The effect of pre-operative cardiorespiratory fitness on functional and subjective outcomes following total hip and knee arthroplasty: a single centre, observational study

Arthroplasty is an effective intervention for treating end-stage osteoarthritis. Despite being a common and relatively safe procedure, there is a wide range of physical function and patient-reported outcomes following arthroplasty.^1,2 Low cardiorespiratory fitness, specifically a low peak oxygen consumption (VO₂; i.e., <15mL/min/kg), and/or low anaerobic threshold (<11mL/min/kg) are associated with poorer outcomes and recovery following non-cardiac surgery;³ however, no data are available on the clinical significance of these metrics in patients scheduled for total joint arthroplasty. As cardiorespiratory fitness provides a systemic and highly integrative assessment of physical function and capacity,⁴ these pre-operative variables are likely associated with a patient’s recovery following total joint arthroplasty (e.g., ability to tolerate surgical stress, early ambulation), as for other surgeries.⁵

Cardiopulmonary exercise testing (CPET) is the gold standard for assessing pre-operative cardiorespiratory fitness.⁵ Despite its utility as an objective pre-operative risk stratification tool, CPET is not universally available, as it is resource intensive (e.g., time, expensive equipment, specialised staff, etc.). Furthermore, clinical contraindications may preclude some patients performing the test.⁶ While assessing the utility of CPET in patients scheduled for arthroplasty, it would, therefore, also be pertinent to know how other forms of assessment compare as a feasible alternative when CPET is not available or appropriate. For example, objectively measured pre-operative physical activity (e.g., accelerometer or pedometer) is an easily implemented tool that is associated with post-operative complications in gastrointestinal surgery patients.^7–9 Although physical activity is correlated with peak VO_2, it may represent an index of fitness independent of peak VO₂; that is, daily physical activity has minimal effect on increasing peak VO₂, but has numerous physiological (e.g., breaking sedentary behaviour, maintaining glucoregulation, neuromuscular function) and psychological benefits for the surgical patient.⁴ Even simpler to use, the Duke Activity Status Index (DASI) questionnaire provides an estimated peak VO₂ using a person’s perceived ability to perform activities of daily living and has been shown to correlate with directly measured peak VO₂ (via CPET) in older adults presenting for major non-cardiac surgery.¹⁰ While these associations show promise in older adults and some peri-operative settings, it remains unclear how well these measures predict recovery from total joint arthroplasty.

Therefore, the aims of this study were to assess: 1) the effect of pre-operative peak VO₂ on functional and subjective recovery at 6-weeks post-operative, and 2) the relationship between other pre-operative assessments, and functional and subjective recovery following total hip or knee arthroplasty.

Materials and methods

Ethics

Ethical approval for this study was obtained from the Health and Disability Ethics Committee (18/NTA/148), and the study conformed to the standards set by the Declaration of Helsinki. The trial was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12618001358235). Written, informed consent was obtained for all participants.

Study design

Participants who completed a prior randomised control trial¹¹ were invited to participate in this study comparing pre-operative measures of cardiorespiratory fitness and physical and subjective health on functional and subjective recovery from total hip or knee arthroplasty (referred to as hip arthroplasty or knee arthroplasty hereafter). This study uses a subset of data from the previous trial, as the COVID-19 pandemic altered the original intervention timeline. Specifically, participants who completed the 12-week prehabilitation intervention but had not yet scheduled surgery were invited for a pre-operative assessment within a week of their surgical procedure, with post-operative follow-up. Data collection took place from September 2019 to July 2021.

Participants completed a pre-operative assessment (see below for details; Figure 1) within 1 week before undergoing hip or knee arthroplasty. Post-operatively, each participant completed the Surgical Recovery Scale (SRS) on day 7.¹² At 6-weeks post-operative, participants repeated the assessment session.

View Figure 1–2, Table 1–4.

Study participants

Patients from Dunedin Public Hospital’s Orthopaedic department waiting list for hip or knee arthroplasty, who were able to travel to the testing location within 1 week of surgery, were invited to participate in this study.

Participants were excluded if they met any of the following criteria: surgery scheduled and insufficient time to perform assessment; a contraindication to non-physician supervised maximal exercise testing;¹³ stable or unstable angina; myocardial infarction within the last 3 months; an implantable cardioverter defibrillator and/or pacemaker; revision arthroplasty; staged bilateral total joint replacement; pathology limiting upper-limb exercise (i.e., shoulder-joint osteoarthritis); and any other medical condition deemed by the study anaesthetist or cardiologist to be a significant risk to study participation.

Experimental procedures

Pre-operative assessment

CPET was performed to measure cardiorespiratory fitness (i.e., peak VO₂, anaerobic threshold) using a previously reported protocol;¹⁴ in brief, the cross trainer was the preferred CPET modality, but arm ergometry was used when this was not feasible (e.g., severe joint pain, joint motion limitations, etc.). A 3–5-minute warm-up was performed, then intensity was increased by 10–20 watts per minute (dependent on participant tolerability and ergometer type) until volitional fatigue. Calibration of gas and flow (Quark CPET; COSMED, Rome, Italy) was conducted per manufacturer instructions before each test. All CPETs were supervised by a registered clinical exercise physiologist (BR). Anthropometric measures of height (stadiometer, Wedderburn WS-HRP, Auckland, New Zealand), body mass (scales, Seca, Hamburg, Germany) and body mass index (BMI; mass [kg]/height² [m]) were collected. Physical function was assessed via the 30-second sit to stand test, timed up and go test and knee joint range of motion, following established procedures.^15,16 Subjective measures were assessed using validated questionnaires, including perceived quality of life (12-Item Short Form Survey¹⁷ [SF-12] and EuroQual-5D¹⁸ [EQ-5D]), functional capacity (DASI¹⁹) and impact of osteoarthritis (Western Ontario and McMaster Universities Osteoarthritis Index²⁰ [WOMAC]). Physical activity was assessed using accelerometry (activPAL3c, Glasgow, Scotland) in the 7 days preceding surgery.

Post-operative measurements

Seven-days post-operative

Length of stay data were acquired from hospital notes and discharge letters. Participants completed the SRS questionnaire at 7-days post-operative.¹² Participants were asked to complete the questionnaire alone and in a quiet area with no distractions.

Six-weeks post-operative

Participants repeated the pre-operative assessments (except CPET) and the SRS. All assessments were conducted by the same researcher (BR) at Dunedin Hospital and the School of Physical Education, Sport and Exercise Sciences, University of Otago, Dunedin, New Zealand.

Standardisation

To ensure test repeatability, participants were asked to comply with recommendations outlined previously;^11,14 in brief, this included abstaining from cigarette smoking 4 hours prior, caffeinated and alcoholic beverages 12 hours prior and moderate- or high-intensity physical activity 24 hours prior to each CPET. The peri-operative team (surgeon, anaesthetist, surgical nurse and administrative team) were not aware participants were being followed up post-operatively, and study participation did not impact inpatient or outpatient care. Following arthroplasty, all participants were offered standard care rehabilitation as typically provided by the hospital. Initially, rehabilitation entailed 2–5 days of inpatient physiotherapy consisting of weight bearing, joint mobilisation, lower-limb flexibility exercises and stair climbing/descending exercise. Upon discharge from hospital, participants were provided (by the hospital) a generic home exercise programme. The exercise programme primarily consisted of joint range of motion and strengthening exercises. Participants were advised by hospital staff to perform the exercises daily.

Data analysis

CPET data were analysed independently by two researchers (BR & HC), and peak VO₂, anaerobic threshold and oxygen uptake efficiency slope were determined using previously reported methods.¹⁴ Accelerometry data were analysed using manufacturer software (PALbatch v8.10, PAL Technologies, Glasgow, Scotland) and data from only “valid wear days” were included in analysis—an entire 24-hour period with 0 minutes of “non-wear time” was considered valid.

Statistical analysis

Given the relatively small sample, observational nature, absence of prospective power and large number of simultaneous statistical comparisons, all statistical analyses were exploratory and should be interpreted with caution. Descriptive data are expressed as mean (standard deviation) or number (proportion). To assess the effect of peak VO₂ and the anaerobic threshold on post-operative outcomes, participants were stratified based on previously defined thresholds in non-cardiac surgery (i.e., peak VO₂: 15mL/min/kg; anaerobic threshold: 11mL/min/kg;³ no previously defined thresholds available for total joint arthroplasty patients). Grouped means were then compared via a repeated-measures Analysis of Variance (ANOVA), with pre-operative osteoarthritis impact as a covariate, or an Independent Samples t-test, as appropriate. For the ANOVA, the main effects of Group (low vs high fitness), Time (pre-operative vs post-operative) and the Interaction (Group×Time) were assessed to determine whether the change from pre-operative to post-operative differed by fitness group. Post-hoc testing using Tukey’s test was performed if statistical significance was observed (i.e., p <0.05). Categorical data were analysed using the Chi-squared test. Simple linear regression and Pearson’s correlation coefficients were used to analyse the relationship between pre-operative variables and post-operative functional and subjective measures. To aid in the interpretation of the correlation coefficients, 0.9–1.0 was considered a very high correlation, 0.7–0.9 high correlation, 0.5–0.7 moderate correlation, 0.3–0.5 low correlation and 0.0–0.3 negligible correlation.²¹ Multiple linear regression was performed to assess pre-operative peak VO₂ for predicting functional (daily step count, timed up and go test) and subjective (SRS and WOMAC score) recovery at 6-weeks post-operative. Pre-operative peak VO₂ was modelled with sex, BMI and age; these were selected due to their potential relationship with post-operative recovery while attempting to avoid multicollinearity with other measures. Complete-case analysis was used, excluding participants with missing data on any variable required for a specific analysis. Statistical analysis was performed using R (version 4.1.1, R Development Core Team) and graphed using Prism (v8.0, GraphPad, San Diego, USA).

Results

Participant characteristics

Fifty-one participants completed pre-operative and post-operative assessments and were included in this analysis (Figure 2). All participants survived the surgery, and 6-week recovery period and descriptive characteristics are listed in Table 1. Participants scheduled for knee arthroplasty had a higher BMI and on average were Class 1 obese.²² There were no differences in any CPET variables between hip and knee arthroplasty participants, except the oxygen uptake efficiency slope was lower (-418 AU, 95% CI: -823 to -12, p=0.044) in participants scheduled for hip arthroplasty.

Association between pre-operative peak VO₂ and post-operative functional and subjective recovery

Low fit vs high fit participants

Participants were categorised into high fit or low fit based on their peak VO₂ (i.e., >15 or <15mL/min/kg respectively). At 6-weeks post-operative, participants with a pre-operative high fit score performed five more sit to stand repetitions (95% CI: 3 to 7, p=0.002) and performed an additional 3,460 steps/day (95% CI: 1,053 to 5,867, p=0.006) compared to low fit participants (Table 2). Low fit participants decreased their timed up and go test duration to a greater extent but remained 2.2 seconds slower (95% CI: 0.9 to 3.5, p=0.002) at 6-weeks post-operative. Estimated peak oxygen consumption was higher (via the DASI; +4.3mL/min/kg, 95% CI: 1.0 to 7.7, p=0.011), and impact of osteoarthritis (WOMAC; -8.5 AU [95% CI: 0.7 to 16.6, p=0.034]) was less for the high fit group at 6-weeks post-operative compared to the low fit group. High fit participants had a better early (i.e., 7-day post-operative; +12 AU, 95% CI: 3 to 22, p=0.014) and later (i.e., 6-weeks post-operative; +13 AU, 95% CI: 5 to 21, p=0.003) subjective recovery from surgery. Length of hospital stay was not statistically different between groups (0.8 days, 95% CI: -0.4 to 1.9, p=0.176). Analyses using alternate stratification thresholds for peak VO₂ (i.e., low fit £14.2mL/min/kg; high fit ³18.2mL/min/kg),²³ and using the anaerobic threshold (i.e., > or <11 mL/min/kg) did not appreciably alter the statistical significance of these results (Appendix Tables 1 and 2).

Correlational and simple and multiple linear regression analysis

Pre-operative peak VO₂ was moderately correlated with post-operative daily step count (r=0.65, p <0.001) and timed up and go test time (r=-0.56, p <0.001). In regression analysis, for every 1.0mL/min/kg increase in pre-operative peak VO₂, the number of daily steps at 6-weeks post-operative increased by 361 (95% CI: 234 to 488; Table 3); this was slightly lower (209 steps; 95% CI: 32 to 386) in the multiple regression model (Table 3, Appendix Tables 3 and 4). Hip arthroplasty participants had a high correlation with pre-operative peak VO₂ and post-operative daily step count (r=0.73, p <0.001), and participants completed 490 (95% CI: 285 to 695) additional steps per day for each 1.0mL/min/kg increase in pre-operative fitness (p <0.001; Appendix Tables 5 and 6).

Pre-operative peak VO₂ was also a significant predictor of subjective recovery (Table 3; Appendix Tables 7 and 8). For every 1.0mL/min/kg higher pre-operative peak VO₂, surgical recovery scale score at 7-days and 6-weeks post-operative was 0.8 (95% CI: 0.2 to 1.4) and 0.7 (95% CI: 0.3 to 1.1) higher, and this was not statistically different for 6-week post-operative score when controlling for age, sex and BMI. Pre-operative peak VO₂ was not associated with length of hospital stay (p=0.204; Table 3) for either arthroplasty type (p ≥0.350; Appendix Tables 5 and 6).

Non-CPET measures for estimating post-arthroplasty recovery

Participants in the high fit group as estimated by the DASI (estimated peak VO₂ >15mL/min/kg) post-operatively performed five more sit to stand repetitions (95% CI: 3 to 7, p <0.001) and performed an additional 2,026 steps/day (95% CI: 558 to 3,492, p=0.007) compared to low fit participants (i.e., <15mL/min/kg; Table 4). High fit participants had a better early (i.e., 7-days post-operative; +11 AU, 95% CI: 1 to 20, p=0.027) and later (i.e., 6-weeks post-operative; +11 AU, 95% CI: 3 to 19, p=0.007) subjective recovery from surgery. Estimated pre-operative peak VO₂ using the DASI was moderately correlated with post-operative functional recovery (r=0.51, p <0.001; Appendix Table 9). For every 1mL/min/kg increase in estimated pre-operative peak VO₂, participants performed 497 (95% CI: 250 to 744) more steps per day, were 0.3 seconds quicker on the timed up and go test (95% CI: -0.5 to -0.1) and scored 1.1 (95% CI: 0.1 to 2.1) points higher on the SRS at 6-weeks post-operative. Estimated pre-operative peak VO₂ was not associated with length of hospital stay (+0.3 days, 95% CI: -0.7 to 1.4, p=0.511).

Pre-operative daily step count was highly correlated with post-operative step count (r=0.83, p <0.001; Appendix Table 10). Low corelations were observed for pre-operative daily step count with other indices of functional and subjective recovery (r ≤0.43, p ≤0.036).

Discussion

Participants who had a low pre-operative peak VO₂ (i.e., <15mL/min/kg) on average performed approximately 50% (5) fewer sit to stand reps, approximately 60% (3,500) fewer daily steps and took approximately 2 seconds longer to perform the timed up and go test at 6-weeks post-operative; they also had poorer subjective surgical recovery at 7-days and 6-weeks post-operative. Pre-operative peak VO₂ had the highest correlation with functional and subjective recovery at 6-weeks post-operative. Pre-operative daily step count was highly correlated with post-operative step count; that is, participants who were more active pre-operatively were more likely to be active 6-weeks post-operative. Lastly, estimated pre-operative peak VO₂ using the DASI was moderately correlated with post-operative functional recovery.

These results reinforce previous work highlighting the clinical importance of pre-operative fitness in patients undergoing non-cardiac surgery and underscore the importance of pre-operative fitness as a prognostic surgical indicator for patients scheduled for hip or knee arthroplasty. Additionally, estimated pre-operative fitness (via the DASI or daily step count) may be as effective for predicting recovery from arthroplasty as the gold standard CPET.

What is the effect of pre-operative peak VO₂ on functional and subjective recovery following hip or knee arthroplasty?

In this study a low pre-operative peak VO₂ was associated with poorer functional and subjective recovery from arthroplasty at 6-weeks post-operative. Early and longer-term perceived recovery from surgery was poorer in low fit participants, and despite the surgery, the impact of osteoarthritis remained significantly worse in these participants at 6-weeks post-operative. The rate of functional and subjective recovery was similar for both groups; however, low fit participants had worse pre-operative scores, resulting in worse 6-week post-operative values. Although the timed up and go test was an exception and improved more than high fit participants, the difference between groups was still up to twice the minimal clinically important difference (i.e., ~1.2 seconds).^24,25 Of all post-operative outcomes, daily step count at 6-weeks had the highest correlation with pre-operative peak VO₂ (r=0.65, p <0.001), and pre-operative peak VO₂ was still predictive when controlling for age, sex and BMI; this correlation was slightly higher in hip arthroplasty participants compared to knee arthroplasty participants (r=0.73 vs 0.59). Pre-operative measures of peak VO₂ had a low–moderate correlation with other measures of physical function at 6-weeks post-operative (e.g., timed up and go test, 30-second sit to stand test).

As this is the first study to compare pre-operative peak VO₂ and post-arthroplasty recovery, no comparative evidence is available. However, theoretical models propose that those with higher pre-operative peak VO₂ retain a higher level of functional capacity following surgery,⁴ and therefore should be able to ambulate and return to normal activities of daily living earlier, contributing to improved functional and subjective outcomes (e.g., quality of life, satisfaction). The findings of the current study support this hypothesis, with higher fit patients having a quicker perceived surgical recovery at 7-days and 6-weeks post-operative and greater physical function and physical activity at 6-weeks post-operative.

Pre-operative peak VO₂ was not associated with length of hospital stay for either hip or knee arthroplasty participants. Previous studies have reported clinician rated functional capacity (i.e., American Society of Anaesthesiologists [ASA], particularly those III/IV) to be associated with longer hospital stay.²⁶ Pre-operative peak VO₂ is considered a more discriminate measure of functional capacity; the lack of association in the present study may be attributable to insufficient statistical power.

When compared to other indices of cardiorespiratory fitness (e.g., anaerobic threshold), peak VO₂ had higher correlations with functional and subjective recovery. When using a binary cut-off for anaerobic threshold, the relationships with post-operative recovery were similar to those for peak VO₂. This is reassuring, as maximal effort is not always possible in patients with osteoarthritis; often the test is terminated prematurely due to the lower-limb limitations. Therefore, in practice, when a representative peak VO₂ is unable to be obtained during pre-operative CPET, determining an anaerobic threshold from sub-maximal exercise appears to have a similar ability to predict how well a patient functionally recovers post-operatively. An individual with either a peak VO₂ <15mL^.kg^-1.min^-1 or anaerobic threshold <11mL^.kg^-1.min^-1 may require more attention pre-operatively to improve cardiorespiratory fitness, or greater care post-operatively, to increase the likelihood of greater physical function after hip or knee arthroplasty.

Are non-CPET pre-operative assessments associated with functional and subjective recovery following hip or knee arthroplasty?

The DASI questionnaire has utility as an alternative to pre-operative CPET for risk stratification.¹⁰ The results from the present study show it may be as effective for stratifying patients as low or high fit and predicting post-operative functional recovery following arthroplasty as CPET. However, there was still a large amount of variability explained by other factors, and similar to CPET it had only low correlations with subjective recovery (i.e., r ≤0.35). Wijeysundera et al.¹⁰ showed the predictive capacity of the DASI for post-operative complication to be as good as CPET and significantly better than a clinician’s subjective assessment of risk. Although cardiopulmonary exercise testing can provide additional insight on how patients respond physiologically to stress,⁴ an advantage of the DASI is that being questionnaire-based, it can be performed anywhere and does not require the patient to be in the presence of a clinician or other healthcare worker. This has utility in any setting where the public hospital serves a large geographical area, with many pre-operative assessments conducted via telemedicine. Furthermore, it may be useful for triaging patients for pre-operative CPET if resources are scarce, and for identifying high priority patients for prehabilitation. Future adequately powered studies should compare estimated peak VO₂ via the DASI questionnaire with CPET-measured peak VO₂ to confirm these findings and its utility for detecting surgical complications.

We reported that low fit participants had a larger improvement in estimated peak VO₂ from pre-operative to 6-week post-operative assessments than high fit participants (Table 4). A limitation of the DASI is that it has poor ability to detect change in peak VO₂ across time.¹¹ Although we did not directly measure peak VO₂ at 6-weeks post-operative, it is likely the reported improvement in estimated peak VO₂ is due to improved function from the surgery and ability to tick “yes” to more questions, rather than an improvement in fitness per se.

The strongest correlation with physical activity at 6-weeks post-operative was pre-operative physical activity. This is likely attributable in part to multicollinearity; nevertheless, a higher pre-operative step count was associated with a higher step count 6-weeks post-operatively (r=0.83, p=<0.001), and this is an important and valid measure of functionality. Consistent with previous research,^27–30 the present study highlighted that despite the surgical procedure, there was no difference between pre-operative physical activity and physical activity at 6-weeks post-operative. Essentially, those most active before surgery are also most active after surgery, and one could argue that their severe osteoarthritis diagnosis was not the limiting factor to physical activity. This relationship persists at 6-months,^27,28 9-months²⁹ and 1-year post-operative^28,30 in hip and knee arthroplasty recipients.

Despite the high correlation between pre-operative and post-operative daily step count in the present study, previous research has shown limited association with pre-operative daily step count and other functional post-arthroplasty outcomes.^31–33 In particular, there appears to be no clear relationship with time to return to work or subjective recovery,^31,32 but increased pre-operative physical activity may be associated with less post-operative pain.³³ Research in other surgery types has shown a relationship between pre-operative physical activity and outcomes following surgery. Richards et al.⁷ reported that in patients aged 65 years and older undergoing major abdominal surgery, those with a median step count <2,500 steps/day had a more than two-fold greater length of hospital stay (14 days vs 6 days, p <0.001) compared to those with a normal step count. Low step count also increased the risk of intensive care unit admission and post-operative complications.

Limitations

The findings of this study must be considered in the context of several limitations. A convenience sample was used for this study and the sample size was relatively small; not all participants recruited into the original RCT could be included, largely due to loss to follow-up related to COVID-19 (e.g., lockdowns, extended surgical waitlisting). Despite this, post hoc between-group power for the main physical and subjective variables was high, ranging from 0.86–0.99. Approximately 10% of patients were excluded from the original study due to medical contraindications to CPET. Therefore, the findings of this study may not be as relevant for high-risk patients. Six-weeks post-operative is still relatively early in the recovery period, particularly for knee arthroplasty. It remains unclear whether the differences between low fit and high fit participants persist later in recovery; nevertheless, fitness appears to be associated with superior early post-operative functional and subjective recovery. All participants were provided in-hospital rehabilitation and standardised advice to continue their rehabilitation at home. Participation in rehabilitation is influential in the recovery from surgery, and as rehabilitation compliance was not assessed in this study, this must be considered when interpreting the study findings.³⁴ CPETs were performed on two different modalities, and based on previous work this will have influenced some CPET variables.¹⁴ It is possible that the presence of comorbidities (i.e., higher rates of obesity, hypertension and chronic kidney disease) or the severity of osteoarthritis contributed to lower fitness and/or the poor post-operative recovery. However, controlling for WOMAC score and BMI in the statistical analysis will have limited confounding or bias. Moreover, the presence of good fitness can offset the deleterious effects of comorbidities,³⁵ underscoring the importance of prehabilitation to minimise the potential impact of pre-operative comorbidities. The potential for overfitting and multicollinearity in regression modelling is acknowledged, particularly the high correlation between pre-operative and post-operative daily step count.

Conclusions

These findings suggest that participants with low pre-operative fitness have poorer functional and perceived outcomes following hip or knee arthroplasty. Directly measured pre-operative peak VO₂ had similar associations with post-operative functional outcomes as pre-operative daily step count and estimated peak VO₂ via the DASI questionnaire; these have potential as a post-arthroplasty quality of recovery prediction tool when pre-operative CPET is not indicated or available. Pre-operative fitness, either measured or estimated, and daily step count are useful clinical tools for predicting functional and subjective recovery following hip or knee arthroplasty. Future work should investigate whether pre-operative optimisation of these predictors translates to improved post-operative functional and subjective recovery, and clinical outcomes.

View Appendix.