Introduction

Knee osteoarthritis (KOA) accounts for the majority of the total OA disease burden1 and is highly prevalent, affecting up to 30% of the older population2. Amidst trends of population aging and rising obesity rates, it is an important healthcare issue, having consistently ranked among the top causes of years lived with disability globally and in Asia1. The impact of KOA on the individual is extensive, ranging from chronic pain, functional impairment and social isolation, to an increase in all-cause mortality owing to its interplay with underlying comorbidities. To date, there is no approved efficacious or safe treatment option for KOA that can reverse the disease process. Knee arthroplasty (KA) is an effective surgical intervention for KOA if conservative treatment has failed3. However, around 20–30% of patients experience suboptimal outcomes after KA, including chronic pain, functional limitations and dissatisfaction4.

In recent years, there has been growing interest in understanding the impact of patient expectations on outcomes after KA. As patient-centric care and shared decision making become linchpins in today’s practice of medicine, the importance of understanding patient expectations becomes increasingly recognized5. Patient expectations towards treatment is a complex construct that encompasses beliefs or cognitions regarding health outcomes following a therapeutic intervention6. They are shaped by a plethora of factors including an individual’s personal experiences and that of those within their social circle. Across various clinical disciplines and contexts7,8, there is emerging evidence that positive expectations play a central role in influencing treatment outcomes. Expectancy theory has also posited that the difference between the actual and expected outcome, for which the fulfillment of expectations is a function of, is a dominant determinant of patient satisfaction9, an important patient-centered metric of quality of care. Multi-faceted as they may be, patient expectations are dynamic and malleable, and the patient encounter represents an important timepoint in the patient journey where clinicians may be instrumental in influencing these expectations6. Being cognizant of the interplay between patient expectations and post-operative outcomes will enable clinicians to better manage, align and close the expectation gap between patient and clinicians, empower the shared decision-making process, and improve post-operative outcomes.

In the setting of KA, several studies conducted in Western populations have found that the fulfilment of pre-operative expectations correlates with post-operative satisfaction and improvements in patient-reported outcomes, while some have demonstrated a positive association between baseline expectations and post-KA outcomes10,11,12,13,14,15. However, there is a paucity of such studies evaluating the relationship between patients’ expectations and post-operative outcomes in the Asian context. Patient expectations regarding KA have been known to vary across different socio-cultural and geographical contexts16. Moreover, phenotypic heterogeneity exists across ethnic groups that accounts for differences in the severity of KOA disease manifestation and post-operative outcomes17. Among patients with osteoarthritis, ethnicity have been shown to influence the degree of pain independent of socioeconomic status18, and specific domains of health were different across ethnic groups within the same country19. Compared with their Western counterparts, the functional demands of Asian patients are generally greater as high flexion activities such as squatting, kneeling and cross-legged sitting are cardinal components of daily living for some ethnic groups20. In addition, it is also well established that Asian patients have distinctly different anatomical and anthropometric characteristics that may influence post-operative outcomes20. In this study, we aimed to explore the relationship between patient expectations and post-KA outcomes in the domains of pain, function and patient satisfaction within a multi-ethnic Asian population in Singapore. We postulate that higher baseline expectations is associated with poorer outcomes whilst expectation fulfilment is associated with better post-KA outcomes.

Methods

Study population

Between May 2017 and May 2019, we recruited consecutive patients with severe KOA scheduled for KA (total knee arthroplasty [TKA] or uni-compartmental knee arthroplasty [UKA]) in Singapore General Hospital during their pre-operative assessment visit, 1–2 weeks before KA. We excluded patients who had cognitive impairment, and those who were listed for revision KA or KA indicated for a diagnosis other than KOA. Participants completed a paper copy of a pre-operative questionnaire during their pre-operative assessment visit. Follow-up questionnaires at 6- and 12-months after surgery were completed over the telephone with a trained interviewer. The study protocol was read and approved by the SingHealth Centralized Institutional Review Board (CIRB Ref: 2016/3168) and conforms to the provisions of the Declaration of Helsinki. Informed written consent was obtained from all participants.

Data collection

We collected socio-demographic information including age, sex, ethnicity and highest education level. Body height and weight were measured at recruitment. Surgical information collected include the type of KA (TKA or UKA) and laterality (unilateral versus bilateral).

All patient-reported outcomes at recruitment were self-completed on paper by patients in either English or Chinese language as per their preference. We assessed patients’ baseline KOA symptoms and functional impairment using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), with reference to the index knee which was designated as the side planned for KA or the dominant side in cases of bilateral KA. The WOMAC is a disease specific, tri-dimensional PRO that comprises a battery of 24 questions designed to assess the 3 core dimensions of pain (5 questions), function (17 questions) and stiffness (2 questions)21. Each question is answered on a 5-point Likert scale from 0 to 4 (0: none; 4: extreme) and aggregated to a WOMAC pain score (0–20), function score (0–68), stiffness score (0–8) and total WOMAC score (0–96), with a higher score indicating more severe symptoms. The WOMAC was developed to detect clinically important changes in overall health status following therapeutic interventions, and has been validated for Asian patients with KOA in Singapore22. To accurately reflect a meaningful change of scores with arithmetic property, we converted the ordinal WOMAC scores to interval scales using a transformation table developed through Rasch model analysis (Supplementary table 1)23. The score range of 0–20 for pain, 0–68 for function, 0–8 for stiffness and 0–96 for overall WOMAC remains unchanged in interval scale, with a higher score reflecting greater impairment.

We assessed the number of co-morbidities with the Functional Comorbidity Index (FCI)24. The FCI includes questions on the presence or absence of 17 comorbid conditions in addition to arthritis and is summated to a total score (0–18). We evaluated anxiety and depression using the Hospital Anxiety and Depression Scale (HADS)25, and social support through the Lubben Social Network Scale (LSNS)26.

Routine weight-bearing anterior–posterior radiographs of the index knee were taken as part of standard care. The radiographic severity of KOA in the medial and lateral tibio-femoral compartments of the index knee was assessed using the Kellgren and Lawrence (KL) classification system27 by a designated radiologist (SBW) who was blinded to the patients’ clinical profile. A random sample of 70 radiographs were read twice 2–4 weeks apart and the intra-class correlation coefficient (ICC) of the KL grading was 0.75 (95% confidence interval, CI: 0.60–0.84).

Pre-operative expectations

We assessed patients’ pre-operative expectations using the Hospital for Special Surgery Knee Replacement Expectations Survey (HSS-KRES)28. The HSS-KRES is a validated 19-item patient-derived instrument that evaluates patients’ expectations in relation to pain relief, baseline activities, high flexion activities, social activities and psychological well-being. Patients indicated the degree of improvement they expected for each of the domains on a 5-point Likert scale from 0 to 4 (0: I do not have this expectation, or this expectation does not apply to me; 1: a little improvement; 2: a moderate amount of improvement; 3: a lot of improvement; 4: back to normal or complete improvement). Their responses to the 19 items were subsequently summed up and transformed to a 0–100-point scale28,29, with a higher score reflecting higher overall expectations.

Fulfilment of expectations

We collected post-operative data at 6-months when clinically important improvements and maximal achievable benefits following KA are expected to be achieved30, and longer term outcomes at 12-months. Patients were contacted and asked to complete questionnaires with a trained interviewer who was blinded to the initial interview over the telephone at 6- and 12-months post-KA. They were asked to rate the extent to which each expectation previously cited on the HSS-KRES was fulfilled, on a scale of 1 to 3 (1 = complete satisfaction, 2 = partial satisfaction, 3 = dissatisfaction)31. A previously cited expectation was defined as a domain on the HSS-KRES for which patients expected at least a little improvement. Domains that patients reported as “0: I do not have this expectation, or this expectation does not apply to me” at baseline were not collected at 6- or 12-months. An expectation was defined as completely fulfilled if the participant gave it a “complete satisfaction” rating post-operatively, in relation to the expectation expressed by the participant pre-operatively. We tabulated the fulfilment of expectations by taking the number of expectations fulfilled completely (at 6 months or 12 months) divided by the total number of expectations cited pre-operatively, and converting to a 0–100 point scale31. A higher score represents greater fulfilment of expectations.

Pain, function, and satisfaction after KA

At 6- and 12-months post-operatively, the WOMAC and Self-Administered Patient Satisfaction Scale (SAPS) were collected over the telephone.

We derived the change in WOMAC pain and WOMAC function by subtracting the pre-operative scores from the follow-up scores. All WOMAC scores were converted to interval scales prior to calculation of changes23. The lower or more negative the value of the change in WOMAC pain or function, the greater the improvement in pain or function respectively.

The SAPS is a 4-item questionnaire that measures patients’ overall satisfaction with the outcome of KA32. Patients rated their satisfaction with KA for overall outcome, pain relief, ability to perform housework and ability to take part in recreational activities on a 4-point Likert scale (100: very satisfied, 75: somewhat satisfied, 50: somewhat dissatisfied, 25: very dissatisfied). Total scores were calculated by taking the unweighted average of the scores from the individual questions, with a higher score reflecting higher satisfaction.

Statistical analysis

Continuous variables were presented as means and standard deviations unless otherwise stated, while categorical variables were presented as frequencies and percentages. Univariable and multivariable analyses were conducted using the generalized linear model. Univariable analyses were first performed to examine the relationship between the study variables (pre-operative expectations and post-KA fulfilment of expectations) and each of the outcome measures (change in WOMAC pain, change in WOMAC function and overall satisfaction) at 6- and 12-months. We also evaluated variables that have been found to be plausible predictors of post-KA outcomes including age, sex, ethnicity, type of KA, pre-operative pain and function, body mass index (BMI), comorbidities, KL grade, anxiety, depression and social support17,33,34,35,36,37. Study variables with p < 0.1 association with the KA outcomes in the univariable analysis were entered into the corresponding multivariable model.

To evaluate for clinical relevance, we conducted sensitivity analyses to assess if the variables in test were statistically significantly associated with the achievement of minimal clinically important difference (MCID) for WOMAC pain or function respectively at 6- and 12-months. The MCID represents the smallest magnitude of improvement in WOMAC pain or function that is clinically important and meaningful from the patient’s perspective38. As there is no consensus cut-off for MCID for WOMAC in knee arthroplasty38, we adopted an MCID of 20.5 and 24.2 for change in WOMAC pain and function respectively which were derived from a large multi-center cohort study using a patient-derived anchor and Receiver Operating Characteristics curve approach39.

A subgroup analysis limited to patients who underwent UKA was conducted to evaluate for consistency in the results. Furthermore, to evaluate the effect of ethnicity, we conducted a subgroup analysis limiting to the non-Chinese patients who together represent all the minority ethnic groups, as the number of patients in each of the minority ethnic groups was small.

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 25 (IBM Corp., Armonk, N.Y., USA) and a significance level of 0.05 was set as the threshold for statistical significance.

Participant characteristics

We approached 1698 patients, of which 1539 were eligible for the study. 1201 patients consented to the study, reflecting a response rate of 78%. The mean (standard deviation, SD) age of the 338 patients who declined participation was 67.0 (6.9) years and 261 (77.2%) were female, which were not dissimilar from that of the included participants. Of the 1201 patients who were recruited for the study, 1136 underwent KA. Amongst those who had undergone KA, 1103 and 1089 patients completed the follow-up at 6- and 12-months post-KA respectively (Fig. 1). The baseline characteristics of the study cohort are presented in Table 1. Of the study population, 794 (69.9%) were female and 954 (84.0%) were Chinese. The mean (SD) age was 65.9 (7.0) years, and the mean BMI was 28.3 (4.7) kg/m2. More than half (58.2%) of the patients had severe KOA on radiography (KL grade 4). The majority (92%) of patients underwent TKA while 8% underwent UKA. Additionally, 94.6% of the patients underwent unilateral KA, while the rest underwent bilateral KA. The baseline demographics of our study cohort were similar to other KA cohorts from Asia40.

Fig. 1
figure 1

Flow diagram detailing recruitment process and follow-up.

Full size image
Table 1 Baseline characteristics of patients who underwent knee arthroplasty (n = 1136).
Full size table

Results

Description of measures

The baseline patient expectations, degree of expectation fulfilment post-KA and outcomes at 6- and 12-months post-KA are described in Table 2. The mean HSS-KRES score was 73.4, whilst the mean post-KA expectation fulfilment at 6- and 12-months were 64.1 and 71.6 respectively. The largest change in WOMAC pain and WOMAC function can be observed during the first 6 months, with the mean change in WOMAC pain and WOMAC function (in ordinal scale) being -5.4 and -14.1 respectively. The overall satisfaction (SAPS) score at 6- and 12-months post-KA were 89.7 and 92.5 respectively.

Table 2 Baseline patient expectations, expectation fulfilment and outcomes at 6 months and 12 months post knee arthroplasty.
Full size table

Associations with change in WOMAC pain and function

In the uni-variable analyses, HSS-KRES was statistically significantly associated with change in WOMAC pain or function at 6 months. However, statistical significance was lost in the multi-variable analyses. Of note, the association between HSS-KRES and change in WOMAC function at 6 months was trending towards significance in the multi-variable model [ß: − 0.03, 95% CI − 0.07 to 0.00, p = 0.069]. In contrast, post-KA expectation fulfilment was statistically significantly associated with the change in WOMAC pain [ß: − 0.06, 95% CI − 0.07 to − 0.06, p < 0.001] and function [ß: − 0.32, 95% CI − 0.34 to − 0.31, p < 0.001] at 6 months (Tables 3 and 4). This suggests that higher post-KA expectation fulfilment was associated with greater improvements in both WOMAC pain and function. Similar results were found in the analyses of the change in WOMAC pain and function at 12 months post-operative (Supplementary tables 2 and 3). We tested and refuted significant multicollinearity between HSS-KRES and post-KA expectation fulfilment at 6 months (r = 0.063, Variance Inflation Factor 1.004).

Table 3 Factors associated with change in WOMAC pain at 6 months (n = 1060).
Full size table
Table 4 Factors associated with change in WOMAC function at 6 months (n = 1060).
Full size table

In the sensitivity analyses, post-KA expectation fulfilment was also found to be statistically significantly associated with the achievement of MCID for WOMAC pain [OR: 1.05, 95% CI 1.04 to 1.06, p < 0.001 (6 months)] and function [OR: 1.08, 95% CI 1.07 to 1.09, p < 0.001 (6 months)] at both 6- and 12-months post-KA (Supplementary tables 5 to 8), supporting that post-KA expectation fulfilment is associated with a clinically relevant change in WOMAC pain and function after arthroplasty.

Consistent results were found in the subgroup analyses for a smaller sample of 83 patients who underwent UKA (Supplementary tables 9 and 10) and 169 non-Chinese patients (Supplementary tables 11 and 12).

Associations with satisfaction

There was no association between pre-operative HSS-KRES and SAPS at 6 months post-operative in the univariable analysis (Table 5). In contrast, post-KA expectation fulfilment was positively associated with SAPS at 6 months in the multi-variable model [ß: 0.38, 95% CI: 0.35 to 0.40, p < 0.001], suggesting that greater post-KA expectation fulfilment was associated with higher overall satisfaction at 6 months post-KA. Similar results were found in the analysis of overall satisfaction at 12 months post-operative (Supplementary table 4).

Table 5 Factors associated with Self-Administered Patient Satisfaction Scale (SAPS) at 6 months (n = 1060).
Full size table

Discussion

In this study, we found that baseline patient expectations were not associated with improvements in WOMAC pain, function and overall satisfaction at both 6- and 12-months post-KA after adjustments in the multi-variable analyses. However, higher post-KA expectation fulfilment was associated with greater improvements in WOMAC pain, function and overall satisfaction at both 6- and 12-months post-KA.

In the current study, the pre-operative expectations of our patients were comparable to that reported in previous studies using the same tool12,14. The most commonly cited expectations in this study population were pertaining to pain relief and return of baseline activity. Notably, we found that higher expectations were associated with greater improvements in both pain and function post-KA in the univariable analysis. This observation is congruent with several other studies13,14,15. Although the association of pre-operative expectation and post-operative WOMAC outcomes were lost in the multi-variable analyses, this should not undermine the importance of individualizing discussion for specific expectation before KA. Across a spectrum of clinical contexts7, there is a growing body of evidence demonstrating the positive influence of patient expectations on treatment outcomes. One of the postulated mechanisms explaining this phenomenon is the “placebo” effect that patient expectations might have on treatment outcomes. A randomized controlled trial involving patients with degenerative tear of the medial meniscus demonstrated the equivalence of all patient outcomes between placebo surgery and arthroscopic meniscectomy, suggesting the power of the placebo effect from patients’ positive expectations to improve regardless of the treatment per se 8. Another possibility for this association is that patients with greater expectations may have a higher level of motivation and self-efficacy, and may therefore be more proactive in their rehabilitation post-KA41, which can contribute to optimizing their post-operative pain and function.

Another important finding in the current study is the association of expectation fulfilment with improvements in WOMAC pain and function, as well as overall satisfaction at both 6- and 12-months post-KA. The association between expectation fulfilment and improvements in pain and function post-KA has been shown from other studies10. Similarly, expectation fulfilment has stood out consistently in previous studies as a key predictor of overall satisfaction post-KA11,12,13. Patient satisfaction is a complex and multi-faceted concept6 that has recently been acknowledged as a key measure of the quality of patient-centric care. It reflects patients’ affective reaction to, and cognitive evaluation of their healthcare experience and has been widely accepted as a function of the degree of congruence or divergence between the expected and actual outcomes of care9. Several studies did not show any correlation between patient expectations and overall satisfaction post-KA11,13,15. While higher patient expectations may have a positive influence on post-operative pain and function as described earlier, the lack of an association with overall satisfaction suggests that improvements in patient-reported outcomes may not necessarily translate to expectations being met. Pre-operative expectations could change post-operatively, and may potentially explain why expectation fulfilment is associated with post-operative outcomes rather than baseline expectations42.

The findings of our study that expectation fulfilment play a more important role than baseline expectations in determining post-operative outcomes were largely concordant with studies conducted in Western socio-cultural contexts. This could possibly be explained by the observations that in both Asian and Western socio-cultural settings, the domains of health-related quality of life (HRQoL) that are regarded as important are mostly similar19,43. Importantly, a qualitative study conducted in Singapore has also previously demonstrated that the main domains of pain, physical disability, other symptoms of OA and mental health were important across all three major ethnic groups (Chinese, Malay and Indian)19, which could have accounted for the similar results seen in the subgroup analyses that included non-Chinese patients only.

The findings from our study also provide insights into how clinical practice may be adapted to drive better KA outcomes. Given the importance of expectation fulfilment, it is essential for clinicians to understand patients’ expectations for KA pre-operatively and engage them in a discussion on whether these expectations may be realistically fulfilled from KA. It has been shown that discrepancies in expectations commonly exist between patients and clinicians44. This discrepancy is often driven by patients’ expectations of higher level activities44. Therefore, clinicians should counsel patients about the expected outcomes post-KA11. Priorities, values and the acceptance threshold differ from patient to patient. The importance of engaging patients in such a conversation and helping them to re-align their expectations at a realistic level cannot be undermined. In a randomized controlled trial, additional education on realistic outcome expectations pre-operatively has been demonstrated to have a modifying effect on patients’ pre-operative expectations, and resulted in higher post-operative expectation fulfilment and satisfaction after KA compared to standard pre-operative counselling45. Conversely, the lack of pre-operative education was a reason cited for unmet expectations and poorer outcomes amongst patients who underwent hip replacement46. While it is paramount to set realistic expectations amongst those undergoing KA, the link between positive expectations and improvements in pain and function also suggests that it is equally important to identify and build up positive expectations in those with low or even negative expectations for KA, who may be less motivated during the rehabilitation phase41 and hence may not be able to derive the fullest benefits of KA.

The strength of our study is the prospective study design, large sample size and high follow up rates up to the 12-month time point. To the best of our knowledge, this is the first study examining the relationship between patient expectations and post-KA outcomes in the Asian context. Comprising of a unique ethnic profile of Chinese, Malays, Indians and others, our multi-ethnic cohort in Singapore captures the diversity within Asia. Singapore is a small city state in Asia where Chinese, Malays, Indians and others have lived together for generations since the 1800s47. It is a place where the society is relatively homogeneous in education system, job opportunities and access to health care, making it ideal to study health expectations and health outcomes with minimized confounding effects from other disparities. Another strength of this study is that patient expectations was assessed using a validated and reliable instrument, the HSS-KRES. Despite the complexities of patient expectations as a concept, the HSS-KRES is a patient-derived questionnaire specifically developed for the setting of KA, encompassing a comprehensive spectrum of expectations that are important and meaningful to patients28. We assessed improvements in pain and function post-KA using the change in WOMAC pain and function measured in interval scale level. The WOMAC is a validated tool that is sensitive in detecting clinically important changes in pain and function following therapeutic interventions48. In our study, we further applied conversion algorithms derived from Rasch model analysis to transform the ordinal WOMAC scores into interval-level scores23. This WOMAC calibrated in interval scale allowed us to reflect a meaningful magnitude of change in WOMAC scores with arithmetic property and reduce the measurement bias associated with ordinal scale measurement. In addition to assessing post-KA outcomes at 6 months when clinically important improvements are expected, we have shown consistency in the results at a longer follow-up time point (12-month). Sensitivity analyses using achievement of MCID for WOMAC pain and function provided support for the clinical relevance of the results. Finally, we also adjusted for an extensive set of variables that may affect post-KA outcomes, including baseline pain and functional status, social support and psychological well-being which were not included in many of the previous studies49. However, we acknowledge some limitations of our study. As the response rate of the current study is 78%, we were unable to fully account for the characteristics of the 22% who were non-responders, albeit similarities between responders and non-responders in age and sex distribution. There were also patients that we excluded from participating in the study due to low literacy (n = 137), hearing impairment (n = 16) and cognitive impairment (n = 6). Notably however, we had high follow-up rates of 97% and 96% at 6- and 12-months, thereby minimizing the effect of selection bias for KA outcomes. In addition, there has been a lack of a validated method for the assessment of expectation fulfilment to date. We closely adapted the method described by Mancuso et al. in assessing expectation fulfilment for patients undergoing KA31, which has been utilized in other studies50. However, challenges remain in drawing comparisons with studies that used other methods for expectation fulfilment. In the assessment of functional outcomes, we did not collect the post-operative knee range of motion nor utilize performance based scales which could provide a more objective assessment and may have allowed for a more granular analysis. Caution is required for the interpretation of the subgroup analysis limiting to patients who underwent UKA, where the sample size was small, although the results were consistent as those found in the main cohort. Moreover, we did not collect information regarding surgical factors (number of surgeons operating and surgical technique), presence of post-operative complications and participation in post-operative rehabilitation program, which could potentially influence post-KA outcomes and satisfaction.

In conclusion, expectation fulfilment was associated with clinically significant improvements in WOMAC pain, function and satisfaction at both 6- and 12-months post-KA. In addition, we found that it was the fulfilment of expectations, rather than expectations at baseline, that mattered more as a determinant of KA outcomes in an Asian population. Our study underscores the importance of realigning patients’ pre-operative expectations at a level that is realistic and achievable to drive better KA outcomes.