Functional capacity at rehabilitation discharge predicts physical activity characteristics twenty-four weeks later for people with total knee arthroplasty: a secondary analysis of a randomized controlled trial

Abstract

Objective:

To determine the association between performance-based and patient-reported functional capacity at the conclusion of 12-week rehabilitation with average daily step counts and peak walking cadence 38 weeks following total knee arthroplasty (TKA).

Design:

Secondary analysis of an RCT.

Setting:

Veterans Affairs Medical Center.

Participants:

87 U.S. military Veterans (age: 67±7 years, 87% male).

Interventions:

12-week rehabilitation beginning two weeks post-TKA plus random assignment to either a telehealth-based physical activity behavior change intervention (PABC) or control group.

Main Outcome Measures:

Performance-based (Timed Up-and-Go (TUG), 30-second sit-to-stand) and patient-reported measures (Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), Veterans Rand 12-Item Health Survey (VR12)) were assessed at rehabilitation discharge (14-weeks post-TKA). Physical activity was measured using thigh-mounted accelerometry 38 weeks post-TKA. Relationships between participant characteristics (age, sex, BMI, group assignment), functional capacity at discharge, and long-term physical activity outcomes (average daily step count and peak walking cadence) were evaluated using single- and multiple-variable linear and logistic regressions.

Results:

Univariate analyses: TUG time (r= −0.33, p=0.002) and VR12 physical health subscore (r= 0.23, p=0.036) were correlated to average daily step count at week 38. TUG time (r= −0.31, p=0.006) was correlated to peak walking cadence. Multivariate analyses: Multiple linear regression controlling for age, sex, and BMI identified TUG (B= −301.25, p=0.039) and VR12 physical health (B= 93.1; p=0.049) as predictors of daily step count. TUG time (B= −1.5, p=0.012) and assignment to PABC intervention (B= 13.7, p<0.001) predicted peak walking cadence. No significant predictors of attaining a 7500 steps/day threshold were identified.

Conclusion:

Functional capacity at discharge is related to physical activity characteristics 38 weeks post-TKA. While behavior-change interventions are needed to address physical activity deficits post-operatively, the link between functional capacity and activity suggests additional need to address functional capacity limitations during TKA rehabilitation.

Physical activity outcomes following total knee arthroplasty (TKA) are below recommended levels for optimal health and often remain unchanged following rehabilitation.^1–7 A recent observational study comparing pre- and post-TKA daily step counts measured via accelerometry identified a small increase in activity after rehabilitation, which plateaued six months post-TKA and remained below recommended activity levels.⁶ Subsequent observational studies noted the majority of patients remained at their pre-TKA activity levels with the remainder demonstrating modestly increased activity that did not meet the minimum activity guidelines.^{5, 7} Conventional rehabilitation focuses on reducing pain and improving range of motion, muscle strength, and functional mobility.⁸ Despite improvements in objectively measured performance and self-reported function with rehabilitation^9–12 evidence suggests that improved physical capacity is insufficient to increase free-living physical activity.^{1, 6} As a result, recent trials have implemented physical activity coaching to improve free-living physical activity following TKA with results indicating promise for increased physical activity outcomes with no adverse effects on rehabilitation progress.^{13, 14}

However, when rehabilitation and coaching are performed simultaneously, a minimum level of functional capacity recovery is likely necessary to achieve recommended physical activity guidelines.¹⁵ In other words, restoring functional capacity alone may not lead to increased activity, but inadequate functional capacity may limit achieving desired physical activity outcomes. Moreover, there is currently insufficient evidence to suggest specific performance metrics that predict people’s ability to achieve physical activity guidelines after TKA.

Physical activity is often impaired before TKA in individuals with knee osteoarthritis.¹⁶ After TKA, physical activity levels do not increase, with many achieving activity outcomes at or below pre-TKA levels.^{1, 17} Activity levels in the typical post-TKA range have negative long-term health consequences and indicate a clear area for rehabilitation improvement.¹⁸ Increased physical activity after TKA may facilitate better functional outcomes¹⁹, fewer comorbidities²⁰, and better long-term joint health.²¹ Thus, the lack of physical activity observed after TKA is of concern, especially considering patients with initial TKA are at increased risk of requiring additional lower extremity joint replacement and associated further declines in mobility and activity.²²

In addition to daily step count, stepping cadence is an important physical activity outcome. As a surrogate measure of walking intensity, assessing cadence can indicate fitness level.²³ A higher walking cadence is associated with maintaining function and improving disability in people with knee osteoarthritis.^{24, 25} Activity below the recommended intensity levels, including light-intensity and shorter duration moderate-intensity activity has been linked to disability in people with knee osteoarthritis.^{24, 26} Faster walking cadence after TKA is associated with improved patient-reported function^{27, 28}, quality-of-life², and walking speed when assessed in clinic or laboratory settings.²⁹ While evidence that free-living cadence improvements from pre- to post-TKA has been noted², there is limited investigation in identifying predictors of walking intensity to inform post-TKA rehabilitation.

The focus on physical activity patterns has led to a need to establish minimum recommended criteria for optimal benefit. Prior work has identified 7500 steps per day as a recommended target to maintain or improve function, improve the risk of falling, improve cognitive health, and improve cardiovascular and metabolic health.^30–33 While this daily step threshold may be unattainable early in TKA rehabilitation, a target of rehabilitation should be to ensure that a patient has sufficient functional capacity to progress towards this goal in the months following rehabilitation. Thus, the purpose of this study is to determine the association of functional capacity, assessed with performance-based and self-reported measures of physical function after rehabilitation, with average daily step counts and peak walking cadence 24 weeks after discharge from rehabilitation (38 weeks after TKA). Additionally, we sought to identify minimal thresholds of physical function outcomes necessary after rehabilitation to achieve an average of >7500 steps per day 24 weeks following discharge from rehabilitation (38 weeks after TKA).

METHODS

Study Design

This is a secondary analysis of a two-arm randomized controlled trial that examined the effect of physical activity behavior change coaching intervention compared to an attention control group on physical activity outcomes in Veterans undergoing TKA. The study protocol and primary outcomes have been published (NCT03226106).^{34, 35} For the purposes of this study, both the experimental and control group data were included, with group assignment used as a covariate during statistical analyses. The study was conducted within the Veterans Affairs (VA) Eastern Colorado Healthcare System and approved by the Colorado Multiple Institutional Review Board. Written informed consent was obtained from all participants.

Participants

Participants were recruited from November 2017 to August 2022 and were between ages 50-85 years old, US Military Veterans, and scheduled for unilateral primary TKA surgery due to end-stage knee osteoarthritis. Potential participants were excluded if they presented with severe non-surgical limb pain (>5/10 with walking), prior contralateral TKA, an unstable orthopedic, neurologic, or cardiopulmonary condition that limited function, uncontrolled hypertension, uncontrolled diabetes, acute systemic infection, active cancer treatment, or experienced a stroke within the last two years.

Interventions

Conventional Outpatient Physical Therapy

Participants completed a conventional rehabilitation protocol delivered by physical therapists at two VA clinics. The protocol involved 12 sessions over 12 weeks and initiated rehabilitation 2 weeks post-TKA. Additional details on the rehabilitation protocol have been published previously.³⁴

Telehealth Intervention

Participants were randomized to one of two telehealth intervention arms: physical activity behavior change (PABC) or an attention control group. Both groups involved ten, 30-minute participant/interventionist 1:1 telerehabilitation sessions. The telerehabilitation sessions were initiated when the participant began conventional outpatient physical therapy and were distributed over 12 weeks. The PABC intervention involved monitoring daily step counts using a Fitbit wrist-worn sensor (Fitbit, Inc.) and each participant working with the interventionist to develop action plans for achieving daily step activity goals. The attention control group telerehabilitation program matched the frequency and duration of the PABC group but did not involve physical activity discussion, instead focusing on non-activity aspects of health. Further details on each telehealth intervention were published previously.^{34, 35}

Measures

Outcome measures reported in this study were collected at baseline (2-4 weeks prior to TKA), post-intervention/rehabilitation discharge (14 weeks post-TKA), and twenty-four weeks post-intervention/rehabilitation discharge (38 weeks post-TKA).

Baseline: Descriptive Measures

Demographic and descriptive information such as age, sex, and body mass index (BMI) were documented approximately two weeks before TKA.

14-week assessment: Functional Capacity

The conclusion of conventional rehabilitation was 14 weeks after TKA surgery, which also marked the conclusion of the randomly assigned telehealth program. Outcomes assessed at 14 weeks included two performance-based measures, the Timed Up-and-Go test (TUG)^{36, 37} and 30-second Chair-Stand test (30-SCST)³⁸, along with two self-reported measures of physical function: the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)³⁹ and Veterans RAND 12-Item Health Survey (VR-12).⁴⁰ The TUG is a functional mobility assessment during which participants are instructed to stand up from a chair (seat height 46 cm), walk 3 meters, turn, walk back to the chair, and sit as quickly and safely as possible.^{36, 37} The 30-SCST is another measure of functional mobility and lower limb strength. Participants are asked to repeatedly sit and stand from a 46cm seat height chair as quickly and safely as possible in 30 seconds.³⁸ The WOMAC is a 24-item self-report measure of knee pain, stiffness, and function commonly used to evaluate patients with knee osteoarthritis and TKA.³⁹ The VR-12 is also a self-report measure that quantifies overall (e.g., not joint- or condition-specific) function and health-related quality of life.⁴⁰ For the current study, VR-12 scores were separated into mental (VR-12 MCS) and physical (VR-12 PCS) health construct scores.

38-week assessment: Physical activity characteristics

This time point represents 24 weeks (e.g., six months) after completion of rehabilitation. Participants were provided an activPAL micro (PAL Technologies, Glasgow, UK) accelerometer-based sensor and instructed to wear the device for ten consecutive days.^{41, 42} The device was wrapped in a waterproof dressing and worn on the non-surgical thigh (one-third the proximal distance between the knee and hip joints). Participants were instructed to wear the device for 24 hours per day. The device collected accelerometry data at 20 Hz. To be included in the analysis, participants needed ≥4 valid days (3 weekdays and one weekend day) of data, defined by at least 10 hours of wear time and at least 100 steps/day.^43–45

Data were analyzed using associated proprietary software (PALanalysis, PAL Technologies), which reported the total steps taken for each recorded stay. Daily step counts were averaged for each participant across all valid days. In addition, the peak stepping cadence of each walking bout was extracted in 1-minute epochs from the PALanalysis software, further analyzed using a custom MATLAB script (MATLAB R2022a, MathWorks, Natick, MA, USA), and the single highest cadence observed for a minimum of a 1-minute walking bout during the device wear period was used for data analysis.^46–48

Statistical Analysis

Candidate predictor variables of 38-week physical activity characteristics were all collected at the 14-week assessment and first evaluated using univariate linear regression. Any variables with a significant relationship (p < 0.05) to 38-week average daily step count or peak walking cadence were included in subsequent multiple linear regression models, respectively, to identify the demographic and 14-week outcomes most associated. Lastly, significant predictors identified in the multiple linear regression were used in logistic regressions to identify which demographic and 14-week measures were predictive for achieving ≥7500 average daily steps at week 38, the amount recommended for maintaining health and function for adults.^{30, 31, 33}

As a secondary analysis, the sample size was determined for the primary outcome of the parent trial. For stability of models, given the moderate sample size, a maximum of 8 predictors were considered in multivariable linear regression models and 2 predictor variables in logistic regression models.^49–51

RESULTS

Complete data from 87 Veterans were available for this secondary analysis of the 92 that were randomized and participated in rehabilitation. Two participants withdrew from the study and one participant experienced a post-operative adverse event prior to data collection at week 14. Two other participants were lost to follow-up between 14-week and 38-week outcome testing. Due to missing data in one or more predictor variables, 82 and 76 participants were included in the multiple linear and logistic regression analyses for daily step count and peak walking cadence, respectively. Participants included in the analysis were, on average, 67 years old and predominantly male (87.4%) (Table 1). The mean daily step count for the sample at 38 weeks was 7056 steps with an average peak walking cadence of 87.3 steps/min, indicating a medium-to-brisk pace. Sample averages of all 14- and 38-week outcomes are presented in Table 2.

Table 1:

Participant Demographics (n = 87)

Variable	Mean (SD)
Age (years)	67 (7)
Sex (n, % Female)	11 (12.64%)
Mass (kg)	89.3 (11.7)^
Height (m)	1.76 (0.08)^
Body Mass Index (kg/m2)	28.9 (3.2)^
Operative Knee (n, % Left)	37 (44.05%)#
Group Assignment (n, % experimental)	41 (47.13%)

^{^:}missing data from 4 participants

^#:missing data from 3 participants

Table 2:

Participant Outcomes (n = 87)

Variable	Mean (SD)
TUG (s)*	9.5 (3)
30SCST (repetitions)*	12.5 (5)
WOMAC*	19.8 (16)
VR12*	PCS12: 40.9 (8.8); MCS12: 53.3 (10)
Mean daily step count (steps/day)^	7,056 (3,456)
Peak walking cadence (steps/min)^	87.3 (15.9)

^*data from 14-week assessment

^{^}data from 38-week assessment (missing step count data from 1 participant; missing cadence data from 7 participants)

TUG: Timed Up-and-Go Test (missing data from 2 participants)

30SCST: 30-second Chair-Stand test (missing data from 2 participants)

WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index

VR12: Veterans RAND 12-Item Health Survey

PCS12: VR12 Physical Health Construct Score

MCS12: VR12 Mental Health Construct Score

Univariate Regression

Average TUG time (r = −0.33, B = −374.5, p = 0.002) and VR-12 PCS (r = 0.23, B = 88.1, p = 0.036) assessed at 14 weeks post-TKA demonstrated significant relationships with average daily step count at 38 weeks. Group assignment to the PABC intervention, 30-SCST, WOMAC, and VR-12 MCS outcomes were not associated with average daily step count at 38 weeks and were not included in the subsequent multivariate regression analyses. Average 14-week TUG time (r = −0.31, B = −1.64, p = 0.006) and assignment to the PABC intervention (yes/no) (B = 13.27, p < 0.001) were significantly associated with higher peak walking cadence at 38 weeks. No significant associations with peak walking cadence were observed for the 30-SCST, WOMAC, VR-12 PCS, or VR-12 MCS outcomes. The 14-week variables with significant associations were included in the subsequent multiple linear and logistic variable regression analyses.

Multivariable Regression

Demographic variables of age, sex, and BMI were included in the regression models for both outcomes of daily stepping and peak walking cadence due to known relationships of these variables with performance-based and patient-reported physical function.^52–54 For daily step count, average TUG time and VR-12 PCS were significant predictors, explaining 10.9% of the variance in the 38-week daily step count (Table 3). In the multiple linear regression, average TUG time and assignment to the PABC intervention (yes/no) were significant predictors of 38-week peak walking cadence, explaining 25.2% of the variance (Table 4).

Table 3:

Multivariable regression analysis predicting mean daily step count at 38-weeks post-TKA

Predictor Variable	Univariate [r; B]	Multivariate Slope [B] (95% CI)	P	VIF
TUG (s)	−0.33; −374.5	−301.25 (−586.48, −16.01)	0.039	1.48
PCS12	0.23; 88.1	93.10 (0.10, 186.10)	0.049	1.25
Age	−0.02; −11.2	4.45 (−116.70, 125.60)	0.942	1.48
Sex	-; 278.3	168.17 (−1980.04, 2316.37)	0.877	1.04
BMI	−0.02; −9.6	−117.67 (−355.23, 119.89)	0.327	1.08
Intercept	-	9057.85 (−823.17, 18938.87)	0.072	NA

Adjusted R² = 0.109

r = Pearson’s correlation coefficient

B = unstandardized coefficient

VIF = Variance Inflation Factor

TUG: Timed Up-and-Go Test

PCS12: Veterans RAND 12-Item Health Survey Physical Health Construct Score

BMI: Body Mass Index

Table 4:

Multivariable regression analysis predicting peak cadence at 38-weeks post-TKA

Predictor Variable	Univariate [r; B]	Multivariable Slope [B] (95% CI)	P	VIF
TUG (s)	−0.31; −1.64	−1.5 (−2.7, −0.3)	0.012	1.26
Group assignment	-; 13.27	13.7 (7.2, 20.3)	<0.001	1.06
Intercept	-	106.4 (62.2, 150.6)	<0.001	NA
Age	−0.17; −0.37	−0.02 (−0.5, 0.5)	0.952	1.30
Sex	-; 10.0	3.3 (−6.1, 12.7)	0.489	1.10
BMI	−0.19; −0.5	−0.3 (−1.4, 0.7)	0.497	1.03

Adjusted R² = 0.252

r = Pearson’s correlation coefficient

B = unstandardized coefficient

VIF = Variance Inflation Factor

TUG: Timed Up-and-Go Test

BMI: Body Mass Index

The logistic regression using 14-week average TUG time and VR-12 PCS score to predict 38-week daily step counts ≥ or <7500 daily steps did not identify a significant predictor (Table 5).

Table 5:

Multivariable logistic Regression (≥ or <7500 mean daily steps at 38-weeks post-TKA)

Predictor Variable	Slope [B] (95% CI)	P	VIF
TUG (s)	−0.031 (−0.066, 0.003)	0.081	1.06
PCS12	0.009 (−0.003, 0.021)	0.143	1.06

B = unstandardized coefficient

VIF = Variance Inflation Factor

TUG: Timed Up-and-Go Test

PCS12: Veterans RAND 12-Item Health Survey Physical Health Construct Score

DISCUSSION

This analysis demonstrates that measures of functional capacity (TUG time and VR-12 PCS) at the time of discharge from TKA rehabilitation predict daily step count 24 weeks later. TUG time, along with participation in the physical-activity behavior-change intervention, was also a significant predictor of peak walking cadence 24 weeks following completion of rehabilitation. While unable to determine causation from the current study design, the clinical implications are that post-operative rehabilitation optimizing functional capacity may provide a foundation for patients to be physically capable of increasing physical activity. In doing so, patients may be more likely to maintain function and prevent disability following TKA rehabilitation. This analysis contributes to the existing body of evidence on optimal rehabilitation utilization and goals of the post-operative rehabilitation period as a means to maximize long-term function and health.

Functional capacity and physical activity relationship

Prior studies have identified conflicting relationships between functional capacity and physical activity following TKA.^{19, 55, 56} Many studies have focused on identifying pre-operative predictors of physical activity, identifying higher BMI, younger age, male sex, and higher activity levels via patient-reported questionnaire and accelerometry as primary predictors.^{5, 7, 57} During the acute post-operative phase (3 weeks post-TKA), achieving a gait speed >1.05 m/s and higher patient-reported physical activity predicted higher patient-reported physical activity levels at 2 years post-TKA.⁵⁸ At two years post-TKA, those with high self-reported knee function were most likely to have increased physical activity from pre-TKA levels.⁶ Many identified predictors are either non-modifiable (age, sex) or are identified at time points during which rehabilitation services are rarely utilized (pre-operative, >1 year post-operative). This study is the first to explore predictors of future post-TKA physical activity at conclusion of rehabilitation (14 weeks post-TKA). Our univariate analyses indicate TUG is negatively correlated, while VR-12 PCS scores positively correlate with daily step count 24 weeks after rehabilitation end. When accounting for age, sex, and BMI variability in the multivariable analysis, TUG performance and patient-reported physical health (as measured by the physical health subscale of the VR-12) were confirmed as the most relevant modifiable functional capacity outcomes associated with increased long-term daily step counts.

Walking cadence as potential indicator of functional mobility

Beyond total daily step counts, cadence during walking has emerged as a potential indicator of lower extremity functional mobility and a surrogate measure of walking intensity.^{23, 30} To date, no predictors of stepping cadence have been identified following TKA. Current activity guidelines recommend adults perform 30 minutes per day at a moderate intensity, which for walking is approximately 100 steps per minute.^{23, 30, 59–61} Meeting this guideline would result in approximately a 3000-step walking bout each day and approximately 21000 steps per week. Peak walking cadence has been linked with numerous benefits, including better functional performance and higher self-reported function, and may be an important outcome independent of daily step count.^{62, 63} Our results indicate that those with better TUG scores at the end of TKA rehabilitation walk at higher cadences 24 weeks later, suggesting that improving physical functional capacity during rehabilitation may allow patients post-TKA to walk with higher cadence to achieve recommended exercise intensity. Of note, being assigned to the PABC group focused on changing physical activity behaviors also predicted a higher walking cadence, suggesting the behavioral intervention may work best in concert with a threshold level of functional capacity to encourage purposeful walking at higher cadences. However, the average peak walking cadence for our sample was 86.9 steps per minute, below the threshold of 100 steps/min for moderate-intensity walking. While predictive of increased peak cadence, the participants still did not achieve peak daily walking at a moderate intensity. Of note, the behavior-change intervention was focused on improving daily step counts but did not focus on the intensity of walking.

Differences between functional capacity and free-living physical activity

The results of these analyses are noteworthy because prior studies have identified improvements in functional capacity following TKA with no concurrent change in physical activity. For example, across multiple studies, self-reported function as measured using the SF-12, SF-36, Knee Society Score, and Nottingham Health Profile instruments improved while measures of physical activity remained near pre-operative levels or below.^64–66 The combination of prior findings with those presented here suggests a threshold level of functional capacity may be necessary to achieve optimal activity levels. However, improving functional capacity alone is not sufficient. Also of note are potential differences in the construct being assessed with self-reported compared to performance-based functional measures. Prior evidence indicates self-report measures improve more rapidly post-TKA than performance-based measures with patients often rating themselves as more functional than their performance-based measures would suggest.^{10, 67, 68} Our findings indicate both self-reported and performance-based measures of function predict physical activity behavior 24 weeks later, highlighting the need to evaluate both constructs during post-TKA rehabilitation. As our data also illustrate, there is likely a role of behavioral interventions to introduce more specific and purposeful changes in physical activity, such as walking bout duration, cadence, and number of bouts. Our findings provide support that behavioral intervention can occur concurrent with traditional rehabilitation to maximize physical function and ensure patients have the capacity to be active at sufficient levels to derive benefit.

This study also sought to identify the minimum outcomes required to achieve >7500 average daily steps. Our analysis was unable to identify predictors for achieving this activity threshold. This is perhaps due to the distribution of daily step counts within the sample as <42% (n=37) of participants achieved >7500 steps at 38 weeks post-TKA. Alternatively, it may also suggest the lack of a clear threshold in physical function to meet the recommended steps. To this point, our study sample had higher step counts when entering the study (before TKA) than those reported in prior studies, and those exceeding the 7500 daily step threshold may have done so by returning to baseline habits regardless of their physical function recovery.

Study Limitations

A primary limitation of this study is the large percentage of males in the sample compared to national trends in TKA by sex (87.4% male). Our sample was also limited to Veterans of the U.S. military and may not generalize to non-Veterans or patients receiving care under different care models. In addition, our sample was more physically active pre-operatively than previously published typical average step counts. Due to the nature of the parent clinical trial, the rehabilitation protocol for this study was also highly standardized and may not reflect usual practice patterns or clinical outcomes. Future studies should prospectively assess samples that include more females, less active individuals, and those receiving less rehabilitation to identify these factors’ influence and necessary thresholds to facilitate meeting physical activity guidelines.

CONCLUSIONS

TUG time and patient-reported physical health at completion of TKA rehabilitation predict physical activity 24 weeks later, along with participation in physical activity behavior change intervention. An emphasis on improving TUG time and the patient’s self-reported physical health symptoms during post-operative rehabilitation may provide a foundation of functional capacity to increase daily step count and walking cadence after TKA.

List of Abbreviations

30SCST

30-second Chair-Stand Test

BMI

body mass index

PABC

physical activity behavior change

TKA

total knee arthroplasty

TUG

Timed-Up-and-Go

VA

Veterans Affairs

VR12

Veterans Rand 12 Item Health Survey

VR12-MCS

Veterans Rand 12 Item Health Survey Mental Component Score

VR12-PCS

Veterans Rand 12 Item Health Survey Physical Component Score

WOMAC

Western Ontario and McMaster Universities Osteoarthritis Index