Reliability and Validity Measures of the Patellofemoral Subscale Koos-Pf in the Greek Patients with Patellofemoral Pain

1. Introduction

Patellofemoral pain (PFP) is one of the most common multifactorial musculoskeletal conditions. It is defined as pain originating from the patellofemoral joint or surrounding soft tissues in individuals, regardless of tissue damage [1]. PFP typically worsens during activities that stress the joint, such as deep squats, running, jumping, or climbing stairs [1,2,3].

The prevalence of PFP is significant, affecting approximately 23% of the general population and impacting specific groups, such as adolescent athletes, where 25% discontinue sports activities. This can lead to serious consequences, including musculoskeletal issues like osteoarthritis and a reduced quality of life, affecting their ability to perform daily activities [4,5]. PFP is more common in specific population groups, including females, runners, military personnel, and young athletes, negatively impacting not only musculoskeletal health but also overall quality of life, as described by the biopsychosocial model [1,6].

Several tools have been developed to evaluate patellofemoral pain (PFP) including the Kujala score, the Patellofemoral Pain Syndrome Severity Scale (PPSSS) [7,8] and the Patellofemoral subscale and Osteoarthritis Outcome Score (KOOS-PF) [9]. The KOOS-PF scale is widely recognized and has been validated in both Spanish and Arabic [10,11]. However, there is no validation of the KOOS-PF scale in the Greek language for the in the Greek population suffering with patellofemoral pain.

The aim of this study was to assess the reliability and validity of the KOOS-PF scale in Greek among the Greek population with Patellofemoral pain.

2. Materials and Methods

This study evaluated the validity and reliability of the Patellofemoral scale (KOOS-PF) for the Greek population following the standard guidelines by [12]. It has been approved by the Ethics Committee of the University of West Attica (No. 68323).

2.1. Participants

Fifty-nine (59) patients aged 18 to 65 years participated in this study, all diagnosed with patellofemoral pain. The inclusion criteria for the study were: a) anterior knee pain diagnosed by the same orthopedic surgeon; b) pain during functional activities in the patellofemoral joint, assessed by a VAS scale of >5-6, which worsened with activities such as squatting, climbing/descending stairs, and running; and c) a duration of pain exceeding 3 months [9]. The exclusion criteria included: a) pathological conditions such as stage 3 or higher knee osteoarthritis or recent surgeries (e.g., knee arthroplasty); b) other pathological conditions; and c) an inability to understand the Greek language.

2.2. Procedure

2.2.1. Translation

Permission has been given from the main authors of the KOOS-PF subscale, [13] to translate it into the Greek. The original version includes 11 self-assessment questions which have been categorized in three groups: stiffness, pain and quality of life. Every question has five possible answers. Each answer has a numeric value from 0 to 4. The final score is on a 0–100 scale, where 0 is the worst health score and 100 represent the best. The following steps were used in the translation:

• Initial translation: A bilingual expert translated the KOOS-PF scale from English to Greek ensuring that the knowledge of medical terminology.
• Back- translated: Another bilingual expert translated the Greek version back into English to check for consistency with the original text.
• Expert Review: A panel of researchers and clinicians reviewed both the Greek and back-translated English versions to ensure that the meaning and context were preserved.
• Pre-testing: The translated questionnaire was administered to a small group of Greek-speaking individuals to assess clarity and relevance.

Final Adjustments: Based on feedback from the pre-testing phase, necessary revisions were made to improve comprehension and cultural appropriateness.

5.

Final Adjustments: Based on feedback from the pre-testing phase, necessary revisions were made to improve comprehension and cultural appropriateness.

6.

Approval: The final Greek version was submitted to the main authors [13] for approval to ensure fidelity to the original subscale.

2.3. Outcome Measures

2.3.1. Visual Analog Scale (VAS)

The VAS scale was administered to the patients during their demographic data collection to describe their pain experience [14].

2.3.2. Patellofemoral Subscale (KOOS-PF)

The KOOS-PF subscale was administered twice to the patients, on the first and third days to evaluate the reliability of the measurements [11,15,16]. The final score was calculated by taking the mean score of the 11 questions, dividing it by the maximum possible score for each question ranging from 0-100, where 0 indicates the worst health status and 100 indicates the best using the following formula: Final score=100 – [(average score ΕΠ1-ΕΠ11)/4 * 100].

2.3.3. Knee Outcome Survey-Activities of Daily Living Scale (KOS-ADLS)

The KOS-ADLS was administered on the first day simultaneously with the KOOS-PF subscale to ensure construct validity of the KOOS-PF scale [15,16]. The KOS-ADLS was used with the official permission of the authors who were validated it in Greek [17]. This scale consists of 14 questions that evaluate the symptoms of the knee joint pathology, and the functional limitations experienced during daily activities.

2.4. Psychometric Assessment

2.4.1. Reliability

Reliability was assessed through internal consistency, test-retest reliability, Standard Error of Measurement (SEM), and Smallest Detectable Difference (SDD) [18]. Internal consistency, evaluated through Cronbach’s alpha, measures in regard assessing whether the items are aligned with the same construct. Test-retest reliability was evaluated across time intervals, and it was quantified through the Intraclass Correlation Coefficient (ICC). The Standard Error of Measurement (SEM) assessed how precise the individual scores were while accounting for measurement error, and the Smallest Detectable Difference (SDD) represented the smallest score difference that indicated an actual change rather than one due to measurement error [19].

2.4.2. Construct Validity

The construct validity of the KOOS-PF scale was evaluated through convergent validity, which examines the degree of the KOOS-PF baseline scores when they were correlated with the KOS-ADL scores using the correlation coefficient method [20].

2.5. Statistical Analysis

Data analysis was conducted using Stata 18.0 (Anon n.d.). For the description of the demographics and baseline data, we reported the mean value, the Standard Deviation (SD), the median value, the first (Q1) and third (Q3) quartile, and the absolute and relative frequencies, as applicable. For the psychometric measurements, we reported the point estimates, the 95% Confidence Intervals (95% CI), and the respective p-values, as applicable. Precision was set to one decimal for the baseline and demographic data description, two for the psychometric measurements, and three for the p-values and four for the factor analysis results. To be consistent with the benchmarks of the subject-to-item ratio method [21], we opted for a ratio of five participants per item [22]. Initially, Exploratory Data Analysis (EDA) was conducted to explore the data distribution and characteristics [23]. The benchmark values and interpretation of Cronbach’s alpha [24], ICC [25], and the correlation coefficient [26] are demonstrated in Table 1. Regarding SEM and SDD, lower values indicate higher reliability [27]. Normality of the KOOS-PF, the VAS, and the KOS-ADL was evaluated through the Shapiro-Wilk test [28] to select the appropriate correlation coefficient for quantifying the convergent validity of the KOOS-PF. Floor and ceiling (F/C) effects were categorized as significant (≥15%), moderate (10% to <15%), minor (5% to <10%), or negligible (<5%) [29]. Data was tested for suitability for factor analysis through the inter-item correlations, the determinant of the correlation matrix, the Kaiser-Meyer-Olkin (KMO) statistic, and Bartlett’s test; inter-item correlations should lie between 0.30 and 0.90, the determinant was acceptable if greater than 0.00001, acceptable KMO values were values above 0.50, and Bartlett’s test was considered significant if p-value was smaller than 0.50. To identify the number of factors to extract, both the Kaiser criterion (eigenvalues greater than 1) and the scree plot method were applied [30]. Factor loadings were evaluated based on thresholds of ≥0.32 for acceptable communalities, ≥0.5 for moderate communalities, and ≥0.8 for high communalities. A minimum of 60% variance was deemed acceptable to account for the total item variance [31]. The statistical significance was pre-specified at p<0.05.

3. Results

Fifty-five (55) patients, primarily women (N=31, 54.6%), with an established diagnosis of patellofemoral pain, aged from 18 to 65 years old (mean age=38.9, SD=14.8), participated in the study. The participants’ average duration of patellofemoral pain was 7.5 (SD=10.7) months, and the average pain intensity was 7/10 (SD=0.9). The mean baseline KOOS-PF score was 36.3 (SD=16.6). The EDA revealed no missing data across all variables. The demographics and baseline data are demonstrated in Table 2.

3.1. Reliability

The Cronbach’s alpha point estimate was 0.87 (95% CI: 0.85 to 0.93, p-value<0.001), the ICC was 0.95 (95% CI: 0.71 to 0.99, p-value<0.001), the SEM was 3.7, and the SDD 13.6. All reliability measurements ranged from good to excellent, indicating a robust reliability of the Greek version of the KOOS-PF scale. The results are demonstrated in Table 3.

3.2. Construct Validity

Construct validity was assessed by evaluating convergent validity between the baseline scores of the KOOS-PF, the VAS for pain [31], and the KOS-ADL, since the KOOS-PF theoretically measures constructs related to those captured by the VAS and KOS-ADL. The correlations between the scales were analyzed through the appropriate correlation coefficient [20]. The validity assessment results demonstrated a high positive linear correlation between the KOOS-PF and Greek KOS-ADL scores (r = 0.72, 95% CI: 0.57 to 0.83, p-value<0.001) and a moderate negative linear correlation between the KOOS-PF and the VAS scores (r = -0.64, 95% CI: -0.45 to -0.77, p-value<0.001) supporting the construct validity of the KOOS-PF scale in this context. The results of construct validity are presented in Table 4.

3.3. Structural Validity

Factor analysis was used to model the relationships among the observed variables using latent variables, the factors; Exploratory Factor Analysis (EFA) was the method of choice as no strong a priori hypothesis for the interrelations between the variables and the factors is available [32].

3.4. Factorial Validity

Table 5 demonstrates the results of the inter-item correlation matrix. The KMO statistic was 0.826, Bartlett’s test of sphericity was significant (n<0.001), and the determinant was 0.001, disproving multicollinearity. The lowest correlation was observed between items 1 and 8 (0.0529), while the highest was found between items 5 and 6 (0.8323). No correlations exceeded 0.9, but numerous were found below the lower threshold (0.3), primarily related to items 8 and 9. Given that the above-mentioned global tests indicate factorability and that factor analysis is used in an exploratory basis in our research, we decided to proceed with the EFA, accommodating the varying inter-item correlations, through the Principal Axis Factoring (PAF) model [33].

The results of the PAF analysis are presented in Table 6. Two factors were retained, which explained 74% of the total variance; factor 1 explains 46.58% of the variance, and Factor 2 explains 27.42%. The moderate correlation (0.6103) between the factors suggests a meaningful but distinct relationship, supporting the use of oblique rotation [34]. Based on the Kaiser criterion and the scree plot (Figure 1) the two-factor model is the best representation of the 11-item Greek version of the KOOS-PF. The rotated factor loadings indicated that most variables related to everyday activities (items 1, 2, 3, 4, 5, 6, 10, 11) load highly on Factor 1. This suggests that Factor 1 could represent the construction of general physical function in everyday activities. In contrast, the two items related to jogging and running (items 8 & 9) had strong loadings on Factor 2, indicating that Factor 2 could capture a dimension of sport-related activities.

3.5. Floor and Ceiling Effects

Floor and ceiling (F/C) effects were evaluated through the percentage of respondents achieving the lowest (floor) or highest (ceiling) possible scores in a domain. High F/C effects, defined by a significant proportion of respondents scoring the lowest or highest possible values, indicate limited instrument range, measurement inaccuracy, and response bias, compromising the questionnaire’s performance. Historically, F/C effects above 15% are considered significant, though some recommend stricter thresholds for optimal sensitivity [29]. None of the subjects reached the worst or the best possible score, therefore the KOOS-PF demonstrated no F/C effects. The results are summarized in Table 7.

4. Discussion

The aim of this study was to assess the reliability and validity of the KOOS-PF scale in Greek among the Greek population with patellofemoral pain. The reliability measurements showed findings indicating scores ranging from above average to excellent, demonstrating a robust reliability of the Greek version of the KOOS-PF scale. Cronbach’s alpha point estimate was found to be 0.87, while the ICC was found to be 0.95, and regarding SEM and SDD values were lower indicating the high reliability of the KOOS-PF in Greek (Table 2). Evidently, all measurements were pertained to reliability were above average to excellent, indicating a robust reliability of the Greek version of the KOOS-PF scale.

These findings provided the internal consistency of the KOOS-PF scale in Greek, ensuring almost similar painful situations in terms of quality of life of Greek patients suffering from patellofemoral pain. However, our findings (Cronbach’s alpha 0.87) agreed with the original version (Cronbach’s alpha 0.86) [13] and, as well as with the Arabic language version (Cronbach’s alpha 0.81) [21], while in Spanish the internal consistency scores were the highest (Cronbach’s alpha 0.93) [11]. Overall, all the above results confirm the consistent way of using this scale in different cultural populations.

The excellent reliability of the Greek version of the KOOS-PF describes the ability of the scale to perform accurately and consistently well. In the Greek and Arabic languages, the reliability of this scale was quite similar and in these languages the time estimate between repeated measurements was the same at 48 hours, even in the Spanish and the original version their reliability was assessed at 1 and 2 weeks also having high values [11]. In addition, all the above studies showed lower SEM and SDD values as in the Greek version, indicating the clarity and precision of the questions contained in the KOOS-PF scale regarding the good understanding of the patients’ knowledge.

The construct validity of the KOOS-PF scale was assessed through convergent validity showing a high positive linear correlation between KOOS-PF and KOS-ADL scores (r = 0.72) and a moderate negative linear correlation between KOOS-PF and VAS scores (r = -0.64). These findings support the construct validity of the KOOS-PF scale in this context for use in the Greek version (Table 4). Therefore, the KOOS-PF did not show floor/ceiling effects of patients’ scores, indicating no limited instrument range, measurement inaccuracy and response bias, compromising the performance of the KOOS-PF scale in terms of use in Greek (Table 5).

Our findings are consistent with previous validity studies in Spanish (r= 0.71), the original version (r= 0.74), and Arabic, although the validity score was lower (r=0.57) than ours [11]. Therefore, considering that the KOS-ADL questionnaire is already validated for its reliability and validity, we considered that since the participants answer in almost the same way to similar questions of the two tools, it means that experience what they describe (pain, stiffness, quality reduction life) to the extent that they describe them.

Regarding the results of the exploratory data analysis of the KOOS-PF scale indicated that there was insufficient sample adequacy, as the Kaiser–Meyer–Olkin (KMO) statistic was less than 0.70 to conduct confirmatory factor analysis [35,36]. According to Kaiser’s criterion [37], the selection of factors that could be used to explain variance was performed based on the criteria of factors that explained a significant percentage of variance (70-80% in total) and had an eigenvalue >1, indicating that this value is very good for use in statistical analysis (Figure Y). However, eight items (1, 2, 3, 4, 5, 6, 10, 11) showed high loading and communalities at the appropriate range (above 0.30). The decision to retain a single-factor solution was further supported by the scree plot. Subsequently, the above 8 items were used for exploratory factor analysis, with a pre-hypothesized single factor. Loadings and item communalities of the 8 items are presented in Table 6.

Regarding the factors mentioned above indicated that most variables related to everyday activities (items 1, 2, 3, 4, 5, 6, 10, 11) load highly on Factor 1, it appeared that almost all questions of the KOOS-PF scale were either positively or negatively correlated with related activities, such as heavy household chores, jumping, running, and kneeling, along with corresponding symptoms like stiffness and pain during participation (Table 7). This suggests that Factor 1 could represent the construction of general physical function in everyday activities. The correlation of these factors indicated the patients’ self-efficacy in their actions and management.

In contrast, the two items related to jogging and running (items 8 & 9) had strong loadings on Factor 2, indicating that Factor 2 could capture a dimension of sport-related activities. This two-factor model supports the differentiation between general physical activities and more specific activities like jogging and running. The findings of this validity and reliability study will be further analyzed when the sample size of patients is larger, allowing for confirmatory factor analysis. Therefore, future studies should involve larger sample size and specific populations, such as athletes, soldiers, and female in Greece.

5. Conclusions

In conclusion, the KOOS-PF scale in Greek appears to be valid and reliable, providing clinical researchers with an effective evaluation tool for patients with Patellofemoral Pain.