Article Text

Extended report
Psychosocial determinants of outcomes in knee replacement
  1. Maria A Lopez-Olivo1,
  2. Glenn C Landon2,
  3. Sherwin J Siff2,
  4. David Edelstein2,
  5. Chong Pak1,
  6. Michael A Kallen1,
  7. Melinda Stanley3,4,5,
  8. Hong Zhang1,
  9. Kausha C Robinson1,
  10. Maria E Suarez-Almazor1
  1. 1The University of Texas, MD Anderson Cancer Center, Houston, Texas, USA
  2. 2St Luke's Episcopal Health System, Houston, Texas, USA
  3. 3Houston VA HSR&D Center of Excellence, Health Services Research and Development Service, Department of Veteran Affairs Medical Center, Houston, Texas, USA
  4. 4Baylor College of Medicine, Houston, Texas, USA
  5. 5VA South Central Mental Illness Research, Education and Clinical Center, North Little Rock, Arkansas, USA
  1. Correspondence to Dr Maria E Suarez-Almazor, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 1467, Houston, TX 77030, USA; msalmazor{at}


Objective To identify potential psychosocial and educational barriers to clinical success following knee replacement.

Patients and Methods The authors evaluated 241 patients undergoing total knee replacement, preoperatively and 6 months after surgery. Outcomes included the Western Ontario McMaster (WOMAC) scale and the Knee Society rating system (KSRS). Independent variables included: the medical outcome study–social support scale; depression, anxiety and stress scale; brief COPE inventory; health locus of control; arthritis self-efficacy scale and the life orientation test–revised. Multiple regression models evaluated associations of baseline demographic and psychosocial variables with outcomes at 6 months, controlling for body mass index, comorbidities and baseline outcome scores.

Results Patients' mean age was 65±9 years; 65% were women. Most patients improved outcomes after surgery. Several psychosocial variables were associated with outcomes. Regression analyses indicated lower education, less tangible support, depression, less problem-solving coping, more dysfunctional coping, lower internal locus of control were associated with worse WOMAC scores (R2 contribution of psychosocial variables for pain 0.07; for function, 0.14). Older age, lower education, depression and less problem-solving coping were associated with poorer total KSRS scores (R2 contribution of psychosocial variables to total KSRS model 0.09). Psychosocial variables as a set contributed from 25% to 74% of total explained variance across the models tested.

Conclusion Patients' level of education, tangible support, depression, problem-solving coping, dysfunctional coping and internal locus of control were associated with pain and functional outcomes after knee replacement. The findings suggest that, in addition to medical management, perioperative psychosocial evaluation and intervention are crucial in enhancing knee replacement outcomes.

Statistics from

Osteoarthritis is among the most prevalent musculoskeletal disorder,1 2 and is a major cause of disability in older Americans, second only to cardiovascular disease. The knee is the most common joint associated with disability in osteoarthritis, and the prevalence of knee osteoarthritis increases with age.3 Knee replacement is an effective intervention for patients with knee osteoarthritis who have not responded to medical and physical therapy. Over 500 000 knee replacements are performed annually in the USA at an average cost of US$25 000 each; approximately 80% of these replacements are performed on patients with osteoarthritis.4,,7

Knee replacement is effective in most patients, with successful reduction of pain and improvement of function.8,,16 However, approximately 25% of individuals who undergo replacement have little or no improvement after surgery, or are unsatisfied with their outcome.12 Although lack of success in knee replacement can sometimes be attributed to surgical complications, in many cases no technical or medical reasons for the lack of success can be identified. Some studies have shown outcome differences among types of prostheses but, in general, these differences are small.17 Over the past decade, a convergence of improved design and surgical techniques has led to less variability in prosthesis-related outcomes, and long-term ‘technical failures’ requiring revision of the prosthesis, such as loosening, fracture or infection, are low (less than 10% over 10 years).18 19

Rehabilitation after knee replacement is far more dependent on the will and cooperation of the patient than for other joint procedures.20 Therefore, individual psychosocial attributes may contribute to functional outcomes.21 The objective of this study was to identify potential psychosocial and educational barriers to surgical success following knee replacement. Previous studies22,,35 have reported an association between specific psychosocial variables and knee replacement outcomes; however, the present study is the first to evaluate the independent contribution of several important psychosocial attributes, including education, social support, depression, anxiety, stress, coping strategies, locus of control, self-efficacy and optimism.

Patients and methods


We conducted a prospective cohort study to identify potentially modifiable psychosocial determinants of outcomes after knee replacement.

Research setting

Four orthopaedic surgeons from two outpatient clinics affiliated with St Luke's Episcopal Hospital in Houston, Texas, USA, participated in the study. Patients scheduled for knee replacement between May 2004 and October 2005 were contacted by telephone on behalf of their surgeons, and if they were eligible and agreed to participate were mailed a consent form and baseline preoperative questionnaires. All patients undergoing knee replacement at St Luke's Hospital are asked by their surgeons to attend a preoperative group educational session informing them about the procedure and what to expect postoperatively. We attended these sessions to enrol additional patients who might have been missed and to perform additional baseline evaluations on participating patients.

All surgeries were performed at the same location (St Luke's Hospital), where all immediate postoperative rehabilitation care was also provided. After discharge, most patients received physical therapy at an outside facility. All patients in the study had health insurance (private or Medicare).


Inclusion criteria were: (1) radiological diagnosis of knee osteoarthritis; (2) first knee replacement (previous hip replacement was allowed); (3) adequate cognitive status; (4) living in the community (not in long-term care facilities) and (5) ability to communicate in English. Exclusion criteria were: (1) revision surgery; (2) inflammatory arthropathies (eg, rheumatoid arthritis); (3) neurological disorders; (4) Paget' s syndrome or bone disorders; (5) litigation process related to surgery and (6) patients seeking or receiving workers' compensation benefits.


Patients were assessed at baseline (within the month before they were scheduled for surgery) and 6 months after surgery. These assessments occurred at face-to-face visits, with trained personnel performing a physical examination of the knee, and through the completion of self-report questionnaires, including demographics, comorbid conditions and the following measures:

Outcome measures

  1. Western Ontario McMaster (WOMAC) scale,36 a widely used self-response questionnaire measuring pain, function and stiffness in the knee during the past 48 h. Subscale scores range from 0 to 100, with higher scores indicating worse outcomes. We established a priori that we would only use the pain and function subscales, because these have shown the best reliability and validity.37

  2. Knee Society rating system (KSRS).38 This instrument, widely used in orthopaedic research, includes two major scales. The first scale, the knee score, includes three components: pain, stability and range of motion (ROM) measured with a goniometer. The maximum possible scores are 50 for pain, 25 for stability and 25 for ROM; component scores are summed to obtain the KSRS knee score, which ranges from 0 to 100. Higher KSRS scores indicate better outcomes (ie, a score of 100 represents a well-aligned knee with no pain, negligible anteroposterior or mediolateral instability and 125 degrees of motion). The other KSRS score measures function, specifically walking and stair climbing, and ranges from 0 to 100, with higher scores representing better function. A total KSRS score (or clinical rating score) can also be calculated by summing the KSRS knee and function scores (range 0–200). Examination of the knee was performed by study research assistants. To ensure reliability and consistency in the measurement of stability and ROM, research assistants received extensive training from the St Luke's physical therapists, who routinely participate in the care of patients undergoing knee replacement. Although the majority of patients underwent a physical examination, some were not available for this baseline face-to-face evaluation; for these, knee stability and ROM were not ascertained and, therefore, KSRS knee and total scores were not available.

Psychosocial attributes

  1. Medical outcome study–social support scale (MOS–SSS),39 a 19-item scale that asks patients how often different types of support are available to them. Four subscales are derived: tangible support; affectionate; positive social interaction and emotional or informational support, each scored from 1 to 5. The total score is a weighted average of all items, rescaled to range from 0 to 100, with higher scores indicating greater social support.

  2. Depression, anxiety and stress measured using the depression, anxiety and stress scale 21 (DASS21).40 This is the validated reduced-item version of the full 42-item DASS and includes 21 items to measure three negative emotional symptoms (seven items for each emotional state). Scores range from 0 to 42, with higher scores indicating worse emotional states.

  3. Coping responses to stressors using the brief COPE inventory,41 which examines people's responses to stressors, such as chronic illness and natural disasters. It has 14 two-item scales and provides three summary scores: emotional coping; problem-solving coping and dysfunctional coping, all ranging from 1 to 4. Higher scores indicate more coping behaviour per scale of coping style.

  4. Multidimensional health locus of control42 measuring beliefs about what determines one's health, with 18 items and three subscales: internal (I am in control of my health); chance (my health is related to chance events) and powerful others (others have control over my health). Each subscale's scores range from 6 to 36, with higher scores indicating stronger beliefs about that particular determinant of one's health.

  5. Arthritis self-efficacy scale (ASES)43 measuring belief in one's own capability to perform tasks or cope with adversity. It is an eight-item scale, with scores ranging from 0 to 10. Higher scores indicate more self-efficacy.

  6. The life orientation test–revised (LOT–R)44 measures optimism. The summary score was rescaled to a possible range of 0–100, with higher values indicating more optimism.

Statistical analysis

Initial bivariate analysis compared presurgery and postsurgery scores using t tests and non-parametric Wilcoxon tests. Pearson and Spearman rank correlation coefficients were used to evaluate associations between baseline predictors and outcomes at 6 months, as appropriate. Baseline predictors were selected for multiple linear regression modelling if the correlation coefficients were significant at the p≤0.30 level. Final multiple regression models were selected using a stepwise method (variable significance entry criterion of p<0.20 and removal criterion of p>0.10). Independent variables included demographic variables and baseline psychosocial domain subscales. Results were adjusted for body mass index (BMI), comorbidities and the baseline score of a model's dependent variable (WOMAC or KSRS subscales) as appropriate for the model. Sensitivity analysis was also performed by running models without baseline outcome scores; because no substantial model differences were observed, these supporting analyses are not reported. Statistical significance was set at p≤0.05 (two-sided). All analyses were performed using SAS version 9.1.3.

We estimated that a sample size of 240 would achieve 97% power to detect, in a multiple regression analysis using an F test with α=0.05, a cumulative R2 of 0.10, with 10 independent variables, adjusting for two additional control variables jointly contributing an R2 of 0.05 to the overall regression model. This sample size of 240 would achieve 80% power to detect a cumulative R2 as small as 0.067 under the same conditions.

The study was approved by St Luke's Episcopal Hospital's and the University of Texas, MD Anderson Cancer Center's institutional review boards. All participants signed a consent form.


Of the 615 patients scheduled for surgery, 253 were ineligible. Of the remaining 362, 90 (25%) declined to participate. Of the 272 who joined the study, 241 (89%) completed the 6-month follow-up assessment and underwent total knee replacement.

Patient baseline characteristics are presented in table 1. Table 2 shows the internal consistency reliability (ie, Cronbach's alpha) of self-report measures at baseline and mean scores at baseline and 6 months. Scale reliability was greater than 0.60 for most subscales, except for KSRS function. Patients showed significant improvement in knee outcomes at 6 months. Social support, self-efficacy and levels of stress, anxiety and depression did improved at 6 months.

Table 1

Patient characteristics

Table 2

Outcome measures and psychosocial domain scores at baseline and 6 months

Bivariate correlations between psychosocial attributes at baseline and study outcomes at 6 months are presented in table 3. Most predictors included in the study showed statistically significant associations with one or more of the outcomes, with correlations ranging from 0.2 to 0.3; these included education, tangible social support, depression, anxiety, stress, problem-solving coping and self-efficacy. Statistically significant correlations were also observed between gender, education, BMI, comorbidities and most outcomes. The correlations among psychosocial variables are presented in supplementary table 1, and the variability in WOMAC and KSRS scales is displayed in supplementary figures 1 and 2 (all available online only).

Table 3

Correlations between baseline predictors and outcomes at 6 months

Table 4 shows the results of stepwise multiple linear regression analyses to determine the independent influence of baseline patient demographic characteristics and psychosocial domains on outcomes at 6 months, adjusting for BMI, comorbidities and baseline outcomes scores. Greater pain, as assessed by the WOMAC, was associated with lower education (less than high school), less problem-solving coping, more dysfunctional coping and lower internal locus of control over health (R2=0.22). Worse WOMAC function scores were associated with less tangible support, depression and decreased problem-solving coping (R2=0.19). Older age, lower education, depression and less problem-solving coping were significantly associated with worse total KSRS scores (R2=0.36). A worse knee score (pain, ROM and stability) was predicted by less problem-solving coping (R2=0.13). Decreased KSRS function was predicted by older age, being female, ethnicity (non-white), depression and less problem-solving coping (R2=0.43). Across these models, approximately 7–14% of the variation in the response variables can be explained by psychosocial variables; 5–31% can be explained by patient characteristics. The remaining 57–87% can be explained by unknown variables or inherent variability. As a result of the poor psychometric characteristics of the KSRS function score (the least reliable measure), the results reflected in our study involving KSRS function should be taken with caution. Use of the total knee rating system with and without the function component has been suggested as a solution addressing the problem that arises when deterioration of a patient's overall health or comorbidities influence their functional status, while at the same time the state of the knee after total knee replacement continues in good condition. However, our findings suggest that the Knee Society function score needs improvement and cannot be used as a reference standard measure without upgrading its psychometric qualities.

Table 4

Stepwise multiple linear regression models


Knee replacement is an effective procedure in patients with knee osteoarthritis who have not responded to conservative therapy. Nevertheless, many patients undergoing knee replacement will experience medically unexplained persistent pain and dysfunction.12 In this study, we evaluated the relationships between baseline patient psychosocial characteristics and outcomes 6 months after undergoing knee replacement. Several psychosocial constructs were evaluated; among them, lower educational level, less tangible support, depression, decreased problem-solving coping, increased dysfunctional coping and lower internal locus of control negatively influenced postoperative outcomes.

Poor functional mental status, as evaluated by the short form 36, has been associated with worse outcomes in knee replacement.22 Several other studies have investigated the relationship between knee replacement outcomes and more specific psychosocial domains, but their results have often been inconclusive.22,,35 Most of those studies have examined specific attributes without adjusting for other important psychosocial covariates, and many have had modest sample sizes or have not used standard measures for pain and functional assessment, such as the WOMAC or the KSRS.

To our knowledge, this is the first study to evaluate the independent role of multiple psychosocial domains comprehensively on recovery after knee replacement in a large cohort of patients, using standard outcome measures such as the WOMAC and the KSRS, which incorporates physical measures such as ROM and knee stability. Furthermore, in our modelling we controlled for potentially important confounders, such as comorbidities and BMI, and adjusted for baseline outcome scores, as lower preoperative health status has been associated with higher pain scores and lower function after surgery.31 45 As a result, we were able to identify several psychosocial factors as being independent determinants of outcomes:

  1. Depression. The deleterious effects of depression were observed across most outcomes. While the mechanisms inherent in this association are unclear, patients feeling depressed may be less inclined to participate actively in rehabilitation, leading to a less than optimum result. Previous reports have suggested that depression might contribute to poor knee replacement outcomes, defined either in terms of pain, function, or quality of life indices,23 24 26 29 although these results have not always been consistent.23 46 A history of depression has also been associated with failure to achieve adequate knee flexion.27

  2. Coping style was a determinant of most outcomes. Coping strategies are the behavioural mechanisms used by individuals to deal with adversity. Individuals who have dysfunctional coping may be less resilient in adapting to the challenges of surgery than those whose problem-solving coping is more effective. Problem-solving coping helps an individual identify goals and stay engaged in trying to attain them, and provides motivation in the midst of adversity.33 Two other studies have reported that pain coping strategies are important predictors of pain behaviour and functional impairment in patients with osteoarthritis.47 48 There are also data indirectly supporting the importance of the coping domain. First, psychological adaptation to a change in health status (response shift) has been identified as an essential component of outcome perception in knee replacement.45 Second, a sense of purpose in life, which can help in coping, has been associated with recovery.33 Finally, some constructs that relate to poor coping, such as pessimism, pain catastrophising and fear of movement, have been associated with deleterious outcomes.32 35 In our study, the largest effect of coping skills was observed in functional improvement, with a difference in WOMAC scores of 18 on a scale of 0–100, which is a substantial and clinically significant effect, between an individual with the worst COPE problem-solving score and an individual with the best possible score. Furthermore, problem-solving and dysfunctional coping were both independently associated with the KSRS knee scores, a measure including ROM and stability.

  3. Social support, specifically tangible, was associated with WOMAC function. Tangible support is the perceived adequacy of assistance from others. Affectionate support is determined by the expression of love and affection from important others. Social support has been found to be related to the earlier achievement of rehabilitation tasks after a knee replacement.28 The impact of social support in joint replacement has mostly been studied with respect to spousal contributions to the strengthening of efficacy beliefs to manage recovery.29 Spousal pressure and persuasion 1 month after surgery were indirectly associated with patient recovery through positive emotional responses to the spouse.34 While these aspects of spousal interaction may not represent tangible support, it is conceivable that spouses are the major providers of support in this context.

  4. Health locus of control examines the degree to which individuals believe their health is controlled by internal or external factors, and could be related to self-efficacy.28 42 Individuals who are more independent and effective in their symptom management, relying less on external help, may be more willing to attain rehabilitation goals. Kendell et al28 found that higher scores on the internal locus of control subscale were associated with the faster achievement of a straight leg raise, which is an early milestone in the postoperative recovery from knee replacement. Of interest, in our study, this psychosocial domain did not improve after surgery, suggesting that it is a persistent individual trait.

In general, most of our study's measured psychosocial domains were associated with outcomes in bivariate analyses; however, not all were independent contributors for outcome measures when included in multivariable models. This is probably because of there being some degree of interrelatedness among our study's set of measured psychosocial variables, and because our study had, as a point of emphasis, the aim of identifying individual psychosocial variables impacting outcomes, rather than psychosocial variables as a class of variables. For example, anxiety and stress were initially significantly associated with function and other outcomes on a bivariate basis but were not independently significant in the multivariate models. Note that anxiety has been associated with worse pain in the first year after surgery but has not been identified as a predictor of other outcomes.23

There are several limitations to our study. Participation was voluntary, and although it was high (75%), those who did not participate might differ from those who did. The study was carried out in two orthopaedic practices associated with St Luke's Episcopal Healthcare Hospital, a private institution in Houston, therefore, the results may only be generalisable to patients in this area who have medical insurance. However, a strength of the study in this specific setting is that patients attend an educational session before their surgery, and are therefore provided with similar knowledge about recovery issues. Furthermore, patients undergo fairly similar rehabilitation programmes, as prescribed by their surgeons. Finally, potentially important mediating factors, such as patient goals, adherence with the rehabilitation programme, or intervening events, which could potentially explain the mechanistic influence of psychosocial domains on outcomes, were not included in the analyses. Further analyses are needed to determine if these findings are mediated through the effect of patient expectancies of improvement, patient goals, and/or compliance with rehabilitation recommendations. Finally, we used the KSRS in addition to the WOMAC scale, because it is one of the most commonly used surrogate measures by orthopaedic physicians. However, as a result of the low reliability of the functional scale, reported findings involving KSRS function should be taken with caution.

In summary, our study shows that various psychosocial constructs are important independent determinants of recovery after knee replacement. Future studies should evaluate the development and implementation of screening procedures to identify individuals at risk of poor outcomes. The role of perioperative interventions targeting depression, coping skills and social support in patients with knee osteoarthritis undergoing joint replacement should also be investigated, given the substantial contributions of these domains to functional and pain outcomes.


The authors are grateful to Vanessa Cox, BS, and to Roy B Smith, MD, for their contributions to the study.


Supplementary materials


  • Funding This study was supported by a grant from the National Institute for Arthritis, Musculoskeletal and Skin Disorders (NIAMS; R01 AR48662). MES-A is the recipient of a K24 career award from the National Institute for Musculoskeletal and Skin Disorders. She is also the Director of the Houston Center for Education and Research on Therapeutics, funded by the Agency for Healthcare Research and Quality. It was also partly supported by the VA HSR&D Houston Center of Excellence (HFP-90-020). The views expressed are those of the authors and do not necessarily reflect those of the Department of Veterans Affairs/Baylor College of Medicine. MES-A has received honoraria from Zimmer as a speaker at the 2010 Summit on Musculoskeletal Health Disparities.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Institutional Review Board of the University of Texas, MD Anderson Cancer Center, St Luke's Episcopal Health System and Baylor College of Medicine.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.