Background
An anterior cruciate ligament (ACL) tear is a frequently reported injury in sports activities that include pivoting movements of the knees, such as soccer, basketball, football, handball, and skiing. The incidence of ACL tears is estimated at 68.6 per 100,000 person years [
1]. An ACL tear is a painful condition and has a large impact on a patient’s functioning [
2]. Both conservative and reconstructive (ACLR) treatment options are available [
3]. However, most athletes in pivoting sports will opt for an ACLR procedure. Rehabilitation after ACLR involves exercises and training to enhance the athlete’s performance in sports activities [
4]. Despite extensive rehabilitation, it is estimated that only 50–65% of the patients with ACLR return to their pre-injury level of sports [
5,
6].
Psychological factors are important predictors of return to sport in injured athletes in general [
7], and this also applies to ACL injured athletes [
8,
9]. Psychological variables related to return to sport in ACLR include catastrophizing cognitions and kinesophobia [
10], levels of stress [
11], and self-efficacy beliefs [
12]. Self-efficacy beliefs also predict outcome after ACL tears [
13]. A recent systematic review showed strong support for the role of self-efficacy in predicting ACLR outcome [
9]. As these psychological variables are relevant for rehabilitation outcome in ACL patients, they should be assessed and addressed in ACL rehabilitation [
14].
To be able to identify self-efficacy in ACL and ACLR patients, researchers have used different instruments [
9]. Bandura suggested that self-efficacy should be measured as specific as possible by measuring self-efficacy beliefs related to the target behaviour [
15]. In line with these recommendations, the Knee Self-Efficacy Scale (K-SES) was developed in Sweden to assess self-efficacy beliefs related to behaviour after ACL tears and is previously published in Swedish and English [
16]. The K-SES consists of 22 items measuring self-efficacy in daily activities, sports and leisure activities, physical activity, and knee function in the future. Based on dimension reduction techniques, two important factors were identified [
16]. The first factor refers to self-efficacy related to present functioning (K-SES
present) measured with items related to daily activities, sports and leisure activities, and physical activities. The second factor measures knee function in the future (K-SES
future). The K-SES has shown good internal validity, face validity, and convergent validity. In a longitudinal study, it was shown that self-efficacy assessed with the K-SES prior to surgery was a predictor of ACLR outcome 1 year after surgery [
13].
However, the K-SES is not available in Dutch, and no data have been reported on the goodness of fit of the two factor model in a confirmatory factor analysis. Therefore, the aim of this study was 1) to cross-culturally adapt the K-SES into Dutch according to international accepted guidelines, and 2) to determine measurement properties of the K-SES Dutch language version (K-SES-D) including goodness of fit for the two factors.
Results
Consensus was reached in the translation panel with only minor differences resolved. The final Dutch version was endorsed by 15 patients receiving physical therapy treatment after an ACLR (9 males (60%), mean age is 21 years (S.D. = 2.8), average number of months since reconstruction is 4 (S.D. = 5.6). The back-translation was endorsed by the author of the original K-SES study (personal communication R. van Cingel).
Table
2 displays the demographic characteristics of both sample 1 and 2 and the scores for self-reported knee function scales.
Table 2
Characteristics of patients in the ACL tear group pre-surgery (sample 1: ACL) and in rehabilitation after reconstruction surgery (sample 2: ACLR), and T value for independent samples
Age (mean; SD) | 23.7 (8.5) | 25.9 (8.2) | n.s. |
Gender: % male | 69.0% | 43.0% | |
KOOS pain (0–100) (mean; SD) | 70.0 (18.2) | 81.8 (17.7) | 4.3** |
KOOS symptoms (0–100) (mean; SD) | 63.5 (20.7) | 63.2 (12.8) | n.s. |
KOOS ADL (0–100) (mean; SD) | 82.0 (17.7) | 89.1 (14.2) | 2.8** |
KOOS Sports and Leisure (0–100) (mean; SD) | 37.7 (26.6) | 59.1 (30.6) | 5.2** |
KOOS QOL (0–100) (mean; SD) | 38.7 (16.7) | 48.3 (14.1) | 3.9** |
Average number of months since ACL for patient on the waiting list in sample 1 was 6.7 months (S.D. = 11.1). In sample 2, 58 (77%) of the 75 eligible patients participated. Time since reconstruction is 12.9 month (range = 2–48). The proportion of males was higher in the ACL group awaiting surgery (Z score for difference between two independent proportions Z = 3.6, p < .001). The ACL group awaiting surgery reported poorer scores in KOOS pain, ADL, Sports and Leisure, and QOL (p < .001) when compared to the ACLR group (sample 2). The difference in KOOS subscale scores may be expected as the average time since ACL between both groups is very different. Item characteristics of the K-SES-D items of sample 1 and 2 were analysed first. Item scores ranged from 0 to 10 for each item. All items showed symmetry of frequency distribution, with measured of Skewness for all items < |1|.
Structural validity
Factor analysis in the combined data from sample 1 and 2 was adequate (KMO = 0.95, and Bartlet test significant < .001). PCA using Harris Kaiser’s factor rotation resulted in 2 important factors. Items of the K-SES-D reflecting present physical performance/function had high loading on the first factor (all factor loadings > 0.70). This factor K-SES-Dpresent explained 56% of the variance of the K-SES-D total. The second factor (K-SES-Dfuture) explained an additional 12% of the variation in item scores with only the four items of the knee function in the future subscale demonstrating high loadings. In a confirmatory factor analysis the two factor model could not be confirmed. Only two of the four indices indicated acceptable fit of the model (SRMR = 0.05; CMIN/DF = 4.9; CFI = 0.85; RMSEA = 0.12). Inspection of the covariance matrix showed that in particular the 18 items relating to K-SES-Dpresent did not fit well into the CFA.
Reliability
Average scale scores were calculated for K-SES-D
present and K-SES-D
future and K-SES-D
total score, both in sample 1 (ACL), and 2 (ACLR). In Table
3 scale statistics are given for both subscales of the K-SES and total score for both samples.
Table 3
Descriptive data on final K-SES-D from two independent samples: sample 1 (ACL patients; N = 200), and sample 2 (ACLR patients; N = 58)
Mean | 4.6 | 7.9 | 6.2 | 5.3 | 5.4 | 7.2 |
SD | 2.3 | 2.0 | 2.1 | 2.6 | 1.7 | 1.9 |
Cronbach’s Alpha | 0.92 | 0.91 | 0.83 | 0.81 | 0.95 | 0.96 |
95% CI lower | 4.3 | 7.4 | 5.9 | 4.6 | 4.7 | 6.7 |
95% CI upper | 4.9 | 8.4 | 6.5 | 6.0 | 5.3 | 7.7 |
Observed range | 0–10 | 0–10 | 0–10 | 0–10 | 0–10 | 0–10 |
Skewness | 0.15 | −1.1 | −0.56 | − 0.25 | 0.12 | − 0.98 |
Cronbach’s Alpha as a measure of internal consistency for each subscale is good, and measures of skewness indicate normal distribution. K-SES-Dpresent in the ACL group was significantly lower compared to mean score in the ACLR group (T = 9.9, p < .0001). With regard to K-SES-Dfuture the relation was reversed: ACL patients reported higher levels of efficacy beliefs with regards to their future knee functioning compared to ACLR patients (T = 2.71, df = 256; p < .001).
Reliability or test-retest reliability for the K-SES-Dpresent and K-SES-Dfuture subscales was computed in sample 3 consisting of 50 ACLR patients. Average age in this sample was 25.8 years (range 16–51), with 57% males, and time since reconstruction was on average 6 months (range 2–21). Average change score for K-SES-Dpresent was − 0.36 (95% CI -0.62,-0.09), with SD = 0.92. Test-retest ICC between both assessments was 0.93 (95% CI 0.88,0.96) and SEM was 0.13. Average change score for K-SES-Dfuture was − 0.14 (95% CI -0.35,0.08), with SD = 0.74. Test-retest ICC between both assessments was 0.92 (95% CI 0.86,0.95) and SEM was 0.46.
Construct validity (using hypothesis)
Correlations were computed between measures of self-efficacy on the one hand, and knee function variables in sample 1 and 2 on the other hand. Table
4 shows the correlations between K-SES-D subscales on the one hand, and self-reported knee functioning assessed with the KOOS subscales on the other. For both K-SES subscales multiple correlation (R), and percentage of variation explained (R
2) are given depicting the strength of the relation between the combined set of KOOS scales with K-SES-D scales.
Table 4
Pearson correlation (95% Confidence interval) between K-SES-D and KOOS self-report knee function scales in sample 1 (ACL) and 2 (ACLR)
Pain | .64(.53,.73) | .82(.89,.67) | .16(.00,.31) | .54(.33,.72) | .61(.48,.72) | .82(.68,.89) |
Sympt | .44(.30,.55) | .27(.05,.59) | .15(.00,.32) | .37(.10,.59) | .43 (31,.56) | .34(.00,.64) |
ADL | .63(.51,.72) | .82(.64,.90) | .16 (.01,.33) | .46(.24,.67) | .60(.47,.71) | .79(.61,.69) |
Sport | .70(.59,.79) | .89(.67,.89) | .13(.00,.28) | .60(.41,.76) | .66(.53,.76) | .82(.70,.90) |
Qol | .64(.41,.38) | .59(.35,.78) | .32(.18,.46) | .39(.13,.63) | .57(.44,.67) | .59(.37,.77) |
R | .75** | .85** | .33** | .66** | .73** | .87** |
R2 | .57** | .74** | .11** | .42** | .54** | .75** |
With the exception of the KOOS symptom subscale, all correlations between KOOS subscales and K-SES-Dpresent scores were moderate to high.
In Table
5, correlations are given between K-SES-D
present, K-SES-D
future, and K-SES-D
total scores on the one hand, and the psychological variables on the other hand in sample 1. Psychological variables include fear of movement, pain catastrophizing, health locus of control, and anxiety and depression.
Table 5
Correlation (95% Confidence Interval) between K-SES-D subscales and self-report psychological variables
TSK | −0.66 (−0.51, −0.78) | −0.65 (− 0.46, − 0.78) | −0.74 (− 0.61, − 0.83) |
PCS | −0.64 (− 0.40, − 0.79) | −0.54 (− 0.34, − 0.72) | −0.67 (− 0.42, − 0.80) |
MHLC Intern | 0.10 (− 0.16, 0.40) | −0.35 (− 0.11, − 0.57) | −0.19 (− 0.19, − 0.48) |
MHLC Extern | 0.32 (0.09, 0.57) | 0.34 (0.11, 0.55) | 0.36 (0.10, 0.59) |
MHLC Physician | 0.15 (− 0.12, 0.44) | 0.28 (0.04, 0.52) | 0.20 (0.10, 0.49) |
HADS | − 0.29 (− 0.06, − 0.56) | −0.34 (− 0.04, − 0.56) | −0.32 (− 0.05, − 0.37) |
R | 0.74** | 0.71** | 0.80** |
R2 | 0.55** | 0.51** | 0.63** |
K-SES-D scale scores had moderate to high (r > |.50|) negative correlations with TSK, and PCS. Negligible to low correlations were found between with MHLC scale scores, and HADS score. Together, these variables explained 55, 51, and 63% of the variation in the K-SES scales. Three stepwise regression analyses were performed with K-SES-Dpresent, K-SES-Dfuture, and K-SES-Dtotal scores as dependent variables. Self-reported knee function assessed using the KOOS subscales explained 74% of the variance in K-SES-Dpresent. Entering the combined psychological variables explained an additional 7% of the variation in K-SES-Dpresent (F change = 2.4; df = 6,46; p < .05), 21% of variation K-SES-Dfuture, and 7% of the variation in K-SES-Dtotal (F change = 3.2; df = 5, 47; p < .05).
All except one of the hypotheses formulated in Table
1 were confirmed in the analysis. Hypotheses 2 was rejected as the FCA did not confirm a 2 factor model underlying the data.
Discussion
Measurement properties of the translated K-SES-D were good, with good reliability and validity. Exploratory factor analysis revealed two factors underlying the scores on the K-SES-D similar to the factors reported in the original study: self-efficacy beliefs about daily activities, sports and leisure activities, and physical activity (K-SES-Dpresent), and self-efficacy beliefs about knee function in the future (K-SES-Dfuture). However, in a CFA, this hypothesised 2-factoral model could not be confirmed. More than 75% of a-priori formulated hypotheses were confirmed.
The negligible to low correlation with Internal Health Locus of Control underlines the importance of measuring self-efficacy beliefs as specific as possible related to the target behaviour [
15]. As rehabilitation after ACLR involves exercises and training to enhance the patient’s performance in sports activities [
30], there is a need to understand the athlete’s beliefs regarding exercise and training. The K-SES assesses these beliefs [
16], but item responses did not fit the hypothesised two factor model in a CFA. This means that the items of this scale are not consistent with a researcher’s understanding of the nature of that construct (or factor) [
29]. In particular, the items on the K-SES-D
present factor do not fit well in the model. These findings are consistent with the way the instrument was developed. The authors generated items to be categorized in four a-priori defined groups: daily activities, sports and leisure activities, physical activities, and knee functioning in the future [
16]. In an exploratory factor analysis of the original data, which was replicated in this study, two important factors emerged, K-SES
present, and K-SES
future. Items developed to measure self-efficacy related to present daily activities, sports and leisure activities, and physical activities, are combined in the composite K-SES
present scale. The meaning of the average K-SES
present score is therefore not easy to interpret. As might be expected, K-SES-D
present is strongly correlated with self-reported functioning of the knee assessed with the KOOS, showing large overlaps in variation. The 4 items reflecting K-SES-D
future did not pose any problem in the CFA, and the scale score showed low correlations with current knee function. In addition, K-SES
future has been shown to be a better predictor of future functioning compared to K-SES
present [
13]. However, as ACLR rehabilitation involves extensive exercises and training [
4], the patients beliefs with regard to training and exercise are also important to the physical therapist.
This study is not without limitations. The selection of patients makes it hard to make direct comparisons between this study and outcomes reported in the original Swedish K-SES validation study. In the original study, ACL and ACLR patients were both used to determine different measurement properties [
16]. However, although the samples of patients included in this study differ from the samples in the original study, average item scores in both factor obtained in the current study are similar to item scores reported earlier [
13]. Furthermore, patients were not assessed at a fixed moment in time after ACL (sample 1) or ACLR (sample 2). It is unclear whether the reported differences in time since ACL or ACLR will impact these results. Another drawback of this study is that there are no data available on responsiveness of the K-SES-D. The original K-SES was shown to be sensitive to change as scores increased during rehabilitation after ACL tear [
31]. Further research is needed to determine to what extent the K-SES-D is responsive to change. Finally, in this study data for test-retest reliability were assessed with an interval of 1 week. This interval was chosen to accommodate patients and therapists, and avoid additional visits to the physical therapist. However, COSMIN recommends an interval of 2 weeks to avoid recall of the items. Therefore, we do not know whether recall of items poses a problem in this study.
Despite these limitations, the study has important findings. The K-SES-D measures self-efficacy beliefs related to functioning in ACL and ACLR patients. Acceptability, internal consistency, and test-retest reliability of the subscales are good. Construct validity was confirmed in principal component analysis and hypothesis testing, but not in a confirmatory factor analysis. These self-efficacy beliefs are only weakly correlated to general self-efficacy beliefs. In ACLR patients, self-efficacy beliefs towards present functioning (in particular towards physical activity exercise) are different from self-efficacy beliefs about future functioning. Addressing these self-efficacy beliefs is important as most patients are motivated to return to previous levels of sports [
32], and self-efficacy predicts future function [
13]. Furthermore, self-efficacy is recognized as one of the main psychological factors associated with return to sports [
33], and satisfaction with knee function after an ACLR [
34]. A number of different available techniques to improve self-efficacy beliefs have been studied [
35]. However, only one study reported on the effects of an intervention addressing self-efficacy in ACL patients [
36]. This study concluded that a strategy to improve self-efficacy beliefs as part of a standard rehabilitation protocol did not result in a better outcome compared to a standard protocol. However, the study was limited in size, and no inclusion criteria related to self-efficacy were used. Interventions are likely to be more effective when addressing only those patients with low self-efficacy. In recent years, studies in ACL and ACLR have concluded that an individual, or tailored, approach is needed to increase return to sports [
37]. Including only patients with low K-SES
future in self-efficacy targeting interventions is likely to increase self-efficacy in these patients. Furthermore, other modes of interventions should be addressed as well. Cognitive behavioural approaches as described in the study of Thomeé [
36] heavily depends on social persuasion as a way to improve self-efficacy. Interventions using an exposure based approach within the fear-avoidance model should be considered as well [
38]. Such interventions might also hold a stronger appeal for patients.