitive appraisal of pain as threatening, combined with motivational processes that promote avoidance behavior. According to contemporary fear-avoidance models, particularly the work of Vlaeyen and Meulders, fear of movement-related pain encompasses threat appraisal processes (eg, catastrophic interpretations of pain) as well as behavioral avoidance tendencies. The fear-avoidance model represents components of this overarching construct rather than fear of movement-related pain being a single element within the model itself [6,7].

Within this framework, kinesiophobia, commonly operationalized by the Tampa Scale for Kinesiophobia (TSK) specifically reflects fear of (re)injury, which represents a subconstruct of fear of movement-related pain rather than the full multidimensional construct [7,8]. Thus, while fear of movement-related pain includes affective, cognitive, and motivational dimensions, the TSK primarily captures beliefs and concerns related to bodily harm, vulnerability, and (re)injury during movement.

In individuals with shoulder pain, elevated fear of (re)injury can manifest as reduced joint range of motion, altered movement strategies, and decreased adherence to rehabilitation programs. Beyond physical consequences, heightened fear responses are associated with anxiety, depressive symptoms, and reduced quality of life [9,10]. Importantly, empirical and conceptual work has emphasized that fear of (re)injury should not be used interchangeably with broader constructs, such as pain-related fear or fear of movement-related pain, as these are related but distinct conceptual levels [7,11].

The measurement of fear of (re)injury in clinical populations frequently relies on the TSK. The Turkish adaptation of the original 17-item TSK (TSK-17) was first validated by Tunca Yılmaz et al in 2011 in patients with chronic low back and neck pain [8]. Although the TSK-17 and its short forms have been validated in various musculoskeletal populations, including temporomandibular disorders and low back pain [12,13], evidence specifically addressing patients with shoulder pain remains limited. Cultural and linguistic differences may influence how beliefs about (re)injury and bodily vulnerability are interpreted, underscoring the need for population-specific validation. The 18-item version of the scale was validated by van Iersel et al in a population of patients with shoulder dislocation [14].

Therefore, the present study aims to examine the reliability and structural validity of the Turkish version of the modified TSK-18 in patients with shoulder pain, specifically focusing on its capacity to measure fear of (re)injury within this clinical population.

Material and Methods

PATIENT SELECTION AND SAMPLING:

Patients aged between 18 and 70 years who presented to the clinic with shoulder pain lasting less than 6 months and a visual analog scale (VAS) score of 4 or higher were included in the study. The diagnosis was confirmed through clinical examination using impingement tests, such as the Neer, Hawkins, Speed, and Jobe tests, with at least 1 positive result considered sufficient. Patients who had undergone ultrasonographic or magnetic resonance imaging examinations confirming a shoulder-related pathology were also included. Those who did not provide informed consent or had cervical disc herniation or uncontrolled systemic diseases were excluded. Additionally, patients who had undergone surgery, injections, or physical therapy on the affected arm within the past 3 months were excluded. Patients with neurological deficits, psychiatric disorders, or language difficulties that prevented them from understanding or responding to the survey were also excluded from the study.

When determining sample size, a subject-to-item ratio of at least 5: 1 is commonly recommended. In this study, a sample size equal to 5 to 10 times the number of items was used, following the recommendation of Watson and Thompson [15]. Considering that the scale included 18 items and that up to 100 participants are suggested for a robust analysis, the study planned to include a minimum of 180 patients.

ADAPTATION OF THE MODIFIED TSK ITEMS INTO TURKISH AND ENSURING LINGUISTIC VALIDITY:

The modified TSK-18 was translated and culturally adapted into Turkish by a team of 3 experts in physical therapy and orthopedics who were experts in the relevant fields and had prior experience in validation studies. The initial translation from English to Turkish was conducted collaboratively by the committee, which carefully selected the most contextually and semantically appropriate equivalents for each original item. Next, 2 bilingual researchers, native English speakers fluent in Turkish, independently performed blind back-translations of the adapted items into English, without access to the original scale. The back-translated versions were then critically reviewed by the committee to identify and resolve any inconsistencies, ensuring linguistic accuracy and conceptual equivalence.

To assess the content validity and linguistic appropriateness of the final version, expert opinions were obtained from a panel consisting of 2 field experts, 2 non-field experts, and a linguist. Content validity was quantified using the Polit and Beck validity index that was completed by the experts [16]. Based on their feedback, the original item 18, “I am worried that I will dislocate my shoulder during my activities of daily life,” was revised to “I am worried that my shoulder complaints will increase when I undertake activities of daily living” to ensure broader applicability across various shoulder pain conditions. The translation process followed the ISPOR guidelines [17].

Linguistic validity was further evaluated through a pilot study in which the Turkish version of the scale was administered to 30 bilingual patients with shoulder pain who were fluent in both Turkish and English. Participants completed the Turkish version of the scale and provided feedback on its clarity and comprehensibility. A physiatrist assisted during this process to note any items that were unclear or difficult to understand. Based on the pilot study results and expert feedback, the final version of the scale was developed, and its linguistic validity was established. The items of the modified TSK-18, which was adapted into Turkish through committee review and expert consultation, are shown in Figure 1.

VALIDITY AND RELIABILITY ASSESSMENT OF THE TURKISH MODIFIED TSK-18:

In this study, 180 patients who met the inclusion criteria were enrolled to evaluate the validity and reliability of the finalized Turkish version of the TSK-18. All patients were examined by a physiatrist, and demographic and clinical data, including sex, age, educational level, occupation, history of trauma, duration of illness, and comorbidities, were recorded. In addition to completing the Turkish version of the modified TSK-18, the patients were assessed using the VAS, Shoulder Pain and Disability Index (SPADI), Constant Score, Quick Disabilities of the Arm, Shoulder, and Hand Questionnaire (QuickDASH), and the 36-Item Short Form Health Survey (SF-36).

Subsequently, factor analysis and hypothesis testing were conducted to determine the construct validity of the scale [17]. Criterion validity was evaluated by examining the correlations between the modified TSK-18 and other established measurement instruments. For this purpose, patients were evaluated using the Turkish version of the modified TSK-18, along with the VAS, SPADI, Constant Score, QuickDASH, and SF-36, to assess concurrent validity.

To evaluate the reliability of the scale, several statistical methods were used, including internal consistency analysis, test-retest reliability, and calculation of the Cronbach’s alpha coefficient [18]. Additional reliability evaluations included item-total correlation and split-half reliability analyses. For the test-retest assessment, patients completed the modified TSK-18 a second time, 2 weeks after the initial administration. In this 2-week period, the patients did not have any treatment, and standardized protocols were used consistently across centers. The methodological flowchart of the study is presented in Figure 2.

PARAMETERS USED IN PATIENT EVALUATION:

The TSK-18, consisting of 18 items, is designed to assess fear of (re)injury associated with movement and physical activity. The scale operationalizes beliefs regarding vulnerability to harm and bodily damage during movement rather than measuring the full construct of fear of movement-related pain. The scale uses a 4-point Likert format, with response options ranging from “strongly agree” to “strongly disagree”. Total scores range from 18 to 72, with higher scores indicating a greater level of fear of (re)injury [14].

The VAS is a widely used and simple tool for assessing the intensity of subjective experiences, particularly pain. It typically consists of a 10-cm horizontal line anchored by descriptors such as “no pain” at one end and “worst imaginable pain” at the other. Patients are asked to mark the point on the line that best represents the intensity of their pain. The distance from the zero point to the patient’s mark is measured in centimeters, providing a score between 0 and 10. This enables a quantitative assessment of pain intensity [19].

The SPADI is a self-report questionnaire designed to assess shoulder pain and related functional impairment. It evaluates the effect of shoulder pain on daily activities and consists of 2 subscales: pain intensity and functional limitation. The pain subscale measures the intensity of pain during various movements, while the functional limitation subscale assesses the extent to which shoulder pain interferes with daily functioning. The SPADI includes 13 items, each scored on a scale from 0 to 10, with higher scores indicating greater pain intensity and disability. The total score represents the overall severity of the patient’s shoulder-related problems. The validity and reliability of the Turkish version of the SPADI have been previously established [20].

The Turkish validation study of the Constant Score was conducted by Çelik et al [21]. This assessment tool evaluates several parameters, including pain intensity, functional capacity in daily activities, shoulder range of motion, and muscle strength. The total score is calculated out of 100, with higher scores indicating better shoulder function, while lower scores reflect greater functional limitations and pain intensity [21].

The QuickDASH questionnaire is a shortened version of the original DASH outcome measure, developed to evaluate physical function and symptom severity in individuals with upper limb musculoskeletal disorders. It consists of 11 items that assess the patient’s ability to perform daily activities, as well as the intensity of pain and other symptoms affecting the arm, shoulder, and hand. Each item is rated on a 5-point Likert scale, and the total score is converted to a scale of 0 to 100, with higher scores indicating greater disability. The Turkish validity and reliability of the QuickDASH were established by Koldaş Doğan et al [22].

The SF-36 is a widely used questionnaire designed to assess health-related quality of life. It consists of 36 items and provides a comprehensive evaluation of physical, mental, and social health. The SF-36 includes 8 subscales: physical functioning, role limitations due to physical health, pain, general health perception, vitality (energy), social functioning, role limitations due to emotional problems, and mental health. Each subscale is scored from 0 to 100, with higher scores indicating better health status and quality of life. The scale has been validated in Turkish by Demiral et al [23].

STATISTICAL ANALYSIS:

The significance level was set at α=0.05, and statistical analyses were performed using SPSS Statistics 25.0 (IBM Corp, Armonk, NY, USA). To assess the suitability of the data for factor analysis, the Kaiser-Meyer-Olkin test and Bartlett’s Test of Sphericity were applied [24,25]. Exploratory factor analysis was conducted using principal component analysis with varimax rotation (orthogonal rotation), and factor loadings of 0.40 or higher were considered acceptable. To determine the dimensional structure of the TSK-18 scale, exploratory factor analysis was performed on the collected data. Confirmatory factor analysis was then conducted using Amos 25.0 to evaluate whether the factor structure of the scale was supported in the study sample. Confirmatory factor analysis was conducted using the maximum likelihood estimation method [26]. The model fit was evaluated using several indices, including the chi-square goodness of fit test (χ2/df), comparative fit index (CFI), Tucker-Lewis Index (TLI; equivalent to the non-normed fit index in Amos), standardized root mean square residual, and root mean square error of approximation (RMSEA) [25–28].

As part of the reliability analysis, the Cronbach’s alpha coefficient was calculated to assess the internal consistency of the measurement tool. The coefficients were interpreted values below 0.60 were considered poor, 0.60 to 0.70 as low-acceptable, 0.70 to 0.80 as acceptable, 0.80 to 0.90 as good, and above 0.90 as excellent [18].

To further evaluate measurement reliability, the standard error of measurement (SEM) and the smallest detectable change (MDC) were calculated. The SEM was determined using the formula SEM=SD(T1)×√(1-ICC), where SD(T1) represents the standard deviation at time point T1, and ICC is the intraclass correlation coefficient. The MDC at the 95% confidence level (MDC95) was then calculated as MDC95=1.96×SEM×√2. For sensitivity analysis, the difference-based SEM was computed using the formula SEM=SD(T2-T1)/√2, where SD(T2-T1) refers to the standard deviation of the change between 2 time points (T2 and T1) [29,30]. Finally, the MDC at the 90% confidence level (MDC90) was reported, calculated as MDC90=1.64×SEM×√2 [31,32].

The ICC was calculated using the test-retest method. A single-measurement, absolute-agreement model was applied, and 95% CIs were reported alongside the ICC values. The thresholds for interpreting the ICC were established as follows: values less than 0.50 were considered poor, values between 0.50 and 0.75 were considered fair, values between 0.75 and 0.90 were considered good, and values greater than 0.90 were considered excellent [33]. Criterion-related validity was evaluated by examining correlations between the TSK-18 and the reference scales. Differences in TSK-18 scores between male and female patients were analyzed using the Mann-Whitney U test.

Results

Evaluation of Validity

EXPLANATORY FACTOR ANALYSIS:

An examination of the inter-item correlations of the TSK-18 revealed values ranging from 0.3 to 0.7, suggesting a moderate level of association among items. These correlation coefficients indicate that the scale is appropriate for factor analysis. All correlation pairs were statistically significant (P<0.001).

To determine the factor structure of the TSK-18 scale, an exploratory factor analysis was conducted based on the collected data. The Kaiser-Meyer-Olkin value was 0.631, indicating that the sample size had a moderately acceptable level of adequacy for factor analysis (0.60 ≤Kaiser-Meyer-Olkin value <0.70). The Bartlett’s Test of Sphericity also demonstrated that the correlation matrix was significantly suitable for factor analysis (χ²=1040.826; df=91; P<0.001). Based on these findings, the data were deemed appropriate for factor analysis.

Within the scope of principal component analysis, the total variance explained by each variable through the extracted components was examined. Communality values ranged between 0.40 and 0.72, indicating that the variables were suitable for factor analysis and made meaningful contributions to the model. Factor loadings in the component matrix ranged between 0.33 and 0.80, demonstrating that the scale was represented under a multifactorial structure. When the item-total statistics were examined, items 2, 4, 6, and 12 were sequentially removed from the analysis in order to improve reliability. The final analysis revealed that the scale had a 2-factor structure. Although 4 components had eigenvalues greater than 1 according to the Kaiser criterion, expert theoretical evaluations determined that a 2-factor structure would be more appropriate, and this solution was adopted. The 2-factor structure explained 43.17% of the total variance. In the literature, it has been reported that an explained variance above 40% is considered satisfactory in social and health sciences [23–25]. In conclusion, the items of the scale were found to represent 2 latent constructs, which measure dimensions associated with activity avoidance and somatic focus, and the communality values (>0.40) indicated that the factor structure adequately represented the scale items.

The 2-factor model (activity avoidance and somatic focus) provided the best explanatory power among the tested structures, showing clearer conceptual distinction and acceptable internal consistency compared with the 1-factor model. Comparative confirmatory factor analysis demonstrated that the 2-factor model (χ²/df=6.12, CFI=0.63, TLI=0.53, RMSEA=0.16) exhibited slightly improved fit indices relative to the 1-factor model (χ²/df=7.64, CFI=0.476, TLI=0.49, RMSEA=0.19). Although the fit indices did not reach optimal thresholds, the model selection was based not only on statistical improvement but also on theoretical coherence and empirical evidence from earlier literature, which consistently reported a 2-dimensional latent structure. Moreover, models with 3 or more factors yielded weak interpretability and overlapping item loadings, reducing semantic coherence. Therefore, the 2-factor model was confirmed as the most theoretically and statistically justifiable structure.

CONFIRMATORY FACTOR ANALYSIS: The model demonstrated poor fit to the data (χ2/df=6.12, P<0.001, CFI=0.63, TLI=0.53, RMSEA=0.16, standardized root mean square residual=0.13). These fit indices suggest that the initial 2-factor model did not satisfy the commonly accepted threshold criteria for a good model fit. Nevertheless, the scale items loaded significantly onto their respective factors, supporting the theoretical justification of the 2-factor structure. Based on the modification indices provided by AMOS, covariances were introduced between selected error terms (e20-e9, e20-e6, e6-e18, and e14-e15) to improve model fit, reflecting shared variance between items with similar content. Although these modifications resulted in a better representation of the data, the fit indices still remained below the ideal thresholds. The 2-factor structure demonstrated that the items clustered under dimensions associated with activity avoidance, somatic focus, fear of movement, and pain-focused/avoidance behaviors. The acceptable factor loadings indicated that the items contributed acceptably to their respective factors. The path diagram of the Turkish TSK-18 is shown in Figure 3.

The internal consistency of the overall scale revealed a Cronbach alpha of 0.75, indicating acceptable reliability in line with values reported in the literature. When examined separately, Cronbach’s alpha coefficients were 0.723 for the activity avoidance subdomain and 0.520 for the somatic focus subdomain. While the activity avoidance dimension demonstrated acceptable internal consistency, the somatic focus dimension showed relatively low reliability. Nevertheless, the 2-factor model (activity avoidance and somatic focus) provided better explanatory power than the 1-factor structure, offering clearer conceptual differentiation between constructs. Overall, these findings suggest that the scale demonstrates acceptable reliability at the total score level and partial reliability within its 2-factor structure.

Although the somatic focus subdomain demonstrated relatively low internal consistency (α=0.520), the Cronbach’s alpha is known to be influenced by the number of items within a subscale. Given the limited number of items and the acceptable factor loadings observed, this value was considered interpretable. Moreover, the 2-factor model showed superior conceptual clarity and explanatory power compared with the 1-factor structure; therefore, the somatic focus dimension was retained despite its modest reliability.

According to the analysis results, the ICC values calculated for the scale items were generally within acceptable ranges, supporting the temporal stability of the measurements. Although a few items showed relatively low ICC values, several demonstrated moderate to good reliability.

In line with the classification proposed by Koo and Li, in which ICC values below 0.50 indicate poor reliability, values between 0.50 and 0.75 indicate moderate reliability, 0.75 to 0.90 indicate good reliability, and values above 0.90 indicate excellent reliability, the reliability of certain items on the scale was found to be weak [33]. Items 2, 6, 7, and 11 showed poor reliability. However, items 12, 13, and 14 showed moderate to good reliability, indicating satisfactory measurement stability for those components. In the present study, item 2, which showed very low test-retest reliability, was excluded from the final model. Further item removal was not found to improve either the internal consistency or the factor structure of the scale. Multiple alternative exploratory factor analyses were performed with different item combinations; however, models with fewer than 17 items did not provide better statistical or theoretical adequacy. Therefore, the analyses were finalized using the remaining items that yielded the most stable and theoretically coherent structure across exploratory factor analyses and confirmatory factor analysis. The ICC values of items and the total score of the TSK-18 are given in Table 2.

For the modified TSK-18, test-retest reliability yielded an ICC of 0.629 (95% CI: 0.532–0.710), indicating moderate temporal stability. The SEM was 3.16 points and the 95% confidence level (MDC₉₅) was 8.75 points, suggesting that score changes below this threshold may reflect measurement error rather than true change.

For the TSK-14 version, the ICC was 0.618, also reflecting moderate reliability. The test-retest distributions were 43.65±4.99 at baseline and 42.26±5.36 at retest. The SEM was 3.08 points, with an MDC95 of 8.53 points and an MDC90 of 7.14 points. The Cronbach’s alpha for the TSK-14 was 0.764. Test-retest reliability and measurement-error indices for both versions are presented in Table 3.

CRITERION-RELATED VALIDITY:

In the present study, an analysis of the correlations between the Turkish TSK-18 and clinical measurements revealed an association between the TSK-18 scores and VAS pain scores (r=0.176, P=0.018). According to the correlation classification proposed by Schober, correlation coefficients in the range of 0.00 to 0.19 indicate a very weak relationship, 0.20 to 0.39 a weak relationship, 0.40 to 0.59 a moderate relationship, 0.60 to 0.79 a strong relationship, and values of 0.80 or more reflect a very strong relationship [34]. Based on this classification, the observed correlation between VAS and TSK scores (r=0.176, P=0.018) can be interpreted as very weak. No statistically significant correlations were found between the TSK-18 and SPADI score, Constant score, QuickDASH, physical functioning, physical role limitation, emotional role limitation, energy/fatigue, pain, or general health subscores (P>0.05). However, weak negative correlations were found between the TSK-18 scores and the emotional well-being (r=−0.148, P=0.047), social functioning (r=−0.220, P=0.003), and health change (r=−0.157, P=0.036) subscales of the SF-36.

These findings indicate that the Turkish version of the modified TSK-18 is particularly associated with the psychosocial dimensions of health, specifically emotional well-being, social functioning, and perceived health change, showing low but statistically significant relationships. This suggests that the TSK-18 reflects psychosocial influences in individuals with shoulder pathology and supports its criterion validity in this context.

CEILING AND FLOOR EFFECTS:

In accordance with COSMIN recommendations, ceiling and floor effects were considered present if 15% or more of the sample achieved the highest or lowest possible total score (range, 14–56) [35]. In our sample (N=180), the observed score range was 31 to 53. No patients achieved the lowest (14) or highest (56) possible scores, resulting in 0% floor (0/180=0.0%) and 0% ceiling (0/180=0.0%) effects. The histogram distribution was unimodal, with scores clustering around the middle values, indicating that the scale did not saturate at either extreme and that no clinically significant ceiling or floor effects were present. Additionally, the “near-ceiling” band was defined as scores of 52 or higher, and the “near-floor” band as 18 or lower. The proportions for these bands were below the 15% threshold. As an additional check, 10.6% of patients (19/180) scored 50 or higher, which remained below the predefined criterion, confirming the absence of a ceiling effect.

MULTIPLE REGRESSION ANALYSIS RESULTS:

Multiple linear regression analyses were conducted to evaluate the predictive power of age, sex, and disease duration on 2 distinct dependent variables: activity avoidance and somatic focus. Prior to interpreting the model parameters, collinearity diagnostics were examined. For all independent variables, variance inflation factor values ranged between 1.03 and 1.06, and tolerance values exceeded 0.94, indicating no multicollinearity issues within the dataset.

PREDICTION OF ACTIVITY AVOIDANCE:

The first regression model assessed the determinants of activity avoidance. The analysis revealed that the demographic and clinical variables included in the model did not significantly predict activity avoidance scores. Specifically, age (β=0.045, P=0.558) was not a significant predictor. Although sex (β=0.131, P=0.088) and disease duration (β=−0.134, P=0.078) approached marginal significance, they failed to reach the conventional threshold of statistical significance (P<0.05).

PREDICTION OF SOMATIC FOCUS:

The second model evaluated the predictors of somatic focus. Similar to the first model, none of the independent variables made a statistically significant contribution to the variance in somatic focus scores. Age (β=0.123, P=0.111), sex (β=0.083, P=0.282), and disease duration (β=−0.058, P=0.442) were not found to be significant predictors in this cohort.

Discussion

In this study, exploratory factor analysis of the Turkish version of the modified TSK-18 revealed a 2-factor structure encompassing activity avoidance and somatic focus, which together explained 43.17% of the total variance. Although confirmatory factor analysis indicated suboptimal model fit (χ2/df=6.12, CFI=0.63, TLI=0.53), the 2-factor solution provided the clearest theoretical interpretation, reflecting the multidimensional nature of kinesiophobia. Internal consistency was partially acceptable (Cronbach’s alpha=0.75), and correlations with pain, emotional well-being, social functioning, and health change were weak, indicating that fear of movement or (re)injury may be present even when pain intensity or functional limitations are mild. Additionally, there was no ceiling effect. No effects of age, sex, or disease duration were detected in the regression analyses.

Importantly, the modified TSK-18 measures fear of (re)injury rather than the broader construct of fear of movement-related pain. Therefore, the present findings should be interpreted as reflecting beliefs about bodily vulnerability and anticipated harm during movement rather than capturing the full emotional-cognitive-motivational construct described in contemporary fear-avoidance frameworks [7,11].

Some individuals experiencing shoulder pain develop an avoidance behavior, refraining from using the affected arm and shoulder due to the fear that movement will exacerbate their pain intensity [36]. This reflects fear of (re)injury rather than fear of movement-related pain in its broader theoretical sense. Such avoidance can contribute to muscular deconditioning and reduced joint mobility [37]. While pain intensity and functional impairment are frequently emphasized in shoulder research, fear of (re)injury operationalized by the TSK has received comparatively less attention as an independent psychological factor influencing recovery [38–40]. Identifying the presence of fear of movement-related pain, implementing appropriate interventions, and encouraging proper arm use, functional impairments, and related work productivity losses [41]. Validating the TSK in patients with shoulder pain may facilitate earlier detection of fear of movement-related pain and help prevent delays in recovery. This, in turn, is expected to improve rehabilitation outcomes [42].

To assess this issue, van Iersel et al in the Netherlands and Motta et al in Italy modified the scale for use in patients with shoulder instability [14,43]. Mintken et al and Kamonseki et al modified the scale for use in patients with shoulder pain in the United States and Brazil, respectively; however, there appears to be no Turkish validity and reliability study conducted specifically for patients with shoulder pain [44,45]. In the study by Mintken et al, significant correlations were observed between TSK-11 and Fear-Avoidance Beliefs Questionnaire (FABQ) scores, indicating conceptual overlap between fear of (re)injury and fear-avoidance beliefs. Both instruments demonstrated meaningful associations with clinical outcomes, although the magnitude of associations differed across measures [44]. In the study conducted by Kamonseki et al, which aimed to validate the FABQ and the TSK-11 in patients with shoulder pain, the FABQ was identified as a multidimensional instrument, whereas the TSK-11 demonstrated a unidimensional structure. Both instruments were found to be psychometrically sound, exhibiting validity and reliability within this patient population [45]. Due to the low inter-item correlation levels, they developed a 7-item version and reported a Cronbach’s alpha of 0.81, indicating moderate reliability, no floor or ceiling effects, and adequate internal consistency. In the present study, the Cronbach’s alpha was 0.75, reflecting partially acceptable internal consistency with no ceiling and floor effect. The main difference between these 2 studies and the present study is that the FABQ parameter was not included, as the questionnaire has not yet been validated for use in individuals with shoulder pain in Turkey, and a 2-factor structure was identified in our analysis. The TSK-18 demonstrated a 2-dimensional structure, consisting of activity avoidance and somatic focus factors. Overall, low reliability with poor fit data and partially acceptable validity were observed in this study. In addition to this, van Iersel et al validated a modified version of the TSK-18 in individuals with anterior shoulder instability [14]. Similarly, in their study validating the TSK-13 among patients with shoulder instability, Motta et al identified 2 underlying constructs: activity avoidance and health-related anxiety. They reported a Cronbach’s alpha of 0.874 and an ICC of 0.915, thereby establishing the instrument’s validity and reliability within this specific patient population [43]. In Motta et al study, which used 13 items, both the Cronbach’s alpha and the inter-item correlations were higher than those observed in our study. Future research could consider designing studies that utilize a 13-item version for individuals with shoulder pain. In the study conducted by Pagels et al, which validated the German version of the TSK in patients experiencing shoulder pain, the scale demonstrated good internal consistency, with a Cronbach’s alpha coefficient of 0.81, and was deemed a reliable assessment tool. The homogeneity of the scale was evaluated as poor, with a Loevinger H coefficient of 0.35. Correlation analyses showed no significant relationships between the numeric rating scale, FABQ, and SPADI. Notably, and in contrast to previous studies, a significant association was found between symptom duration and the level of kinesiophobia, indicating that kinesiophobia increased in parallel with the length of symptom persistence [46]. The main difference between the present study and Pagel et al’s study is that our study included acute and subacute cases, and a weak correlation was detected with the VAS. In this study, exploratory factor analysis revealed factor loadings between 0.33 and 0.80, suggesting that the scale has a multidimensional composition, primarily reflecting 2 constructs: avoidance of activity and focus on physical symptoms. This dual-factor framework explained 43.17% of the total variance. Nonetheless, the model demonstrated a suboptimal fit with the dataset, as indicated by the chi-square statistic. The internal consistency was found to be adequate, with a Cronbach’s alpha of 0.75. A weak association was observed between the modified TSK-18 and VAS scores, emotional well-being, social function, and health change, while correlations with other variables were not statistically significant. Overall, findings support that the Turkish adaptation of the modified TSK-18 is a partially dependable, acceptably valid, and multidimensional instrument suitable for evaluating kinesiophobia in patients experiencing shoulder pain. Since it appears the TSK has yet to be validated for patients with shoulder pain in Turkey, this study was aimed to validate the modified TSK-18 into Turkish, specifically for patients with shoulder pain.

To date, no other scale specifically assessing fear of movement-related pain in individuals with shoulder pain has been validated in Turkish. The primary strength of this study lies in its novelty, as it addresses a significant gap in the existing literature. A notable limitation of this study is the absence of an alternative fear of movement-related pain scale specific to shoulder pain with which to compare the modified TSK-18. The TSK is the only scale that can accurately measure fear of movement-related pain, and no other scale provides a fully comparable assessment. While other scales have been validated in Turkish, the Pain Catastrophizing Scale has been validated for general populations, and the FABQ has been validated for patients with low back pain; neither has been validated for shoulder pain [47,48]. Therefore, this study included the VAS, SPAID, Constant Score, QuickDASH, and SF-36, which are commonly used in shoulder validation studies and have been validated in patients with shoulder pain. Likely as a result of this choice, only a weak correlation was observed with the VAS, emotional well-being, social function, and health change, and no significant correlations were found with any of the other parameters. Consequently, the validity of the validated TSK-18 form was limited, representing one of the study’s most important limitations. However, in their validation study, Kamonseki et al modified the FABQ by replacing the word “back” with “shoulder” and suggested that the FABQ might be more suitable than the TSK for assessing shoulder conditions, as it produced more meaningful results [45]. For this reason, although these scales were not used in our study, their exclusion represents a limitation. Future research should focus on validating these scales in patients with shoulder pain and conducting comprehensive studies using all of them.

In addition, this study examined parameters related to pain intensity, function, and quality of life; however, it did not include alternative psychological assessments, such as the State-Trait Anxiety Inventory or the Beck Depression Inventory. This omission is another important limitation of the study.

Arguably, the most significant limitation of the study lies in the results of the confirmatory factor analysis, which indicate that the model’s fit indices do not meet the threshold values commonly accepted in the literature. Consequently, the statistical fit of the proposed model appears to be limited. It is important to emphasize that multiple structural models (eg, 1-factor and 2-factor solutions) were tested; however, none yielded sufficiently high fit indices. The analysis showed that distinguishing between activity avoidance and somatic focus offered the clearest and most coherent explanation of the data. Compared with a single-factor approach, this structure demonstrated better conceptual separation and acceptable consistency. Although the statistical indicators were not ideal, the choice was supported by theoretical considerations and previous research, which has repeatedly identified a 2-part underlying structure. Models with additional factors were less meaningful and showed overlapping interpretations, reducing clarity. As a result, the 2-factor structure was judged to be the most sound and defensible option. This finding suggests that the original factor structure of the scale was not fully replicated in the present sample, potentially due to cultural influences, sample heterogenity or characteristics specific to the clinical population studied, which may have adversely impacted model fit. Nevertheless, the factor loadings of the scale items were not substantially affected, and the observed 2-dimensional structure aligns with theoretical expectations. Furthermore, the internal consistency of the scale, as indicated by a Cronbach’s alpha coefficient of 0.75, was found to be within an acceptable range, supporting the partial reliability of the instrument for use in this context.

The limited correlations among the items may be explained by the presence of reverse-scored items within the scale, as well as by linguistic and cultural factors specific to the population in which the scale was validated. Based on the feedback received, the necessary revisions were made to item 18. During the pilot study, participants experienced difficulty understanding the reverse-scored items, likely due to cultural characteristics; however, efforts were made to adhere to the original items as closely as possible. Major structural modifications were deemed inappropriate during the validation process. Furthermore, cultural influences, such as beliefs about pain and healthcare-seeking behaviors, may have affected the inter-item correlations. Future studies should therefore aim to develop new instruments for assessing fear of movement-related pain that better reflect the cultural and linguistic context of the target population. The fact that fear of movement-related pain is currently measured exclusively through the TSK underscores the need for additional, culturally adapted assessment tools.

Shorter versions of the TSK, such as the TSK-11 and TSK-13, were not considered, even though they are potential alternatives. In the study by Mintken et al [44], the TSK-11 was used. Although 4 explanatory factors were initially identified, the authors ultimately adopted a single-factor model due to weak inter-item correlations. Conversely, Motta et al [43] used the TSK-13 and identified 2 explanatory factors, activity avoidance and harm, with strong inter-item correlations. Therefore, if the TSK-13 had been used instead of the TSK-18 in the present study, the inter-item correlations would likely have been higher.

Also, the use of the Delphi technique to gather expert input remotely prevented direct, real-time dialogue among participants. Another concern is the possibility of selection bias in choosing the experts, which could have influenced the outcomes. Since the assessments relied on personal judgments, it is also difficult to determine how applicable the results are to other settings. Conducting the study across multiple centers increases the generalizability of the findings; however, differences in the assessment methods used by various physiatrists may have introduced inconsistencies in the data. Additionally, all data were collected from individuals experiencing shoulder pain in Antalya, Turkey, which limits how well the findings apply to populations in different geographical areas. The fact that both the participants and experts were exclusively Turkish further narrows the relevance of the outcomes to more diverse populations. Finally, since the study concentrated exclusively on patients with shoulder-related conditions, the results may not be applicable to individuals experiencing other types of musculoskeletal disorders.

Although fear-related constructs are often studied in chronic pain populations, investigating fear of (re)injury in acute and subacute shoulder pain is clinically relevant. Early maladaptive beliefs about bodily harm may contribute to the transition from acute to persistent disability. Therefore, validating the TSK-18 in an acute/subacute population allows examination of whether fear of (re)injury can already be reliably identified at earlier stages of the pain trajectory. Nevertheless, the absence of a chronic cohort limits generalizability, and future studies should directly compare populations with acute and chronic shoulder pain. Because of this issue, the findings cannot be generalized to individuals with chronic shoulder conditions. Although no effect of age, sex, or disease duration was detected in the regression analysis, future research should focus specifically on patients with chronic pain. Conducting multicenter studies with larger sample sizes, particularly in patients with chronic pain and specific shoulder pathologies, would further enhance the scale’s validity, reliability, and specificity. In future studies, developing an original scale tailored to the cultural and linguistic characteristics of a specific region may be more appropriate in cases in which sufficient correlations between items cannot be achieved.

Conclusions

These findings suggest that the Turkish version of the modified TSK-18 demonstrates acceptable, although not optimal, validity and reliability, as indicated by the limited model fit and weak correlations. However, to enhance the reliability and generalizability of the findings, future research should include multicenter studies with larger sample sizes and participants representing specific shoulder pathologies.