Patients’ perceived walking abilities, daily-life gait behavior and gait quality before and 3 months after total knee arthroplasty

A prospective observational cohort study was performed in a large teaching hospital in the Netherlands between 1 October 2017 and 30 November 2018. Patients were approached via telephone and asked to participate in the study. They then received an informed consent form through the mail (which they completed before starting measurements) together with an accelerometer and questionnaires. They were asked to wear the accelerometer on their lower back for 1 week, day and night, except during aquatic activities such as showering. Patients wore the accelerometer and completed questionnaires both before surgery and 3 months after surgery. This follow-up moment was chosen since at 3 months almost all patients are expected to have recovered to a level close to their final functional level [21]. Similarly, in total hip arthroplasty (THA) patients, most gain in gait function as measured with accelerometers in a clinical setting was found in the first 3 months after surgery [22]. Institutional Review Board approval was received from the Medical Ethical Committee Noord-Holland (number M017.011).

Patients with grade 3–4 knee osteoarthritis, debilitating complaints, and who had previously been evaluated in the outpatient clinic and were on the waiting list for a primary TKA were eligible for inclusion. To increase external validity and to provide results indicative of the everyday function of the general orthopaedic population, we decided to include all patients who were ambulatory and willing to cooperate with study requirements regardless of comorbidity, previous surgeries or osteoarthritis in other joints besides the affected knee. Using G*Power, (Heinrich-Heine-University, Dusseldorf, Germany) [23] a population of 30–50 patients was deemed sufficient to determine a medium-sized effect of 0.5 on instrumented gait quality measures, with a correlation between repeated measures of r = 0.6, based on a power of 0.8 and a p-value of 0.05 [24]. All patients’ general characteristics such as age, gender, body heights, body mass index (BMI), smoking status, and American Society of Anesthesiologists (ASA) classification were collected from their patient files.

All patients received a Genesis II fixed-bearing total knee prosthesis (Smith & Nephew, London, United Kingdom), with a fast-track protocol with local infiltration anesthesia, early mobilization and short-acting opiates only when necessary. Procedures were performed by a team of eight surgeons, each of whom performs at least several dozens of TKA procedures per year. Patients were discharged home when they were able to ambulate independently with crutches or other walking aids or were discharged to a rehabilitation center. They were prescribed a standard physical therapy regimen, which they performed with their own physical therapist.

To evaluate patients’ gait function before and after TKA, three aspects of gait were analyzed: patients’ perception about their own walking abilities, gait behavior (quantity and speed), and gait quality (regularity and stability).

For quantification of patients’ perception of gait, we used The Dutch version of the modified Gait Efficacy Scale (mGES) [25]. It comprises 10 questions about daily walking tasks, with patients indicating how confident they feel regarding executing each task on a scale from 1 to 10 points. The total score ranges from 0 to 100, with 100 indicating full confidence in all tasks. Additionally, patients were asked to complete a questionnaire before surgery and 3 months postoperatively. The main PROM was the Oxford Knee Score (OKS) [26], a 12-item questionnaire with scores ranging from 0 (worst) to 48 (best). The questionnaire also included a numeric rating scale (NRS) for pain, the short version of the Knee Osteoarthritis Outcome Scale (KOOS-PS) [27], the EuroQol 5D (EQ-5D) [28], and the High Activity Arthroplasty Score (HAAS) [29].

Daily life physical activity was assessed during seven consecutive days using a tri-axial trunk-worn accelerometer (Fig. 1) (MoveMonitor, McRoberts, The Hague, The Netherlands). All patients received standardized written instructions on how to wear the device. Accelerations were measured in three directions: anterior–posterior (AP), medial–lateral (ML), and vertical (VT) using a range from − 6 to 6 g, with the sample rate set to 100 samples/s. Patients were instructed to wear the device at their lower back at the level of the fifth lumbar vertebrae using a Velcro belt. Accelerometer data were analyzed using MATLAB (Mathworks, Natick, USA). We used previously validated algorithms [20] to determine epochs of 8 s or more in which continuous gait was detected.

To express gait behavior, first the average number of 8-s epochs per day that the accelerometer was worn was calculated as the amount of gait. If patients reported they had inadvertently been unable to wear the accelerometer for the whole week, the actual number of days the device was worn was noted and the number of epochs per day corrected. Second, gait speed was calculated from the accelerometer data by using the leg length of patients (calculated as 53% from their body height [30]). In a previous study, we established three factors representing three different, prominent aspects of gait quality in TKA patients: AP/VT gait quality, ML gait quality, and Symmetry [16]. Therefore, in the current study we calculated the most representative variable of each of these three factors (stride regularity-VT for AP/VT gait quality, stride regularity-ML for ML gait quality, and harmonic ratio-AP for symmetry) for each of the 8 s epochs of daily-life gait. Because of the longer bouts of ambulation in the current study, additional gait quality parameters that describe the regularity and stability of the gait pattern could be calculated that were not used in the previous study: sample entropy (SE) [31] and the local divergence exponents (LDE) [32] calculated using the methods described by Wolf [33] and Rosenstein [34].

Statistical analysis was done using IBM SPSS statistics, version 23 (IBM, Armonk, USA). For all questionnaire, gait behavior, and gait quality variables, the normality of the differences between the two assessments (before and after TKA) was checked by visual inspection of their q–q plot and box plot. A Shapiro–Wilks test was carried out on the differences. If the differences were from a normally distributed population, a paired t-test was used to determine if there was a significant difference between the two assessments. If the differences were not from a normal distribution, a Wilcoxon signed-rank test was used to determine if there was a significant difference between the two assessments. The gait quality parameters that were not included in the factor analysis of the previous study, i.e. SE and LDE, were also tested separately.

After being approached by phone, 50 patients agreed to participate in the study and completed the informed consent form. Out of these, a total of 38 patients completed the entire protocol and were included in the analysis. All patient characteristics are reported in Table 1, and a flow chart of the inclusion of the study population is shown in Fig. 2. Of the 12 patients that did not complete the protocol, five dropped out due to missing gait assessment because of a device malfunction, four patients had their surgery postponed or cancelled, two patients did not want to partake in follow-up measurements, and one patient did not receive the accelerometer due to logistical problems.

Patients’ perceived ability of gait function was measured with the mGES and PROMs questionnaires. The mGES showed a significant increase after 3 months [median (IQR) 67 (24) vs 79 (25), p = 0.001], indicating that patients felt more confident after TKA. The postoperative score on OKS showed a statistically significant increase [med (IQR) 29 (10) vs 38.5 (8), p < 0.001], which indicated better outcome after surgery as reported by patients themselves. When looking at gait behavior, the quantity of gait as measured with the median number of epochs per day was four epochs higher after 3 months, which was not statistically significant [med (IQR) 219 (179) vs 223 (123), p = 0.421]. In addition, there was no significant difference in gait speed between before surgery and 3 months after surgery [med (IQR) 0.85 (0.03) vs 0.79 (0.01) m/s, p = 0.126]. Of the three key gait quality variables and the dynamic regularity and stability measures (SE and LDE) tested, none showed statistically significant differences between the pre- and postoperative measurements (Table 2). Values before and after TKA of all questionnaire, gait behavior, and daily life gait quality measures are presented in Supplementary Table 1.

Both the OKS and mGES showed significant increases 3 months after surgery, indicating that patients subjectively reported that their surgery was beneficial improved their gait efficacy. No minimally clinically important difference (MCID) or patient acceptable symptomatic state (PASS) have been published for the mGES, making it challenging to determine the clinical relevance of these findings. The psychometric properties, however, have been well established [35, 36]. The improvement found in OKS scores in the current study well exceeds the known MCID [37].

Earlier research suggested that an increased score on the mGES can be predictive of physical activity after TKA [38]. Therefore, the fact that in the current study patients did neither increase the amount nor speed of walking, nor improve their gait quality, is surprising. This implies that patients did not alter their gait behavior after surgery, even with more confidence in walking. It is feasible that once back home after their hospital stay, patients continued their old behavioral patterns of walking. This could be a target for postoperative rehabilitation, and provides an argument for more attention to the behavioral/perceptual aspects of patients’ rehabilitation after TKA, to supplement the functional training aimed at a range of motion and muscle strength.

In contrast to our expectations, the average number of walking epochs per day did not significantly increase in our patients. Similarly, other studies also did not find any changes in physical activity levels 1 month after TKA in the number of steps [15], after 6 months in the average amount of activity [39], and even at almost 2 years after TKA surgery in the average amount or intensity of activity [40]. When compared to healthy age-matched control patients, TKA patients appear to remain below or at most at similar levels of activity [41]. There are indications that the quantity of walking is correlated with patient satisfaction after TKA [42]. The fact that gait quantity does not improve in the first 3 months after TKA surgery could therefore be considered a possible target for increasing satisfaction rates in TKA patients. However, as patient satisfaction is strongly associated with other factors [43], the effect of increasing gait quantity on improving patient satisfaction is probably limited.

Gait speed did not change significantly after surgery in the present study. The finding that patients do not walk faster after TKA is surprising and very relevant since a higher gait speed has been associated with better gait quality and a lower risk of falling [44]. Furthermore, lower gait speed has even been shown to be a predictive factor in patients’ hazard ratio on mortality [6]. Since targeted therapy can help patients achieve higher gait speeds [45], postoperative rehabilitation after TKA could help to address the lack of increase in daily life gait speed found in this study.

Our previous study using trunk-based accelerometry showed significant improvements in gait quality parameters 1 year after TKA in a controlled setting [16]. In contrast to what would be expected based on these previous results as well as TKA outcome measured with PROMs [21], 3 months after surgery the gait quality variables tested in the current study did not show statistically significant differences compared to baseline. One explanation for these findings could be that gait quality measured with accelerometers differs between a controlled clinical setting and daily life gait [9]. This raises the question whether studies that evaluate gait quality only in a clinical setting assess improvement of function and performance in optimal circumstances, and/or whether patients are motivated to perform better after surgery when observed by surgeons or researchers. For future studies, it would be interesting to have gait quality measurements of the same patients both in a clinical setting and in a daily life domestic setting at 3 months and 1 year after surgery. If there is indeed a discrepancy between the clinical and domestic setting, demonstrating this to patients could provide treating physicians and therapists with a way to motivate patients to walk more and better in their home environment. Another reason for the lack of significant differences in gait quality variables could be that patients have altered their walking behavior during their time suffering from knee OA, and that 3 months is not enough for patients to significantly readjust their behavior and thereby their gait quality (and quantity) in daily-life. In contrast to positive effects on gait function 3 months after THA [22], there are other studies that also found insufficient or no improvement of gait quality 3 months after TKA. For example, Alice and coworkers found that knee function and gait velocity had not improved to satisfactory levels 3 months after TKA as measured with 3D motion analysis in a clinical setting [46].

A strength of our study was the high reliability of the daily life gait characteristics, as patients wore the accelerometers at home, day and night, for 7 days. The consensus is that increasing the number of bouts of gait and, thereby, the number of steps measured, increases the reliability of gait parameters [17, 18]. Second, external validity was increased through including a heterogeneous sample of patients that had an indication for primary unilateral TKA.

We added several gait quality parameters that were not part of the factor analysis of our previous study, possibly increasing the chance of a type I error. However, since these parameters did also not show a statistically significant difference, this does not seem to be an issue.

Of three patients, it was not possible to analyze the accelerometry data. It appears that patients used the device inadequately, despite explicit instructions. For future studies, this might be prevented through the use of a smartphone app, comparable to what is already used in improving efficacy and compliance of training programs after TKA [47]. However, algorithms are at present not sufficiently accurate to validly and reliably calculate gait quality measures and gait speed from smartphone data, mostly because of the large variation in the way smartphones are carried on the body.

Information on contralateral knee joint or hip joint pain was not collected, which is a limitation of this study. Kahn et al. reported that in their analysis of physical activity before and after TKA measured with an accelerometer, comorbidities did not appear to influence the measurements in their stratified analysis [40]. They did report that this might be underpowered, so it is not possible to fully exclude the possibility that degenerative joints and previous arthroplasties in the ipsi- and contralateral leg could have influenced the walking capabilities of patients and therefore the results of this study.