Background
Knee pain is common complaint affecting people of all ages. In adult populations, the reported prevalence of knee pain has been 25–28% depending on the age of the examined persons and the level of pain chronicity [
1‐
3]. In the younger age groups, knee pain is commonly secondary to increased activity, injury or contact sports [
4‐
6]. In older people, however, knee pain that gets progressively worse over time is often sign of osteoarthritis (OA) [
7,
8]. Knee OA in particular is associated with severe disability, owing to the weight-bearing function of the knee and the large range of movements it performs.
To be able to detect and treat individuals with a variety of knee problems effectively requires reliable and valid outcome measures, preferably at low cost. During recent decades, a few well-validated outcome measures have been developed for the assessment of symptoms and function in subjects with knee or hip OA [
9‐
11]. In the elderly population, the Western Ontario and McMaster Osteoarthritis Index (WOMAC) is the self-administered instrument most commonly used to measure pain, stiffness and function in daily living. However, in younger and/or more physically active subjects, joint injuries cause knee problems more often than primary knee OA per se. Younger patients also often have higher expectations regarding physical functioning. Thus, the WOMAC may not be appropriate for these subjects. Partly for this reason, the Knee injury and Osteoarthritis Outcome Score (KOOS) was developed in the late 1990s as an extension of the WOMAC index to address problems associated with knee injuries and/or knee OA [
12].
KOOS is a disease-specific, patient-reported outcome (PRO) measure assessing perceived pain, other symptoms, activities of daily living, sport and recreation functions, and knee-related quality of life. It is freely accessible, and intended for use over the short- and long-term for both research and clinical purposes. KOOS has been found to be a valid, reliable and responsive outcome measure in different patient populations with varying knee injuries (of menisci, ACL or cartilage) [
13,
14], knee OA [
15‐
19] and total knee replacement [
20]. Currently, KOOS is available in 50 different languages and language variants [
21]. However, the validity and reliability of the Finnish version of KOOS in subjects with different knee problems or OA have not been previously reported. Hence, the purpose of this study was to produce a Finnish version of KOOS and to evaluate its construct validity and reliability in a sample of middle-aged patients with knee injuries.
Discussion
This study reports on the cross-cultural adaptation and translation of KOOS into Finnish and its reliability, construct validity and floor/ceiling effects in patients with knee injuries and/or OA. The results indicated that the Finnish version of KOOS has good construct validity, and that the questionnaire is a reliable measure of pain, symptoms, activities of daily living, sport and recreation, and quality of life in Finnish-speaking patients with knee injuries of different kinds.
The psychometric properties of the Finnish version of KOOS were in line with those of the original KOOS [
13,
34,
35] and with the Persian version of KOOS [
14], which studied a group of patients with knee injuries, similar to those in the present study, although not including knee OA. The mean scores for the Sport and Recreation Function and QOL subscales were considerably lower than the scores for the other subscales, as previously reported [
12,
15,
16,
34,
36]. The likely reason for these low values, especially for the Sport and Recreation Function, is that knee injury patients tend to avoid risky activities in their daily lives. The present study demonstrated neither floor nor ceiling effects, as the proportion of worst or best possible percentage scores for the KOOS subscale with the highest scores was only 9%. This demonstrates the appropriateness and comprehensiveness of the questionnaire for a patient population with relatively moderate knee pain and other knee injury symptoms.
The construct validity of the KOOS was determined by comparing the KOOS subscales with the subscales of the WOMAC, SF-36 and Pain-NRS. The KOOS subscales are as representative as those of the WOMAC for measuring pain, stiffness and function. It was therefore expected that strong or very strong correlations between the KOOS and WOMAC subscales would be found. However, in interpreting these coefficients it must be acknowledged that all these measurements overlap to a certain extent. The WOMAC Pain items are included in the KOOS subscale of Pain, the WOMAC Stiffness items are included in the KOOS subscale of Symptoms, and the WOMAC Physical Function items are identical to the KOOS ADL items. Thus, in the latter, the subscale response options (VAS vs Likert) are compared rather than the constructs. Due to overlapping subscales between the KOOS and WOMAC, it was essential in this study to compare and correlate the subscales of the KOOS also with other questionnaire’s subscales assessing similar constructs than those of the WOMAC.
We found, somewhat surprisingly, that the KOOS subscale of Symptoms and the WOMAC subscale of Stiffness showed “only” a moderate correlation. This may partially be due to fact, as referred to above, that the KOOS Symptoms subscale contains five items in addition to the two original WOMAC items, as it also takes symptoms related to knee movement into account. Also noteworthy, alongside the KOOS subscales of Pain, Symptoms and Activities of Daily Living, were the subscales of Sport and Recreation Function and QOL, which showed a moderate or strong correlation with all the WOMAC subscales.
When the KOOS and SF-36 were compared for construct validity, we found strong correlations between the KOOS subscales and those of the SF-36 that measured similar constructs. The highest correlations were observed between the SF-36 subscale of Physical Functioning and the KOOS subscales of ADL and Pain. The SF-36 subscale of Bodily Pain and the KOOS subscales of Pain and ADL also showed strong correlations. In contrast, the KOOS subscale of Symptoms showed the lowest correlations with all the SF-36 subscales. This is in line with the findings of Salavati et al. (2008) in patients with knee injuries, and of Roos et al. (1998) in subjects with knee OA. In fact, Roos et al. concluded that the KOOS Symptoms subscale is not as important as the other four subscales as a determinate of Physical Health. The authors suggested that symptoms and functional limitations should be reported separately and not aggregated into a single score [
13]. All in all, the construct validity for the patients in our study was at more or less the same level as observed in patients with knee injuries [
14] and less severe forms of OA [
18], but higher than that obtained in elderly patients with advanced OA eligible for total joint replacement [
15,
34].
The numeric pain rating scale used in this study measures the intensity of pain experienced in general as well as in specific body regions. It contains subscales that make it possible to explore correlations with the KOOS subscales. As expected, we found strong and moderate correlations between all the KOOS and Pain-NRS subscales, particularly those focusing on the lower extremities and the knee region. These findings, in conjunction with the fact that self-reported knee pain was more severe than pain in any other body region, confirm the utility of the KOOS as a lower-extremity PRO measure in subjects with diverse knee problems. Also of note was the finding of strong correlations between the Pain-NRS Knee and lower limb subscales and the KOOS Activities of Daily Living subscale. This is understandable, given that the knee is a large weight-bearing joint with a large range of movements, and that managing the activities of daily living presumes an extensive repertoire of pain-free weight-bearing movements. However, somewhat surprisingly, the KOOS Pain subscale correlated only strongly instead of, as expected, very strongly with the Pain-NRS Knee subscale. This is most likely due to only partial similarity between the constructs in the two subscales. While the nine-item KOOS Pain subscale is designed to elicit the prevalence and degree of pain during different activities and rest, the Pain-NRS Knee subscale consists of a single item in asking about the intensity of pain experienced during the past week. Nevertheless, the use of the Pain-NRS yielded new information, since to our knowledge no previous KOOS validation studies have investigated the correlations between the KOOS and Pain-NRS subscales. From our pre-defined hypotheses altogether 89% could be confirmed.
Internal consistency was good for all five subscales, exceeding the cutoff value of 0.70. This result is comparable to observations in other languages and patient populations, such as in the Swedish version with preoperative patients [
13], the Dutch version with patients with focal cartilage defects [
37], the Persian version with patients with knee injuries [
14], and the Singapore English and Singapore Chinese [
15], Dutch [
17], Portuguese [
18] and French [
16] versions with knee OA patients. Two recent Polish validation studies with test-retest intervals of one to 2 weeks found Cronbach’s αs higher than 0.90 in patients undergoing ACL reconstruction [
38] and total knee replacement [
36]. The authors concluded that this might be due to the relative homogeneity of their patient group. Accordingly, it has been claimed that, for clinical application, high Cronbach’s α values, of at least 0.90, are needed [
39]. However, given that, our Cronbach’s α values were, with the exception of the Symptoms subscale, all around borderline significance, we consider them reasonable for clinical purposes. The result of the item analysis also suggests that all five subscales are acceptable for inclusion in the Finnish version of KOOS.
The ICC values for test-retest reliability for all the KOOS subscales were good, ranging from 0.72 to 0.86, thereby indicating sufficient standardization of the KOOS questionnaire. Overall, our results are comparable to those of previous methodological studies of KOOS [
12,
13] and other studies [
14,
17‐
19,
36,
37,
40,
41] where ICC values over 0.70 have indicated good or excellent test-retest reliability. Noteworthy, the ICC value for Sport and Recreation Function in our study was somewhat lower than expected. It is commonly seen that Sport and Recreation Function has higher ICC value than Symptoms [
17,
36,
40,
41].We cannot find an exact reason why Sport and Recreation Function had relatively low value, but it is possible that some patients may have changed their activity level during 2 week interval.
The MDC value of 4.4 points at the group level indicates that the Finnish version of the KOOS has an ability to detect a minimum change of 4.5 points between the measurements. The MDC should be smaller than the minimal important change (MIC), which is regarded as the smallest change score needed for the effect to be considered clinically relevant [
42]. For the different KOOS subscales a MIC of 8–10 points has been considered to be appropriate [
34]. Thus the Finnish version of the KOOS is applicable to detect such a change without difficulties. However, it is important to note that the mentioned MDC values apply to patient groups and not to individual patients. The MDC values at the individual level were considerably higher ranging from 16.6 to 33.8 for the different KOOS subscales. The MDC values in our study are of same magnitude or somewhat higher than those found in two recent studies in people with OA awaiting arthroplasty [
36,
43].
We recognize some limitations of this study. First, our sample size was rather small. The response rate for the re-test remained low (~ 45%), meaning that a relatively small number of participants were investigated. The reasons why some subjects did not respond to the second KOOS questionnaire remain unknown. Nevertheless, according to the Consensus-based Standards for the selection of health status Measurement Instruments (COSMIN) group [
44], the sample size of 50 to 99 is considered sufficient. Second, participants in the present study may not represent patients with entire spectrum of knee OA and some common knee complaints affecting pain, such as patellofemoral pain syndrome, rheumatoid arthritis, plica syndrome, Bakers cyst and bursitis. However, it must be borne in mind that the KOOS questionnaire is intended to be used particularly for knee injuries that can result for a variety of reasons, including OA. In addition, the initial participant group recruited to this study was representative of their population, as all knee patients attending the outpatient clinic were invited to take part to the study. Third, due to its cross-sectional design, the responsiveness of the questionnaire was not assessed in this study, which may limit the instrument’s ability to detect clinically important changes over time. For these reasons, we recommend further validation of this preliminary Finnish version of KOOS and consideration of its responsiveness with a larger number of patients with knee complaints, including patients across the full radiographic spectrum of knee OA.