Improving patients' experience and outcome of total joint replacement: the RESTORE programme

Body mass index

There is no clear evidence linking high BMI with the development of hip osteoarthritis,137 but people with higher BMI are more likely to require THR. For example, in a UK study including over 490,000 women, those with a BMI of > 30 kg/m² had nearly 2.5 times the risk of requiring a THR of those with a BMI of < 22.5 kg/m. 138 Patients included in the National Joint Registry for England and Wales in 2012 had an average BMI of 28.71 kg/m² (SD 5.29 kg/m²) and about 39% had a BMI of ≥ 30 kg/m². 3 In the RESTORE APEX RCT, the mean BMI in 322 patients receiving THR was 29.1 kg/m² (SD 5.5 kg/m², range 18.6 to 47.8 kg/m²) and 4.3% of patients had a BMI of ≥ 40 kg/m². The distribution of BMI of patients in the APEX cohort of patients with THR is shown in Figure 9.

FIGURE 9.

Distribution of BMI in the APEX THR cohort.

Our searches of MEDLINE and EMBASE considered long-term patient-reported outcomes after THR. Searches identified eight studies that specifically focused on the relationship between BMI and patient-reported outcomes at ≥ 12 months after THR. Three studies included patients from multiple centres,127–129 and five studies were conducted at a single centre. 103,104,132,133,135 Study details are summarised in Table 4. In studies with data that allowed estimation, rates of follow-up ranged from 6% to 38%.

Jones and colleagues followed up approximately 167 patients (72% eligible) from a Canadian health region 3 years after a THR. 127 The authors used WHO criteria to classify patients into groups of BMI (< 25 kg/m², 25–29.9 kg/m², 30–34.9 kg/m², ≥ 35 kg/m²). In the cohort, 13.9% of patients had a BMI of ≥ 35 kg/m². In an analysis with adjustment for age, sex, diabetes and cardiac disease, the authors reported that similar long-term WOMAC pain and function scores were achieved in patients with different levels of BMI. Considering data collected at 6 months, the authors noted that recovery of function and reduction of pain was slower in patients with a high BMI (≥ 35 kg/m²) than groups with lower BMI.

Judge and colleagues128 followed up 1375 patients (64% eligible) who had received a specific design of THR prosthesis at seven UK centres 5 years after surgery. The mean BMI of patients in the cohort followed up was 27.6 kg/m² (SD 4.8 kg/m²) and it was treated as a continuous variable. In multivariable analyses including age, sex, primary diagnosis, occupation, comorbidities, health-related quality of life (HRQoL), hip ROM, surgical variables and Oxford Hip Score (OHS), there was a relationship between increasing BMI and poorer long-term function and pain as measured by the OHS. The authors considered the differences in function and pain associated with BMI to be small.

In a study at three orthopaedic centres in the Netherlands, Stevens and colleagues129 followed up 653 patients (77% eligible) 12 months after receiving a THR. The mean BMI was 27.0 kg/m² (SD 4.1 kg/m²) and the authors defined three groups (< 25 kg/m², 25–30 kg/m², > 30 kg/m²). After adjusting analyses for age, sex, comorbidities and complications, increased BMI was associated with worse long-term function but the size of the effect was low, particularly in comparison to that of presence of comorbidities and complications.

Moran and colleagues135 followed up 687 patients (86% eligible) with THR at a single UK centre. The mean BMI in this cohort was 27.8 kg/m² (SD 5 kg/m²) and the authors reported that only 9 out of 687 patients (1.3%) had a BMI of > 40 kg/m² and that no conclusions could be drawn for these patients. After adjusting for sex, comorbidities, OHS and SF-36, BMI treated as a continuous variable was not a significant predictor for any SF-36 domains.

Nilsdotter and colleagues103 reported the follow-up of 198 patients (94% eligible) with THR from a single centre in Sweden at a mean of 3.6 years. The authors analysed BMI as a continuous variable but did not report mean or categorical values. Increasing BMI was associated with poorer long-term WOMAC function in univariate analysis. After adjustment for sex, comorbidities, WOMAC, SF-36 (including mental health), employment, marital status, contralateral osteoarthritis, need of walking assistance, walking distance, analgesic use and regional or widespread pain, BMI was not associated with long-term WOMAC function.

Davis and colleagues132 followed up 1095 patients (68% eligible) from a single UK centre 60 months after THR. The mean BMI in the cohort was 28 kg/m² at baseline and 9.2% of patients had a BMI of ≥ 35 kg/m². Patients were divided into groups according to their BMI (< 25 kg/m², 25–29.9 kg/m², 30–34.9 kg/m², ≥ 35 kg/m²). In multivariable analysis with age, sex, pre-operative hip score, SF-36, comorbidities and consultant, increasing BMI predicted poorer long-term SF-36 physical function and bodily pain. However, the authors acknowledged that although absolute levels of long-term pain and function were poorer than in patients with low BMI, there were dramatic improvements in patient outcomes in those with high BMI.

Singh and Lewallen104 reported a 2-year follow-up of 5707 patients with THR (62% eligible) at a single US centre. Four per cent of patients had a BMI of ≥ 40 kg/m² or greater. The authors divided patients into five groups according to their BMI (< 25 mg/m², 25–29.9 mg/m², 30–34.9 kg/m², 35–39.9 kg/m², ≥ 40 kg/m²). In an analysis adjusted for age, sex, comorbidities, depression, anxiety, operative diagnosis, distance from centre and implant design, patients with BMI of ≥ 35 kg/m² reported significantly greater long-term moderate or severe pain than the lowest BMI group.

In a study from a single centre in Canada, Gandhi and colleagues133 followed up 636 patients (per cent eligible not reported) for an average of 3.3 years. The mean BMI in this cohort was 27.6 kg/m² (SD 4.9 kg/m²) and was treated as a continuous variable in analyses. After adjusting for age, sex, comorbidities, WOMAC and SF-36 scores there was a non-significant trend for less improvement in WOMAC score with increasing BMI.

With the increasing levels of BMI in developed countries, it is important that patients and health-care providers are aware of any factors that influence the long-term outcome of THR. We identified eight studies that reported the association between BMI and a long-term patient-reported outcome in multivariable analysis. 103,104,127–129,132,133,135 If reported in the studies we identified, patients had broadly similar average BMIs to those reported in the National Joint Registry for England and Wales3 and the APEX cohort described in Chapters 6 and 7.

In three studies, the absolute long-term OHS,128 WOMAC function129 and a simple measure of pain severity104 were more favourable in lower BMI groups. In another study this was observed for WOMAC function in univariate but not multivariate analysis. 103 In two studies there were no long-term differences in WOMAC pain or function,127 WOMAC score or SF-36. 135 Further to the selection of studies according to specific quality criteria, studies had either one or no additional marker of quality based on centres studied and losses to follow-up. Differences in results of studies were not explained by issues relating to these additional markers of study quality.

In the four studies for which BMI was treated as a continuous variable in multivariable analysis, authors reported no strong association between BMI and long-term function,103,128,133,135 or pain. 128,135 Two studies reported changes in function which may be a more appropriate method of analysis as patients with higher BMI generally have poorer function before surgery. There were greater improvements in OHS128 and WOMAC133 in patients with lower BMI. Associations observed were not limited to studies according to additional markers of quality.

In four studies, authors focused on the relationship between categorical levels of BMI and patient-reported outcomes. There was some evidence that patients classified as obese (≥ 30 kg/m²) according to WHO classifications139 had poorer function129,132 or pain outcomes,104 but only one study had an additional marker of good quality. 127 In a fourth study with one additional marker of good quality, no association was noted between pre-operative BMI and long-term function or pain, although the authors noted a slower recovery in patients with high BMI. 127

Overall, the absolute levels of physical function and pain achieved after THR in patients with particularly high BMI may be somewhat lower than that achieved by other patients. However, there is a clear indication that many patients with high BMI benefit from THR with long-term improvements to physical function and reduction in long-term pain.

Pre-surgical mental health

The period between being placed on the waiting list and the day of surgery can be a time of distress for patients and is characterised by pain, poor physical function and uncertainty. Parsons and colleagues140 identified six major themes describing patients’ experiences of waiting for joint replacement: living and coping with pain; not being able to walk; coping with everyday activities; body image; help, advice and support; and the effect on family, friends and helpers.

Anxiety and depression are common in people with osteoarthritis. 141,142 In the APEX cohort of patients with osteoarthritis waiting for THR, pre-surgical anxiety and depression was identified using the Hospital Anxiety and Depression Scale (HADS) questionnaire. Definite or potential anxiety was reported by 33% of patients and definite or potential depression by 30% of patients.

One previous systematic review explored the relationship between pre-surgical anxiety and long-term patient outcomes in patients with THR. 143 Vissers and colleagues143 searched MEDLINE and EMBASE to January 2011 and identified nine studies including 8823 patients receiving THR. The authors reported that there was limited and conflicting evidence on the relationship between psychological factors and postoperative function and pain.

Our literature searches identified seven studies with ≥ 100 patients with THR followed up for ≥ 12 months with pre-surgical mental health included in multivariable analyses. One study reported data from a joint registry,125 one study included patients from multiple centres,128 and five studies collected data from patients treated at a single centre. 103,104,130,131,133 Study details are summarised in Table 4. In studies that reported the number of patients eligible, between 6% and 38% of patients were not followed up.

In a 12-month follow-up study of 6158 patients (92% eligible) from the Swedish Hip Arthroplasty Register, Rolfson and colleagues125 assessed the impact of the pre-operative European Quality of Life-5 Dimensions (EQ-5D) measure of anxiety and depression on long-term pain and satisfaction. In analysis of covariance adjusting for EQ-5D domains, comorbidities and age and sex, anxiety and depression were strong predictors of poor long-term pain relief and low patient satisfaction. Furthermore, in patients with persistent anxiety and depression, only 24% of patients showed improvement in the EQ-5D mobility dimension compared with 59% in those unaffected by high levels of anxiety or depression.

Judge and colleagues128 followed up 70% of 1375 patients eligible from seven UK centres 5 years after they had received a specific THR prosthesis. The multivariable model included the SF-36 mental health score, age, sex, primary diagnosis, occupation, comorbidities, HRQoL and pre-surgical OHS. Poorer mental health measured by the SF-36 mental health score was associated with a less favourable long-term patient outcome as measured by the OHS.

Anakwe and colleagues130 followed up 850 patients (94% eligible) 12 months after THR at a single UK centre. The Short Form questionnaire-12 items (SF-12) mental health component, diabetes, hypertension, history of depression, age, sex, SF-12 physical components, OHS, and musculoskeletal comorbidities were included in analyses. Although significant in univariate analyses, neither a history of depression nor the SF-12 mental health component predicted the level of long-term patient satisfaction in multivariable analysis.

Nilsdotter and colleagues103 followed up 198 patients with THR (94% eligible) at a single centre in Sweden after a mean of 3.6 years. Multivariable analysis included sex, comorbidities, WOMAC, SF-36 (including mental health), employment, marital status, contralateral osteoarthritis, need of walking assistance, walking distance, analgesic use and regional or widespread pain. In preliminary univariate analysis, pre-operative SF-36 mental health status was not a significant predictor of long-term WOMAC function and was not entered into multivariable analysis.

In a single-centre UK study, Clement and colleagues131 followed up 1312 patients with THR (per cent eligible not reported) at 12 months. In a multivariable analysis with age, deprivation, Charlson comorbidities, OHS, length of stay and SF-12 physical health, SF-12 mental health was a significant predictor of long-term change in OHS. Depression was the comorbidity with the strongest prediction of poor improvement in OHS.

Gandhi and colleagues133 followed up 636 patients (per cent eligible not reported) from a single centre in Canada for an average of 3.3 years. In multivariable analysis with age, sex, comorbidities, BMI and fixation (cemented or uncemented), SF-36 mental health was not associated with long-term changes in WOMAC score or SF-36 physical function.

In a single-centre US study, Singh and Lewallen104 followed up 5707 patients (62% eligible) 2 years after THR. Multivariable analyses included anxiety, depression, age, sex, comorbidities, operative diagnosis, distance from centre and implant design. Patients with depression but not anxiety [International Classification of Diseases (ICD) classifications] before surgery were more likely to report moderate to severe long-term pain at 2 years. A trend in a similar direction was not significant at 5 years.

All studies we included in the review reported multivariable analyses with inclusion of comorbidities and other factors in the statistical model. Authors examined outcomes in patients using generic measures of mental health (SF-12 mental health and SF-36 mental health), combined measures of anxiety and depression (EQ-5D anxiety/depression) and specific diagnoses of anxiety and depression (ICD code or clinical history).

In five studies in which generic mental health scores were measured before surgery using SF-36 or SF-12, results of multivariable analyses were inconsistent. 103,128,130,131,133 In two studies, patients with worse pre-surgical mental health scores had a poorer long-term outcome128 or less improvement measured by the OHS. 131 However, in two studies, SF-36 mental health scores measured before surgery did not predict long-term WOMAC function103 or change in overall WOMAC score. 133 Similarly pre-surgical SF-36 mental health did not predict long-term change in SF-36 physical function score. 133 In one study, the authors reported that long-term satisfaction was not predicted by pre-surgical SF-12 mental health score. 130 Inconsistencies between studies were not explained by differences in additional markers of quality.

Unlike the SF-36, which measures psychological distress and well-being, the mental health component of the EQ-5D relates specifically to anxiety and depression. In one study considered to be of good quality based on both additional markers, pre-operative anxiety/depression measured with the EQ-5D was associated with poorer long-term pain relief, satisfaction and mobility. 125 Three studies specifically reported outcomes in patients with anxiety or depression before surgery. In one study, patients with depression before surgery had poorer long-term pain outcomes but this was not the case for patients with anxiety. 104 In another study, depression before surgery was associated with poor improvement in OHS. 131 Both these studies were single centre and reported high losses to follow-up. In one study from a single centre but with low losses to follow-up, except in univariate analysis, patients with a history of depression did not report lower levels of long-term satisfaction after THR. 130

Overall, there was some evidence that patients with depression before THR may have poorer long-term outcomes but evidence for pre-surgical anxiety was weaker. For general mental health measures evidence was equivocal.

Pre-surgical pain

Pain is the principal indication for THR144 and the key patient expectations of surgery are good long-term functional and pain outcomes. 145,146 We identified two single-centre studies that had explored the relationship between pre-operative pain levels and long-term patient-reported outcomes after THR. 103,134 Details of studies are shown in Table 4.

In a study of 198 patients (94% eligible) from a single Swedish centre, Nilsdotter and colleagues103 reported a multivariable analysis with follow-up at a mean of 43 months. Increased levels of SF-36 bodily pain domain measured before surgery were associated with poorer long-term WOMAC function outcomes in a statistical model including age, sex, comorbidities, BMI and SF-36 physical function and mental health components.

Garbuz and colleagues134 followed up 147 patients (73% eligible) at a single Canadian centre 12 months after THR. Greater pre-operative pain measured with the WOMAC pain score was predictive of a poorer long-term WOMAC pain outcome in univariate analysis and after inclusion of age, sex and comorbidities in multivariable analysis.

Few studies have reported the relationship between pain levels before THR and long-term patient outcomes. In two studies with appropriate multivariable analyses, an increased level of pre-operative joint specific pain was associated with a greater risk of long-term pain and general pre-operative pain was predictive of poorer long-term functional outcome.

Pre-surgical physical function

As noted previously, the key expectations of patients undergoing total joint replacement are good long-term functional and pain outcomes. 145,146

Searches identified six studies that had followed up patients at 12 months and conducted multivariable analysis with pre-operative physical function included in the statistical model. Three studies were conducted in multiple centres,31,126,128 and three in single centres. 130,131,134 Characteristics of studies are summarised in Table 4.

In this classification we included four studies that reported overall WOMAC or OHS. As well as reflecting physical function these scores also include joint specific pain (OHS) or pain and stiffness (WOMAC).

Hajat and colleagues126 followed up 3600 patients (77% eligible) 12 months after THR at multiple UK centres. In multivariable analysis with age, sex, waiting time, comorbidities, housing and surgical factors, a worse pre-operative OHS was predictive of a worse long-term OHS.

In a European multicentre study, Judge and colleagues31 followed up 845 patients (64% eligible) with THR at 12 months. The authors included the WOMAC score, EQ-5D, age, sex, BMI, education, comorbidities and radiographic status in a multivariable model. A worse pre-operative WOMAC score was predictive of a poorer long-term outcome as judged by the Outcome Measures in Rheumatology-Osteoarthritis Research Society International (OMERACT-OARSI) rated good WOMAC response.

Judge and colleagues128 reported the 5-year follow-up of 1375 patients (70% eligible) who had received a specific THR prosthesis at seven UK centres. 128 In multivariable analyses including age, sex, BMI, primary diagnosis, occupation, comorbidities and HRQoL, the pre-operative OHS was the strongest determinant of long-term OHS.

Anakwe and colleagues130 reported a single UK centre study with 850 patients (94% eligible) followed up 12 months after THR. SF-12 physical and mental health components, diabetes, hypertension, history of depression, age, sex, OHS and musculoskeletal comorbidities were included in multivariable analysis. The SF-12 physical component score (PCS) measured before surgery did not predict long-term satisfaction. The pre-surgical OHS was associated with greater long-term dissatisfaction but only in univariate analysis.

In a study at a single Canadian centre, Garbuz and colleagues134 followed up 147 patients (73% eligible) 12 months after THR. In univariate analysis a 10-out-of-100-point difference in pre-surgical WOMAC function was associated with a 35% increase in long-term WOMAC function. In multivariable analysis age, sex, and comorbidity did not change the association.

Clement and colleagues131 reported 12-month follow-up of 1312 patients (per cent eligible not reported) with THR at a single UK centre. In a multivariable analysis with age, deprivation, comorbidities, length of stay and SF-12 physical and mental health, the pre-operative OHS was a strong predictor of long-term improvement in OHS.

The studies we identified suggested that better physical function before THR is associated with a better long-term functional outcome. This was apparent in one study with a specific functional measure134 and in four studies with a more general patient-reported outcome. 31,126,128,131 This observation was supported when considering studies with an additional marker of good quality. In one study with a more general measure of pre-operative physical function, there was no association with long-term satisfaction. 130

Body mass index

There is a strong association between high BMI (≥ 30 kg/m²) and the development of knee osteoarthritis. 137 People with high BMI are also more likely to require a TKR than those with lower BMI. 138,167 For example, in a cohort of 315,495 people in Norway, men and women in the top quartile of BMI (≥ 27.3 kg/m²) had relative risks compared with the lowest quartile (< 21.6 kg/m²) of undergoing a TKR of 6.16 (95% CI 4.23 to 8.95; p < 0.0001) and 11.06 (95% CI 7.83 to 15.62; p < 0.0001), respectively. 167

Analyses of the National Joint Registry for England and Wales3 show that 56% of patients undergoing TKR in 2012 had a BMI of 30 kg/m² or greater. In the RESTORE APEX RCT including 311 patients with TKR the mean BMI was 32.6 kg/m² (SD 6.5 kg/m², range 17.0–56.2 kg/m²). The distribution is shown in Figure 10. Thirteen per cent of patients had a BMI of 40 kg/m² or greater.

FIGURE 10.

Distribution of BMI in the APEX TKR cohort.

Although clearly of potential importance as a treatment for severe knee pain, irrespective of a patient’s BMI, concern has been expressed about the outcomes of TKR in patients considered to be obese. 137,168,169 Previous reviews have suggested that patients with a high BMI receiving TKR are at increased risk of deep infection,170–172 complications171–173 and need for revision surgery. 170,173 Long-term outcomes measured by surgeon assessed scores after TKR may be poorer in patients with a BMI of ≥ 40 kg/m² but not in patients with BMI of 30–40 kg/m². 173 No previous systematic review has considered the association of pre-operative BMI and long-term patient-reported outcomes.

Our systematic literature search identified 13 articles that met the inclusion criteria and were included in the review. 46,127,147,149,151–155,159–162 Details of these studies are summarised in Table 5. As a marker of generalisability, one study reported a registry analysis,147 eight studies included multiple centres,46,127,149,151–155 and four studies included patients from a single centre. 159–162

Pre-surgical mental health

In their systematic review with searches of MEDLINE and EMBASE to January 2011, Vissers and colleagues143 identified 19 studies including 6274 patients receiving TKR. There was strong evidence that pre-operative pain catastrophising was associated with increased pain in the first year after surgery and that poorer pre-operative mental health was associated with worse long-term physical function and pain. The authors identified no strong evidence that pre-operative depression influenced functioning in the year after TKR.

In our systematic review, we identified 12 studies with ≥ 100 patients with TKR followed up for ≥ 12 months with pre-surgical mental health included in multivariable analyses. 50,147,148,150,155–158,160,163–165 Two studies reported data from joint registries,147,148 five studies included patients from multiple centres150,155–158 and five studies collected data from patients treated at a single centre. 50,160,163–165 Study details are summarised in Table 5.

In an analysis of the National Joint Registry for England and Wales, Baker and colleagues148 reported on 22,691 patients (55% eligible) with TKR followed up for at least 6 months and up to 12 months after surgery. The authors included age, OKS, EQ-5D, disability, general health, comorbidities, and surgical and hospital variables in a multivariable model. Pre-operative anxiety and depression measured using the EQ-5D were associated with poorer long-term improvement in OKS and EQ-5D.

Franklin and colleagues147 followed up 8050 patients (47% eligible) from a US joint registry 12 months after TKR. In multivariable analysis, the authors included SF-12 PCS, sex, age, BMI, osteoarthritis diagnosis and poor quadriceps strength in the statistical model. The pre-operative SF-36 mental component score was an independent predictor of poor long-term physical function measured with the SF-36 PCS.

Heck and colleagues157 followed up 268 patients (92% eligible) 24 months after TKR at multiple US centres. The authors conducted a multivariable analysis with age, ethnicity, sex, poverty, patient health status, WOMAC scales, SF-36 physical component, knee ROM, comorbidities, surgical factors and joint problems in the other knee. Poor pre-operative SF-36 mental health was associated with less improvement in the SF-36 PCS.

Lingard and Riddle150 followed up 628 patients (70% eligible at 2 years) who had received a specific TKR prosthesis at 12 international centres. 150 In a multivariable model with age, sex and comorbidities, patients with psychological distress identified using the SF-36 mental health component had worse WOMAC pain at 12 and 24 months than non-distressed patients. There was no strong evidence to support such a relationship with long-term WOMAC function although there was a trend in a similar a direction. The authors reported that there were no strong associations between pre-operative psychological distress and changes in function and pain over either follow-up period.

In the study of Merle-Vincent and colleagues,158 264 patients (87% eligible) with TKR at multiple centres in France were followed up after 24 months. In a multivariable analysis with age, sex, BMI, Lequesne index and joint space narrowing, there was no strong evidence that patients with feelings of depression before surgery were more dissatisfied with their TKR.

Papakostidou and colleagues155 followed up 204 patients (90% eligible) at two centres in Greece, 12 months after TKR. In a multivariable analysis with age, sex, BMI, education, social support, place of residence and baseline status of knee, the extent of pre-operative depressive symptoms, measured with the 10-Item Center for Epidemiologic Studies Short Depression Scale, were predictive of long-term VAS pain. The authors concluded that depressed mood had a strong positive correlation with long-term pain and functional limitation.

Singh and Lewallen156 followed up 7139 patients (65% eligible at 2 years) with TKR at multiple US centres 2 years after surgery. 156 Pain outcome was measured using a standardised Mayo Clinic questionnaire. In a multivariable analysis with age, sex, BMI, operative diagnosis and comorbidities, anxiety identified in the comorbidity assessment was an independent predictor of moderate to severe pain at 2 years, whereas depression was marginally not. The authors also followed patients up at 5 years but losses to follow-up were greater than at 2 years. There was evidence for an association between both pre-operative anxiety and depression and moderate to severe pain at this longer follow-up.

In a study in 165 patients (per cent eligible not reported) with TKR at a single US centre, Ayers and colleagues163 reported follow-up at 12 months. In multivariable analysis with age, sex, pre-operative function and comorbidities, poorer pre-operative SF-36 emotional health was associated with smaller improvements in SF-36 PCS and WOMAC physical function scores.

Brander and colleagues50 followed up 116 patients (per cent eligible not reported) at a single US centre at 12 months after TKR. In multivariable analysis with age, sex, other demographics and physiological, psychometric and pain variables, pre-operative depression and anxiety were associated with greater long-term VAS pain.

In their study of 966 patients (per cent eligible not reported) with TKR followed up for 12 months at a single UK centre, Clement and colleagues164 reported a multivariable analysis including age, sex, comorbidities, socioeconomic deprivation, OKS and the SF-12 components. Poorer pre-operative SF-36 mental health was associated with a poorer long-term improvement in OKS but was not related to satisfaction.

Gandhi and colleagues160 reported the follow-up of 551 patients (per cent eligible not reported) with TKR at a single Canadian centre at a mean of 3 years. In multivariable analysis including age, sex, ethnicity, BMI, education and comorbidity, patients with poorer pre-operative mental health according to the SF-36 had worse long-term WOMAC and SF-36 functional outcomes.

Scott and colleagues165 reported 12-month follow-up at a single UK centre of 1414 patients (87% eligible) who had received a specific TKR prosthesis. 165 In multivariable analysis with age, sex, SF-12 physical component, OKS and comorbidities, patients with depression or a poor SF-12 mental health status were more likely to be dissatisfied with their long-term outcome.

Studies investigating the relationship between pre-operative mental health and long-term patient outcomes used generic measures, specifically the SF-36 and SF-12 mental health components or more specific measures of anxiety or depression. For generic measures, poor mental health status before TKR was a predictor of long-term increased pain or poorer function in all seven studies that reported it. 147,150,157,160,163–165 There was a consistent suggestion in three studies that patients with anxiety had worse long-term pain or other patient outcomes. 50,148,156 In five studies, patients with depression or depressive symptoms had poorer long-term pain or functional outcomes. 50,148,155,156,158 Associations between aspects of pre-operative mental health and long-term patient outcomes were apparent in studies with one or two additional markers of good-study quality. For satisfaction as an outcome, the associations with pre-operative mental health measures were inconsistent.

Pre-surgical pain

We identified five studies with ≥ 100 patients with TKR followed up for ≥ 12 months with pre-surgical pain included in multivariable analyses. Four studies included patients from multiple centres,46,153,155,157 and one study was based on analyses of patients from a single centre. 165 Study details are summarised in Table 5.

Heck and colleagues157 followed up 268 patients (92% eligible) at multiple US centres 24 months after TKR. The authors conducted a multivariable analysis with the SF-36 mental health component, age, ethnicity, sex, poverty, patient health status, WOMAC scales, SF-36 physical component, knee ROM, comorbidities, surgical factors and joint problems in the other knee. In this analysis pre-operative WOMAC pain was not a predictor of the long-term SF-36 physical component.

Lingard and colleagues46 followed up 678 patients (79% eligible) from multiple international centres at 12 and 24 months after TKR. In multivariable analysis with age, sex, income, education, BMI, flexion, country, centre and comorbidities, one of the strongest determinants of long-term WOMAC pain was the pre-operative WOMAC pain score. The authors considered the difference in WOMAC pain scores to be clinically important.

In a study of 204 patients (90% eligible) from multiple centres in Greece, Papakostidou and colleagues155 reported follow-up at 12 months after TKR. In multivariable analysis including depression, sex, BMI, education, social support, age, place of residence and baseline status of knee, pre-operative WOMAC pain predicted long-term WOMAC pain.

Sullivan and colleagues153 followed up 120 patients (per cent eligible not reported) 12 months after TKR conducted at multiple Canadian centres. Greater pain catastrophising and pain-related fear of movement were predictors of poorer long-term WOMAC pain and function outcomes in multivariable analysis including age, sex, BMI, comorbidities, surgical factors, pain catastrophising, pain-related fear of movement and depression.

Scott and colleagues165 followed up 1414 patients (87% eligible) at a single UK centre 12 months after receiving a specific TKR prosthesis. Although significantly associated with long-term dissatisfaction in univariate analysis, there was no association between pre-operative OKS pain and dissatisfaction in a multivariable mode including OKS function, SF-12 physical and mental components, age, sex and comorbidities.

Studies comparing pre- and postoperative pain measures consistently showed that patients receiving TKR with higher levels of pre-operative pain had worse long-term pain. All studies had one or two additional markers of good quality. Associations with long-term physical function and satisfaction were inconsistent.

Pre-surgical physical function

We identified five studies with ≥ 100 patients with TKR followed up for ≥ 12 months with pre-surgical physical function included in multivariable analyses. 46,157,163–165 Details of studies are summarised in Table 5. Two studies were conducted in multiple centres46,157 and three studies in a single centre each. 163–165

Heck and colleagues157 followed up 268 patients (92% eligible) 24 months after TKR at multiple US centres. The authors included WOMAC function in a multivariable model with SF-36 mental health, age, ethnicity, sex, poverty, patient health status, SF-36, knee ROM, comorbidities, surgical factors and joint problems in the other knee. Patients with poorer WOMAC function had the greatest long-term improvement in SF-36 PCS. The authors noted that patients who were more likely to show improvement to general health had functional impairment at the time of surgery.

Lingard and colleagues46 followed up 678 patients (86% at 1 year) from multiple international centres at 12 and 24 months after TKR. The authors included age, sex, income, education, BMI, flexion, country, centre and comorbidities in multivariable analysis. Poor pre-operative WOMAC function was an independent predictor of poor long-term WOMAC function outcome.

In a study including 165 patients (per cent eligible not reported) with TKR from a single US centre, Ayers and colleagues163 reported follow-up at 12 months. The authors included pre-operative WOMAC physical function, age, sex and comorbidities in multivariable analysis. The model with increasing age and poorer physical function predicted a poorer SF-36 and WOMAC physical function outcome.

Clement and colleagues164 followed up 966 patients (per cent eligible not reported) for 12 months after TKR at a single UK centre. 164 Patients with a better OKS score before surgery had better long-term OKS but were more likely to be dissatisfied with their operation in multivariable analyses with age, sex, comorbidities, socioeconomic deprivation, SF-12 physical component and back pain.

Scott and colleagues165 reported a 12-month follow-up of 1414 patients (87% eligible) who had received a specific TKR prosthesis at a single UK centre. In multivariable analysis with SF-12 mental component, age, sex, depression and comorbidities, lower pre-operative OKS function was associated with greater satisfaction.

The relationship between pre-operative physical function and long-term patient outcomes after TKR was complex. In three studies there was a simple association between low pre-operative function before surgery and a poor long-term functional outcome. However, in another study patients with worse pre-operative function had a greater improvement in physical function. In two studies those with lower pre-operative function were more likely to be satisfied with their operation. Results of studies were consistent in those with one or two additional markers of good quality.

Diabetes

In the APEX study, 6.7% of patients receiving a THR reported that they had type 1 or type 2 diabetes.

Searches identified four studies reporting long-term patient-reported outcomes in patients according to their pre-surgical diabetic status. Two studies each included patients from multiple centres,127,185 or a single centre. 130,166 The proportion of patients followed up varied from 6.3% to 49.9%, which was largely explained by lower follow-up rates in studies of longer duration.

In a study in two Canadian hospitals, Jones and colleagues127 followed up an estimated 167 patients (72% eligible) after THR. Diabetes mellitus was not a significant independent factor predicting recovery measured by WOMAC function or pain scores in a multivariable analysis including cardiac disease, age, sex, BMI, education, principal diagnosis, living arrangements, type of living accommodation, previous joint surgery, ambulatory status and number of comorbid conditions.

Judge and colleagues127 followed up 249 patients (50% eligible) from two UK health districts at a mean of 96 months. 185 In multivariable analysis with diabetes, hypertension, thyroid disease, age, sex, BMI, smoking habit, previous knee injury, Heberden’s nodes, number of painful joints and radiographic grade, the OR for no improvement in SF-36 physical function was 5.45 (95% CI 0.99 to 29.89) in patients with diabetes compared with non-diabetics. This was close to statistical significance. In another analysis of this cohort, the MD in change in SF-36 physical function was 25.8 points (95% CI 6.8 to 44.9 points) favouring patients with no diabetes. 184 However, both analyses were based on only eight patients with diabetes.

In a study at a single UK centre, Anakwe and colleagues130 followed up 850 patients (84% eligible) with THR for 12 months. Diabetes was not associated with long-term patient satisfaction in multivariable analyses including other comorbidities, age, sex, OHS, SF-12 physical and mental components and musculoskeletal comorbidities (p = 0.227).

Gandhi and colleagues166 followed up 707 patients (83% eligible) after a THR at a single Canadian centre. In a multivariable analysis with BMI, hypertension, hypercholesterolaemia, age, sex, WOMAC score, and cumulative illness rating scale, self-reported diabetic status did not predict the extent of WOMAC function improvement 12 months after THR (p = 0.46). However, in analyses considering the combination of medical conditions relating to metabolic syndrome (hypertension, obesity, hypercholesterolaemia and diabetes), presence of all four conditions was associated with a poorer long-term WOMAC score (p = 0.04).

The four studies we identified did not provide any conclusive evidence on the impact of diabetic status on long-term patient-reported outcomes. 127,130,166,185 This may be explained by good glycaemic management in patients with diabetes receiving THR in the cohorts we identified.

Patients with diabetes may be more likely to have complications after their THR. During the hospital stay, patients with diabetes may be at greater risk of stroke, pneumonia and requirement for a blood transfusion but not prosthetic joint infection. 186 In a literature search relating to infection including all studies reporting outcomes in patients according to diabetic status we identified 10 longitudinal studies with 376,138 patients in which 2487 deep infections were recorded. 186–195 As shown in Figure 11, the relative risk of developing a deep infection was 2.14 (95% CI 1.44 to 3.18) in patients with diabetes compared with non diabetics. We used a random-effects model owing to the high extent of heterogeneity (I² = 74%). This was largely explained by one large study with follow-up only to hospital discharge. 186

FIGURE 11.

Summary plot of deep infection complications after THR by diabetic status. df, degrees of freedom; M–H, Mantel–Haenszel.

In an analysis of the US National hospital discharge survey including 43,215 patients receiving any orthopaedic surgery, diabetes was associated with an increased risk of inpatient mortality in univariate analysis but this was not statistically significant in multivariable analysis (OR 1.23; p = 0.18). 196

In a study of 16,317 patients with total hip or knee replacement, there was an increased risk of myocardial infarction and venous thromboembolism after joint replacement in patients with diabetes or hypertension. 197 Considering the cluster of hypertension, diabetes, obesity and dyslipidaemia that constitute metabolic syndrome, the risk of myocardial infarction was increased by 128% and the risk of venous thromboembolism more than tripled.

In a US cohort of 1,030,013 patients receiving total hip or knee replacement, patients with diabetes were compared by the level of glycaemic control. 198 In those patients with uncontrolled diabetes, the risk of surgical and systemic complications was significantly higher and mortality greater than in those with controlled diabetes. Danish hip registry data from 3278 patients also suggests increased risk of revision due to deep infection in patients with diabetes, particularly in those with short diabetes duration or diabetic complications probably reflecting poor blood glucose control. 194

Cardiovascular disease

In the APEX study, 12.0% of patients receiving THR reported a cardiovascular-related condition.

Our searches identified one study in multiple centres with multivariable analysis assessing patient-reported outcomes according to presence of cardiovascular disease. 127

In the study of Jones and colleagues127 at two Canadian centres, an estimated 167 patients (72% eligible) with THR were followed up at 3 years. In an analysis adjusted for age, sex, BMI, education, principal diagnosis, living arrangements, type of living accommodation, previous joint surgery, ambulatory status and number of comorbid conditions, patients with cardiac disease had worse long-term WOMAC pain (p = 0.014) and function (p = 0.012), and a slower recovery.

The mortality rate within 90 days after THR is about 1%199 and cardiovascular disease is generally the leading cause of death. 43 Research in patients receiving a THR with comorbid cardiovascular disease has mainly focused on adverse events after surgery.

Ackland and colleagues200 developed a risk index based on patient history of ischaemic heart disease, heart failure and cardiac risk factors. This method was able to stratify patients with elective orthopaedic surgery by risk of in-hospital adverse events. However, with an area under the receiver operating characteristic curve of 0.62 (95% CI 0.57 to 0.67), the predictive ability was modest.

Bozic and colleagues201 developed a risk calculator for 90-day mortality and prosthetic joint infection in patients with THR. The risk model included age, sex, ethnicity and socioeconomic status and 29 pre-operative comorbidities including cardiopulmonary conditions. The authors concluded that levels of risk could be used to counsel patients with heart disease before surgery. Sanders and colleagues202 explored the effect of time since a vascular event (stroke, myocardial infarction and unstable angina) on outcomes of total hip and knee replacement in 414,985 patients. The authors concluded that patients with a vascular event in the year before surgery were at greatest risk of death within 30 days of surgery.

In an analysis considering long-term outcomes, Singh and Lewallen203 noted a greater risk of periprosthetic fractures in patients with heart disease identified before THR.

Hypertension

In a US study including 53,252 Medicare patients with THR, 66% of patients had hypertension before surgery. 201

Our searches identified three studies that included hypertension before surgery in multivariable analysis for the prediction of long-term patient-reported outcomes. 130,166,185

Cushnaghan and colleagues151 reported a multivariable analysis in 249 patients (50% eligible) from two UK health districts followed up for 96 months after THR. Hypertension was not a significant predictor of change in SF-36 physical function score, MD in change 0.5 points (95% CI –6.5 to 7.6 points).

In a single-centre UK study with 850 patients (94% eligible) followed up for 12 months, Anakwe and colleagues130 compared levels of satisfaction with THR in patients with and without hypertension. A total of 8.7% of hypertensive patients were dissatisfied with their THR compared with 6.4% of those without hypertension. This did not reach the specified level of statistical significance (p < 0.10) for inclusion in multivariable analysis.

In their study of metabolic syndrome, Gandhi and colleagues166 followed up 707 patients (83% eligible) 12 months after THR at a single Canadian centre. In multivariable analysis with BMI, self-reported diabetic status, hypercholesterolaemia, age, sex, WOMAC and comorbidity score, hypertension was a significant predictor of lower overall WOMAC improvement at 1 year (p = 0.006).

Overall, there was limited evidence that patients with hypertension identified before surgery have poorer long-term patient outcomes. As described previously, its importance may be as part of the cluster of cardiovascular risk factors that constitute metabolic syndrome.

Considering adverse outcomes after THR, Bozic and colleagues201 reported that patients with hypertension had an increased risk of mortality and joint infection. In a multivariable analysis, hypertension was an independent variable included in a risk equation to identify patients at risk of joint infection and mortality.

Anaemia

We did not identify any studies reporting the association between pre-surgical anaemia and long-term patient-reported outcomes after THR. Searches identified one systematic review by Spahn204 looking at studies of perioperative anaemia and clinical outcomes. The author concluded that pre-operative anaemia in patients with hip and knee surgery was associated with increased need for blood transfusion, infection and death, poorer physical functioning and recovery, and longer hospital stay.

More recent analyses of registries and large cohort studies support this review. Jämsen and colleagues205 reported a multivariable analysis comparing mortality rates in 1998 patients with hip and knee replacement. After extensive adjustment for patient and clinical characteristics including other comorbidities, the hazard ratio for death was 1.47 (95% CI 1.08 to 1.99) in patients with pre-operative anaemia compared with those without. In another multivariable analysis including 15,722 patients reported by Greenky and colleagues,206 pre-operative anaemia was associated with a greater risk of prosthetic joint infection but not mortality. In a multivariable analysis including 40,919 patients with THR followed up for 90 days, Bozic and colleagues207 reported that patients with pre-operative anaemia had a greater risk of prosthetic joint infection (hazard ratio 1.36, 95% CI 1.15 to 1.62). Pre-operative anaemia was not associated with mortality within 90 days in multivariable analysis in this cohort.

O’Malley and colleagues208 identified factors associated with major complications after THR leading to delays in hospital discharge. In multivariable analysis including BMI, pre-operative bleeding disorder, comorbidities and surgical factors, pre-operative anaemia was associated with increased risk of major complications before discharge. Major complications resulted in a mean increase in hospital stay of 62 days.

In a prospective analysis of data from patients with THR at a single centre, Myers and colleagues209 noted a higher infection rate and need for transfusion in patients with pre-clinical anaemia on admission. Inpatient stay was, on average, 18 days in patients with anaemia compared with 11 days in patients with no pre-clinical anaemia.

Research on anaemia and THR has focused on early adverse events and death. Patients with anaemia may be at greater risk of infection and other major complications, are more likely to require blood transfusion and may have a substantially increased length of hospital stay.

Thyroid disease

In the Whickham survey of men with a median age of 58 years and women with a median age of 59 years, the prevalence of hypothyroidism was 1.3% and 9.3%, respectively. 210 This compares with the prevalence of treatment for hypothyroidism in people with severe osteoarthritis of 7.0% and 14.0% in men and women, respectively, in the APEX cohort.

Our searches identified one study with long-term patient outcomes after THR according to whether or not patients had thyroid disease. 185 Details of the study are summarised in Table 7.

In the study of Judge and colleagues,185 249 patients (50% eligible) with THR in two UK health districts were followed up for a mean of approximately 96 months. In univariate analysis, there was no suggestion of a difference in long-term SF-36 physical function in patients with thyroid disease compared with people without thyroid disease (OR for no improvement or worse SF-36 physical function 0.96, 95% CI 0.36 to 2.53). The OR for a meaningful improvement in SF-36 physical function score (at least 30 points) was 0.33 (95% CI 0.07 to 1.49), a non-significant trend favouring patients with no thyroid disease.

Diabetes

Diabetes is a common comorbid condition in patients receiving TKR. In an Australian cohort of 1214 patients with knee replacement, the prevalence of diabetes was 17.0%. 211 Of 3,672,247 patients discharged from US hospitals after unilateral TKR, about 13% were diabetic. 212 In the APEX study, 14.1% of patients with TKR recruited into a UK RCT had type 1 or type 2 diabetes.

Six studies reported long-term patient outcomes according to pre-operative diabetic status. Of these, two were from multiple centres127,151 and four included patients from a single centre. 163–166

Cushnaghan and colleagues151 followed up 259 patients (39% eligible) from three UK health districts at a mean of 72 months after TKR. In multivariable analysis, a trend for greater improvement in SF-36 physical function in diabetic patients was not statistically significant (mean relative change 3.8, 95% CI –8.5 to 16.2). The analysis included only 16 patients with diabetes.

Jones and colleagues127 followed up an estimated 209 patients (72% eligible) after TKR at two Canadian centres. At 3 years, there was little to suggest that WOMAC pain and function differed according to diabetic status; however, the authors observed that pain and function scores worsened slightly after 6 months among diabetic patients but not in non-diabetics.

Ayers and colleagues163 followed up 165 patients (per cent eligible not reported) from a single US centre at 12 months after TKR. In multivariable analysis there was no association between diabetic status and long-term physical function measured by WOMAC and SF-36 scales.

In a study from a single UK centre with 1141 patients (87% eligible) followed up at 12 months, Scott and colleagues165 found no association between diabetic status and long-term satisfaction after TKR.

In a study from a single Canadian centre, Gandhi and colleagues166 followed up 889 patients (83% eligible) after TKR. Self-reported diabetic status showed a trend for poorer WOMAC score at 12 months but this was not statistically significant (p = 0.07). In analyses considering the combination of medical conditions relating to metabolic syndrome (hypertension, obesity, hypercholesterolaemia and diabetes), in the presence of all four conditions there was a trend for poorer long-term WOMAC score but this was not statistically significant (p = 0.08).

Clement and colleagues164 followed up 966 patients (per cent eligible not reported) at a single UK centre 12 months after TKR. In multivariable analysis, diabetes was not associated with a poorer long-term patient outcome (p = 0.47).

There is no strong evidence to suggest that long-term patient outcomes after TKR differ according to diabetic status; however, much research has focused on the risk of adverse events. In our literature search on the incidence of infection by diabetic status, we obtained data from 14 studies including 633,813 patients in which 3988 deep infections were recorded. 186,191,192,207,211,213–221 In a random-effects meta-analysis shown in Figure 12, the relative risk for infection in patients with diabetes compared with non-diabetics was 2.03 (95% CI 1.54 to 2.67).

FIGURE 12.

Summary plot of deep infection complication after TKR by diabetic status. df, degrees of freedom; M–H, Mantel–Haenszel.

In studies with more general orthopaedic inclusion criteria, patients with diabetes had an increased risk of myocardial infarction and venous thromboembolism. This was particularly evident in patients with the metabolic syndrome cluster of risk factors. 197 There was also a suggestion of an increased risk of inpatient mortality for patients with diabetes, but this was not statistically significant. 196

Cardiovascular disease

In the APEX study, 20.2% of patients receiving TKR reported a cardiovascular-related condition.

Our literature searches identified four studies reporting multivariable analysis of patient-reported outcomes according to presence of cardiovascular disease. One study was conducted in multiple centres127 while three studies included patients from a single centre. 163–165

Jones and colleagues127 followed up an estimated 209 patients (72% eligible) at 3 years after TKR at two centres. Cardiac disease before surgery was not a predictor of long-term WOMAC pain or function.

In the study of Ayers and colleagues,163 165 patients (per cent eligible not reported) with TKR at a single centre were followed up at 1 year. Only univariate analyses were reported for individual comorbid conditions. There was a lower improvement in the SF-36 physical component in patients with cardiovascular comorbidity than those without but the relationship was not strong. For WOMAC physical function, there was a suggestion that patients with cardiovascular comorbidity had a lesser improvement in physical function than those without.

Scott and colleagues165 followed up 1141 patients (87% eligible) with TKR at a single centre 12 months after surgery. In univariate analysis, there was little difference in long-term satisfaction in those with heart disease compared with those with no heart disease.

In the study of Clement and colleagues,164 966 patients (per cent eligible not reported) with TKR at a single centre were followed up for 12 months. Patients with a history of heart disease had a poorer improvement in OKS compared with those with no history of heart disease.

Overall, there was a suggestion that people with heart disease had somewhat poorer long-term improvement in pain and function, but this was only evident in two studies with no additional markers of good quality. Further research has studied the association between comorbid cardiac disease and adverse events.

Variables in the risk calculator for in-hospital adverse events described by Ackland and colleagues200 included specific cardiac diseases and risk factors but the discriminatory ability of the model was modest. Gill and colleagues222 looked at factors predicting mortality within 90 days of TKR in a cohort of 3048 patients, of whom 14 died. Patients with any major cardiovascular disease before surgery had a greater risk of death within 90 days of surgery. In the study of Sanders and colleagues,202 a greater risk of death within 30 days of surgery was observed in patients with a vascular event in the year preceding total hip and knee replacement.

Hypertension

Searches identified four studies reporting long-term patient outcomes according to whether or not patients had hypertension before surgery. Studies were in multiple151 or single centres. 164–166

In a study in three UK health districts, Cushnaghan and colleagues151 followed up 259 patients (39% eligible) from three centres at a mean of 72 months after TKR. In multivariable analysis, hypertension was not a significant predictor of change in SF-36 physical function score.

Gandhi and colleagues166 followed up 889 patients (83% eligible) after a TKR at a single Canadian centre. Similar WOMAC function improvement was observed at 1 year, irrespective of presence of hypertension. The authors noted a trend for poorer long-term WOMAC score in patients with the four conditions relating to metabolic syndrome but this was not statistically significant.

Scott and colleagues165 followed up 1141 patients (87% eligible) from a single UK centre 12 months after TKR. In univariate analysis, more patients with hypertension were dissatisfied with their outcome than those with no hypertension.

In a study at a single UK centre, Clement and colleagues164 followed up 966 patients (per cent eligible not reported) 12 months after TKR. There was no difference in change in OKS in patients with hypertension compared with those with no hypertension.

Considering adverse events, Bozic and colleagues223 followed up 83,011 patients with TKR 90 days after surgery. In multivariable analysis, hypertension identified pre-operatively was not associated with infection or mortality.

There was no clear evidence that patients with hypertension had worse long-term pain and function outcomes after TKR. As observed with diabetes, management of blood pressure may affect studies of the association between pre-operative hypertension and post-surgical outcomes. Most people with a diagnosis of hypertension will be treated before surgery; indeed management of undiagnosed hypertension may be required in the preparation for elective surgery.

Anaemia

We identified one study with patients from a single centre that reported long-term patient outcomes after TKR according to pre-operative anaemia status. 164

Clement and colleagues164 followed up 966 patients (per cent eligible not reported) 12 months after TKR at a single UK centre. There was a small difference in improvement in OKS between patients favouring those with no anaemia but the relationship was not strong.

The systematic review of Spahn204 considered clinical outcomes of patients with hip and knee surgery. Pre-operative anaemia was associated with post-surgical mortality and infection, increased need for blood transfusion, poorer physical functioning and recovery, and longer hospital stay.

More recently, Bozic and colleagues223 followed up 83,011 patients with TKR for 90 days after surgery. In multivariable analysis, pre-operative anaemia was associated with an increased risk of infection but not mortality.

Jämsen and colleagues205 reported an increased risk of death in 1998 patients receiving knee or hip replacement with pre-operative anaemia compared with those without. In a multivariable analysis including patients with total hip or knee replacement, Greenky and colleagues206 reported a greater risk of infection but not mortality.

In summary, there was no strong evidence relating pre-operative anaemia to long-term patient outcomes. However, by reducing adverse events and limiting hospital stay, strategies to manage anaemia before surgery may have substantial benefits for both patients and health-care providers.

Thyroid disease

Searches identified one study conducted in multiple centres reporting patient-reported outcomes in patients with or without thyroid disease who received a TKR. Cushnaghan and colleagues151 followed up 259 patients (39% eligible) in three English health districts at a mean of 72 months after TKR. In multivariable analysis, thyroid disease was not a significant predictor of change in the SF-36 physical component.

Psychological impact

Whether or not they experienced delay, participants all described emotional reactions to the experience of waiting.

Participants on pathways B, C and E who experienced postponement of their surgery date experienced a range of emotional reactions. Frustration and disappointment were frequently reported; however, some participants expressed understanding and acceptance of postponement. They tried to rationalise the news and described their gratitude when delay was a matter of weeks rather than months. The wait for surgery can also have more detrimental emotional consequences, for instance leading to a feeling of helplessness and utter desperation. Participants talked of feeling as if they were ‘left to linger’ in the secondary care system and forced to ‘live in limbo’.

Participants on pathway A without any delay or changes to their surgery date also described some disappointment at the length of time that they had had to wait for their operation. They recalled their dissatisfaction that their operations were not scheduled as soon as they had hoped and the disruption to their lives caused by this. The impact of delay on physical well-being and functioning also had some psychological effects. Mr Rayner’s experience provides a useful example. He experienced recurrent postponement of his operation owing to investigative but inconclusive tests for additional health problems. While waiting for THR he then experienced a rapid deterioration in his general health and functional well-being, and expressed ‘frustration’ at the situation.

Wider impact

Accounts illustrated how cancellation of operations could have wider impact, particularly affecting support networks. Participants explained how their friends and family also had to cope with and manage the participants’ own deteriorating health as well as share in their disappointment about a delay to the surgery date. Participants described how family and friends put their lives on hold and had to cope with the detrimental impact of ‘living in limbo’. Cancellation and delay also had a more practical impact as friends and family had to renegotiate their own plans to remain supportive during the perioperative period.

Consequences of preceding pathway on current experience

Participants’ views of their wait for surgery and the detrimental impact of the waiting period were influenced by their experience of time already spent living with pain and discomfort. Many participants described complex and sometimes slow journeys through health care from initial onset of their problems through to eventual referral to secondary care.

Two key factors were central in accounts of patients’ journeys through care. First, participants reported that they had initially delayed seeking advice and support from primary care practitioners. Reasons cited for this included a fear of the undesirable inevitable (i.e. surgery); other priorities (e.g. caring for a sick spouse); and choosing to ‘put up’ with pain and discomfort. Second, many participants reported the sense that referral from primary to secondary care had been delayed. Explanations provided for this delay included receiving an incorrect diagnosis; that their GP saw them as ‘too young’ to undergo THR; and that their GP strongly advocated alternative strategies (e.g. weight loss, use of pain relief). Participants’ accounts highlighted the interaction between frustration with management in primary care and subsequent impatience with the time spent waiting in secondary care.

Unavoidable changes to use and passage of time in the lead-up to surgery

The progression of time from onset of osteoarthritis towards THR was marked by the experience of pain and patients made unavoidable changes to their use of time (Box 2).

Participants’ accounts highlighted how they had been living with pain for long periods of time and many had experienced deterioration and acceleration of their problems in the lead-up to surgery. As well as describing the long periods for which they had lived with pain, participants also described the experience of pain in terms of time. They described fleeting spasms (a ‘horrible twang’) that lasted seconds as well as constant, unrelenting pain or ‘throbbing ache’. Living a life in continuous pain with no respite appeared to provide the sense that time was drawn out during the lead-up to, and wait for, surgery.

Participants described making unavoidable and considerable changes to the way in which they spent their time because of pain and functional difficulties. When waiting for their operation, participants described withdrawal from their everyday activities and no longer actively engaging with life. For example, participants talked of inability to walk or stand for long because doing so resulted in ‘unbearable pain’. Many talked of giving up pastimes that had previously provided much pleasure, such as golf and gardening. Most of those who had been working had stopped doing so. Participants found themselves progressively unable to fill and enjoy their time as they once had done and, instead, described how they often found themselves ‘sitting down and doing nothing’ in their homes rather than actively engaging with life and ‘doing’. They talked about feelings of ‘lost time’ and a sense that time had slowed. The sense was in part due to the long process before diagnosis and then referral to secondary care. For some, such as Mr Horton, this was seen as a failure to diagnose the problem and working ‘along the wrong lines’.

In addition, the accounts of some participants indicated that day and night became conflated in the lead-up to surgery. While they would have previously been awake in the daytime, some reported resting and sleeping during the day to seek relief from the exhaustion of living in pain and due to side effects of pain medication. Participants also described pain onset or increasing intensity at night, which regularly woke them up or kept them awake. Unable to sleep well at night and sleeping during the day, participants experienced time slowing down as they waited for surgery and expressed distress, isolation and upset.

Time in the context of health care

Participants reported that their journeys through health care to arrive at surgery were lengthy, partly because they did not necessarily seek help for joint problems, but also because of slow processes of referral from primary to secondary care (Box 3). Such delays earlier in their journey through health care could influence the experience and impact of the wait for surgery once within the secondary care system.

Once in the secondary care system and before admission to hospital for surgery, participants’ time was also increasingly filled with activities relating to surgery. These included trips to hospital for pre-operative education, assessments and consultations, and tasks relating to psychological and physical preparation. These activities focused participants’ attention on, and heightened their awareness of, the upcoming date in their calendar and also meant that they had to arrange their other commitments and activities.

Participants also described how time in the lead-up to surgery did not always pass at a regular, steady pace. They felt that the date of their planned admission to hospital could appear closer or further away. Some participants felt that the date was approaching too quickly and this evoked anxiety and nervousness. Others were eager and impatient to have surgery and expectations about how long their wait would be were influenced by discussions with health-care professionals in secondary care. However, some were also unsure about how long they would have to wait and thought that the timing of their operation was not static, but would be changed. Participants described uncertainty and, in relation to their experience of time, how they had to put their lives on hold when waiting for surgery.

Participants also experienced changes to their admission date and this could be due to hospital factors, ill health or the option of a date change by choosing to change surgeon. However, the impact of these factors was sometimes complicated; for instance, one participant accepted the offer to change her surgeon because it would mean that she did not have as long to wait for her operation. The new date for her operation was then postponed because her glycaemic control for type 2 diabetes was not deemed adequate for surgery. A referral ‘back’ to primary care aimed to ensure support to achieve better glycaemic control, which was then followed by a wait to re-enter the secondary care system and to learn her surgery date.

‘I’ve got faith in him’: trust and confidence in the surgical team

Participants who were undergoing knee replacement expressed considerable ‘trust’ and ‘faith’ in surgical teams. This seemed to relate to their experience of living with osteoarthritis, in which participants dealt with increasing pain and impaired mobility. By the time that they had reached secondary care, many felt that they had no choice but to rely on medical opinion and expertise, and that surgery was inevitable. This was rooted in previous positive encounters in consultations as well as experiences (their own and others) of successful outcomes after other types of surgery.

Contact with secondary care team

After having their operation, participants’ contact with secondary care health professionals shifted from the surgical team to a team of nurses, physiotherapists and auxiliaries. Relationships with health professionals also changed; participants wanted to receive support and guidance rather than the total control that they had wanted from surgical teams. After returning home, all participants had some contact, although often infrequent, with secondary care professionals – predominantly consultants and sometimes physiotherapists. They were ‘keen’ to receive follow-up appointments and ‘eager’ to obtain clinical opinion about aspects of the recovery process, for instance when they should stop using walking aids or return to leisure activities. These interactions bolstered confidence and offered reassurance.

‘You were sort of cut adrift’: unmet support needs during the recovery process

Participants also spoke of unmet support needs during the recovery process. For example, many felt that input from physiotherapists was received too late in the recovery process and that earlier involvement would have helped to reduce feelings of abandonment, enhanced motivation to exercise and facilitated earlier recovery. Postoperative aftercare in the community was also described as lacking. After discharge from hospital one participant, a widow, was not offered the support of a district nurse. Instead she described struggling with her own care, such as changing her surgical stockings, and had to implement her own support by paying for help. This participant felt that more formal support would have provided reassurance and reduced her feelings of isolation.

Differing perceptions of expertise: primary versus secondary care health professionals

Although confidence in secondary care health professionals was consistently high, participants trust in and willingness to seek support from primary health-care professionals was more mixed. Perceptions of support on offer and that received from primary care during the postoperative period was influenced by experience of care received before surgery. For example, some participants expressed dissatisfaction with primary care before surgery, describing their sense that care and advice had been inconsistent. When patients felt that primary care had not been helpful before their surgery, they were less likely to seek support actively from primary care on return home from hospital (Box 10).

Changes in level and type of assistance

There were changes in level and type of assistance provided and received at different points in the journey through knee replacement. Before surgery, participants described the importance of help provided by family and friends for everyday activities (such as fetching groceries and household chores). Immediately after surgery, the need for this kind of support sharply increased, with support needed for many more daily tasks, such as carrying a drink. Undergoing surgery also led to changes in participants’ roles in their relationships and family units. For example, one participant described how, when recovering from surgery, he was looked after by his children who were ‘running errands and things’ and felt that his wife treated him ‘like a baby’. These changes sometimes evoked negative emotions, including despondency and helplessness. However, as recovery from surgery progressed, the need for support and associated sense of helplessness reduced.

‘She’s always there you know when I want her’: the assumption that family will help

For participants who were married, help often came from spouses. Married participants (n = 7) who all lived with their spouses initially turned to their spouses for assistance during the journey through knee replacement. After surgery, in addition to spouses taking on increased responsibility for tasks relating to everyday living and functioning, they also assumed additional caring responsibilities, including help with personal care (e.g. helping to bathe). Some also played a role in medical aspects of the recovery process. For example, Mr Armstrong’s wife administered postoperative anticlotting injections and Mr Clark’s wife, a retired nurse, removed his stiches. Although many participants appeared comfortable in accepting that their spouse was occupying this novel role, others felt ‘awkward’ and embarrassed at asking spouses to undertake duties that they felt should be provided by paid professionals.

Use of extended informal support networks

Participants looked outside their immediate household to other family members (e.g. children, grandchildren, siblings) and friends to meet their postoperative support needs when they could not be fulfilled by a spouse. For example, several participants were the only driver in a household and this meant that friends and family were asked to drive on their behalf while postoperative restrictions were still in place (patients are currently told not to drive for 6–8 weeks after knee replacement). Participants who did not live with spouses asked friends and family for help, particularly in the early postoperative period.

‘I’m lucky’: willingness to accept help

Participants expressed mixed emotions about the help they received from their friends and family around the time of surgery. Several viewed such support as ‘helpful’ and ‘invaluable’ and felt fortunate to be ‘spoilt’ by friends and family who ‘rallied’ around them. Although surgery was often seen as a way of maintaining independence the time around surgery was a period when help and support from others was a necessity. Many participants craved their return to self-sufficiency and most participants did eventually regain the independence that they had sought (Box 11).

Hopes and expectations relating to long-term pain after joint replacement

All participants hoped that hip or knee replacement would reduce their pain (Box 13). For most, this was a key motivating factor for their decision to have the operation. Some participants described awareness that long-term pain was a potential issue after surgery. However, some thought this might be mild while many hoped to achieve complete freedom from pain in their operated joint in the year after surgery. Expectations relating to pain were based on previous personal experience of undergoing joint replacement surgery, knowledge of others’ experiences – both successful and unsuccessful – and information resources provided by the hospital. Most participants also described a hope that they would be able to stop use of pain relief and anti-inflammatory medication in the year after surgery. However, some also thought that they would have to continue using medication to manage pain in other parts of their body. Participants also described their hopes for the positive benefits of reduced pain in the longer term after surgery, for instance improved mood and enhanced sleep quality.

Expectations relating to postoperative function

Before surgery participants described living with restrictions on movement and mobility, reliance on walking aids and inability to ‘do a lot physically nowadays’ (Box 14). They anticipated that joint replacement would confer better function and bring about a future in which they would be able to ‘get around easier’, walk further and with a ‘normal stride’, navigate stairs and steps with more confidence and ease, and achieve independence from walking aids. They also spoke of hoping to once again be able to kneel down, bend down and reach their toes and have the capability to return to riding a bike again.

As a result of these changed functional abilities, participants anticipated that they would, in the year following surgery, be able to become more active and enhance their general level of health and fitness. However, several participants also acknowledged that there would continue to be some restrictions on their physical capabilities. For example, they spoke of how they would have to continue to avoid lifting heavy objects even in the long term and would also be unable to return to playing high-impact sports. They also believed that although walks to the local shops would become a future reality, hill trekking and climbing mountains would not. In addition, expectations relating to postoperative function were tempered by a sense that they may not achieve the level of function that they had before the onset of the problems with their hip or knee joint. This was informed by the sense that they continued to age and that some lived with problems in other parts of their body.

Like their expectations for pain, hopes for postoperative function were informed and shaped by their own and others’ experiences of undergoing similar medical interventions, in addition to information received around the time of surgery.

Expectations for changed engagement in social, work and life activities

Participants described expectations for participation in social and work activities after hip or knee replacement (Box 15). These expectations were related to anticipation of reduced pain and increased function and were particularly driven by observations of how well others had recovered from similar operations. For example, participants hoped to return to work and looked forward to attending social clubs again, meeting friends for lunch and other valued activities including golf, bowls, bell ringing and ballroom dancing. They also hoped to be able to take holidays once more, have day trips out and travel to see friends and family, both within the UK and abroad. Participants had to stop or limit these kinds of activities before surgery because of difficulties relating to their osteoarthritis.

Participants’ accounts also showed how they hoped that joint replacement would provide them with the chance to engage with life once again: ‘to be able to get out again’, to ‘go out and enjoy themselves’, ‘get on with their life’ and to ‘get back their quality of life’. For some, this also meant taking up new hobbies and interests such as joining a walking club and starting voluntary work. Many participants talked of their hope to be able to participate in these activities by certain points in time, for instance Christmas or their birthday. However, a few participants did not put time markers on when, and if, their goals would be achieved, talking of ‘just having to wait and see’ and ‘just depending on how I get on’.

Fulfilment of expectations relating to pain

Twelve months after surgery, 15 participants described complete relief from pain in their operated joint (Box 16). This group talked of being ‘absolutely over the moon’ with this outcome and of how they now ‘felt a hundred times better’. The operation had surpassed their expectations and they revealed how it was a ‘wonderful’ relief to be free of long-term pain. They described how it was ‘lovely’ to ‘move around without pain’; for example, how they were now able to bend down and walk without experiencing a ‘terrible pain’. As a result, as hoped, those in this group who were not living with pain in other parts of their body had halted their use of pain relief. However, 12 participants said that although they were free from the pre-operative intensity of pain arising from the grating of ‘bone on bone’, they continued to experience discomfort, soreness, tenderness, a dull ache or twinges in the area of their operated joint. They attributed this pain to a variety of causes. For example, some said that the discomfort was caused by their muscles tightening, while others thought that their recovery from surgery was not yet complete. Participants with ongoing problems were also able to identify triggers that intensified these sensations, such as exercise, ‘stretching themselves too much’, moving from sitting to standing, or standing in the same position for too long. Although living with ongoing discomfort, most members still said that they were ‘grateful’ for the treatment that they had. In keeping with attitudes to pain relief medication before surgery, participants in this group did not see pain to be ‘bad enough’ to warrant taking pain relief medication; however, two knee replacement participants described continued experience of a more ‘severe pain’ and they expressed frustration and unhappiness with this outcome. Again, as they did during the pre-operative interview, participants reflected on how previous experiences of joint replacement – their own and others’, in addition to information received around the time of surgery, played an important role in their expectations relating to pain after surgery.

Fulfilment of expectations relating to postoperative function

Participants spoke of how a reduction in the pain experience meant that at 12 months post operation they were able to move their body around with less difficulty and to walk further than they could before having joint replacement surgery (Box 17). They were also able to ride a bicycle, more confidently navigate steps and stairs and had experienced the anticipated independence from walking aids. However, this was not a universal outcome, with walking aids still used by some participants when they walked for any distance as they continued to offer reassurance, as they had done before surgery. Participants also talked of ways in which they continued to experience a lack of freedom and restrictions on the way in which they were able to move their body. For example, some highlighted how they were unable to run or experienced difficulties when bending down to the floor to, for example, pick up objects. Accounts also showed how participants continued to experience difficulties in walking up hills, how their walking pace had slowed and that they were unable to cope with longer walks, as they had hoped. Many of those who had undergone knee replacement surgery spoke of how they were now unable to kneel down. Nonetheless, these restrictions on movement, for most, did not seem to interfere with satisfaction with recovery and were perceived as ‘no major hindrance’. A perception among participants that general ageing also played a significant role in limiting the overall potential for movement and mobility can perhaps help to explain this view.

Fulfilment of expectations relating to engagement in social, work and life activities: ‘I’ve got my life back’

Participants talked of the ways in which their lives had positively changed since undergoing joint replacement surgery (Box 18). They described how they were now ‘more active’ and how life had become ‘more enjoyable’ as a result of having the operation. They spoke of having ‘a new lease of life’, of ‘making up for lost time’ and of how they were able to actively engage with more activities in each day now. As hoped, they had returned to many of the activities that they undertook prior to the onset of the difficulties with their hip or knee joint – a return to employment, social clubs, gardening, playing skittles, improved intimate relationships with partners and had already enjoyed holidays and trips to see family and friends. Planning for, and engaging in, these activities provided them with a psychological ‘lift’ and they talked of feeling ‘more positive’ and ‘optimistic’ about the future. However, for some participants, their pre-operative expectations to engage in particular social activities (e.g. taking holidays and games of golf) once they had recovered from their joint replacement surgery, were tempered, or had to be put on hold, because of other health conditions. In addition, a few participants who did not talk of additional health conditions also revealed that they continued to hold themselves back from undertaking the hobbies and activities that they had previously enjoyed and had hoped to return to (e.g. ballroom dancing). This was attributed to a lack of confidence in their new joint and concerns relating to falling.

Results

Relationship between measures of impairment and participation restrictions

Correlations between measures of impairment and participation restrictions (Ab-P) are displayed in Table 23. Hip and knee ROM correlated poorly with participation restrictions (|0.06| to |0.32|). In comparison, correlations between pain and participation restrictions were high in patients listed for hip replacement (–0.71) and moderate in patients listed for knee replacement (–0.53), and these correlations remained strong after adjustment for demographic factors.

Comparison of functional measures between patients with low and high active flexion

Patients listed for knee replacement with low flexion had significantly worse results on all measures of impairment, activity limitations and participation restrictions than patients with high flexion (Table 24). Patients listed for hip replacement with low flexion had significantly worse activity limitations than patients with high flexion.

Change in scores

The scores for the various functional measures, at baseline and 12 months post operation, are shown in Tables 25 and 26. Tables 27 and 28 also record the changes in each of the scores between baseline and 12 months.

Table 25 demonstrates that, as expected, scores on the function measures improve from baseline to 12 months. People in this cohort had better function after surgery than before they had the surgical intervention.

Table 26 shows similar results, suggesting that people’s function improves after surgery. However, a particular feature of these data is the number of participants who were able to complete certain tasks. Nearly everyone could complete most of the tests, both before and after surgery. However, two of the tests appear to be more discriminatory – the 30-cm step test and the balance test. As shown in Table 26, only 61.5% of people with hip disease could do the 30-cm step test prior to surgery, and this improved to 82.4% postoperatively; the equivalent figures for those with knee disease were 40.0% and 70.3%, respectively. The balance test was also difficult for patients with knee disease; only 38.6% of patients could perform the test pre-operatively and 46.0% postoperatively. Those with hip disease could do a little better – 48.1% were able to do it pre-operatively, and 60.6% postoperatively.

Change for better or worse and clinically important improvements

The above scores tell only a part of the story. They do not provide information about what proportion of people got better or worse, or whether or not the improvements were clinically, as well as statistically, important. In order to answer these questions, we investigated the following:

1. the number of patients with scores that indicated improvement, deterioration or no change in function between baseline and 12 months

2. the numbers of patients in the improved or deteriorated function categories whose changes were statistically significant

3. the proportion of patients for whom these changes reached clinical significance, using the anchoring satisfaction question.

The results of these analyses are shown in Tables 27 and 28. The data also allow us to further compare the degree of improvement following knee or hip replacement.

As shown in Tables 27 and 28, there were a small number of people whose function did not change or deteriorate after surgery. Overall, each of the functional assessment methods is telling a similar story of a few people getting worse (but rarely significantly so) and most getting better, often significantly better. The walking time was more likely to show deterioration than other tests and arguably the ‘get-up-and-go test’ showed more differentiation between changes for the better or worse than other tests. Overall patients having hip surgery were more likely to improve functionally than those having knee replacement, with most of the measures used, although it is interesting to note that walking speed changes were very similar in both groups.

We also looked for ceiling effects on each of the functional measures at baseline and 12 months, as shown in Tables 29 and 30.

The striking finding here is that many of the self-assessment questionnaires that are routinely used to assess function in people with lower limb osteoarthritis show an important ceiling effect in response to joint replacement. The problem is particularly evident in the assessment of function after hip replacement; the WOMAC function subscale, SF-12 physical function measure and all three domains of the Aberdeen scale all reach a maximum score in between 20% and 50% of patients postoperatively. Rather fewer patients reached the maximum score after knee surgery, but the problem is a very real one for this intervention as well (between 8% and 22% of patients reaching the maximum on one or other of the scores). Timed tests such as the walk time or ‘get-up-and-go’ test, cannot, by definition, suffer from this problem and it is interesting to note that very few people achieved ‘top marks’ on either the HHS or AKSS.