Knee extension range of motion and self-report physical function in total knee arthroplasty: mediating effects of knee extensor strength

Our study used a longitudinal pre- and post-TKA design that took advantage of the substantial differences in change for knee impairments and self-report physical function. The study involved 836 consecutive patients aged 50 years or older who underwent a primary TKA for knee OA performed by three high-volume surgeons at Singapore General Hospital, Singapore, from 3 January 2006 to 29 January 2009. To avoid potential confounding effects from a contralateral TKA operation, we excluded patients if they had a contralateral TKA within the 12 months before or 6 months after their index TKA (n = 152). We also excluded patients who (i) had a history of stroke or other neurological disorders (n = 30), (ii) had a history of lower limb fracture (n = 12), (iii) had previously undergone a hip arthroplasty or high tibial osteotomy (n = 26), (iv) had previously undergone a unicompartmental knee arthroplasty (UKA) on their index knee (n = 4), or (v) developed medical or surgical complications prior to the follow up session (n = 64). Finally, because this exploratory study is concerned with the effects of knee flexion contractures, we excluded patients with missing knee data (n = 48) and patients with knee hyperextension (extension ROM <0°) at the preoperative or follow-up assessment (n = 59). (The recruitment process is summarized in a flowchart in Additional file1.) Thus, the sample for analysis comprised the remaining 441 patients who underwent a preoperative evaluation within 5 weeks prior to their operation. Of note, this sample size was not based on a formal power calculation but on all eligible patients in our database. Follow-up assessment was conducted approximately 6 months after the operation. All data were collected, as part of the clinical process, by physical therapists and entered into an electronic database per routine practice policies of our institution. All patients in this study were managed using a coordinated clinical pathway to ensure standardized medical, pharmacological, and rehabilitation care. Within two weeks post TKA, these patients began a 4- to 6-week rehabilitation program at the Singapore General Hospital outpatient physiotherapy clinic. The Centralized Institutional Review Board of Singhealth (CIRB), Singapore, approved the study and waived the need for informed consent due to the retrospective and anonymous nature of the study.

The demographic information collected in this study included age (in years), sex, ethnicity (Singaporean Chinese, as compared with others), and education (1 = primary school or less, 2 = vocational or secondary school, 3 = college or university). Data about comorbidities were retrieved from patients’ medical records by a research assistant using a checklist modeled after the Self-Administered Comorbidity Questionnaire[17]. Patients’ height and weight were also obtained and body mass index (BMI) was calculated as the ratio of weight to squared height (kg/m²).

At each assessment, patients were interviewed in either English or Mandarin using the Short Form-36[18], of which we used the physical function subscale as the outcome variable and the bodily pain and mental health subscales as the study covariates. All subscales range from 0–100, with higher scores representing better health state. The English and Chinese versions of the SF-36 have been previously validated for use in Singapore[19].

A hand-held dynamometer (Lafayette, IN, USA) was used to measure isometric knee extensor strength based on the testing procedures described by Martin et al.[20]. A total of five raters – two physiotherapists and three technicians – obtained the knee strength (and ROM, vide infra) measurements. A “make” test was used as it was deemed more reliable than a “break” test[21]. Knee extensor strength was measured with the patients in supine position, and the knee was positioned in approximately 30^° flexion by a firm wedge. The dynamometer force pad was placed just proximal to the ankle joint and knee extensor strength was quantified in kilogram force. All patients performed two maximal trials for 3 to 5 seconds with a 30-second rest interval. Additional measurements were taken if the patient reported a failure to achieve maximum effort. The higher of two valid trials was recorded and normalized as a direct percentage of body weight (%BW) because we believe that this index of knee strength made intuitive sense to our patients. Although we[22] and others[23] have shown, using allometric analyses, that the association between muscle force generation and body weight is not a strict 1:1 ratio, analyzing our data using allometrically-scaled knee extensor strength did not qualitatively alter our study conclusions (data not shown). Also of interest, we performed strength testing in the supine position – and not in the seated position – with the intent to reduce the influence of the rater’s strength on the patient’s measurements[20]. We attempted to ensure that the same rater evaluated the same patient at each time point although this was not fully achievable due to the large patient volume in our hospital (~1,500 TKAs per annum[24]). Nevertheless, all technicians were trained and audited by authors HCC and WY who have a combined experience of over 30 years in orthopaedic and sports physiotherapy. Furthermore, for the supine knee strength measurements, previous studies have demonstrated good test–retest reliability (intraclass correlation coefficients, 0.80 to 0.86) and concurrent validity with Biodex dynamometry measurements[20, 25].

A large standard goniometer (Jamar, Clifton, NJ, USA) was used to measure passive knee extension ROM. Knee extension ROM was measured with the patients in supine position with the heel elevated on a firm wedge. The axis of the goniometer was placed on the femoral lateral epicondyle. The proximal arm of the goniometer was directed toward the greater trochanter of the femur whilst the distal arm of the goniometer was directed toward the lateral malleolus of the ankle. Patients were asked to relax to allow the knee to passively extend. Two sets of measurements were taken, and the lower measurement (greater knee extension) was recorded. For the knee extension ROM measurements, one previous study in patients with knee OA has demonstrated good test–retest reliability (intraclass correlation coefficient = 0.85)[26].

We used descriptive statistics to characterize the study sample: we used means (SDs) for continuous variables and percentages for categorical variables. Analysis of variance (ANOVA) for repeated measures was used to test for change over time in the SF-36 and knee measures. Pearson correlation was used to quantify the correlation between knee measures.

We used a mediation model to examine whether knee extensor strength mediated the influence of knee extension ROM on SF-36 physical function at the change scores (pre- to post TKA) level (Figure 1). Covariates in each model included age, sex, BMI, the number of comorbidities, and changes in SF-36 bodily pain and mental health scores. In the mediation model, the total effect (path c) of the independent variable (changes in knee extension ROM) on the dependent variable (changes in SF-36 physical function) comprises a direct effect (path c’) of the independent variable on the dependent variable and an indirect (mediation) effect of the independent variable on the dependent variable through the mediator (changes in knee extensor strength). The mediation effect through the mediator is quantified by the product of the regression coefficient of knee extension ROM on knee extensor strength (path a) and the regression coefficient of knee extensor strength on SF-36 physical function (path b). We used the INDIRECT macro[27] to estimate the mediation effect (path axb) and its bootstrapped (1,000 samples), bias-corrected and accelerated 95% confidence interval (95% CI). Mediation effects are considered statistically significant when zero is not contained within the 95%CIs[27]. Finally, in sensitivity analysis, we tested a reversed mediation model – wherein changes in knee extension ROM were due to changes in knee extensor strength – to assess the direction of the proposed mediation effects. All statistical analyses were done with PASW software, version 18, and R software, version 2.15.0.

Table 1 summarizes the participants’ descriptive characteristics, while Table 2 provides the descriptive statistics for the SF-36 physical function and knee measures. The patients were predominantly female (82%) and were on average moderately overweight (mean (SD) BMI, 28(4.8) kg/m²). All SF-36 and knee measures increased significantly 6 months following a TKA. Specifically, mean SF-36 physical function scores were 39(SD = 23) points preoperatively and 67(SD = 18) points 6 months postoperatively. With respect to the knee measures, knee extensor strength increased from preoperative level by around 40% at the 6 month postoperative assessment (18%BW vs. 26%BW; P < 0.001). Knee extension ROM improved significantly (P < 0.001) from 8.2° (knee flexion contracture) preoperatively to 4.3° postoperatively. Improvement in knee extension ROM was positively associated with increased knee extensor strength (r = 0.22, P < 0.001, Figure 2). Increased SF-36 physical function was positively associated with improvement in knee extension ROM (r = 0.15, P < 0.001) and with increased knee strength (r = 0.27, P < 0.001) (scattergrams are shown in Additional file1).

Figure 1 shows the results of the mediational analyses. Overall, the mediator model accounted for ~29% of the variance in physical function (P < 0.001). Improving knee extension ROM was significantly associated with increasing physical function (P < 0.01 for path c, Figure 1). Controlling for covariates, the mediation effect of knee extensor strength on the association between changes in knee extension ROM and physical function was statistically significant (path axb = 0.13; 95% CI, 0.05 to 0.23). In contrast, no mediation effect was observed in the reversed mediation model (path axb = 0.04; 95% CI, -0.01 to 0.11).