Osteoarthritis and Cartilage 27 (2019) 1627e1635 Lifetime risk of knee and hip replacement following a GP diagnosis of osteoarthritis: a real-world cohort study E. Burn y, D.W. Murray z, G.A. Hawker x k ¶, R. Pinedo-Villanueva y *, D. Prieto-Alhambra y # y Centre for Statistics in Medicine, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, UK z Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, UK x Department of Medicine, University of Toronto, Toronto, Canada k Women's College Research Institute, Women's College Hospital, Toronto, Canada ¶ Institute for Clinical Evaluative Sciences, Toronto, Canada # GREMPAL Research Group, Idiap Jordi Gol and CIBERFes, Universitat Autonoma de Barcelona and Instituto de Salud Carlos III, Barcelona, Spain a r t i c l e i n f o s u m m a r y Article history: Objective: The aim of this study was to estimate lifetime risk of knee and hip replacement following a GP Received 17 April 2019 diagnosis of osteoarthritis and assess how this risk varies with patient characteristics. Accepted 10 June 2019 Methods: Routinely collected data from Catalonia, Spain, covering 2006 to 2015, were used. Study par- ticipants had a newly recorded GP diagnosis of knee or hip osteoarthritis. Parametric survival models Keywords: were specified for risk of knee/hip replacement and death following diagnosis. Survival models were Osteoarthritis combined using a Markov model and lifetime risk estimated for the average patient profile. The effects of Orthopedic surgery age at diagnosis, sex, comorbidities, socioeconomic status, body mass index (BMI), and smoking on risk Epidemiology were assessed. Results: 48,311 individuals diagnosed with knee osteoarthritis were included, of whom 2,561 underwent knee replacement. 15,105 individuals diagnosed with hip osteoarthritis were included, of whom 1,247 underwent hip replacement. The average participant's lifetime risk for knee replacement was 30% (95% CI: 25e36%) and for hip replacement was 14% (10e19%). Notable patient characteristics influencing lifetime risk were age at diagnosis for knee and hip replacement, sex for hip replacement, and BMI for knee replacement. BMI increasing from 25 to 35 was associated with lifetime risk of knee replacement increasing from 24% (20e28%) to 32% (26e37%) for otherwise average patients. Conclusion: Knee and hip replacement are not inevitable after an osteoarthritis diagnosis, with average lifetime risks of less than a third and a sixth, respectively. Patient characteristics, most notably BMI, influence lifetime risks. © 2019 The Authors. Published by Elsevier Ltd on behalf of Osteoarthritis Research Society International. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). Introduction interventions are available as initial treatments after diagnosis2. However, if patients develop persistent pain and functional Osteoarthritis is a clinical syndrome of failure of the joint impairment, they may be offered a knee or hip replacement.3,4 characterised by joint pain, functional limitation, and reduced One of the barriers to implementing recommended osteoar- quality of life1. The hip and knee are the principal large joints thritis management after diagnosis is inaccurate beliefs about the affected by osteoarthritis. Following diagnosis of knee or hip oste- disease process and progression among health professionals and oarthritis, patient care is primarily managed in primary care. A patients5. There is general negativity around the condition, with a broad range of non-pharmacological and pharmacological prevailing belief that the development of osteoarthritis is an un- avoidable consequence of aging and that further deterioration is inevitable6. This pessimism may lead to apathy or avoidance when * Address correspondence and reprint requests to: R. Pinedo-Villanueva, Botnar managing people with osteoarthritis7, which may explain the Research Centre, Windmill Road, OX37LD, Oxford, UK. limited adherence to treatment guidelines.8 E-mail address: rafael.pinedo@ndorms.ox.ac.uk (R. Pinedo-Villanueva). https://doi.org/10.1016/j.joca.2019.06.004 1063-4584/© 2019 The Authors. Published by Elsevier Ltd on behalf of Osteoarthritis Research Society International. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1628 E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 Findings from previous research, however, have indicated that higher degree of comorbidity18. All observation time prior to the structural deterioration is not a predictable consequence of aging1. index diagnosis of osteoarthritis were used to identify generate this Further progression of osteoarthritis also does not appear to be score. Socioeconomic status was summarised using MEDEA, a inevitable. In two community-based studies, for example, 39% of census-based indicator that has a distinct category for those living knees with radiographic knee osteoarthritis had not worsened after in rural areas19. MEDEA has previously been used to assess the 15 years9 while 35% of hips with radiographic hip osteoarthritis had effect of socioeconomic status on the risk of hand, hip, and knee not progressed after 8 years10. Individuals with osteoarthritis osteoarthritis in Catalonia20. Participants were grouped by MEDEA experience different symptom trajectories. In one study, disability quintile (first indicating least deprived and fifth most deprived), or progressively worsened over time for only 24% of patients. The as living in a rural area. remaining patients were stable, had short-term fluctuations, or We extracted the most recent record of body mass index (BMI) steadily improved over time.11 and smoking status (non-smoker, ex-smoker, or current smoker) Better understanding of prognosis has been identified as a before the osteoarthritis diagnosis and kept only those recorded particular area of unmet need in patients with osteoarthritis12, and within a year of diagnosis. For the purposes of summarising the patients desire clear and understandable information13. The life- observed outcomes, continuous variables were categorised. BMI time risks of knee and hip replacement would satisfy these re- was categorised following the World Health Organisation cate- quirements and give patients and healthcare providers a better idea gories: normal or underweight (BMI <25), overweight (25 of an individual's future healthcare needs. Understanding the effect and < 30), obese class I (30 and < 35), obese class II (35 of modifiable and non-modifiable patient characteristics on life- and < 40), and obese class III (40). Age was categorised by splitting time risk can help instigate self-management and patient-driven the data into quartiles (with first indicating the youngest and fourth treatments following diagnosis. the oldest). In this study, we estimated the lifetime risks of knee and hip replacement following a GP diagnosis of knee and hip osteoar- Outcomes thritis, respectively. We then assessed the effect of patient char- acteristics on lifetime risk. Instances of total knee and hip replacement (ICD-9 procedure codes 8154 and 8151) were identified using linked hospital records, Methods with the database used collecting details of admissions to every public and private hospital of Catalonia21. All knee and hip re- Setting placements were included in the analysis, and so procedures were not necessarily attributable to the prior osteoarthritis diagnosis. This analysis was based on actual practice data from patients Deaths were identified based on recorded date of death in SIDIAP. from Catalonia, Spain. Individual-level primary care data were In the absence of a knee or hip replacement or death during the extracted from the Sistema d’Informacio  pel Desenvolupament de la study period, subjects were censored at the date of exit from a GP Investigacio a l’Atencio ria (SIDIAP). SIDIAP (www.sidiap.org)  Prima practice linked to SIDIAP (e.g., due to moving out of Catalonia or is a database containing patient records of 80% of the Catalan changing practice within Catalonia to one not linked to SIDIAP) or population and is highly representative of the population in terms the end of the study (31 December 2015). of geographical, age, and sex distributions14. It has been used extensively for research15. About 30% of the contributing practices Statistical methods are also linked to the regional Conjunt Mínim Ba sic de Dades (CMBD), a database containing details of admissions to every hos- Comparison of cumulative incidence of knee/hip replacement pital of Catalonia. The linked records gave us a total source popu- Observed knee and hip replacement over 9 years following lation of 1.7 million active subjects for the period 1 January 2006 to diagnosis of knee or hip osteoarthritis were summarised by their 31 December 2015. cumulative incidence. Cumulative incidence allows the incidence of an event to be estimated while taking competing risk into ac- Study participants count22. We summarised cumulative incidence for the study pop- ulations as a whole and compared cumulative incidence stratified Individuals were eligible for inclusion into the study if they had by participant characteristics of interest (age, sex, Charlson score, an incident diagnosis of knee or hip osteoarthritis. Diagnostic MEDEA quintile or living rurally, BMI, and smoking status). coding in SIDIAP is based on ICD-10 codes. The recording of knee and hip osteoarthritis have previously been validated, with a Estimating parametric survival models for risks of knee/hip sensitivity of 94% and specificity of 71% when compared to self- replacement and mortality reported physician diagnosed osteoarthritis16. A further validation Parametric survival models were estimated for cause-specific using free text records review confirmed the quality of osteoar- risks of knee/ hip replacement and death following diagnosis of thritis diagnosis in SIDIAP17. Individuals were excluded if they were knee or hip osteoarthritis, respectively. Such models require an younger than 50 at their date of diagnosis, had less than 1year of assumption of the underlying distribution of the events of interest. observation time prior to their diagnosis, had a prior diagnosis of Alternative distributions were compared and chosen on the basis of inflammatory arthritis, or had a knee or hip replacement recorded their fit to the observed data and the plausibility of extrapolation23. before or on their date of diagnosis. The identification of knee and Further details on the choice of distributions is provided in Ap- hip replacement is summarised in the outcomes section below. pendix Section 3, with the chosen distributions shown in Appendix Fig. A2-A4. Participant characteristics at diagnosis Univariable and multivariable survival models were estimated for each of the participant characteristics of interest. Non-linearity Participant characteristics at diagnosis were extracted from the in continuous variables was incorporated through the use of primary care records. Age and sex were extracted. Participants’ polynomials, if their inclusion improved fit relative to specifying a comorbidities were summarised using the Charlson score, with linear relationship. Fit was assessed by comparing Akaike infor- possible scores of 0, 1, 2, or 3þ, and a higher score indicating a mation criteria (AIC). Non-proportionality in hazards was assessed E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 1629 using a visual inspection of a log-log plot and by testing the Observed risks of knee/hip replacement and mortality weighted residuals24. As there was evidence of non-proportionality in age for risks of knee and hip replacement, the models were Cumulative incidence of knee and hip replacement in the 9-year estimated separately for each of the age groups, with age included period following primary care diagnosis was 9.4% and 11.6% as an explanatory variable within each of the stratified models.25 respectively. Cumulative incidences stratified by participant char- Missing data in MEDEA, BMI, and smoking status was addressed acteristics of interest are summarised in Table II. Those patients in using multiple imputation, with 50 imputed datasets generated. the oldest age quartile had a substantially lower cumulative inci- Both explanatory variables and outcomes were used as predictors dence of both knee and hip replacement than younger patients, for missing data, with predictive mean matching used for BMI and males had a substantially higher cumulative incidence of hip multinomial logit models used for MEDEA and smoking status. replacement than females, and cumulative incidence of knee Hazard ratios and corresponding 95% confidence intervals (CIs) replacement was greater for those with a higher BMI. The hazard were calculated using Rubin's rules. Estimates from models based ratios estimated for explanatory factors in each of the cause- on complete case data were also provided for comparison. specific survival models for knee replacement and death or hip replacement and death are detailed in Appendix Tables A2 and A3. Estimating lifetime risk Estimates for models based on complete case data were similar, To estimate lifetime risk, parametric survival models for knee Appendix Tables A4 and A5. replacement and death or hip replacement and death were com- bined using a state-based cohort Markov model. For lifetime risk of Average lifetime risks of knee and hip replacement knee replacement, a cohort of individuals began as being diagnosed with knee osteoarthritis and then they remained in the diagnosis At diagnosis, the average participant with knee osteoarthritis state or progressed to either the knee replacement or death state as was a non-smoking 69-year-old woman in the fourth MEDEA time progressed in yearly cycles. An equivalent structure was used quintile, with a Charlson score of 0 and a BMI of 30. At diagnosis, the for lifetime risk of hip replacement following a diagnosis of hip average participant with hip osteoarthritis was a non-smoking 70- osteoarthritis. year-old woman living rurally, with a Charlson score of 0 and a BMI Transition probabilities for knee and hip replacement were of 29. For participants with these characteristics at diagnosis, the based exclusively on the parametric survival models estimated for parametric models indicated that the risks of knee and hip these events. These models were extrapolated beyond the 9 years replacement peaked in the second year after diagnosis, then fell of observed data to participants’ total remaining lifetime. Transition over time (Fig. 1). After accounting for the competing risk of mor- probabilities for death were based on the relevant parametric tality, these translated into a lifetime risk of knee replacement models for the first 9 years of the model, after which they were following a diagnosis of knee osteoarthritis of 30% (95% CI: 25e36%) assumed to revert to estimates based on age- and sex-specific and a lifetime risk of hip replacement following a diagnosis of hip lifetables for Spain26. Estimated hazard ratios were similar across osteoarthritis of 14% (10e19%). multiply imputed datasets and so to reduce computational time, lifetime risks were estimated using the survival models estimated Participant characteristics and lifetime risks of knee/hip replacement on the first imputed dataset. Parameter uncertainty was incorpo- rated using 1,000 bootstrapped models. Lifetime risk of knee and hip replacement following a diagnosis The models were first run for cohorts of individuals with of knee or hip osteoarthritis generally fell as age at diagnosis average characteristics (median for continuous variables and mode increased (Fig. 2). Younger women generally had a slightly higher for categorical ones) for those diagnosed with knee replacement lifetime risk of knee replacement than younger men. For example, a and hip replacement, separately. The partial effect of explanatory 60-year-old woman had a 37% (27e50%) lifetime risk of knee factors on lifetime risk was assessed by re-running the models with replacement, while a 60-year-old man had a 30% (22e46%) risk. participant profiles varying in the explanatory factor of interest, However, men had a substantially higher lifetime risk of hip while holding other characteristics constant at their average. For replacement than women at younger ages. An average 60-year-old the partial effect of continuous variables, a smoothed line was fitted man, for example, had a 30% (25e36%) lifetime risk of hip across the lifetime risk estimates for the different simulated values replacement after a diagnosis of hip osteoarthritis, while a 60-year- of the variable. old woman had a 17% (12e24%) lifetime risk. A higher BMI was associated with a substantially higher lifetime Results risk of knee replacement, but relatively little difference in lifetime risk of hip replacement (see Fig. 3 for the partial effect of BMI on Study participants transition probabilities, and Fig. 4 for the partial effect on lifetime risks). Holding other explanatory variables fixed at their average, 48,311 and 15,105 individuals were included in the knee and hip lifetime risk of knee replacement after a diagnosis of knee osteo- osteoarthritis cohorts, respectively. A study inclusion flow chart is arthritis was 24% (20e28%) for a BMI of 25 which increased to 32% provided in Appendix Fig. A1. The characteristics of these in- (26e37%) for a BMI of 35. Meanwhile, the lifetime risk of hip dividuals are summarised in Table I. The median prior observation replacement after a diagnosis of hip osteoarthritis was 12% (9e17%) time over which prior diagnoses and events could be observed was for a BMI of 25% and 15% (11e19%) for a BMI of 35. Differences in 5 years (with an interquartile range of 3e8 years). A comparison of comorbidities, smoking status, and socioeconomic status and those with and without missing data in socioeconomic status rurality had relatively little effect on lifetime risk of knee or hip (MEDEA), BMI, or smoking status is given in Appendix Table A1, and replacement (Appendix Figures A9-A11). combinations of missing data are summarised in Appendix Figs. A2 and A3. Individuals with missing data were generally younger and Discussion had fewer comorbidities than those with complete data. A com- parison of observed and imputed values for BMI, the one contin- This is to our knowledge the first study to assess the lifetime uous variable that was imputed, are also summarised in Appendix risks of knee or hip replacement for a patient diagnosed with knee Figs. A4 and A5. or hip osteoarthritis in a primary care setting. Despite the 1630 E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 Table I Participant characteristics at the time of a knee or hip osteoarthritis diagnosis Knee osteoarthritis cohort Hip osteoarthritis cohort N 48,311 15,105 Age at diagnosis (median [IQR]) 69 [62, 77] 70 [62, 78] Age at diagnosis group* (%) 1st (youngest) 11,436 (23.7) 3,503 (23.2) 2nd 11,212 (23.2) 3,666 (24.3) 3rd 13,034 (27.0) 3,944 (26.1) 4th (oldest) 12,629 (26.1) 3,992 (26.4) Gender: male (%) 16,076 (33.3) 6,024 (39.9) Charlson score (%) 0 25,967 (53.7) 7,634 (50.5) 1 12,258 (25.4) 3,779 (25.0) 2 5,840 (12.1) 1,982 (13.1) 3þ 4,246 (8.8) 1,710 (11.3) BMI (median [IQR]) 30.4 [27.5, 33.9] 29.1 [26.5, 32.3] BMI group (%) Normal or underweight (<25) 2,401 (5.0) 1,170 (7.7) Overweight (25 and < 30) 9,550 (19.8) 3,444 (22.8) Obese class I (30 and < 35) 8,770 (18.2) 2,372 (15.7) Obese class II (35 and < 40) 3,538 (7.3) 788 (5.2) Obese class III (40) 1,437 (3.0) 2,10 (1.4) Missing 22,615 (46.8) 7,121 (47.1) MEDEA quintile or rural (%) 1st (least deprived) 4,708 (9.7) 1,726 (11.4) 2nd 6,971 (14.4) 2,326 (15.4) 3rd 8,676 (18.0) 2,666 (17.6) 4th 9,472 (19.6) 2,736 (18.1) 5th (most deprived) 7,298 (15.1) 2,010 (13.3) Rural 9,244 (19.1) 2,938 (19.5) Missing 1,942 (4.0) 703 (4.7) Smoking status (%) Non-smoker 32,128 (66.5) 9,328 (61.8) Ex-smoker 6,307 (13.1) 2,334 (15.5) Current smoker 4,140 (8.6) 1,656 (11.0) Missing 5,736 (11.9) 1,787 (11.8) * Age groups based on quartiles for knee osteoarthritis cohort: 50 to 62, 62 to 69, 69 to 77, 77 to 105, and quartiles for hip osteoarthritis: 50 to 62, 62 to 70, 70 to 78, 78 to 103. MEDEA is a measure of socioeconomic status developed for Catalonia (Spain), with those living in rural areas having a distinct category. BMI: body mass index; IQR: interquartile range. prevailing belief that further deterioration is inevitable following Risk of knee and hip replacement appears to be highest in the diagnosis, we find that the lifetime risks of knee and hip replace- second year for those at average age at time of osteoarthritis ment are less than one-third and less than one-sixth following a diagnosis. This implies that there is a sizeable proportion of pa- diagnosis of knee osteoarthritis and hip osteoarthritis, respectively. tients whom referral to surgery is made shortly following diagnosis. The risk of undergoing a knee or hip replacement peaks in the This may be explained to some degree by rapid progression of second year after diagnosis, then steadily falls. osteoarthritis for some patients1. It is likely though to be in large Older participants generally had lower lifetime risks of both part because of a proportion of patients being diagnosed at a late- knee and hip replacement than younger participants. Young stage in the disease process, at which point knee or hip replace- women had a lower lifetime risk of hip replacement than young ment was already merited. men. The lifetime risk of knee replacement increased as BMI We identified participant characteristics associated with differ- increased. ences in lifetime risk. These differences could be due to variation in need, disease progression, time at risk (i.e., risk of mortality), or access to care. Study findings in context Age at diagnosis had a substantial effect on lifetime risks of knee and hip replacement. With mortality as a competing risk, younger We found that joint replacement was not inevitable following a age at diagnosis was associated with a longer time at risk due to GP diagnosis of osteoarthritis. This finding is consistent with pre- greater life expectancy. All else being equal, lifetime risks of knee vious research into the structural and symptomatic progression of and hip replacement can be expected to be higher for younger osteoarthritis9e11. As would be expected, our estimates for lifetime patients. Age also appeared to influence cause-specific risks of knee risks of joint replacement following diagnosis of osteoarthritis were and hip replacement, with those older at diagnosis generally having higher than the lifetime risks previously estimated for the general a reduced risk. This finding is consistent with previous research population27e29. One previous study combined prevalence and that found those over 82 to have less than half the risk of knee and progression estimates from the literature to estimate that a 25-year hip replacement than younger patients, even after controlling for old with no history of knee injury would have a 6% lifetime risk of severity of osteoarthritis symptoms31. Indeed, individuals aged over knee replacement30. This latter study estimated that 13.5% of the 85 have previously been found to receive less knee and hip cohort would develop osteoarthritis, implying a 45% lifetime risk of replacement relative to need than younger patients32, which is may knee replacement for those diagnosed. Although above the esti- be due to a perception of a perceived worse risk-benefit trade-off mate for average lifetime risk found in our study, it is not for surgery in the elderly. dramatically so. E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 1631 Table II Cumulative incidence of knee or hip replacement at 9 years follow-up Knee OA Hip OA KR/PY 9-year cumulative incidence (% (95% CI)) HR/PY 9-year cumulative incidence (% (95% CI)) Total 2561/209508 9.4% (8.9e9.9%) 1247/60723 11.6% (10.9e12.3%) Age at diagnosis group* (%) 1st (youngest) 452/54625 7.7% (6.9e8.8%) 359/14779 15.0% (13.3e16.9%) 2nd 730/49643 13.0% (11.5e14.7%) 329/15124 13.0% (11.5e14.7%) 3rd 1002/56855 12.3% (11.4e13.2%) 371/16491 12.8% (11.5e14.4%) 4th (oldest) 377/48386 4.4% (3.7e5.2%) 188/14328 5.9% (5.0e6.8%) Gender Male 790/69000 8.4% (7.6e9.2%) 609/37361 14.5% (13.3e15.7%) Female 1771/140508 9.9% (9.2e10.6%) 638/23362 9.7% (8.8e10.7%) Charlson 0 1416/120475 9.7% (8.9e10.5%) 718/32869 13.2% (12.1e14.3%) 1 702/52364 9.4% (8.6e10.3%) 300/14917 10.8% (9.6e12.3%) 2 284/22691 9.5% (7.8e11.6%) 149/7450 10.6% (8.8e12.8%) 3þ 159/13978 6.3% (5.2e7.5%) 80/5487 6.0% (4.8e7.5%) BMI group (%) Normal or underweight (<25) 62/9556 6.3% (3.0e13.5%) 67/4361 7.8% (6.1e10.0%) Overweight (25 and < 30) 424/39791 8.5% (7.3e9.8%) 254/13534 11.1% (9.6e12.8%) Obese class I (30 and < 35) 543/37410 10.5% (9.5e11.7%) 205/9630 12.6% (10.7e14.8%) Obese class II (35 and < 40) 278/15260 13.9% (11.9e16.3%) 82/3074 13.7% (11.0e17.1%) Obese class III (40) 97/6015 14.3% (10.9e18.8%) 15/886 10.1% (6.0e16.9%) Missing 1157/101476 8.6% (7.9e9.4%) 624/29238 11.9% (10.9e13.0%) MEDEA quintile or rural (%) 1st (least deprived) 218/20180 7.4% (6.4e8.5%) 146/6946 11.7% (9.9e13.9%) 2nd 367/30180 10.0% (8.5e11.7%) 210/9095 12.7% (11.0e14.7%) 3rd 478/37924 10.3% (8.8e12.0%) 204/10565 10.4% (9.0e12.2%) 4th 606/41932 11.7% (10.3e13.3%) 246/11088 13.2% (11.5e15.3%) 5th (most deprived) 448/31745 11.0% (9.7e12.3%) 189/8,634 14.1% (11.8e16.9%) Rural 377/39555 6.7% (5.8e7.6%) 213/11587 9.8% (8.4e11.4%) Missing 67/7992 4.6% (3.6e5.9%) 39/2807 6.5% (4.7e8.9%) Smoking status Non-smoker 1729/133845 9.7% (8.9e10.5%) 697/36458 10.4% (9.5e11.3%) Ex-smoker 285/23815 9.4% (7.4e11.9%) 208/8,150 13.1% (11.4e15.1%) Current smoker 179/17950 7.4% (6.3e8.8%) 167/6393 16.2% (13.4e19.5%) Missing 368/33897 8.9% (8.0e10.0%) 175/9721 11.5% (9.9e13.3%) Cumulative incidence of total knee or hip replacement 9 years after a diagnosis of knee or hip osteoarthritis. * Age groups based on quartiles for knee osteoarthritis cohort: 50 to 62, 62 to 69, 69 to 77, 77 to 105, and quartiles for hip osteoarthritis cohort: 50 to 62, 62 to 70, 70 to 78, 78 to 103. MEDEA is a measure of socioeconomic status developed for Catalonia (Spain), with those living in rural areas having a distinct category. BMI: body mass index; CI: confidence interval; HR: hip replacement; KR: knee replacement; OA: osteoarthritis; PY: person years. Being a woman has consistently been linked with an increased willingness to undergo surgery31, which may be due to differences risk of developing osteoarthritis10, as reflected in our study cohorts in perceptions of the risks and benefits of knee and hip replacement with both being majority women. However, our findings suggest and general preferences for surgery33. Gender bias has also been that once diagnosed, young men have a substantially higher life- observed, with physicians more likely to recommend knee time risk of hip replacement than young women. Previous research replacement to a male patient than an otherwise equivalent female has found that the effect of sex on risk of surgery is mediated by patient.34 Fig. 1. Annual transition probabilities of knee and hip replacement following a diagnosis of knee and hip osteoarthritis for average patient profiles. Point estimates with 95% confidence intervals (CIs). OA: osteoarthritis. 1632 E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 Fig. 2. Partial effect of age and sex on lifetime risks of knee and hip replacement after diagnosis of knee and hip osteoarthritis. Point estimates with 95% CIs. OA: osteoarthritis. Higher BMI was associated in this study with a substantially Strengths and limitations of this study increased lifetime risk of knee replacement following an osteoar- thritis diagnosis. Higher BMI has previously been associated with This study was based on a large, representative sample from an increased risk of developing knee osteoarthritis35, greater routinely collected data, with lifetime risks of knee and hip structural and clinical progression36e38, and an increased risk of replacement estimated from time of GP diagnosis of knee or hip undergoing knee replacement among the general population39. osteoarthritis. The routinely collected data used, however, only Higher BMI has also previously been shown to be associated with covered a window of time and so historical diagnoses and pro- an increased risk of knee replacement over 6 years after a GP cedures may have been missed, leading to a possible underesti- diagnosis of osteoarthritis using data from SIDIAP40. In contrast, mation of comorbidities and a failure to exclude some patients, for higher BMI at time of diagnosis of hip osteoarthritis appeared to example those who had a diagnosis of inflammatory arthritis prior have little effect on the risk of hip replacement following that to their observation time. In addition, further research into the diagnosis. This discordance has previously been found in a large generalisability of our findings would be useful. Previous research population-based cohort study, and may be explained by differ- has shown that risks of joint replacement for the general ences in biomechanical effects.41 Fig. 3. Partial effect of body mass index (BMI) on annual transition probabilities of knee and hip replacement following a diagnosis of knee and hip osteoarthritis. Point estimates with 95% CIs. OA: osteoarthritis. E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 1633 Fig. 4. Partial effect of BMI on lifetime risks of knee and hip replacement after diagnosis of knee and hip osteoarthritis. BMI: BMI; OA: osteoarthritis. population vary across countries27,28, and this will likely also be the osteoarthritis diagnosis would therefore likely lead to substantial case for risks for those diagnosed with osteoarthritis. health benefits for patients and cost savings for the health system. The use of parametric survival models with flexible distributions allowed us to estimate lifetime risk, which is an understandable, Contributors and possibly the most pertinent, description of risk. By using EB, DWM, RPV, and DPA made substantial contributions to the parametric models instead of lifetable methods, we were able to conception and design of the study. EB, DWM, GH, RPV, and DPA thoroughly analyse the effect of patient characteristics on the risk of made substantial contributions to the interpretation of the data for knee or hip replacement and mortality. However, this approach the work. EB, RPV, and DPA undertook the statistical analysis. EB, required extrapolation well beyond the end of study follow-up, RPV, and DPA drafted the manuscript, with GH and DWM revising it particularly for younger patients. This extrapolation necessarily for important intellectual content. All authors read and approved had a high degree of uncertainty, as reflected in the wide CIs around the final manuscript. estimates for younger patients. When analysing the relationship between patient characteris- tics and lifetime risks of knee and hip replacement, we were limited Conflict of interest to those factors available in routinely collected data. A wide range of All authors have completed the ICMJE uniform disclosure form at other factors are likely to influence lifetime risk, such as willingness www.icmje.org/coi_disclosure.pdf and declare: DPA reports grants to undergo knee or hip replacement and disease severity31,42. In from Amgen, Servier and UCB Biopharma, and non-financial sup- addition, this analysis was limited to risk of first knee or hip port from Amgen, all outside the submitted work. RPV reports replacement following a diagnosis of osteoarthritis with no data consultancy fees from Kyowa Kirin, UCB, and Mereo, all outside the available on laterality at diagnosis or at knee or hip replacement. If submitted work. DWM reports grants and personal fees from data on laterality were available for future research, analyses Zimmer Biomet. In addition, DWM has various patents related to incorporating this information could provide a more detailed Unicompartmental Knee Replacement (Zimmer Biomet) with roy- assessment of prognosis. alties paid, all outside the submitted work. Funding DPA is funded by a National Institute for Health Research Clinician Conclusion Scientist award (CS-2013-13-012). This article presents indepen- dent research funded by the National Institute for Health Research Knee and hip replacement are not inevitable following a GP (NIHR). The views expressed are those of the authors and not diagnosis of knee or hip osteoarthritis. Those with knee osteoar- necessarily those of the NHS, the NIHR, or the Department of thritis have a lifetime risk of less than a third for knee replacement, Health. This work was supported by the NIHR Biomedical Research and those with hip osteoarthritis have a lifetime risk of less than a Centre, Oxford. GH receives salary support as the Sir John and Lady sixth for hip replacement. These findings provide a clear indication Eaton Professor and Chair of Medicine at the University of Toronto. of prognosis for doctors and patients, which should help to inform treatment choices after diagnosis. Risk of knee and hip replacement generally peaked in the sec- Ethical approval ond year following diagnosis in this study. This is likely because of a late diagnosis for a proportion of the study participants, with Approval for all observational research using SIDIAP data is diagnosis made at a point where knee or hip replacement was obtained from a local ethics committee (Clinical Research Ethics already merited. This finding underscores the importance of timely Committee of the IDIAP Jordi Gol). diagnosis, following which non-operative treatments can be pursued. Data sharing Lifetime risks of knee and hip replacement vary depending on patient characteristics at diagnosis. In particular, higher BMI is Data were provided under a licence that does not permit associated with an increased risk of knee replacement. Effective sharing. Data are obtainable from the SIDIAP subject to a full weight loss interventions provided at the time of a knee application. 1634 E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 Transparency osteoarthritis in the elderly. The framingham study. Arthritis Rheum 1997;40(4):728e33, https://doi.org/10.1002/ The senior and corresponding authors (DPA and RPV) affirm that art.20213. this manuscript is an honest, accurate, and transparent account of 11. Leffondre K, Abrahamowicz M, Regeasse A, Hawker GA, the study being reported; that no important aspects of the study Badley EM, McCusker J, et al. Statistical measures were pro- have been omitted; and that any discrepancies from the study as posed for identifying longitudinal patterns of change in planned have been explained. quantitative health indicators. J Clin Epidemiol 2004;57(10): 1049e62, https://doi.org/10.1016/j.jclinepi.2004.02.012. Acknowledgements 12. Victor C, Ross F, Axford J. Capturing lay perspectives in a randomized control trial of a health promotion intervention The authors would like to thank Miss Susan Thwaite (National for people with osteoarthritis of the knee. J Eval Clin Pract Rheumatoid Arthritis Society) for her role as the patient and public 2004;10(1):63e70. representative and her role on the study steering committee. We 13. Chou L, Ellis L, Papandony M, Seneviwickrama KLMD, also thank Dr Jennifer A. de Beyer of the Centre for Statistics in Cicuttini FM, Sullivan K, et al. Patients' perceived needs of Medicine, University of Oxford, for English language editing. osteoarthritis health information: a systematic scoping review. PLoS One 2018;13(4):1e24, https://doi.org/10.1371/ Supplementary data journal.pone.0195489. 14. García-Gil MDM, Hermosilla E, Prieto-Alhambra D, Fina F, Supplementary data to this article can be found online at Rosell M, Ramos R, et al. Construction and validation of a https://doi.org/10.1016/j.joca.2019.06.004. scoring system for the selection of high-quality data in a Spanish population primary care database (SIDIAP). Inf Prim References Care 2011;19(3):135e45, http://www.ncbi.nlm.nih.gov/ pubmed/22688222. 1. Cooper C, Adachi JD, Bardin T, Berenbaum F, Flamion B, 15. Ramos R, Comas-Cufí M, Martí-Lluch R, Ballo  E, Ponjoan A, Jonsson H, et al. How to define responders in osteoarthritis. Alves-Cabratosa L, et al. Statins for primary prevention of Curr Med Res Opin 2013;29(6):719e29, https://doi.org/ cardiovascular events and mortality in old and very old adults 10.1185/03007995.2013.792793. with and without type 2 diabetes: retrospective cohort study. 2. Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, BMJ 2018;362:1e4, https://doi.org/10.1136/bmj.k3359. Arden N, et al. OARSI recommendations for the management of 16. Prieto-Alhambra D, Nogues X, Javaid MK, Wyman A, Arden NK, hip and knee osteoarthritis, Part II: OARSI evidence-based, Azagra R, et al. An increased rate of falling leads to a rise in expert consensus guidelines. Osteoarthritis Cartilage fracture risk in postmenopausal women with self-reported 2008;16(2):137e62, https://doi.org/10.1016/ osteoarthritis: a prospective multinational cohort study j.joca.2007.12.013. (GLOW). Ann Rheum Dis 2013;72(6):911e7, https://doi.org/ 3. Pivec R, Johnson AJ, Mears SC, Mont MA. Hip arthroplasty. 10.1136/annrheumdis-2012-201451. Lancet 2012;380(9855):1768e77, https://doi.org/10.1016/ 17. Prieto-alhambra D, Judge A, Javaid MK, Cooper C, Diez-perez A, S0140-6736(12)60607-2. Arden NK. Incidence and risk factors for clinically diagnosed 4. Carr AJ, Robertsson O, Graves S, Price AJ, Arden NK, Judge A, knee, hip and hand osteoarthritis: influences of age, gender et al. Knee replacement. Lancet 2012;379(9823):1331e40, and osteoarthritis affecting other joints. Ann Rheum Dis https://doi.org/10.1016/s0140-6736(11)60752-6. 2014;73:1659e64, https://doi.org/10.1136/annrheumdis- 5. Egerton T, Nelligan R, Setchell J, Atkins L, Bennell KL. General 2013-20335. practitioners' perspectives on a proposed new model of ser- 18. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method vice delivery for primary care management of knee osteoar- of classifying prognostic comorbidity in longitudinal studies: thritis: a qualitative study. BMC Fam Pract 2017;18(1):1e12, development and validation. J Chronic Dis 1987;40(5):373e83, https://doi.org/10.1186/s12875-017-0656-7. https://doi.org/10.1016/0021-9681(87)90171-8. 6. Paskins Z, Sanders T, Hassell AB. Comparison of patient expe- 19. Ramos R, Comas-Cufí M, Martí-Lluch R, Ballo  E, Ponjoan A, riences of the osteoarthritis consultation with GP attitudes and Alves-Cabratosa L, et al. Construccio n de un índice de privacio n beliefs to OA: a narrative review. BMC Fam Pract 2014;15(1): a partir de datos censales en grandes ciudades espan ~ olas 1e10, https://doi.org/10.1186/1471-2296-15-46. (Proyecto MEDEA). Gac Sanit 2008;22(3):179e87, https:// 7. Egerton T, Diamond LE, Buchbinder R, Bennell KL, Slade SC. doi.org/10.1157/13123961. A systematic review and evidence synthesis of qualitative 20. Reyes C, Garcia-gil M, Elorza JM, Mendez-boo L, Hermosilla E, studies to identify primary care clinicians' barriers and en- Javaid MK, et al. Socio-economic status and the risk of devel- ablers to the management of osteoarthritis. Osteoarthritis oping hand , hip or knee osteoarthritis : a region-wide ecological Cartilage 2017;25(5):625e38, https://doi.org/10.1016/ study. Osteoarthritis Cartilage 2015;23(8):1323e9, https:// j.joca.2016.12.002. doi.org/10.1016/j.joca.2015.03.020. 8. Healey EL, Afolabi EK, Lewis M, Edwards JJ, Jordan KP, 21. CatSalut. Minimum basic data set https://catsalut.gencat.cat/ Finney A, et al. Uptake of the NICE osteoarthritis guidelines in ca/proveidors-professionals/registres-catalegs/registres/cmbd/ primary care: a survey of older adults with joint pain. BMC index.html#googtrans(ca%7Cen). Accessed May 26, 2019. Muscoskelet Disord 2018;19(1):1e9, https://doi.org/10.1186/ 22. Austin PC, Lee DS, Fine JP. Introduction to the analysis of s12891-018-2196-2. survival data in the presence of competing risks. Circulation 9. Leyland KM, Hart DJ, Javaid MK, Judge A, Kiran A, Soni A, et al. 2016;133(6):601e9, https://doi.org/10.1161/ The natural history of radiographic knee osteoarthritis: a CIRCULATIONAHA.115.017719. fourteen-year population-based cohort study. Arthritis Rheum 23. Latimer NR. Survival analysis for economic evaluations 2012;64(7):2243e51, https://doi.org/10.1002/art.34415. alongside clinical trialsdextrapolation with patient-level data 10. Felson DT, Zhang Y, Hannan MT, Naimark A, Weissman B, inconsistencies, limitations, and a practical guide. Med Decis Aliabadi P, et al. Risk factors for incident radiographic knee E. Burn et al. / Osteoarthritis and Cartilage 27 (2019) 1627e1635 1635 Mak 2013;33(6):743e54, https://doi.org/10.1177/ 34. Borkhoff CM, Hawker GA, Kreder HJ, Glazier RH, Mahomed NN, 0272989X12472398. Wright JG. The effect of patients' sex on physicians' recom- 24. Grambsch PM, Therneau TM. Proportional hazards tests and mendations for total knee arthroplasty. Can Med Assoc J diagnostics based on weighted residuals. Biometrika 1994, 2008;178(6):681e7, https://doi.org/10.1503/cmaj.071168. https://doi.org/10.1093/biomet/81.3.515. 35. Cooper C, Snow S, McAlindon TE, Kellingray S, Stuart B, 25. Harrell FE. Regression Modeling Strategies. Cham: Springer Coggon D, et al. Risk factors for the incidence and progression International Publishing; 2015, https://doi.org/10.1007/978-3- of radiographic knee osteoarthritis. Arthritis Rheum 319-19425-7. 2000;43(5):995, https://doi.org/10.1002/1529-0131(200005) 26. INE. Life Tables for the Population of Spain 1991-2009. http:// 43:5<995::AID-ANR6>3.0.CO;2-1. www.ine.es/jaxi/tabla.do?path¼/t20/p319a/serie/l1/ 36. Bastick AN, Belo JN, Runhaar J, Bierma-Zeinstra SMA. What are &file¼01001.px&type¼pcaxis&L¼1. the prognostic factors for radiographic progression of knee 27. Ackerman IN, Bohensky MA, de Steiger R, Brand CA, osteoarthritis? A meta-analysis. Clin Orthop Relat Res Eskelinen A, Fenstad AM, et al. Lifetime risk of primary total 2015;473(9):2969e89, https://doi.org/10.1007/s11999-015- hip replacement surgery for osteoarthritis from 2003 to 2013: 4349-z. a multinational analysis using national registry data. Arthritis 37. Wright AA, Cook C, Abbott JH. Variables associated with the Care Res 2017;69(11):1659e67, https://doi.org/10.1002/ progression of hip osteoarthritis: a systematic review. Arthritis acr.23197. Rheum 2009;61(7):925e36, https://doi.org/10.1002/ 28. Ackerman IN, Bohensky MA, de Steiger R, Brand CA, art.24641. Eskelinen A, Fenstad AM, et al. Substantial rise in the lifetime 38. Bastick AN, Runhaar J, Belo JN, Bierma-Zeinstra SMA. Prog- risk of primary total knee replacement surgery for osteoar- nostic factors for progression of clinical osteoarthritis of the thritis from 2003 to 2013: an international, population-level knee: a systematic review of observational studies. Arthritis analysis. Osteoarthritis Cartilage 2017;25(4):455e61, https:// Res Ther 2015;17(1):152, https://doi.org/10.1186/s13075- doi.org/10.1016/j.joca.2016.11.005. 015-0670-x. 29. Culliford DJ, Maskell J, Kiran A, Judge A, Javaid MK, Cooper C, 39. Franklin J, Ingvarsson T, Englund M, Lohmander LS. Sex dif- et al. The lifetime risk of total hip and knee arthroplasty: re- ferences in the association between body mass index and total sults from the UK general practice research database. Osteo- hip or knee joint replacement resulting from osteoarthritis. arthritis Cartilage 2012;20(6):519e24, https://doi.org/ Ann Rheum Dis 2009;68(4):536e40, https://doi.org/10.1136/ 10.1016/j.joca.2012.02.636. ard.2007.086868. 30. Suter LG, Smith SR, Katz JN, Englund M, Hunter DJ, Frobell R, 40. Leyland KM, Judge A, Javaid MK, Diez-Perez A, Carr A, et al. Projecting lifetime risk of symptomatic knee osteoar- Cooper C, et al. Obesity and the relative risk of knee replace- thritis and total knee replacement in individuals sustaining a ment surgery in patients with knee osteoarthritis: a prospec- complete anterior cruciate ligament tear in early adulthood. tive cohort study. Arthritis Rheum 2016;68(4):817e25, Arthritis Care Res 2016;69(2):201e8, https://doi.org/10.1002/ https://doi.org/10.1002/art.39486. acr.22940. 41. Reijman M, Pols HAP, Bergink AP, Hazes JMW, Belo JN, 31. Hawker GA, Guan J, Croxford R, Coyte PC, Glazier RH, Lievense AM, et al. Body mass index associated with onset and Harvey BJ, et al. A prospective population-based study of the progression of osteoarthritis of the knee but not of the hip: the predictors of undergoing total joint arthroplasty. Arthritis Rotterdam Study. Ann Rheum Dis 2007:158e62, https:// Rheum 2006;54(10):3212e20, https://doi.org/10.1002/ doi.org/10.1136/ard.2006.053538. art.22146. 42. Huynh C, Puyraimond-zemmour D, Maillefert JF, Conaghan PG, 32. Judge A, Welton NJ, Sandhu J, Ben-Shlomo Y. Equity in access Davis AM, Gunther KP, et al. Factors associated with the or- to total joint replacement of the hip and knee in England: cross thopaedic surgeon's decision to recommend total joint sectional study. BMJ 2010;341(aug11 1), https://doi.org/ replacement in hip and knee osteoarthritis: an international 10.1136/bmj.c4092. c4092-c4092. cross-sectional study of 1905 patients. Osteoarthritis Cartilage 33. Borkhoff CM, Hawker GA, Wright JG. Patient gender affects the 2018;26(10):1311e8, https://doi.org/10.1016/ referral and recommendation for total joint arthroplasty. Clin j.joca.2018.06.013. Orthop Relat Res 2011;469(7):1829e37, https://doi.org/ 10.1007/s11999-011-1879-x.