Rheumatology 2019;58:1168–1175 RHEUMATOLOGY doi:10.1093/rheumatology/key424 Advance Access publication 10 January 2019 Original article Impact of TNF inhibitor therapy on joint replacement rates in rheumatoid arthritis: a matched cohort analysis of BSRBR-RA UK registry data Samuel Hawley1, M. Sanni Ali1,2, Rene Cordtz3,4, Lene Dreyer5,6, Christopher J. Edwards7, Nigel K. Arden8,9, Cyrus Cooper8,9, Andrew Judge1,9,10, Kimme Hyrich 11,12 and Daniel Prieto-Alhambra1,13 Abstract Objectives. Previous ecological data suggest a decline in the need for joint replacements in RA patients following the introduction of TNF inhibitor (TNFi) therapy, although patient-level data are lacking. Our primary aim was to estimate the association between TNFi use and subsequent incidence of total hip replacement (THR) and total knee replacement. Methods. A propensity score matched cohort was analysed using the British Society for Rheumatology Biologics Registry (2001–2016) for RA data. Propensity score estimates were used to match TNFi users to similar conventional synthetic DMARD users (with replacement) using a 1:1 ratio. Weighted multivariable Cox regression was used to estimate the impact of TNFi on study outcomes. Effect modification by baseline age and disease severity were investigated. Joint replacement at other sites was also analysed. An instrumental variable sensitivity analysis was also performed. Results. The matched analysis contained a total of 19 116 patient records. Overall, there was no significant association between TNFi use vs conventional synthetic DMARD on rates of THR (hazard ratios = 0.86 [95% CI: 0.60, 1.22]) although there was significant effect modification by age (P < 0.001). TNFi was associated with a reduction in THR among those >60 years old (hazard ratio = 0.60 [CI: 0.41, 0.87]) but not in younger patients. No significant associations were found for total knee replacement or other joint replacement. Conclusion. Overall, no association was found between the use of TNFi and subsequent incidence of joint replace- ment. However, TNFi was associated with a 40% relative reduction in THR rates among older patients. Key words: epidemiology, rheumatoid arthritis, biologics, TNF inhibitor, joint replacement, total hip replace- CLINICAL SCIENCE ment, total knee replacement, comparative effectiveness Rheumatology key messages . Overall TNFi use was not associated with subsequent rates of joint replacement. . Among elderly patients, TNFi use was associated with a 40% reduction in subsequent THR rates. . Given prior ecological data, future studies are needed to confirm and/or further elucidate the relationship between TNFi and joint replacement. Introduction patients’ joints, subsequent hallmarks of which include structural damage to cartilage and bone [1–3]. Joint Rheumatoid arthritis (RA) is a chronic autoimmune dis- damage is a central feature of RA and has been estimated ease driven by pathological inflammatory processes in Lifecourse Epidemiology Unit, University of Southampton, Southampton, 10 Translational Health Sciences, University of Bristol, Bristol, 11NIHR 1 Pharmaco- and Device-Epidemiology Group, Centre for Statistics in Manchester Biomedical Research Centre, NHS Foundation Trust, Medicine, NDORMS, University of Oxford, Oxford, 2Faculty of Manchester University, 12Division of Musculoskeletal & Dermatological Epidemiology and Population Health, London School of Hygiene and Sciences, Arthritis Research UK Centre for Epidemiology, University of Tropical Medicine, London, UK, 3Centre for Rheumatology and Spine Manchester, Manchester, UK and 13GREMPAL Research Group, Idiap Diseases, Gentofte, Rigshospitalet, 4The Parker Institute, Copenhagen Jordi Gol and CIBERFes, Unviersitat Autonoma de Barcelona and University Hospital Copenhagen, 5Department of Rheumatology and Insituto de Salud Carlos III, Barcelona, Spain Clinical Medicine, Aalborg University Hospital, 6Department of Clinical Medicine, Aalborg University, Aalborg, Denmark, 7NIHR Clinical Submitted 20 July 2018; accepted 3 November 2018 Research Facility, University Hospital Southampton, Southampton, 8 Correspondence to: Daniel Prieto-Alhambra, Botnar Research Centre, Nuffield Department of Orthopaedics, Rheumatology and Windmill Road, OX3 7LD Oxford, UK. Musculoskeletal Sciences, University of Oxford, Oxford, 9MRC E-mail: daniel.prietoalhambra@ndorms.ox.ac.uk ! The Author(s) 2019. Published by Oxford University Press on behalf of the British Society for Rheumatology. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Impact of TNFi therapy on joint replacement to account for 25% of disability in established disease with each treatment lasting 56 months. The BSRBR-RA [4]. The permanent and often progressive nature of joint contains an additional comparator cohort of non-biologic damage, in conjunction with associated pain, loss of func- treated RA patients on csDMARDs, entry into which was tion and failure to adequately respond to therapeutic op- dependent on having active disease (guide DAS28 > 4.2). tions are strong indications for eventual joint replacement Recruitment to this cohort closed in 2008. surgery [5, 6]. Outcomes for hip and knee replacement in Participants in all study cohorts are followed up indef- RA are generally considered good [7], although it has been initely using physician questionnaires sent from BSRBR- observed that such patients are at increased risk of vari- RA to the patients’ rheumatology clinic. These were sent ous adverse events compared with patients undergoing every six months for a patients’ first three years of follow- these procedures for osteoarthritis, including dislocation, up, and annually thereafter. In addition to collecting data infection, myocardial infarction and revision [8, 9]. on changes to therapy and disease details, physicians Furthermore, there are significant healthcare costs, were asked what serious/adverse events had occurred which in the UK are £6000 – £7000 per operation [10]. since the last follow-up date. In addition, patients were The prevention of irreversible joint damage through early asked to complete a health diary every six months for and aggressive management using pharmacotherapy has the first three years in the study in which they detailed been well demonstrated and is therefore recommended in any hospital admissions. These free text responses were numerous national guidelines [11–13]. First-line therapy coded by BSRBR-RA staff using the MedDRA hierarchy options include various conventional synthetic DMARDs (Medical Dictionary for Regulatory Activities [18]). (csDMARDs). Mortality data for participants were obtained from the Over recent decades, the emergence of biologic thera- Health and Social Care Information Centre (now merged pies such as TNF inhibitors (TNFi) has revolutionized the into NHS digital) via alerts generated from Office for management of RA as these drugs are widely recognized National Statistics records. The BSRBR has ethical ap- to improve numerous outcomes of the disease, including proval from the North West Multicentre Research Ethics joint damage [14, 15]. Despite this, to our knowledge there Committee (reference number MREC 00/08/053) and pa- are no randomised controlled trials (RCT) studies address- tients gave written informed consent to participate in the ing the issue of whether biologic therapies prevent/delay BSRBR; no further ethical approvals were required to ultimate joint failure, as indicated by the need for a joint undertake this analysis. replacement. Recent ecological data from the UK and Denmark have indicated a reduction in the incidence of Outcomes knee replacement among RA patients following the intro- The primary outcomes of interest were first occurrence of duction of TNFi in 2002/2003 [16, 17], however concurrent THR or TKR, analysed separately. Patients were followed to this has been an increasing emphasis on early and up from date of registration into BSRBR-RA until the ear- more aggressive usage of csDMARDs. Patient-level data liest date of either outcome event, follow-up form date on this topic is required to disentangle these issues, yet indicating change in TNFi exposure status (stopping such data remain scarce. among TNFi users or starting biologics among csDMARD Our current aim was to estimate the comparative effect- users), last follow-up form or death. Other joint replace- iveness of TNFi vs csDMARDs on subsequent rates of ment (OJR) (a composite outcome consisting of elbow, total hip replacement (THR) and total knee replacement shoulder, hand or other small joint replacement) was a sec- (TKR) among a large cohort of RA patients. ondary outcome. Study population Methods Our study sample (Supplementary File 1, available at Data sources and exposures Rheumatology online) consisted of all biologic-naı̈ve RA We obtained data from the British Society for patients, either in the control cohort or those initiating a Rheumatology Biologics Register for RA (BSRBR-RA). TNFi (etanercept, infliximab or adalimumab) no more than This register contains prospectively collected observational 6 months prior to registration within BSRBR-RA. Patients data on over 20 000 RA patients recruited from 2001 on- with a THR or TKR recorded prior to registration were wards, primarily in order to evaluate the real-world safety of excluded, as were patients with <6 months of follow-up TNFi. The initial UK national institute for health and care (i.e. those who did not return at least one follow-up ques- excellence (NICE) guidance on use of TNFi stipulated clin- tionnaire). In analyses of OJR, further exclusion was made icians initiating such therapy must register the patient into of patients who had undergone an OJR prior to baseline. the BSRBR-RA, and recruitment continued originally until the target of at least 4000 patients per TNFi cohort (etaner- Statistical analysis cept, infliximab or adalimumab) was reached (last ‘original’ Owing to confounding by indication, i.e. TNFi users vs patient recruited 2008) but reopened recruitment to these csDMARD users likely having a different baseline risk of three originator drugs from 2010 onwards. NICE guidance THR/TKR, we decided a priori to match TNFi users to restricts National Health Service prescribing of TNFi to pa- csDMARD users based on their propensity for receiving tients with a sustained 28-joint DAS (DAS28) >5.1 who treatment. Propensity scores (PS), i.e. the probability of have failed to adequately respond to two csDMARDs, receiving treatment conditional on observed baseline https://academic.oup.com/rheumatology 1169 Samuel Hawley et al. characteristics (including those predictive of outcome) matching with replacement [20] owing to fewer available were estimated for all patients using logistic regression. csDMARD patients than TNFi users in the register. The list of potential confounders included in the PS equa- Missing data were imputed using chained equations and tion (and described in Table 1) consisted of: age, gender, 10 imputed datasets were created. Statistical analyses ethnicity, index of multiple deprivation (socio-economic were carried out in Stata 15.1 and R. status), BMI, smoking status, year of registration (quin- Baseline characteristics of the TNFi user and csDMARD tiles), time since RA diagnosis, DAS28, HAQ score, quality cohorts were summarized and differences assessed by of life (SF36 domains), 1987 ACR criteria, systemic in- way of standardized mean differences [21, 22], with smal- volvement, co-morbidities and co-medications. A full de- ler values indicative of greater similarity between cohorts. scription of these variables are included in Supplementary This assessment was carried out for the cohort prior to File 2, available at Rheumatology online. We matched matching and in the 10th imputed dataset. Incidence rates each TNFi patient to the csDMARD patient with the of each outcome event with 95% CIs were calculated most similar PS within a caliper distance of 0.2 standard among matched TNFi users and csDMARD users. deviations of the logit of the PS [19]. Patients falling out- Weighted Cox regression was used to compare THR, side this common support region remained unmatched TKR and OJR rates, taking into account the number of and were excluded from further analysis. We used times each individual csDMARD user was included by TABLE 1 Baseline characteristics of propensity score matched cohortsa: stratified by use of TNFi vs conventional synthetic DMARDs CS-DMARD (n = 9558) (1644 unique patients) TNFi (n = 9558) Characteristic n % n % SMD Age, mean (S.D.) 55.2 (12.1) 55.2 (12.3) 0 Gender: female, % 7289 76.3 7259 75.9 0.01 Ethnicity: white/Caucasian, % 9114 95.4 9118 95.4 0.00 Index multiple deprivation Quintile 1 1282 13.4 1322 13.8 0.01 Quintile 2 1420 14.9 1485 15.5 0.02 Quintile 3 1413 14.8 1640 17.2 0.07 Quintile 4 1544 16.2 1710 17.9 0.05 Quintile 5 1284 13.4 1650 17.3 0.11 Unknown 2615 27.4 1751 18.3 0.22 BMI 26.8 (5.9) 27.1 (6.3) 0.06 Smoking? % Current 2448 25.6 2259 23.6 0.05 % Ex 3272 34.2 3577 37.4 0.07 Calendar period of registration Oct 01–Oct 03 1481 15.5 2262 23.7 0.21 Nov 03–Aug 04 2037 21.3 2037 21.3 0 Sept 04–Aug 05 1823 19.1 1802 18.9 0.01 Aug 05–May 07 1664 17.4 1462 15.3 0.06 May 07–May 16 2553 26.7 1995 20.9 0.14 Years since diagnosis, median (IQR) 10.8 (10.7) 11.0 (8.8) 0.02 DAS 28, mean (S.D.) 6.47 (1.09) 6.43 (0.98) 0.04 Overall HAQ score 1.91 (0.63) 1.91 (0.62) 0.01 ACR: ever rheumatoid positive, % 5532 57.9 6055 63.4 0.11 ACR deformity of 53 joint areas?, % 7425 77.7 8118 84.9 0.19 ACR: erosions on hands/feet, % 4651 48.7 5282 55.3 0.13 ACR: ever had nodules, % 4034 42.2 3988 41.7 0.01 ACR: symmetry, % 7468 78.1 7883 82.5 0.11 ACR: deformity of hand joint, % 6771 70.8 7602 79.5 0.20 ACR: morning stiffness >1 h, % 8901 93.1 8966 93.8 0.03 Non-major prior joint replacementb 1742 18.2 1989 20.8 0.07 a Results shown are for the 10th imputed dataset. Matching was performed using replacement of the cs-DMARD users. 9558 biologic users were each matched to one of the 3229 cs-DMARD users (with replacement). Number of csDMARD patients represented in final matched sample was 1644. bComposite variable consisting of: shoulder, elbow, neck or other small joint replacement (e.g. hand). SMD: standardized mean difference (smaller values indicative of better balance). 1170 https://academic.oup.com/rheumatology Impact of TNFi therapy on joint replacement the matching with replacement, and adjusting the stand- Total hip replacement ard error of the estimates accordingly. We censored all A total of 589 THRs were reported during follow-up patients at 12 years due to the small and unstable size (median = 4.94 years [IQR: 1.52–10.04] for TNFi and of the csDMARD cohort after this time. Baseline charac- 5.97 years [IQR: 2.05–9.55] for csDMARD) of the propen- teristics that were not sufficiently similar post-matching, sity-matched cohorts. Incidence rates (per 1000 PYs) defined as standardized mean difference >0.1 (21), were were 5.22 [95% CI: 4.66, 5.88] and 6.30 (95% CI: 4.24, entered into a final multivariable Cox model for further 9.76) among TNFi users and csDMARD users, respect- adjustment [23]. Final models were run for all 10 datasets ively (Supplementary File 4, available at Rheumatology created in the multiple imputation process and hazard online). ratios (HRs) were pooled using Rubin’s rules. Comparing TNFi to csDMARDs yielded a pooled HR = Age and disease severity were a priori specified as po- 0.91 [95% CI: 0.64, 1.31; P = 0.62], which when adjusted tential effect modifiers of the association between TNFi for any remaining post-matching imbalance in baseline use and subsequent need for joint replacement. We covariates was 0.86 [95% CI: 0.60, 1.22; P = 0.39] (Fig. 1). tested for these by including interaction terms for approxi- mately median age (5.1 NICE cut-off) in the weighted Cox model and used the Total knee replacement likelihood ratio test to assess model fit. In the event of a Among the matched sample, a total of 864 TKRs were significant interaction (P < 0.1), matching and survival reported during followup (median = 4.85 years [IQR: models were re-run stratified by the significant effect 1.50–10.01] for TNFi and 5.98 years [IQR: 2.03–9.55] for modifier. csDMARD) of the propensity-matched cohorts. Incidence rates (per 1000 PYs) were 8.89 [95% CI: 8.13, 9.72] and Sensitivity analyses 8.09 [95% CI: 5.32, 12.89] among TNFi users and csDMARD users, respectively (Supplementary File 4, We addressed the potential of unobserved confounding available at Rheumatology online). This yielded a pooled through the use of an instrumental variable (IV) approach, HR = 1.18 [95% CI: 0.90, 1.56; P = 0.24], which when using physician preference as the instrument, as has been adjusted for any remaining post-matching imbalance in done previously [24, 25]. Details of this sensitivity analysis baseline covariates was 1.11 [95% CI: 0.84, 1.47; P = are described in Supplementary File 3, available at 0.46] (Fig. 1). Rheumatology online. We also repeated the main PS ana- lysis after excluding TNFi users recruited into the registry after the csDMARD cohort had closed, in order to max- Other joint replacement imize comparability between groups. Among the matched sample, a total of 336 OJRs occurred during follow-up (median = 4.93 years [IQR: 1.52–10.02] for TNFi and 5.98 years [IQR: 2.05–9.12] for csDMARD) Results among the propensity-matched cohorts. Incidence rates Of 13 126 eligible RA patients identified in BSRBR-RA, (per 1000 PYs) were 4.34 [95% CI: 3.76, 5.02] and 3.87 97% (9558) of the TNFi users and 51% (1644) of the [95% CI: 1.97, 8.73] among TNFi and csDMARD users, csDMARD users were retained following PS matching respectively (Supplementary File 4, available at (Supplementary File 1, available at Rheumatology Rheumatology online). There was no significant difference online). Given the 1:1 matching with replacement, a total in OJR rates between the exposure cohorts (Fig. 1) of 19 116 patient records were used in subsequent ana- lyses, with each csDMARD user being used a median of three (interquartile range (IQR): 1–6) times. Baseline characteristics of TNFi users were markedly FIG. 1 Estimated impact of TNFi on subsequent joint different in the unmatched study sample compared with replacement rates among matched TNFi and csDMARD the csDMARD cohort (Supplementary File 2, available at patients Rheumatology online), especially in aspects of disease severity. Specifically, the TNFi cohort had on average higher DAS28, HAQ score, proportion fulfilling the 1987 ACR RA criteria, lower health-related quality of life (as per SF36), longer disease duration and a higher preva- lence of prior non-major joint replacement. Conversely, baseline characteristics between exposure cohorts were much more similar post-matching (Table 1). The only per- sisting differences (standardized mean difference > 0.1) between the matched cohorts were calendar period of registration, low deprivation and the proportion of patients fulfilling ACR criteria. csDMARD: conventional synthetic DMARD. https://academic.oup.com/rheumatology 1171 Samuel Hawley et al. FIG. 2 Estimated impact of TNFi on subsequent joint re- (P = 0.008) in THR incidence among patients 560 years placement rates among matched TNFi and csDMARD old (Fig. 2), although non-significant increases were patients: stratified by age observed in TKR for the same age group and in THR for those <60 years old. Findings in context Our overall incidence rates of THR and TKR (results not shown) of 4.95/1000 PYs and 7.84/1000 PYs, respect- ively, align well with previous estimates of joint replace- ment among RA patients within the UK [16] and elsewhere [17, 26, 27]. Emerging observational data indicate that the number and/or incidence of RA-related joint surgery has been in decline across numerous developed countries [16, 17, 27–37], although this has primarily been seen for smaller joints [35, 37–39]. Many of these studies have inferred a possible role of biologics in this decline. Indeed, a reduc- csDMARD: conventional synthetic DMARD. tion in need for joint replacement associated with TNFi use would be an expected finding given previous evi- dence of TNFi use reducing joint damage as measured Interactions radiographically [15]. A previous meta-analysis of 70 RCT studies reported that annual radiographic progres- Age was found to be a significant (P < 0.001) effect modi- sion was 0.6% less in patients treated with biologics com- fier for both THR and TKR outcomes, although disease pared with those on a single DMARD [40]. Similarly, severity was not (P > 0.1). In subsequent stratified ana- another meta-analysis has demonstrated that patients lyses (Supplementary File 5, available at Rheumatology on initial combination therapy (methotrexate plus a bio- online), TNFi was associated with an estimated 40% re- logic agent) are 30% more likely to experience non-pro- duction in incidence of THR among older patients (HR = gression at 1-year than those on methotrexate alone [41]. 0.60 [95% CI: 0.41, 0.87; P = 0.008]) (Fig. 2). Differences in In this context, a 40% reduction in THR rates associated THR or TKR incidence rates between TNFi and csDMARD with TNFi use among a more elderly subgroup of patients cohorts among younger patients were non-significant as found in the present study is quite plausible, and a lack (Fig. 2, Supplementary File 5, available at Rheumatology online). of translation of positive findings on joint erosion from prior RCTs into a ‘real world’ reduction in rates of joint Sensitivity analyses replacement within our main study sample is initially surprising. Results were unchanged when PS matching and subse- However, a more detailed examination of prior RCT quent survival analysis was repeated following exclusion findings indicate that an expectation of widespread reduc- of (n = 1213) patients recruited into the TNFi cohort after tion in joint replacement associated with TNFi use is po- the csDMARD cohort had closed. Comparative effective- tentially unwarranted. There is a large degree of variation ness estimates were HR = 0.81 (95% CI: 0.55, 1.18) for in the nature of comparator groups used in prior studies THR and 1.11 (95% CI: 0.83, 1.50) for TKR. Results of [42, 43], and this could be an important factor in consider- the sensitivity IV analysis (Supplementary Files 6 & 7, ing the lack of effect as described in our main findings. For available at Rheumatology online) confirmed main find- instance, while RCT studies have shown reduced radio- ings, with a borderline reduction in risk of THR (absolute graphic progression among biologic users vs csDMARD risk reduction of –1.88 per 100 patients [95% CI: –3.86, monotherapy, an almost equal reduction has been 0.10; P = 0.063]) but no association between biologics use achieved among combination csDMARD users relative and TKR risk (–0.57 per 100 patients [95% CI: –2.69, 1.54; to csDMARD monotherapy [40]. Similarly, while TNFi has P = 0.60]). been shown to confer early benefits over combination csDMARD therapy, these benefits have been reported to Discussion disappear during the second year of follow-up [44], pos- sibly due to time-to-efficacy and time to achieve maximal Main findings dose of csDMARDs. The use of etanercept (vs oral triple We have sought to address the scarcity of patient-level therapy) resulted in only small radiographic benefits in an- data comparing use of TNFi vs csDMARDs on rates of other trial [45], with another showing triple therapy to be joint replacement in RA. Using a large UK-based RA bio- non-inferior to biologics in terms of change in DAS28 at logics register, overall we found no difference in subse- 48 weeks [46]. Given these previous data, it could be that quent rates of joint replacement between PS matched the reduced THR rates in older patients may reflect a gen- TNFi and csDMARD users. When stratified by age, TNFi eral improvement in management of RA over the past was associated with a significant 40% reduction 20 years and earlier and more aggressive use of 1172 https://academic.oup.com/rheumatology Impact of TNFi therapy on joint replacement csDMARDs rather than solely the effect of TNFi. obtain unbiased estimates in the presence of unmeas- Intriguingly, a registry-based study on the topic [47] re- ured confounding. The PS analytical approach taken, in cently found increased rates of major joint replacement which comparable csDMARD matches were found for associated with use of biologic therapy, although the au- each TNFi user—while much improving the internal val- thors concluded residual confounding was an issue given idity—does mean our findings are not average treatment the small number of confounding factors for which the effects generalizable to the entire RA population but an analysis was adjusted. estimate of the ‘average treatment effect on the treated’ We found a reduction in THR incidence among an older [48]. This is evident given that the csDMARD users patient subgroup, but no significant impact on TKR inci- included in the matched sample had more established dence, which is interesting as one might expect any effect and severe disease (thereby making them comparable to to be expressed approximately equally at hip and knee. It the TNFi sample) than the unmatched csDMARD sample could be that the relatively long disease duration at our (Supplementary File 2, available at Rheumatology baseline meant there was greater potential for prevention online). The study findings should also be interpreted in of joint destruction at the hip over knee, although details the context of a relatively long median disease duration of differential natural history of RA disease at these two at baseline (Table 1), which means they are not neces- joints are not well established. It is also very difficult to sarily generalizable to the context of early RA. Finally, we disentangle the impact of TNFi on improved function and relied on a combination of physician-reported and self- overall quality of life and how this may have mediated reported incidence of THR, TKR and OJR for our study effects on longer-term progression of joint damage, po- outcomes as per BSRBR-RA follow-up questionnaires. tentially differentially at the knee and hip. Another factor This may have introduced bias if events were under-re- could be the role of trauma related THRs among the older ported, although this would likely act non-differentially in subgroup and whether there may be some pathway to regard to TNFi status and minimally during the early reduced THR rates via TNFi associated improvement in years of follow-up, during which time study participants bone quality. It should be emphasized, however, that the were sent questionnaires every six months. These vari- positive impact on THR incidence was only observed in a ous limitations to the present analysis may partly explain subgroup of patients that have not been well studied in differences in the results obtained here compared with this regard previously and that prior studies at the popu- previous ecological data on widespread reductions in lation-level have identified different patterns in this regard, joint replacement rates in RA during the biologic era. some finding reduced rates of TKR [16, 17] and others While a reduction in THR amongst older TNFi users THR [27, 34] following introduction of TNFi offers some support for biologics playing a role in redu- cing need for joint replacement, it must also be noted Limitations that the lack of an overall protective effect is suggestive The potential for residual confounding by indication is a that other factors apart from TNFi are likely to be key limitation of the current study. Given that prior to involved in the aforementioned downward population matching there was a much higher disease severity trends in joint replacement rates in RA. among the TNFi group (Supplementary File 2, available at Rheumatology online), we cannot rule out that the Strengths overall lack of reduction in joint replacement rates Our study’s key strengths are that BSRBR-RA is one of among TNFi users may be due in part to a greater preva- the largest RA registers in the world, which made it pos- lence of unmeasured aspects of disease severity and sible for us to adjust for many potential confounders and unresponsiveness to therapy in this group, thereby main- stratify analyses where there was significant effect modi- taining a baseline ‘disadvantage’ even after PS match- fication. We were also able to accurately censor follow- ing. On the other hand, estimates of the impact of TNFi up given the linkage to Health and Social Care exposure may be subject to a general healthy user bias Information Centre mortality data. The use of PS match- in that a clinician perceives sufficient patient ability to ing is a strong method for dealing with bias [19] arising tolerate and benefit from more intensive therapy regi- from likely confounding by TNFi indication, and we were mens, which may have here contributed to the reduced able to reach good balance between the two exposure rate of THR in the older TNFi cohort. The use of an IV groups in terms of baseline characteristics across most approach as a sensitivity analysis sought to address the variables. issue of unmeasured confounding, in which the treat- ment effect was estimated using clinician preference for biologics as an IV, assuming this to be a strong pre- Conclusion dictor of exposure but unrelated/weakly-related to con- In this large prospective study, we found no overall asso- founders of the TNFi–joint replacement relationship. ciation between TNFi vs csDMARD therapy and subse- However, the findings of this IV approach should be in- quent incidence of joint replacement among RA terpreted with caution given the instrument was here patients, although a 40% relative reduction in THR rates associated with several measured confounders was found among older patients. Future studies are (Supplementary File 3, available at Rheumatology needed to confirm and/or further elucidate the relationship online) which may undermine its validity as a means to between TNFi use and joint replacement. https://academic.oup.com/rheumatology 1173 Samuel Hawley et al. Acknowledgements science overview. BMC Musculoskelet Disord 2016;17:463. The authors acknowledge and thank all the consultant 6 Nikiphorou E, Norton S, Young A et al. Association be- rheumatologists in the UK who contributed data to the tween rheumatoid arthritis disease activity, progression of BSRBR: RA. The data that support the findings of this functional limitation and long-term risk of orthopaedic study are available from the British Society for surgery: combined analysis of two prospective cohorts Rheumatology Biologics Registry, but restrictions apply supports EULAR treat to target DAS thresholds. Ann to the availability of these data, which were used under Rheum Dis 2016;75:2080–6. license for the current study, and so are not publicly avail- 7 Nelissen RG. The impact of total joint replacement in able. Study concept: all authors; data acquisition: all au- rheumatoid arthritis. Best Pract Res Clin Rheumatol thors; data analysis: S.H., M.S.A., D.P.A.; drafting 2003;17:831–46. manuscript: S.H.; revising and approving final manuscript: 8 Burn E, Edwards CJ, Murray DW et al. The impact of all authors. rheumatoid arthritis on the risk of adverse events following joint replacement: a real-world cohort study. Clin Funding: D.P.A. is funded by a National Institute for Health Epidemiol 2018;10:697–704. Research (NIHR) Clinician Scientist Award (project 9 Ravi B, Escott B, Shah PS et al. A systematic review and number CS-2013–13-012). Support was also received meta-analysis comparing complications following total from Oxford NIHR Biomedical Research Unit. A.J. is sup- joint arthroplasty for rheumatoid arthritis versus for ported by the NIHR Biomedical Research Centre at the osteoarthritis. Arthritis Rheum 2012;64:3839–49. University Hospitals Bristol NHS Foundation Trust and the 10 Burn E, Edwards CJ, Murray DW et al. Trends and deter- University of Bristol. The views and opinions expressed in minants of length of stay and hospital reimbursement fol- the manuscript are those of the authors and do not ne- lowing knee and hip replacement: evidence from linked cessarily reflect those of the NIHR, NHS or the primary care and NHS hospital records from 1997 to 2014. Department of Health. BMJ Open 2018;8:e019146. 11 American College of Rheumatology Subcommittee on Disclosure statement: S.H., R.C., C.C., M.S.A. and K.H Rheumatoid Arthritis Guidelines. Guidelines for the man- have nothing to disclose. L.D. reports personal fees agement of rheumatoid arthritis: 2002 update. Arthritis from UCB Pharma. and from MSD. C.J.E. has been a Rheum 2002;46:328–46. speaker for, received honoraria or research support from 12 Luqmani R, Hennell S, Estrach C et al. British society for Abbvie, BMS, Celgene, Pfizer, Biogen, Mundipharma, rheumatology and british health professionals in rheuma- UCB Pharma, Roche, MSD. N.K.A. reports grants from tology guideline for the management of rheumatoid arth- Bioberica and personal fees from Bioventus, from ritis (the first two years). Rheumatology 2006;45:1167–9. Regeneration, and from Smith&Nephew. A.J. has received 13 Kwoh CK, Simms RW, Anderson LG et al. Guidelines for consultancy fees from Freshfields Bruckhaus Deringer, the management of rheumatoid arthritis. Arthritis Rheum and is a member of the Data Safety and Monitoring 1996;39:713–22. Board (which involved receipt of fees) from Anthera 14 Curtis JR, Singh JA. Use of biologics in rheumatoid arth- Pharmaceuticals outside the submitted work. D.P.A.’s ritis: current and emerging paradigms of care. Clin Ther group have received unrestricted research grants from 2011;33:679–707. Servier Laboratoires, Amgen and UCB Pharma. 15 Scott DL. Biologics-based therapy for the treatment of rheumatoid arthritis. Clin Pharmacol Ther 2012;91:30–43. 16 Hawley S, Cordtz R, Lene D et al. Association between Supplementary data NICE guidance on biologic therapies with rates of hip and knee replacement among rheumatoid arthritis patients in Supplementary data are available at Rheumatology online. England and Wales: an interrupted time-series analysis. Semin Arthritis Rheum 2018;47:605–10. References 17 Cordtz RL, Hawley S, Prieto-Alhambra D et al. Incidence of hip and knee replacement in patients with rheumatoid 1 Scott DL, Wolfe F, Huizinga TW. Rheumatoid arthritis. arthritis following the introduction of biological DMARDs: Lancet 2010;376:1094–108. an interrupted time-series analysis using nationwide 2 Feldmann M, Brennan FM, Maini RN. Rheumatoid arthritis. Danish healthcare registers. Ann Rheum Dis Cell 1996;85:307–10. 2018;77:684–9. 3 Schett G, Gravallese E. Bone erosion in rheumatoid arth- 18 MedDRA. Welcome to MedDRA. https://www.meddra.org ritis: mechanisms, diagnosis and treatment. Nat Rev (12 September 2018, date last accessed). Rheumatol 2012;8:656–64. 19 Austin PC. The performance of different propensity-score 4 Scott DL, Pugner K, Kaarela K et al. The links between methods for estimating relative risks. J Clin Epidemiol joint damage and disability in rheumatoid arthritis. 2008;61:537–45. Rheumatology 2000;39:122–32. 20 Stuart EA. Matching methods for causal inference: a 5 Gademan MG, Hofstede SN, Vliet Vlieland TP, Nelissen review and a look forward. Stat Sci 2010;25:1–21. RG, Marang-van de Mheen PJ. Indication criteria for total 21 Austin PC. Using the standardized difference to compare hip or knee arthroplasty in osteoarthritis: a state-of-the- the prevalence of a binary variable between two groups in 1174 https://academic.oup.com/rheumatology Impact of TNFi therapy on joint replacement observational research. Commun Stat Simul Comput in clinical practice (1995-2010). Rheumatology 2009;38:1228–34. 2015;54:666–71. 22 Austin PC. Balance diagnostics for comparing the distri- 37 Singh J, Young B, Watson S, Perez J, McGwin G, Ponce B. bution of baseline covariates between treatment groups in Trends in joint replacements surgery in patients with propensity-score matched samples. Stat Med rheumatoid arthritis. Arthritis Rheumatol 2016;68(Suppl 10). 2009;28:3083–107. https://acrabstracts.org/abstract/trends-in-joint-replace- ment-surgery-in-patients-with-rheumatoid-arthritis/ (22 23 Nguyen TL, Collins GS, Spence J et al. Double-adjustment December 2018, date last accessed). in propensity score matching analysis: choosing a threshold for considering residual imbalance. BMC Med 38 Nikiphorou E, Carpenter L, Morris S et al. Hand and foot Res Methodol 2017;17:78. surgery rates in rheumatoid arthritis have declined from 1986 to 2011, but large-joint replacement rates remain 24 Ertefaie A, Small DS, Flory JH, Hennessy S. A tutorial on unchanged results from two UK inception cohorts. Arthritis the use of instrumental variables in pharmacoepidemiol- Rheumatol 2014;66:1081–9. ogy. Pharmacoepidemiol Drug Saf 2017;26:357–67. 39 Cordtz RL, Hawley S, Prieto-Alhambra D, et al. Changes in 25 Brookhart MA, Wang PS, Solomon DH, Schneeweiss S. incidence of shoulder, elbow, wrist and finger replacement Evaluating short-term drug effects using a physician- surgery among rheumatoid arthritis patients following the specific prescribing preference as an instrumental introduction of biological dmards: an interrupted time variable. Epidemiology 2006;17:268–75. series analysis using Danish health care registers. Ann 26 Rodriguez-Rodriguez L, Ivorra-Cortes J, Leon L et al. Rheum Dis 2018;77:903. Temporal trends of total joint replacement in rheumatoid 40 Graudal N, Jürgens G. Similar effects of disease-modify- arthritis patients: a survival study. Arthritis Rheumatol ing antirheumatic drugs, glucocorticoids, and biologic 2016;abstract 1530. agents on radiographic progression in rheumatoid arth- 27 Hekmat K, Jacobsson L, Nilsson JA et al. Decrease in the ritis: meta-analysis of 70 randomized placebo-controlled incidence of total hip arthroplasties in patients with or drug-controlled studies, including 112 comparisons. rheumatoid arthritis–results from a well defined population Arthritis Rheum 2010;62:2852–63. in south Sweden. Arthritis Res Ther 2011;13:R67. 41 Kuriya B, Arkema EV, Bykerk VP, Keystone EC. Efficacy of 28 National Joint Registry for England and Wales: 1st Annual initial methotrexate monotherapy versus combination Report. 2004. http://www.njrcentre.org. therapy with a biological agent in early rheumatoid arth- uk/njrcentre/Portals/0/Documents/England/Reports/NJ- ritis: a meta-analysis of clinical and radiographic remis- R_AR_1.pdf (June 2018, date last accessed). sion. Ann Rheum Dis 2010;69:1298–304. 29 National Joint Registry for England and Wales: 5th Annual 42 Thorlund K, Druyts E, Avina-Zubieta JA, Wu P, Mills EJ. Report. 2008. http://www.njrcentre.org. Why the findings of published multiple treatment compari- uk/njrcentre/Portals/0/Documents/England/Reports/5t- son meta-analyses of biologic treatments for rheumatoid h%20Annual.pdf (June 2018, date last accessed). arthritis are different: an overview of recurrent methodo- logical shortcomings. Ann Rheum Dis 2013;72:1524–35. 30 Louie GH, Ward MM. Changes in the rates of joint surgery among patients with rheumatoid arthritis in California, 43 Estellat C, Ravaud P. Lack of head-to-head trials and fair 1983-2007. Ann Rheum Dis 2010;69:868–71. control arms: randomized controlled trials of biologic treatment for rheumatoid arthritis. Arch Intern Med 31 Jamsen E, Virta LJ, Hakala M et al. The decline in joint 2012;172:237–44. replacement surgery in rheumatoid arthritis is associated with a concomitant increase in the intensity of anti- 44 Graudal N, Hubeck-Graudal T, Faurschou M, Baslund B, rheumatic therapy: a nationwide register-based study Jürgens G. Combination therapy with and without tumor from 1995 through 2010. Acta Orthop 2013;84:331–7. necrosis factor inhibitors in rheumatoid arthritis: a meta- analysis of randomized trials. Arthritis Care Res 32 Mertelsmann-Voss C, Lyman S, Pan TJ et al. US trends in 2015;67:1487–95. rates of arthroplasty for inflammatory arthritis including 45 Moreland LW, O’Dell JR, Paulus HE et al. A randomized rheumatoid arthritis, juvenile idiopathic arthritis, and comparative effectiveness study of oral triple therapy spondyloarthritis. Arthritis Rheumatol 2014;66:1432–9. versus etanercept plus methotrexate in early aggressive 33 Shourt CA, Crowson CS, Gabriel SE, Matteson EL. rheumatoid arthritis: the treatment of early aggressive Orthopedic surgery among patients with rheumatoid arth- rheumatoid arthritis trial. Arthritis Rheum ritis 1980-2007: a population-based study focused on sur- 2012;64:2824–35. gery rates, sex, and mortality. J Rheumatol 2012;39:481–5. 46 O’Dell JR, Mikuls TR, Taylor TH et al. Therapies for active 34 David G, Tandon N, Waters H, Gunnarsson C, Kavanaugh rheumatoid arthritis after methotrexate failure. N Engl J A. Rheumatoid arthritis and joint replacement: impact of Med 2013;369:307–18. biologics. Am J Pharm Benefits 2014;6:256–64. 47 Aaltonen KJ, Virkki LM, Jamsen E et al. Do biologic drugs 35 Nystad TW, Fenstad AM, Furnes O et al. Reduction in affect the need for and outcome of joint replacements in orthopaedic surgery in patients with rheumatoid arthritis: a patients with rheumatoid arthritis? A register-based study. Norwegian register-based study. Scand J Rheumatol Semin Arthritis Rheum 2013;43:55–62. 2016;45:1–7. 48 Pirracchio R, Carone M, Rigon MR et al. Propensity score 36 Harty L, O’Toole G, FitzGerald O. Profound reduction in estimators for the average treatment effect and the aver- hospital admissions and musculoskeletal surgical pro- age treatment effect on the treated may yield very different cedures for rheumatoid arthritis with concurrent changes estimates. Stat Methods Med Res 2016;25:1938–54. https://academic.oup.com/rheumatology 1175