Reply to the Letter to the Editor: Translation and Validation of the German New Knee Society Scoring System

To the Editor,

We thank Mr. Özden for giving us the opportunity to more comprehensively introduce our work, in which we translated and validated the new Knee Society Score (KSS) in German [5]. There are several very interesting comments we will address on a point-by-point basis.

In his letter, Mr. Özden noted that we should not have used the WOMAC as a scoring tool to comparatively analyze the new KSS in terms of construct validity because it doesn’t include an expectation and satisfaction section. Although it is a thoughtful critique, one should keep in mind that the new KSS has a unique satisfaction and expectation section, which no other scoring tool includes. But not all parameters can be present in all scoring tools. As stated in our study, construct validity is analyzed to reflect the consistency in the corresponding domains [5]. We analyzed construct validity using the well-known and widely used scoring tools such as WOMAC and SF-36 because: (1) There is no accepted reference method to reflect the status of patients before and after TKA and (2) these tests have been validated in previous studies [6, 8, 11]. Similar to our study, Noble and colleagues [8] used the Knee injury and Osteoarthritis Outcome Score and SF-12 to evaluate the construct validity of the new questionnaire. Neither measure includes any expectation or satisfaction sections. Other high-quality validation studies of the new KSS translated into Dutch and Korean languages used the WOMAC score for the same purpose as in our study [6, 11].

Mr. Özden also criticized the low correlation of patient expectations in our study. Indeed, we found a low correlation of patients’ expectations; this low correlation was already anticipated, as a similar result had been previously observed by others [6], which we noted in our paper [5].

We agree with Mr. Özden’s second comment, that long questionnaires may burden patients. As discussed in our paper: “like in all adaptation and validation studies, patients had to complete three separate scoring tools simultaneously, which may have resulted in missing or invalid responses as a result of an increase in responders’ burden” [5]. Previous studies have noted this issue [6, 11]. Indeed, there is a tension between respondent burden and the amount of detail one seeks in a patient-reported outcomes questionnaire; more detail can be informative, but too much can represent an unreasonable imposition on patients’ time.

Next, the author of the letter also mentioned total score calculations, which we noted in our Discussion section [5]. We presented total scores alongside all subdomain scores so as to make our findings comparable with other studies [6, 11].

The last two concerns are about sample size and the study design. We find the former comment valuable and appreciate the author’s concern regarding one of the most challenging issues in validation studies. Sample size is vitally important in terms of statistical strength and is sometimes overlooked in the Turkish, Japanese and French validation studies of the new KSS [3, 4, 9]. The most important aspect in this regard is to include a sufficiently strong sample size for internal consistency as well as test-retest reliability.

Although it was thoroughly explained in our paper [5], it must be noted that there is no general consensus regarding the sample size to validate a scoring tool [1]. However, it is recommended to include at least 100 patients for internal consistency [10]. Recommendations vary between two to 20 patients per item [1, 2, 7], although the author quoted only one book chapter, which recommended 10 patients per item [9]. Even if this would be a rule, as the author suggested, other than a recommendation, our study would still be powerful enough with 100 patients, because 10 patients per item suggestion would be met considering the domains of the new KSS. It should not be forgotten, that our analysis was on domain scores as suggested by Noble and colleagues [8], not on total scores. Therefore, a total number of items is of less importance compared to patients per items in individual domains. Additionally, considering the test-retest reliability, our sample size was more than adequate for the analyses we performed as also explained with detail in our paper [5]. Statistical power doesn’t depend on sample size alone but also on the study design. Evaluating the same patients in the pre- and post-operative periods allow for a more-consistent analysis, since these studies are psychometrical studies performed on individual patients. This is more important in the case of the new KSS because the scoring tool has two separate forms for pre- and post-operative patients. Although some validation studies overlooked this detail [9], we performed separate analysis for both forms, which ensures an additional strength to our study. Therefore, the concern about the sample size cannot be applied to our study.

Finally, the reverse correlation between the mental health and vitality domains of SF-36 with the symptoms domain of the new KSS caused some concern. However, as mentioned in our paper, these results were anticipated. Similar results were also previously reported [8, 11], and referenced in our paper [5].