Cross-cultural adaptation and measurement properties of the Brazilian–Portuguese version of the Cincinnati Knee Rating System

Abstract Purpose: To translate, culturally adapt, evaluate the measurement properties, and propose a new scoring system of the Cincinnati Knee Rating System for the Brazilian population. Materials and Methods: One hundred fifty people with anterior cruciate ligament injury completed Cincinnati Knee Rating System and Lysholm Knee Scoring Scale on three occasions: 1-week prior to surgery, 90-days after and 95-days after reconstruction. The measurement properties of the Brazilian–Portuguese Cincinnati Knee Rating System, internal consistency, construct validity, reproducibility, ceiling and floor effect and responsiveness, were tested. Results: Using the original scoring system (multiple scores), the internal consistency (Cronbach’s α) varied between 0.54–0.79 (if item deleted = 0.07–0.73); the construct validity (Pearson’s r) varied between 0.19–0.82 (related to Lysholm); the reliability (intraclass correlation coefficient) varied between 0.96–0.99; the standard error of measurement varied between 0.2–1.3 points; the minimum detectable change varied between 0.4–3.5 points; no ceiling or floor effect was detected, and responsiveness (effect size) varied between −0.3–2.7. Using the new proposed scoring system (single score), Cronbach’s α was 0.78 (if item deleted = 0.65–0.71); the Pearson’s r was 0.79 (related to Lysholm); the intraclass correlation coefficient was 0.99; the standard error of measurement was 0.5 points, the minimum detectable change was 1.3 points; no ceiling or floor effect was detected, and effect size was 1.4. Conclusions: The Brazilian–Portuguese Cincinnati Knee Rating System has adequate property measurement and can be used in a Brazilian population. The new proposed scoring system is appropriate. Implications for rehabilitation The Brazilian–Portuguese Cincinnati Knee Rating System is a valid and reliable instrument which can identify consistent clinical changes over the time. The Brazilian–Portuguese Cincinnati Knee Rating System can be used to evaluate and to follow-up a Brazilian population with anterior cruciate ligament injuries and postoperative reconstruction. The new proposed score for Cincinnati Knee Rating System is appropriate and did not compromise the quality of the measurement properties. The new scoring system generates a unique score, creating a potentially simpler and faster clinical understanding of the patient’s condition.


Introduction
Fifty percent of the knee injuries involve the anterior cruciate ligament (ACL), which represents approximately 120,000 cases annually in the USA [1,2]. ACL injuries can lead to serious consequences at medium and long-term, such as recurrent pain and swelling, instability, meniscal injuries, and finally osteoarthritis [3,4].
The main objective during rehabilitation after an ACL injury is to recover patient's complete function [5], and monitoring of the process can be performed through clinical and functional tests, imaging tests and questionnaires [1]. One of the instruments that can be applied to follow-up the effectiveness of the rehabilitation in patients with ACL injury is the Cincinnati Knee Rating System (CKRS) [6][7][8]. The CKRS is used to evaluate changes in clinical condition after surgeries and other treatments, and is considered a highly sensitive instrument to detect these changes [9][10][11].
Risberg et al. [12] investigated the usefulness of the CKRS and other two instruments in patients that had been submitted to ACL reconstruction and assessed the patients in different time points over two years. The CKRS was the only sensitive instrument to detect consistent clinical changes over the time frame investigated [12], perhaps that being the main reason for the common use of this tool in studies related to ACL injury [10,11,[13][14][15].
The CKRS can be used to assess other knee injuries [16][17][18][19][20], and has also been used as reference tool for the creation and validation of other assessment instruments [21]. In the studies by Laboute et al. [22] and Hoher et al. [6], the CKRS was used to develop the PPLP scoring scale and Knee Disorders Subjective History questionnaires, respectively.
The CKRS has never been translated into other languages, which limits its use and comparison of results between studies. For clinical practice and research, it is important to use instruments that are valid for the population of interest; therefore instruments that have been adapted and validated should be used as they are expected to have similar measurement properties as the original instrument [23]. The clinical and scientific relevance of the CKRS justifies the need for the cross-cultural adaptation of this instrument into Brazilian-Portuguese, thus allowing Brazil and potentially other Portuguese-speaking countries to have at their disposal a low-cost and accurate questionnaire that can measure functionality of the knee to monitor progress of rehabilitation after ACL injury or reconstruction. Therefore, the objective of this study was to cross-culturally adapt the CKRS into Brazilian-Portuguese and to investigate the validity of that version. As a secondary objective, we proposed a new score for the CKRS that includes a final combined score, as the original version presents several independent scores based on each subscale/item. Some clinicians and researchers may not use the CKRS for evaluation and monitoring of patients with ACL injury due to its complex scoring system. The CKRS with a unique score possibly makes the clinical interpretation easier and facilitate a broader statistical analysis.

Study design
This was a prospective study for cross-cultural adaptation and validation of clinical assessment tool.

Cincinnati knee rating system (CKRS)
The CKRS is composed of several items divided into five sections, with each section presenting its own independent scoring systems [9]. In general, the lowest score means the worst condition and the greatest score means the best condition:

Section 1
Subscale "Symptom Rating": four questions -0 to 10 points per question -"pain" question is multiplied by 2 and then the total sum is divided by 5. Item "Patient Perception": one question for overall knee condition -1 to 10 points.

Section 2
Item "Sports Activity": one question -0 to 100 points. Subscale "Activities of Daily Living Function": three questions assessing functionality -0 to 40 points per question and the average among them is used. Subscale "Sport Function": three questions assessing functionality -40 to 100 points per question and the average among them is used.

Section 3
Subscale "Occupational Rating": seven questions assessing functionality -0 to 100 points (summing each answer and then multiplying by 2).

Section 4
Overall Rating Scheme: six questions regarding clinical assessments and specific tests -0 to 100 points along with a final classification of excellent, good, fair or poor. Specific for patients with ACL injury and has to be filled by a clinician.

Section 5
Modifications for "Overall Rating Scheme"two subscales "Symptom" and "Instability Ratings": (1) four symptom rating questions -0 to 5 points per question and the sum is used (0 to 20 points); (2) instability -0 to 20 points. Section 5 is an adaptation of Section 1 to assess patients with chronically affected knees once Section 1 could not be applied to chronic patients as they generally do not return to high level sports. Section 5 also has to be filled by a clinician.

Translation
The translation process was performed based on proposed guidelines [24,25] and it is presented in Figure 1.

Procedures for measurement properties assessment
Recruitment of the participants occurred after the surgery (ACL reconstruction) was scheduled. Two questionnaires (the Brazilian-Portuguese CKRS and Lysholm Knee Scoring Scale) were applied on three occasions: one-week prior to surgery, 90 days after surgery and 95 days after surgery. Participants were instructed to seek clarification from the researchers if they had any doubts on how to complete the questionnaires. The Lysholm Knee Scoring Scale, was applied to analyze the Brazilian-Portuguese CKRS for construct validity. The Lysholm Knee Scoring Scale is a questionnaire specifically designed to assess function and stability of the knee and has been previously translated into Brazilian-Portuguese [26,27]. The Lysholm Knee Scoring Scale has eight questions and scores between 0 and 100, with higher scores indicating better condition.
To complete sections 4 and 5, specific equipment are required (KT-1000, effusion measure equipment); furthermore, these sections are designed to collect results from specific functional tests (hop tests) and other physical signs (effusion, crepitation, x-ray, and mechanical instability). Therefore, these sections are not only to assess the patient's perception of the condition. Thus, it was decided with the authors of the original CKRS not to validate sections 4 and 5 in the present study, although translation was performed and the translated versions are presented in Supplementary Material 1.

New scoring proposal
The new proposal does not change the questions of the original version of CKRS, only other manner to score the sections is used. The changes were based on existing instruments [28,29] using a scale with a final score between 0 and 100 (Supplementary Material 1): Section 1: sum of the four questions of the subscale "Symptom Rating" with the patient-perception item. It generates a final score between 1 and 50. Aiming to use a maximum score of 100 points, the final score is multiplied by 2. Section 2: the three questions of the subscale "Activities of Daily Living Function" were kept with the same scores (0 to 40) and the average is used. The three questions of the subscale "Sport Function" were readjusted, so that each question scored from 0 to 60, not from 40 to 100 as in the original, and the average is used. The modification was made to facilitate the calculation reaching a maximum score of 100 points by summing both averages. The item "Sports Activity" was not changed, scoring from 0 to 100. These three parts now generate a single score that goes from 0 to 200, which is the sum of the individual scores. Section 3: scoring did not change, from 0 to 100 points. Final Score: from the sum of the final score of the three sections above, this value is divided by 4, in order to give equal weight and importance to all sections, and to standardize the maximum final score in 100 pointsthe higher is the score, the better is the function.

Participants
The

Data analysis
The statistical analyzes were performed based on the original analysis [9] and the new proposed scoring system. The analysis was performed with the results of the questionnaires applied 90-days after surgery. Data were analyzed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). The following measurement properties were analyzed: Internal Consistency: Cronbach's alpha was calculated to determine whether there is correlation among the items of the Brazilian-Portuguese CKRS [30,31]. Values between 0.70 and 0.95 are considered as adequate [23]. Construct validity: It determines whether one instrument relates to another instrument with a similar construct [31]. In the present study, the Pearson correlation (CI 95%) between the Brazilian-Portuguese CKRS and Lysholm Knee Scoring Scale was analyzed [32,33]. To verify the validity of the new score in relation to the original scores, the correlation between the original scores and the new score also was analyzed. An alpha of 0.05 was adopted.
Ceiling and floor effect: It is done to evaluate the sensitivity of the instrument in detecting different levels of severity of the impairment by looking at the number of individuals that  receive the minimum or maximum scores for the instrument [30,31,33]. To consider these effects absent, less than 15% of the individuals evaluated should reach maximum or minimum scores [31].
Responsiveness: It verifies the instrument's ability to detect clinical changes by calculating the effect size (ES) [31][32][33]. This measure was performed by calculating the change in scores between the results of the Brazilian-Portuguese CKRS applied one-week before surgery and 90-days after surgery. During this time, participants made rehabilitation following a treatment protocol formulated by the team of physiotherapists from the three participating clinics and the treatment was adapted according to the needs of each participant. The ES was obtained by Cohen' d [34] and values smaller than 0.2 are considered small, between 0.21 and 0.5, considered moderate and above 0.8 considered large [32,33]. Positive ES indicate an improvement in the condition and negative ES indicate worsening of the condition. Reproducibility: refers to the degree to which measurements collected at different times in stable individuals will provide similar responses. Reliability was calculated by the intraclass correlation coefficient (ICC 2,1two-way mixed model for consistence in SPSS) with the results from measurements at 90 and 95-days after surgery [22,33]. The minimum value recommended for ICC 2,1 is 0.7 [31,35]. From the ICC 2,1 values, the standard error of measurement (SEM) was calculated with the formula SDͱ1-ICC 2,1 , with SD (standard deviation) being the average of the SD of the test and retest [31,33,36,37]. Based on the ICC 2,1 and SEM, the minimum detectable change was also calculated at the confidence level of 95% (MDC 95 ): [37,38].

Results
A total of 165 participants were recruited throughout the study. Of these, 15 (9%) did not complete the treatment and/ or did not want to remain in the study, totaling 150 participants who answered the instruments on the three occasions (Table 1).

Original scoring system
The subscales of Brazilian-Portuguese CKRS had an internal consistency (Cronbach's alpha) varied between 0.54-0.79, and the analysis "if item deleted" of each subscale between 0.07-0.73. Construct validity (Pearson's r) for the subscales/ items of Brazilian-Portuguese CKRS in relation to the Lysholm Knee Scoring Scale was between 0.19-0.82 (p < 0.05). The calculated ICCs 2,1 were 0.96 to 0.99. The SEMs were between 0.2-1.3 points and the MDCs 95 were between 0.4-3.5 points. In addition, ceiling and floor effect were not significantly observed; ceiling effect was between 0-15% and floor effect was between 0-2%. Responsiveness measured via ESs varied between À0.3-2.7 for the subscales/items of Brazilian-Portuguese CKRS and 1.8 (CI 95% 1.5 to 2.2) for Lysholm Knee Scoring Scale (Table 2).

New scoring proposal
The new proposed scoring system for the Brazilian-Portuguese CKRS had an adequate construct validity (Pearson's r), in  (Table 3).

Discussion
The Brazilian-Portuguese CKRS showed to be a valid instrument with good measurement properties and therefore can be used in the Brazilian population using the new proposed scoring system. This was the first study that aimed to translate and validate the CKRS questionnaire for another language and population. The results of the present study may serve as a basis for further validations and translations of the CKRS as some measurement properties are not presented in the original version of the CKRS. In this study, we presented additional information on internal consistency, construct validity, SEM and MDC 95 for the CKRS. Using a similar approach from the original CKRS, the Brazilian-Portuguese CKRS presented similar or superior measurement properties compared to the original CKRS. However, some of the measurement properties for the Brazilian-Portuguese CKRS were still considered below ideal when compared to the recommended values suggested by Terwee et al. [39] The subscales "Occupational rating", "Symptoms rating" and "Sports function" were considered of below ideal internal consistency. One possible reason for these results is due to the fact that these subscales are composed of only three to four questions each, with possible redundancies among the questions in each subscale. However, when we analyzed the internal consistency for the Brazilian-Portuguese CKRS using the new scoring system the result was considered adequate (Table 3). It is possible that, with the new system, the weight of individual questions are reduced, showing that the CKRS needs to be interpreted as a whole and not as a pool of subscales/items.
For the construct validity, parts of the Brazilian-Portuguese CKRS presented a weak correlation to the Lysholm scale, particularly for the item "Sports activity" (r ¼ 0.42) and the subscale "Occupational rating" (r ¼ 0.19). We did not expect to see a strong correlation between the subscale "Occupational rating" and the Lysholm scale as the Lysholm scale does not address specific issues related to occupational activities. However, the weak correlation between the item "Sports activity" and the Lyshom scale seems to indicate that the Brazilian-Portuguese CKRS (and the CKRS) covers performance related to sports activities in a more specific way when compared to the Lyshom scale, which is in its conception a more generic tool. Nevertheless, the results for construct validity using the new proposed scoring system and the Lysholm scale shows a strong correlation, indicating that the Brazilian-Portuguese CKRS, when analyzed as a single tool, can assess general function of the knee.
Regarding responsiveness, the subscales/items of the Brazilian-Portuguese CKRS showed moderate to high ESs. The ESs for the subscale "Sports function" (ES ¼ 0.5) and "Occupational rating" (ES ¼ 0.3) suggest that these subscales have limited ability to detect clinical changes. For the original version [9], all evaluated subscales/items presented a large ES, except for the subscale "Activities of daily living function", which presented an ES of 0.69. These are similar to the responsiveness levels observed in the new proposed scoring system for the Brazilian-Portuguese CKRS, where a large ES was found (ES ¼ 1.4), showing that the Brazilian-Portuguese CKRS is responsive to change.
In regards to reliability and ceiling and floor effects, both the original scoring system and the new proposed scoring system were considered positive. All ICCs 2,1 scores were above 0.96, with similar levels of reliability when compared to the original CKRS; and the ceiling and floor effects were below 15%. The results for ceiling and floor effects were the ones with the most evident difference between the original and both scoring systems for the Brazilian-Portuguese CKRS. The original version has a floor effect that varies from 0 to 37% and a ceiling effect that varied from 0 to 76% [9]. A possible explanation for the difference in floor/ceiling effects between the translated version (below 15%) and the original version (up to 76%) is the way in which the analyses were performed. The original version analyzed the floor/ceiling effects for each question of the first section. This approach to the analysis may have increased the number of maximal and minimal scores, having a direct influence in the floor/ceiling effect results.
Different ways to score the CKRS have been used and this creates difficulties in comparing results among patients and studies. These difficulties can be clearly seen in the study by Çelik and Turkel [40], who used the CKRS and described that the questionnaire scores ranged from 120 to 420; however, their results showed scores lower than 120 points. Furthermore, in a study by Hohmann et al. [41], the authors state to have normalized CKRS score values on a scale of 0 to 100, which is a unique way to look at the results, and not comparable to other studies. Thus, due to the scoring system for the original CKRS, there is a tendency for authors to create their own scoring system to overcome limitation of the current system. Therefore, the new proposed scoring, which presented adequate measurement properties, allows a standardization and formulation of a single score that can be used in any version of the CKRS. For the current study, we only used the Brazilian-Portuguese CKRS for participants with an ACL injury; however the CKRS was not originally designed only for this population. Future studies are necessary to verify whether the Brazilian-Portuguese CKRS can be used for patients with other knee injuries. Future studies with longer follow-ups would also be beneficial in further understanding the properties related to responsiveness.

Conclusions
After the process of translation and validation, the Brazilian-Portuguese CKRS was considered a valid and reliable instrument with appropriate measurement properties. The CKRS can be considered suitable for the use in the Brazilian population for the assessment and monitoring of ACL reconstruction recovery. Overall, the comparison of the results for the Brazilian-Portuguese CKRS according to both scoring systems suggests that some subscales/items may be useful as part of the whole CKRS tool; however, need to be used with care if analyzed individually. The new proposed score was also appropriate and did not compromise the quality of the measurement properties of Brazilian-Portuguese CKRS, creating a potentially simpler and faster interpretation.