Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
Contents lists available at ScienceDirect
Best Practice & Research Clinical Rheumatology journal homepage: www.elsevierhealth.com/berh
9
Selecting those to refer for joint replacement: Who will likely benefit and who will not? Michelle M. Dowsey a, b,1, Jane Gunn c, 2, Peter F.M. Choong a, b, * a
The University of Melbourne, Department of Surgery, St. Vincent’s Hospital Melbourne, 29 Regent Street, Fitzroy, Victoria, 3065, Australia St. Vincent’s Hospital Melbourne, Department of Orthopaedics, 41 Victoria Parade, Fitzroy, Victoria 3065, Australia c The University of Melbourne, Department of General Practice, 200 Berkeley Street, Carlton, Victoria, 3053, Australia b
a b s t r a c t Keywords: Osteoarthritis Total joint replacement Pain and function Risk prediction
Osteoarthritis (OA) is one of the 10 most disabling diseases in developed countries and worldwide estimates are that 10% of men and 18% of women aged over 60 years have symptomatic OA, including moderate and severe forms. Total joint replacement (TJR) is considered the most effective treatment for end-stage OA in those who have exhausted available conservative interventions. The demand for TJR is continually rising due to the ageing population; in the United States, more than 1 million TJRs were performed in 2010 and the number of procedures is projected to exceed 4 million in the US by 2030. It has been estimated that of all hip and knee replacements performed, approximately one quarter of the patients may be considered inappropriate candidates. Predicting who will benefit from TJR and who will not would seem critical in terms of containing the current and projected expenditure as well as improving satisfaction in TJR recipients. Few formal predictive tools are available to aid referring clinicians to determine those likely to be good or poor responders to surgery and current available tools tend to focus on disease severity alone with little consideration of risk factors that may predict a poor outcome or impede an effective response to surgery. This review examines
* Corresponding author. The University of Melbourne, Department of Surgery, St. Vincent’s Hospital Melbourne, 29 Regent street, Fitzroy, Victoria, 3065, Australia. Tel.: þ61 3 9288 3980; fax: þ61 3 9416 3610. E-mail addresses: [email protected] (M.M. Dowsey), [email protected] (J. Gunn), [email protected] (P.F.M. Choong). 1 Tel.: þ61 3 9288 3980; fax: þ61 3 9416 3610. 2 Tel.: þ61 3 8344 4530; fax: þ61 3 9347 6163. 1521-6942/$ – see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.berh.2014.01.005
158
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
the tools available to assist with assessing appropriateness for TJR; investigates the modifiable risk factors associated with poor outcome; and identifies areas for future research in selecting those appropriate for joint replacement. Ó 2014 Elsevier Ltd. All rights reserved.
Background Osteoarthritis (OA) is one of the 10 most disabling diseases in developed countries and worldwide estimates are that 10% of men and 18% of women aged over 60 years have symptomatic OA, including moderate and severe forms [1]. Globally, musculoskeletal (MSK) disorders are the second largest contributor to years lived with disability, with OA of the hips and knees combined being the third most prevalent MSK disorder [2]. Age is the strongest predictor of the development and progression of OA, and as such the number of people suffering with OA is expected to increase over the coming years due to the ageing of the population [3]. Based on best available evidence, the American College of Rheumatology (ACR) strongly recommends aquatic and aerobic and/or land-based resistance exercise as first-line interventions for individuals with hip or knee OA [4]. Weight loss in those who are overweight is also strongly recommended. Pharmacological interventions are conditionally recommended and include: paracetamol, nonsteroidal anti-inflammatory drugs (NSAIDs), tramadol and intra-articular corticosteroid injections [4]. However, for those who have exhausted available conservative interventions, total joint replacement (TJR) is considered the most effective treatment for end-stage OA [5,6]. The demand for TJR is continually rising due to the ageing population and between 2000 and 2009, the average rate of hip replacement increased by more than 25% and by nearly 100% for knee replacement [1]. In the United States, more than 1 million hip and knee replacements were performed in 2010 at an average per case cost of US $17,000 and $15,000, respectively [7], with an estimated total financial outlay for TJR procedures in this one year alone at $16.5 billion dollars in health-care costs. With the number of procedures projected to exceed 4 million in the US by 2030 [8] and many other Organisation for Economic Co-operation and Development (OECD) countries demonstrating similar increasing trajectories for hip and knee replacement [1], preparing to meet this challenge will be a major priority for health services globally. TJR improves quality of life by reducing pain, joint deformity and loss of function [9,10]. It is regarded as a successful procedure with a revision rate of only 0.5% per annum from index surgery [11]. However, revision surgery alone as a sole index of failure has been called into question because of the potential for underestimating the problem [12]. There is a significant proportion (20–40%) of patients who endure years of dissatisfaction and disability despite procedures appearing technically and radiologically satisfactory [13,14]. Many of these patients do not undergo revision surgery, but all add to the community’s burden of health care as clinicians and allied services strive to remedy their dissatisfaction. It has been estimated that of all hip and knee replacements performed, approximately one quarter may be considered inappropriate candidates when assessed against a validated pain and functional threshold above which surgery would be indicated to achieve a clinically relevant improvement [15,16]. Predicting who will benefit from TJR and who will not would seem critical in terms of containing the current and projected expenditure as well as the level of dissatisfaction expressed by a significant proportion of TJR recipients. It would seem reasonable that clinicians and hospital management adopt criteria for both selecting candidates and for assessing appropriateness for surgery. Yet there are currently very few formal predictive tools available to aid referring clinicians to determine those likely to be good or poor responders to surgery. Determining appropriateness for surgery should not be based on disease severity alone but also take into account an individual’s capacity to benefit from TJR [17]. The decision to refer for surgery, should be based on a number of key considerations: i) When measured against validated criteria, does
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
159
the individual’s objective (radiographic) and subjective (clinical) presentation warrant specialist referral, ii) To what extent have effective conservative treatments have been trialled, iii) To what extent is an individual’s risk factors likely to impede their capacity to benefit from surgery and iv) What strategies can be applied prior to specialist referral to mitigate the risk in those individuals identified as likely to be poor responders to surgery. However, current available tools to guide specialist referral tend to focus on disease severity alone with little consideration of risk factors that may predict a poor outcome or impede an effective response to surgery. This review examines the tools available to assist with assessing appropriateness for TJR; investigates the modifiable risk factors associated with poor outcome; and identifies areas for future research in selecting those appropriate for joint replacement. Selecting TJR candidates based on disease severity It is generally recommended that patients with severe symptomatic OA, who have pain that has failed to respond to conservative treatments and who have progressive limitation in activities of daily living should be referred to an orthopaedic surgeon for evaluation [18]. However, opinions about the indications for TJR based on the severity of joint disease and symptomatology vary greatly between referring physicians and surgeons [19,20]. While there are no universally accepted criteria to determine the indications for TJR, there are algorithms that have been developed to assist with determining appropriateness of surgery based on clinical presentation. Although these algorithms may be used to aid the decision to refer for specialist evaluation, they fail to take into account individual risk factors which may impact on patient outcomes after TJR and few of these algorithms have undergone rigorous evaluation [21–23]. Algorithms based on appropriateness criteria The most widely validated algorithms are those that have used the RAND/UCLA appropriateness method (RAM) [21,23], which combines expert opinion with available scientific evidence to create explicit criteria for the appropriateness of a medical or surgical intervention [24]. Quintana et al. and Escobar et al. have developed explicit criteria for hip and knee replacement, respectively, based on this method. Appropriateness is defined as meaning that the expected health benefit exceeds the expected negative consequences by a sufficiently wide margin to make a treatment worth performing [23]. Quintana et al. developed explicit criteria for appropriateness in total hip replacement (THR) based on a comprehensive list of indications for THR which were initially derived from an extensive literature review [23]. Individual patient variables included: age, bone quality, surgical risk, prior non-surgical procedures performed and pain and functional limitation with each variable categorised based on accepted classification systems published in the literature. All possible combinations of these variables and categories (n ¼ 216) were reviewed by expert panels who rated appropriateness for surgery for each of the 216 indications using a modified Delphi process. Each scenario was rated as appropriate, uncertain or inappropriate. To determine the validity of the panel’s findings, a random cohort of patients undergoing THR completed the Western Ontario and McMaster Universities Arthritis Index (WOMAC) [25], a diseasespecific questionnaire, and the SF-36 [26], a generic health questionnaire, prior to and 3 months after surgery. Patients were classified into one of the three appropriateness groups (appropriate, uncertain and inappropriate) and the improvement in health scores was assessed for each group. Significantly greater improvements in scores were demonstrated for those who were considered appropriate for surgery compared to those considered inappropriate and the differences were clinically meaningful. Quintana’s appropriateness criteria are expressed as a regression tree which outlines six broad scenarios where referral for consideration for THR would be deemed appropriate (Table 1). For a complete view of the decision tree including appropriate, uncertain and inappropriate scenarios, see Quintana et al. [23]. The evaluation of appropriateness for THR is initially based on the degree of pain on presentation. Where pain is severe and prior conservative measures have been undertaken correctly, THR is deemed appropriate in those with functional limitation, whether minor moderate or
160
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
Table 1 Scenarios in which referral for THR surgery is deemed appropriate. Paina
Prior conservative treatment undertakena
Functional Limitationa
American Society of Anaesthesiologist (ASA) Scorea
Mild Moderate Severe Severe Severe Severe
Correctly Correctly Correctly Correctly Correctly Incorrectly
Severe Severe Minor Moderate Severe Severe
1–3 1–3 Not required Not required Not required Not required
a
For classification definitions of variables in the algorithm, see appendix in Quintana et al. [23].
severe. In cases where a conservative treatment path has not been followed correctly, surgery is only deemed appropriate in those with severe functional limitation. In those whom pain is mild to moderate and prior conservative treatments have been undertaken correctly, surgery is deemed appropriate for those with severe functional limitation and low surgical risk. Using a similar process, Escobar et al. have used the RAND method to develop explicit criteria for appropriateness in total knee replacement (TKR), with a few notable differences; radiographic assessment is included in the TKR algorithm and the analyses exclude those who have undergone prior surgical management, arthroscopy and/or tibial osteotomy [21]. Initially, there were 864 indications for TKR identified through a literature review which was reduced to 624 as a result of modifying the variables included and finally 156 of the most frequent scenarios were reviewed by an expert panel for appropriateness assessment. For a complete view of the decision tree including appropriate, uncertain and inappropriate scenarios, see Escobar et al. [21]. For TKR, surgery was only determined to be appropriate in those with intense to severe pain using a defined set of criteria (Table 2). In those with severe pain and Ahlback classification of I, surgery was deemed appropriate in those who were aged 66 years or older with limited mobility. In those with both intense or severe pain and Ahlback classification of II–III, surgery was deemed appropriate in those aged 55 years or older with limited mobility, whereas in those with Ahlback classification of IV–V, surgery was deemed appropriate in those aged 55 years or older irrespective of mobility. In those younger than 55 years of age with Ahlback classification of IV–V, surgery was only deemed appropriate in those with bicompartmental involvement (including the patella–femoral joint) or tricompartmental OA and again this was irrespective of mobility status. While the above algorithms have undergone a validation process [27], several limitations diminish their usefulness as a clinical guide for specialist referral and these warrant mentioning. The THR algorithm does not take into account radiographic severity, which has, in recent literature, been revealed as an important predictor of the THR outcome in the short and longer term [28,29]. The use of the American Society of Anaesthesiologist (ASA) score designed for anaesthesiologists to classify surgical risk [30] may restrict the tool’s usefulness in the primary care setting. The algorithm for TKR excludes those who have undergone prior surgical management, diminishing its broad applicability to the population who present with late-stage knee OA, given that arthroscopy is common in the preceding 1–2 years amongst those presenting for TKR [31,32].
Table 2 Scenarios in which referral for TKR surgery is deemed appropriate. Paina
Ahlbacka
Agea
Severe Intense–severe Intense–severe Intense–severe Intense–severe
I II–III IV–V IV–V IV–V
>65 >55 >55 35.0 kg/m2 did not achieve the level of improvement in pain compared to patients with a BMI 35.0 kg/m2). Fewer severely obese patients responded to surgery in terms of function, with a response rate of 84% of normal weight, 74% of overweight, 72% of obese and 68% of severely obese patients experiencing a clinically important improvement at 3 years. Again, morbidly obese patients were not analysed due to insufficient numbers. While evidence from larger studies demonstrates that obesity is associated with an increased likelihood of ongoing pain and/or poorer function after TJR, the above studies examining the impact of obesity on overall response rates to TJR surgery report small differences only between non-obese and obese groups. Larger studies that are adequately powered are required to determine if these differences are statistically significant. The impact of morbid obesity on response to TJR remains unclear and warrants investigation given that the average BMI of patients presenting for TJR is also increasing over time [50]. While the literature around pain and functional outcomes following TJR remain equivocal, there is strong evidence of a negative association between obesity and post-operative complications and patients should be cautioned about the risk prior to considering specialist referral. Several systematic reviews and meta-analyses have reported significantly higher complication rates in obese patients undergoing both total hip and knee replacement with infection being the most consistently noted complication [59–61]. Obesity was associated with a twofold to threefold risk of both superficial and deep infection, according to these reviews. The risk of complication for morbidly obese patients undergoing TJR can be even greater [62,63]. In a post hoc analysis data from a study of rivaroxaban and enoxaparin which comprised four phase III randomised double-blind clinical trials, serious adverse events at 6–8 weeks post-surgery were nearly twice as common in morbidly obese patients when compared to non-morbidly obese patients [62]. The risk of both wound complications and respiratory infections was at least twice as high in morbidly obese patients. In a systematic review of TKR in morbidly obese patients, Samson et al. also reported a noted prevalence of deep prosthetic infection three to nine times higher in morbidly obese patients compared to controls [63]. As the rate of obesity continues to rise, the average BMI of patients presenting for TJR also seems to be increasing over time [50] and as such TJR surgery is now being performed for ‘super-obese’ patients and while little data have been published on outcomes in this group, the substantial increased risk of complication does warrant mentioning [57,64]. In a retrospective review comparing TJR in patients with a BMI > 45 kg/m2 to non-obese patients, super-obese patients had more than eight times the risk of incurring an in-hospital complication. A dose effect was also noted with every 5-point increase in BMI above 45 kg/m [2], resulting in a 69% increased odds for the in-hospital complication and 171% increased odds for complications within the first year [64]. Due to small numbers (n ¼ 137), complications were analysed as a composite and included both medical and surgical events. Contrary to popular belief amongst patients, very few patients experience clinically meaningful weight loss after surgery [40,43]. Using 5% change from preoperative weight as the cut-off, we have shown that 89% of patients undergoing THR did not lose weight at 12 months post TJR and one-quarter had gained weight [40]. Similarly, 87% of patients undergoing TKR did not lose weight at 12 months with weight gain occurring in 20% of patients [43]. Given the higher risk of complication and the economic burden of treating these complications in obese patients undergoing TJR [65], denying surgery for obese patients who require TJR may seem a valid option. However, the likely outcome for these patients is continued functional decline, worsening health and loss of independence, potentially creating an even greater demand on patients, their carers and the community. Recommending weight loss prior to referral for TJR would seem prudent in obese patients and dietary intervention in obese patients with OA has been trialled with moderate success [66–68]. Two
164
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
separate trials examining weight loss programmes in obese patients with knee OA, reported losses between 4.9% (dietary programme) and 8.7% (intensive weight loss programme) of baseline weight over 6–18 months [67,68]. A further study involving low-energy diets in individuals with knee OA reported a 12% reduction in weight at 16 weeks co-inducing with a significant improvement in symptoms [69]. A meta-analysis of weight reduction in obese patients diagnosed with knee OA concluded that a weight loss of >5% achieved within a 20-week period is associated with significant improvements in patient reported pain and disability [70]. As low-energy diets appear to induce the greatest amount of weight loss in the shortest period of time, they would seem the most appropriate dietary option for overweight and obese individuals with end-stage OA who warrant specialist referral for consideration of TJR. As sustained weight loss remains a challenge in obesity management, ongoing weight management should continue after TJR surgery [71]. A recent meta-analysis concluded that meal replacements, high-protein diets and pharmacological therapy, were all associated with improved weight loss maintenance over controls at 1–3 years after completing an initial low-energy or very low-energy weight loss programme [71]. Similar results were reported for both pharmacological therapy and meal replacement and therefore an individual’s general state of health as well as lifestyle factors would need to be considered when deciding on the best weight management option post TJR surgery. Intensive preoperative dietary programmes may not induce sufficient weight loss to influence the rate of complications incurred by morbidly and super-obese patients undergoing TJR. Compared with the non-surgical treatment of obesity, bariatric surgery does lead to greater body weight loss [72] and therefore may result in sufficient weight loss so as to reduce the risk of post-operative complications in morbidly obese patients undergoing TJR. However, bariatric surgery is not without risk and this hypothesis remains largely untested. While a retrospective study reported that the wound infection rate was 3.5 times lower and readmission to hospital at 30 days seven times lower in patients who had bariatric surgery prior to joint replacement [73] and a smaller retrospective study reporting no difference in complications [74], there is currently insufficient published data to make any recommendations regarding bariatric surgery prior to TJR. Psychological well-being Psychological co-morbidities and traits reported in TJR patients include depression, anxiety, neuroticism, catastrophising, poor self-esteem and low self-efficacy. These co-morbidities and traits have been associated with poorer function and greater pain after TJR in short and longer term followup studies [75–81]. Stronger associations between psychological factors and TJR outcomes are reported in knee replacement than in hip replacement [81,82] which may, in part, contribute to higher dissatisfaction rates and poorer response to surgery reported in recipients of TKR compared to TJR [83]. A recent systematic review examining the influence of psychological factors on TJR outcomes reported that strong evidence was found that TKR patients with pain catastrophising reported more pain within the first year post-operatively. In studies reporting follow-up beyond 1 year, strong evidence was also found that lower preoperative mental health was associated with lower scores on function and pain in TKR patients [82]. However, depression had no influence on post-operative functioning and the evidence for THR was conflicting. There is a dearth of literature examining the impact of psychological factors on actual response rates in TJR surgery. Only one study was identified whereby preoperative anxiety and depression were assessed using the European Quality of Life 5-dimensions (EQ5D) questionnaire [84] in a cohort of 1991 patients undergoing TKR [85]. In this study, patients were dichotomised into responders and nonresponders, based on achieving an MCID in Oxford Scores of >30 points at 6 months. Seventy-two percent of the patients achieved the MCID. Patients who reported feeling moderately anxious or depressed prior to TKR were less likely (odds ratio (OR) 0.67, 95% confidence interval (CI): 0.54–0.84) to report a response to surgery and those who reported feeling extremely anxious or depressed were only half as likely (OR 0.51, 95% CI: 0.31–0.84) to report a response to surgery, compared to those who indicated they were not anxious or depressed [85]. Despite the prevalence of psychological distress and its association with poorer outcomes, particularly in patients presenting for TKR, the psychological needs of the individual are not generally
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
165
considered in preparation for TJR surgery. While preoperative screening and optimisation are considered mandatory for patients listed for surgery, preparation tends to focus on optimising physical health. Systematic reviews support the use of clinical pathways and preoperative physiotherapy programmes for optimisation of medical health and physical function prior to surgery and to ensure the best possible clinical outcomes [86–89]. However, we could find no literature evaluating psychological interventions in patients awaiting TJR. A recent Cochrane Review did report that preoperative education has a modest beneficial effect on preoperative anxiety and that education tailored according to anxiety may also have a beneficial effect on recovery [90]. Patient expectation It has been shown that more than one-half of the patients undergoing TJR have expectations of surgery that exceed those of their surgeons and one quarter expect less than their surgeons [91]. The rate of dissatisfaction amongst TJR recipients with unmet expectations is as high as 49% compared to 6% in those whose expectations have been met [13]. Several studies have investigated the predictive capacity of expectations for outcomes of TJR; however, a recent systematic review reported that there was no consistency in the association between patients’ preoperative expectations and treatment outcomes for hip or knee replacement. This was largely due to heterogeneity in measures of expectation and outcomes, the variations in follow-up time frame and the poor reporting standards of included studies [92]. One study examined patient expectation and response to surgery at 12 months in a cohort of 1327 patients undergoing THR. Patients were dichotomised into responders and non-responders based on the OARSI/OMERACT criteria [49] and logistic regression modelling was used to explore associations between patient expectations and response to surgery [93]. Patients were questioned about their expectations in terms of their ability to walk, reduction in pain and return to normal. Patients with a larger number of preoperative expectations were more likely to report a clinically important response to TJR. Each individual expectation was associated with a 34% increase in the probability of responding to surgery and the authors suggest that further research is required to determine whether building patient expectations, such as telling patients their joint replacement will be highly successful, positively influences the outcome of surgery [93]. No studies could be found examining the effect of expectation on response to TKR surgery. Given the well-documented differences between total hip and knee replacement surgery in terms of both risk factors and outcomes, without further research, we would caution against extrapolating the above findings when discussing expectations with patient considering TKR. Radiographic OA severity To date, a majority of the algorithms designed to select appropriate candidates for TJR are based on patient-reported symptomology without consideration of radiographic OA severity [23,36,37]. While several studies have failed to demonstrate an association between baseline radiographic OA severity and clinical severity [94,95], there is consistent evidence of a definite inverse relationship between preoperative radiographic severity of OA and pain and function outcomes in the intermediate term after both hip and knee joint replacement [28,29,93,95,96]. Of these studies (Table 3), two have reported higher response rates for total hip and knee replacement surgery accordingly in those with more severe radiographic OA [29,93]. Keurentjes et al. assessed the probability of achieving an MCID in health-related quality of life scores as well as satisfaction at 2 and 5 years post hip (n ¼ 445) and knee (278) replacements [29]. The probability of reporting a satisfactory outcome was twice as high in patients undergoing THR for those with a KL grade 3–4, compared to KL grade 1–2. For TKR, the probability of reporting an MCID in the physical function scale of the SF-3626 was five times more likely for those with KL grade 3–4 compared to KL grade 1–2. For the general health scale of the SF-36, the likelihood of reporting an MCID was nearly four times higher for KL grade 3–4, compared to KL grade 1–2. A notable limitation of this study is an ascertainment rate of only 46%. Judge et al. examined predictors of patient expectation and response to surgery at 12 months in a cohort of 1327 patients undergoing THR. Patients were dichotomised into responders and non-
166
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
Table 3 Radiographic OA severity and patient outcomes after total joint replacement. Author
Cohort size
Follow-up
Radiographic severity
Outcome measure
Main findings
Dowsey et al. [46]
478 TKR
1 and 2 years
M_KL grade [94]
IKSS [97] Pain and Function
Keurentjes et al. [98]
445 THR 278 TKR
2 and 5 years
KL grade [99]
SF-3626 and NRS [29]
Cushnaghan et al. [28]
282 THR
8 years
Croft grade [100]
SF-3626
Valdes et al. [96]
860 TKR 928 THR
Average 3.2 years
KL grade JSW
WOMAC [101]
Judge et al. [93]
908 THR
1 year
KL grade
OMERACT OARSI [49]
People with less severe radiographic changes prior to TKR are less likely to experience major improvement in pain and function than those with more severe changes. KL grade 2 mm for hips) 1.56 times increased risk of high pain amongst post-TJR cases. Lower odds (0.61; 95% CI 0.38–0.95) of responding to surgery in KL grade
Contents lists available at ScienceDirect
Best Practice & Research Clinical Rheumatology journal homepage: www.elsevierhealth.com/berh
9
Selecting those to refer for joint replacement: Who will likely benefit and who will not? Michelle M. Dowsey a, b,1, Jane Gunn c, 2, Peter F.M. Choong a, b, * a
The University of Melbourne, Department of Surgery, St. Vincent’s Hospital Melbourne, 29 Regent Street, Fitzroy, Victoria, 3065, Australia St. Vincent’s Hospital Melbourne, Department of Orthopaedics, 41 Victoria Parade, Fitzroy, Victoria 3065, Australia c The University of Melbourne, Department of General Practice, 200 Berkeley Street, Carlton, Victoria, 3053, Australia b
a b s t r a c t Keywords: Osteoarthritis Total joint replacement Pain and function Risk prediction
Osteoarthritis (OA) is one of the 10 most disabling diseases in developed countries and worldwide estimates are that 10% of men and 18% of women aged over 60 years have symptomatic OA, including moderate and severe forms. Total joint replacement (TJR) is considered the most effective treatment for end-stage OA in those who have exhausted available conservative interventions. The demand for TJR is continually rising due to the ageing population; in the United States, more than 1 million TJRs were performed in 2010 and the number of procedures is projected to exceed 4 million in the US by 2030. It has been estimated that of all hip and knee replacements performed, approximately one quarter of the patients may be considered inappropriate candidates. Predicting who will benefit from TJR and who will not would seem critical in terms of containing the current and projected expenditure as well as improving satisfaction in TJR recipients. Few formal predictive tools are available to aid referring clinicians to determine those likely to be good or poor responders to surgery and current available tools tend to focus on disease severity alone with little consideration of risk factors that may predict a poor outcome or impede an effective response to surgery. This review examines
* Corresponding author. The University of Melbourne, Department of Surgery, St. Vincent’s Hospital Melbourne, 29 Regent street, Fitzroy, Victoria, 3065, Australia. Tel.: þ61 3 9288 3980; fax: þ61 3 9416 3610. E-mail addresses: [email protected] (M.M. Dowsey), [email protected] (J. Gunn), [email protected] (P.F.M. Choong). 1 Tel.: þ61 3 9288 3980; fax: þ61 3 9416 3610. 2 Tel.: þ61 3 8344 4530; fax: þ61 3 9347 6163. 1521-6942/$ – see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.berh.2014.01.005
158
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
the tools available to assist with assessing appropriateness for TJR; investigates the modifiable risk factors associated with poor outcome; and identifies areas for future research in selecting those appropriate for joint replacement. Ó 2014 Elsevier Ltd. All rights reserved.
Background Osteoarthritis (OA) is one of the 10 most disabling diseases in developed countries and worldwide estimates are that 10% of men and 18% of women aged over 60 years have symptomatic OA, including moderate and severe forms [1]. Globally, musculoskeletal (MSK) disorders are the second largest contributor to years lived with disability, with OA of the hips and knees combined being the third most prevalent MSK disorder [2]. Age is the strongest predictor of the development and progression of OA, and as such the number of people suffering with OA is expected to increase over the coming years due to the ageing of the population [3]. Based on best available evidence, the American College of Rheumatology (ACR) strongly recommends aquatic and aerobic and/or land-based resistance exercise as first-line interventions for individuals with hip or knee OA [4]. Weight loss in those who are overweight is also strongly recommended. Pharmacological interventions are conditionally recommended and include: paracetamol, nonsteroidal anti-inflammatory drugs (NSAIDs), tramadol and intra-articular corticosteroid injections [4]. However, for those who have exhausted available conservative interventions, total joint replacement (TJR) is considered the most effective treatment for end-stage OA [5,6]. The demand for TJR is continually rising due to the ageing population and between 2000 and 2009, the average rate of hip replacement increased by more than 25% and by nearly 100% for knee replacement [1]. In the United States, more than 1 million hip and knee replacements were performed in 2010 at an average per case cost of US $17,000 and $15,000, respectively [7], with an estimated total financial outlay for TJR procedures in this one year alone at $16.5 billion dollars in health-care costs. With the number of procedures projected to exceed 4 million in the US by 2030 [8] and many other Organisation for Economic Co-operation and Development (OECD) countries demonstrating similar increasing trajectories for hip and knee replacement [1], preparing to meet this challenge will be a major priority for health services globally. TJR improves quality of life by reducing pain, joint deformity and loss of function [9,10]. It is regarded as a successful procedure with a revision rate of only 0.5% per annum from index surgery [11]. However, revision surgery alone as a sole index of failure has been called into question because of the potential for underestimating the problem [12]. There is a significant proportion (20–40%) of patients who endure years of dissatisfaction and disability despite procedures appearing technically and radiologically satisfactory [13,14]. Many of these patients do not undergo revision surgery, but all add to the community’s burden of health care as clinicians and allied services strive to remedy their dissatisfaction. It has been estimated that of all hip and knee replacements performed, approximately one quarter may be considered inappropriate candidates when assessed against a validated pain and functional threshold above which surgery would be indicated to achieve a clinically relevant improvement [15,16]. Predicting who will benefit from TJR and who will not would seem critical in terms of containing the current and projected expenditure as well as the level of dissatisfaction expressed by a significant proportion of TJR recipients. It would seem reasonable that clinicians and hospital management adopt criteria for both selecting candidates and for assessing appropriateness for surgery. Yet there are currently very few formal predictive tools available to aid referring clinicians to determine those likely to be good or poor responders to surgery. Determining appropriateness for surgery should not be based on disease severity alone but also take into account an individual’s capacity to benefit from TJR [17]. The decision to refer for surgery, should be based on a number of key considerations: i) When measured against validated criteria, does
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
159
the individual’s objective (radiographic) and subjective (clinical) presentation warrant specialist referral, ii) To what extent have effective conservative treatments have been trialled, iii) To what extent is an individual’s risk factors likely to impede their capacity to benefit from surgery and iv) What strategies can be applied prior to specialist referral to mitigate the risk in those individuals identified as likely to be poor responders to surgery. However, current available tools to guide specialist referral tend to focus on disease severity alone with little consideration of risk factors that may predict a poor outcome or impede an effective response to surgery. This review examines the tools available to assist with assessing appropriateness for TJR; investigates the modifiable risk factors associated with poor outcome; and identifies areas for future research in selecting those appropriate for joint replacement. Selecting TJR candidates based on disease severity It is generally recommended that patients with severe symptomatic OA, who have pain that has failed to respond to conservative treatments and who have progressive limitation in activities of daily living should be referred to an orthopaedic surgeon for evaluation [18]. However, opinions about the indications for TJR based on the severity of joint disease and symptomatology vary greatly between referring physicians and surgeons [19,20]. While there are no universally accepted criteria to determine the indications for TJR, there are algorithms that have been developed to assist with determining appropriateness of surgery based on clinical presentation. Although these algorithms may be used to aid the decision to refer for specialist evaluation, they fail to take into account individual risk factors which may impact on patient outcomes after TJR and few of these algorithms have undergone rigorous evaluation [21–23]. Algorithms based on appropriateness criteria The most widely validated algorithms are those that have used the RAND/UCLA appropriateness method (RAM) [21,23], which combines expert opinion with available scientific evidence to create explicit criteria for the appropriateness of a medical or surgical intervention [24]. Quintana et al. and Escobar et al. have developed explicit criteria for hip and knee replacement, respectively, based on this method. Appropriateness is defined as meaning that the expected health benefit exceeds the expected negative consequences by a sufficiently wide margin to make a treatment worth performing [23]. Quintana et al. developed explicit criteria for appropriateness in total hip replacement (THR) based on a comprehensive list of indications for THR which were initially derived from an extensive literature review [23]. Individual patient variables included: age, bone quality, surgical risk, prior non-surgical procedures performed and pain and functional limitation with each variable categorised based on accepted classification systems published in the literature. All possible combinations of these variables and categories (n ¼ 216) were reviewed by expert panels who rated appropriateness for surgery for each of the 216 indications using a modified Delphi process. Each scenario was rated as appropriate, uncertain or inappropriate. To determine the validity of the panel’s findings, a random cohort of patients undergoing THR completed the Western Ontario and McMaster Universities Arthritis Index (WOMAC) [25], a diseasespecific questionnaire, and the SF-36 [26], a generic health questionnaire, prior to and 3 months after surgery. Patients were classified into one of the three appropriateness groups (appropriate, uncertain and inappropriate) and the improvement in health scores was assessed for each group. Significantly greater improvements in scores were demonstrated for those who were considered appropriate for surgery compared to those considered inappropriate and the differences were clinically meaningful. Quintana’s appropriateness criteria are expressed as a regression tree which outlines six broad scenarios where referral for consideration for THR would be deemed appropriate (Table 1). For a complete view of the decision tree including appropriate, uncertain and inappropriate scenarios, see Quintana et al. [23]. The evaluation of appropriateness for THR is initially based on the degree of pain on presentation. Where pain is severe and prior conservative measures have been undertaken correctly, THR is deemed appropriate in those with functional limitation, whether minor moderate or
160
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
Table 1 Scenarios in which referral for THR surgery is deemed appropriate. Paina
Prior conservative treatment undertakena
Functional Limitationa
American Society of Anaesthesiologist (ASA) Scorea
Mild Moderate Severe Severe Severe Severe
Correctly Correctly Correctly Correctly Correctly Incorrectly
Severe Severe Minor Moderate Severe Severe
1–3 1–3 Not required Not required Not required Not required
a
For classification definitions of variables in the algorithm, see appendix in Quintana et al. [23].
severe. In cases where a conservative treatment path has not been followed correctly, surgery is only deemed appropriate in those with severe functional limitation. In those whom pain is mild to moderate and prior conservative treatments have been undertaken correctly, surgery is deemed appropriate for those with severe functional limitation and low surgical risk. Using a similar process, Escobar et al. have used the RAND method to develop explicit criteria for appropriateness in total knee replacement (TKR), with a few notable differences; radiographic assessment is included in the TKR algorithm and the analyses exclude those who have undergone prior surgical management, arthroscopy and/or tibial osteotomy [21]. Initially, there were 864 indications for TKR identified through a literature review which was reduced to 624 as a result of modifying the variables included and finally 156 of the most frequent scenarios were reviewed by an expert panel for appropriateness assessment. For a complete view of the decision tree including appropriate, uncertain and inappropriate scenarios, see Escobar et al. [21]. For TKR, surgery was only determined to be appropriate in those with intense to severe pain using a defined set of criteria (Table 2). In those with severe pain and Ahlback classification of I, surgery was deemed appropriate in those who were aged 66 years or older with limited mobility. In those with both intense or severe pain and Ahlback classification of II–III, surgery was deemed appropriate in those aged 55 years or older with limited mobility, whereas in those with Ahlback classification of IV–V, surgery was deemed appropriate in those aged 55 years or older irrespective of mobility. In those younger than 55 years of age with Ahlback classification of IV–V, surgery was only deemed appropriate in those with bicompartmental involvement (including the patella–femoral joint) or tricompartmental OA and again this was irrespective of mobility status. While the above algorithms have undergone a validation process [27], several limitations diminish their usefulness as a clinical guide for specialist referral and these warrant mentioning. The THR algorithm does not take into account radiographic severity, which has, in recent literature, been revealed as an important predictor of the THR outcome in the short and longer term [28,29]. The use of the American Society of Anaesthesiologist (ASA) score designed for anaesthesiologists to classify surgical risk [30] may restrict the tool’s usefulness in the primary care setting. The algorithm for TKR excludes those who have undergone prior surgical management, diminishing its broad applicability to the population who present with late-stage knee OA, given that arthroscopy is common in the preceding 1–2 years amongst those presenting for TKR [31,32].
Table 2 Scenarios in which referral for TKR surgery is deemed appropriate. Paina
Ahlbacka
Agea
Severe Intense–severe Intense–severe Intense–severe Intense–severe
I II–III IV–V IV–V IV–V
>65 >55 >55 35.0 kg/m2 did not achieve the level of improvement in pain compared to patients with a BMI 35.0 kg/m2). Fewer severely obese patients responded to surgery in terms of function, with a response rate of 84% of normal weight, 74% of overweight, 72% of obese and 68% of severely obese patients experiencing a clinically important improvement at 3 years. Again, morbidly obese patients were not analysed due to insufficient numbers. While evidence from larger studies demonstrates that obesity is associated with an increased likelihood of ongoing pain and/or poorer function after TJR, the above studies examining the impact of obesity on overall response rates to TJR surgery report small differences only between non-obese and obese groups. Larger studies that are adequately powered are required to determine if these differences are statistically significant. The impact of morbid obesity on response to TJR remains unclear and warrants investigation given that the average BMI of patients presenting for TJR is also increasing over time [50]. While the literature around pain and functional outcomes following TJR remain equivocal, there is strong evidence of a negative association between obesity and post-operative complications and patients should be cautioned about the risk prior to considering specialist referral. Several systematic reviews and meta-analyses have reported significantly higher complication rates in obese patients undergoing both total hip and knee replacement with infection being the most consistently noted complication [59–61]. Obesity was associated with a twofold to threefold risk of both superficial and deep infection, according to these reviews. The risk of complication for morbidly obese patients undergoing TJR can be even greater [62,63]. In a post hoc analysis data from a study of rivaroxaban and enoxaparin which comprised four phase III randomised double-blind clinical trials, serious adverse events at 6–8 weeks post-surgery were nearly twice as common in morbidly obese patients when compared to non-morbidly obese patients [62]. The risk of both wound complications and respiratory infections was at least twice as high in morbidly obese patients. In a systematic review of TKR in morbidly obese patients, Samson et al. also reported a noted prevalence of deep prosthetic infection three to nine times higher in morbidly obese patients compared to controls [63]. As the rate of obesity continues to rise, the average BMI of patients presenting for TJR also seems to be increasing over time [50] and as such TJR surgery is now being performed for ‘super-obese’ patients and while little data have been published on outcomes in this group, the substantial increased risk of complication does warrant mentioning [57,64]. In a retrospective review comparing TJR in patients with a BMI > 45 kg/m2 to non-obese patients, super-obese patients had more than eight times the risk of incurring an in-hospital complication. A dose effect was also noted with every 5-point increase in BMI above 45 kg/m [2], resulting in a 69% increased odds for the in-hospital complication and 171% increased odds for complications within the first year [64]. Due to small numbers (n ¼ 137), complications were analysed as a composite and included both medical and surgical events. Contrary to popular belief amongst patients, very few patients experience clinically meaningful weight loss after surgery [40,43]. Using 5% change from preoperative weight as the cut-off, we have shown that 89% of patients undergoing THR did not lose weight at 12 months post TJR and one-quarter had gained weight [40]. Similarly, 87% of patients undergoing TKR did not lose weight at 12 months with weight gain occurring in 20% of patients [43]. Given the higher risk of complication and the economic burden of treating these complications in obese patients undergoing TJR [65], denying surgery for obese patients who require TJR may seem a valid option. However, the likely outcome for these patients is continued functional decline, worsening health and loss of independence, potentially creating an even greater demand on patients, their carers and the community. Recommending weight loss prior to referral for TJR would seem prudent in obese patients and dietary intervention in obese patients with OA has been trialled with moderate success [66–68]. Two
164
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
separate trials examining weight loss programmes in obese patients with knee OA, reported losses between 4.9% (dietary programme) and 8.7% (intensive weight loss programme) of baseline weight over 6–18 months [67,68]. A further study involving low-energy diets in individuals with knee OA reported a 12% reduction in weight at 16 weeks co-inducing with a significant improvement in symptoms [69]. A meta-analysis of weight reduction in obese patients diagnosed with knee OA concluded that a weight loss of >5% achieved within a 20-week period is associated with significant improvements in patient reported pain and disability [70]. As low-energy diets appear to induce the greatest amount of weight loss in the shortest period of time, they would seem the most appropriate dietary option for overweight and obese individuals with end-stage OA who warrant specialist referral for consideration of TJR. As sustained weight loss remains a challenge in obesity management, ongoing weight management should continue after TJR surgery [71]. A recent meta-analysis concluded that meal replacements, high-protein diets and pharmacological therapy, were all associated with improved weight loss maintenance over controls at 1–3 years after completing an initial low-energy or very low-energy weight loss programme [71]. Similar results were reported for both pharmacological therapy and meal replacement and therefore an individual’s general state of health as well as lifestyle factors would need to be considered when deciding on the best weight management option post TJR surgery. Intensive preoperative dietary programmes may not induce sufficient weight loss to influence the rate of complications incurred by morbidly and super-obese patients undergoing TJR. Compared with the non-surgical treatment of obesity, bariatric surgery does lead to greater body weight loss [72] and therefore may result in sufficient weight loss so as to reduce the risk of post-operative complications in morbidly obese patients undergoing TJR. However, bariatric surgery is not without risk and this hypothesis remains largely untested. While a retrospective study reported that the wound infection rate was 3.5 times lower and readmission to hospital at 30 days seven times lower in patients who had bariatric surgery prior to joint replacement [73] and a smaller retrospective study reporting no difference in complications [74], there is currently insufficient published data to make any recommendations regarding bariatric surgery prior to TJR. Psychological well-being Psychological co-morbidities and traits reported in TJR patients include depression, anxiety, neuroticism, catastrophising, poor self-esteem and low self-efficacy. These co-morbidities and traits have been associated with poorer function and greater pain after TJR in short and longer term followup studies [75–81]. Stronger associations between psychological factors and TJR outcomes are reported in knee replacement than in hip replacement [81,82] which may, in part, contribute to higher dissatisfaction rates and poorer response to surgery reported in recipients of TKR compared to TJR [83]. A recent systematic review examining the influence of psychological factors on TJR outcomes reported that strong evidence was found that TKR patients with pain catastrophising reported more pain within the first year post-operatively. In studies reporting follow-up beyond 1 year, strong evidence was also found that lower preoperative mental health was associated with lower scores on function and pain in TKR patients [82]. However, depression had no influence on post-operative functioning and the evidence for THR was conflicting. There is a dearth of literature examining the impact of psychological factors on actual response rates in TJR surgery. Only one study was identified whereby preoperative anxiety and depression were assessed using the European Quality of Life 5-dimensions (EQ5D) questionnaire [84] in a cohort of 1991 patients undergoing TKR [85]. In this study, patients were dichotomised into responders and nonresponders, based on achieving an MCID in Oxford Scores of >30 points at 6 months. Seventy-two percent of the patients achieved the MCID. Patients who reported feeling moderately anxious or depressed prior to TKR were less likely (odds ratio (OR) 0.67, 95% confidence interval (CI): 0.54–0.84) to report a response to surgery and those who reported feeling extremely anxious or depressed were only half as likely (OR 0.51, 95% CI: 0.31–0.84) to report a response to surgery, compared to those who indicated they were not anxious or depressed [85]. Despite the prevalence of psychological distress and its association with poorer outcomes, particularly in patients presenting for TKR, the psychological needs of the individual are not generally
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
165
considered in preparation for TJR surgery. While preoperative screening and optimisation are considered mandatory for patients listed for surgery, preparation tends to focus on optimising physical health. Systematic reviews support the use of clinical pathways and preoperative physiotherapy programmes for optimisation of medical health and physical function prior to surgery and to ensure the best possible clinical outcomes [86–89]. However, we could find no literature evaluating psychological interventions in patients awaiting TJR. A recent Cochrane Review did report that preoperative education has a modest beneficial effect on preoperative anxiety and that education tailored according to anxiety may also have a beneficial effect on recovery [90]. Patient expectation It has been shown that more than one-half of the patients undergoing TJR have expectations of surgery that exceed those of their surgeons and one quarter expect less than their surgeons [91]. The rate of dissatisfaction amongst TJR recipients with unmet expectations is as high as 49% compared to 6% in those whose expectations have been met [13]. Several studies have investigated the predictive capacity of expectations for outcomes of TJR; however, a recent systematic review reported that there was no consistency in the association between patients’ preoperative expectations and treatment outcomes for hip or knee replacement. This was largely due to heterogeneity in measures of expectation and outcomes, the variations in follow-up time frame and the poor reporting standards of included studies [92]. One study examined patient expectation and response to surgery at 12 months in a cohort of 1327 patients undergoing THR. Patients were dichotomised into responders and non-responders based on the OARSI/OMERACT criteria [49] and logistic regression modelling was used to explore associations between patient expectations and response to surgery [93]. Patients were questioned about their expectations in terms of their ability to walk, reduction in pain and return to normal. Patients with a larger number of preoperative expectations were more likely to report a clinically important response to TJR. Each individual expectation was associated with a 34% increase in the probability of responding to surgery and the authors suggest that further research is required to determine whether building patient expectations, such as telling patients their joint replacement will be highly successful, positively influences the outcome of surgery [93]. No studies could be found examining the effect of expectation on response to TKR surgery. Given the well-documented differences between total hip and knee replacement surgery in terms of both risk factors and outcomes, without further research, we would caution against extrapolating the above findings when discussing expectations with patient considering TKR. Radiographic OA severity To date, a majority of the algorithms designed to select appropriate candidates for TJR are based on patient-reported symptomology without consideration of radiographic OA severity [23,36,37]. While several studies have failed to demonstrate an association between baseline radiographic OA severity and clinical severity [94,95], there is consistent evidence of a definite inverse relationship between preoperative radiographic severity of OA and pain and function outcomes in the intermediate term after both hip and knee joint replacement [28,29,93,95,96]. Of these studies (Table 3), two have reported higher response rates for total hip and knee replacement surgery accordingly in those with more severe radiographic OA [29,93]. Keurentjes et al. assessed the probability of achieving an MCID in health-related quality of life scores as well as satisfaction at 2 and 5 years post hip (n ¼ 445) and knee (278) replacements [29]. The probability of reporting a satisfactory outcome was twice as high in patients undergoing THR for those with a KL grade 3–4, compared to KL grade 1–2. For TKR, the probability of reporting an MCID in the physical function scale of the SF-3626 was five times more likely for those with KL grade 3–4 compared to KL grade 1–2. For the general health scale of the SF-36, the likelihood of reporting an MCID was nearly four times higher for KL grade 3–4, compared to KL grade 1–2. A notable limitation of this study is an ascertainment rate of only 46%. Judge et al. examined predictors of patient expectation and response to surgery at 12 months in a cohort of 1327 patients undergoing THR. Patients were dichotomised into responders and non-
166
M.M. Dowsey et al. / Best Practice & Research Clinical Rheumatology 28 (2014) 157–171
Table 3 Radiographic OA severity and patient outcomes after total joint replacement. Author
Cohort size
Follow-up
Radiographic severity
Outcome measure
Main findings
Dowsey et al. [46]
478 TKR
1 and 2 years
M_KL grade [94]
IKSS [97] Pain and Function
Keurentjes et al. [98]
445 THR 278 TKR
2 and 5 years
KL grade [99]
SF-3626 and NRS [29]
Cushnaghan et al. [28]
282 THR
8 years
Croft grade [100]
SF-3626
Valdes et al. [96]
860 TKR 928 THR
Average 3.2 years
KL grade JSW
WOMAC [101]
Judge et al. [93]
908 THR
1 year
KL grade
OMERACT OARSI [49]
People with less severe radiographic changes prior to TKR are less likely to experience major improvement in pain and function than those with more severe changes. KL grade 2 mm for hips) 1.56 times increased risk of high pain amongst post-TJR cases. Lower odds (0.61; 95% CI 0.38–0.95) of responding to surgery in KL grade
