# paper.pdf ## Metadata - PDFFormatVersion=1.4 - Language=en-US - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260707084048+05'30' - Creator=Adobe InDesign 15.1 (Windows) - ModDate=D:20260707084050+05'30' - Producer=Adobe PDF Library 15.0 - Trapped=False - Custom.jobID=2a85ebbe-a24e-4a2c-b2e5-251cd56a4dbc - dc:identifier=doi:rmdopen-2026-006866 - dc:format=application/pdf - xmp:createdate=2026-07-07T08:40:48+05:30 - xmp:metadatadate=2026-07-07T08:40:50+05:30 - xmp:modifydate=2026-07-07T08:40:50+05:30 - xmp:creatortool=Adobe InDesign 15.1 (Windows) - xmpmm:instanceid=uuid:ab2053be-378b-489c-8140-c0d3ace086c6 - xmpmm:originaldocumentid=xmp.did:B2C59A163194E611BD009A7CB80B8EC7 - xmpmm:documentid=xmp.id:06ab123b-54c2-7344-9544-515e227fb157 - xmpmm:renditionclass=proof:pdf - xmpmm:derivedfrom=xmp.iid:a0d887a3-b81c-bb4a-939c-f56836a38e44xmp.did:28d958fa-6c01-fe4b-9866-135d1a799b2exmp.did:B2C59A163194E611BD009A7CB80B8EC7default - xmpmm:history=convertedfrom application/x-indesign to application/pdfAdobe InDesign 15.1 (Windows)/2026-07-07T08:40:48+05:30 - pdfx:jobid=2a85ebbe-a24e-4a2c-b2e5-251cd56a4dbc - pdf:producer=Adobe PDF Library 15.0 - pdf:trapped=False ## Contents ### Page 1 1Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866ORIGINAL RESEARCHGlucocorticoid in giant cell arteritis: real-world treatment patterns over time and associated burden in a nationwide hospital- based study within the French health insurance databaseMaxime Beydon ,1 Diane Lacaille,2,3 Bruno Fautrel ,1,4 Raphaèle Seror ,5,6,7 David Hajage,1 Florence Tubach1To cite: Beydon M, Lacaille D, Fautrel B, et al. Glucocorticoid in giant cell arteritis: real-world treatment patterns over time and associated burden in a nationwide hospital- based study within the French health insurance database. RMD Open2026;12:e006866. doi:10.1136/ rmdopen-2026-006866► Additional supplemental material is published online only. To view, please visit the journal online (https:// doi. org/ 10. 1136/rmdopen- 2026- 006866). Received 24 February 2026 Accepted 9 June 2026 For numbered affiliations see end of article.Correspondence toDr Florence Tubach;florence. tubach@ aphp. frVasculitis© Author(s) (or their employer(s)) 2026. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ Group.ABSTRACTObjective Recent advances in giant cell arteritis (GCA) management aim to reduce glucocorticoid (GC). We described changes in treatment patterns over time in real life and assessed associations between cumulative GC use and adverse outcomes.Methods This retrospective population-based study used the French national health insurance database (Système National des Données de Santé) and included 18 301 incident GCA cases between 2010 and 2022. Latent class growth modelling identified GC use trajectories. We described trends in cumulative dose and sparing agent use. Multivariable survival models assessed factors associated with sustained GC-free remission and association between GC cumulative exposure and mortality, serious infections, major adverse cardiovascular events (MACE) and osteoporotic fractures.Results Four GC use trajectories were identified. Two high-dose trajectories declined over time. Yet, mean 2-year cumulative GC dose remained high, reaching 7.6 g among patients diagnosed in 2022. Methotrexate use increased until 2020 and plateaued thereafter while tocilizumab prescriptions rose, reaching over 25% in 2022. Male sex and baseline methotrexate or tocilizumab use were associated with faster GC tapering, while age >75 years, hypertension, kidney and neurodegenerative diseases were associated with slower tapering. Each gram of GC was associated with higher mortality (HR per gram 1.024, 95% CI 1.021 to 1.027). Yearly cumulative exposure to doses ≤5 mg/day was associated with increased infection (HR 1.13, 95% CI 1.00 to 1.29) and MACE (HR 1.04, 95% CI 1.01 to 1.08).Conclusion Despite reductions in high-dose use, cumulative GC exposure in GCA remains substantial and associated with adverse outcomes, even at the lowest doses.INTRODUCTIONGlucocorticoids (GC) are still to date the cornerstone of giant cell arteritis (GCA) treat- ment, due to their rapid onset of action and remarkable efficacy, particularly in preventing ischaemic complications. However, disease relapses are common during tapering.1 2 As a result, patients with GCA have substantial cumulative GC exposure, both in terms ofWHAT IS ALREADY KNOWN ON THIS TOPIC⇒ Glucocorticoids (GC) are the cornerstone of giant cell arteritis (GCA) management, although their use is expected to decrease with the growing availability of GC-sparing drugs.⇒ Some features have been associated with risk of relapse, but factors associated with prolonged GC exposure are not known.⇒ GC- related toxicity has been established in chronic rheumatic conditions; however, few studies have quantified the effect of cumulative exposure.WHAT THIS STUDY ADDS⇒ Despite decrease of high-dose regimens in recent years and increased use of tocilizumab and meth- otrexate, cumulative GC exposure in the 2 years following GCA diagnosis remains high at 7.6 g in patients diagnosed in 2022.⇒ Female sex, age ≥75 years at diagnosis, hyperten- sion, chronic kidney disease and neurocognitive dis- orders were associated with slower tapering of GC.⇒ Each gram of GC was associated with mortality (HR 1.024, 95% CI 1.021 to 1.027), and cumulative ex- posure to doses as low as ≤5 mg/day was associat- ed with serious infection and major cardiovascular events.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY⇒ GC use in GCA remains significant and is associated with mortality as well as multiple morbidities.⇒ This study underscores the need to limit GC use in the management of GCA, in particular in patients most at risk of toxicity such as elderly patients. ### Page 2 2 Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866RMD OpenRMD OpenRMD Opendose and treatment duration. In a monocentric descrip- tive study conducted in France, Castan et al estimated a total median cumulative GC dose exceeding 9 g during follow- up,3 similar to a large British population-based study spanning from 2000 to 2011.4 These studies were conducted before the increasing usage of GC-sparing agents aimed at reducing GC-related morbidity.2 5 A more recent retrospective study published in 2022 including 77 patients found a median cumulative dose of 4.8 g in relapse-free patients, although half of the cohort expe- rienced a subsequent relapse.6 Likewise, a retrospective study of 47 patients with GCA diagnosed between 2020 and 2022 found that cumulative GC dose at 1 year ranged from 3.9 g in patients without relapse to 4.6 g in patients with at least one relapse.7 With the growing availability of disease-modifying anti- rheumatic drugs (DMARDs), there is renewed hope that GC exposure in the treatment of GCA could be reduced or even avoided in the foreseeable future.8 In France, their use is recommended in patients at high risk of GC-related toxicity in order to accelerate GC tapering. However, there is to date little real-world evidence describing shifts in practice patterns since the availability of new agents. In addition, while some studies have iden- tified factors associated with clinical relapse such as sex and large vessel involvement,9 robust investigations with large sample sizes investigating factors associated with prolonged GC exposure are lacking. Although the adverse effects associated with cumu- lative GC exposure are well recognised and have been described across multiple inflammatory diseases, including GCA,10–16 studies quantifying the relationship between duration of exposure at a given daily dose and adverse outcomes remain scarce. It therefore remains unclear how high-dose regimens or prolonged exposure to lower doses are associated with adverse outcomes such as mortality, infections, cardiovascular events or osteopo- rotic fractures. Quantifying these associations may help inform therapeutic decisions, including the early intro- duction of GC- sparing agents in patients at highest risk. Moreover, current GCA guidelines recommend tapering to ≤5 mg/day by 1 year and discontinuation by 2 years.17 However, there is no evidence that the objective of 5 mg/ day is safe. Therefore, the objectives of this nationwide retrospec- tive cohort study were first to characterise trajectories of oral GC use in the 2 years following GCA diagnosis and how these trajectories evolved over time. Second, to assess factors associated with time to sustained GC-free remission. Third, to evaluate the association between cumulative GC dose and cumulative duration of expo- sure time according to predefined dose categories (>30 mg/day, 15–30 mg/day, 5–15 mg/day, 1–5 mg/ day), with risk of mortality, major adverse cardiovascular events (MACE), serious infections and major osteopo- rotic fractures.PATIENTS AND METHODS Setting and designWe designed a retrospective, population-based cohort study within the French national health data (Système National des Données de Santé (SNDS)), which links the health insurance claims database, the hospital discharge database and the death certificates database. It provides real- world, quasi-exhaustive nationwide data, as in France all individuals are enrolled in the same social security system and any reimbursed service, at least partially, is captured in the database. It notably includes all in-hos- pital diagnostic codes on discharge summaries using International Classification of Diseases 10 (ICD-10) codes, drug deliveries including Anatomical Therapeutic Chemical classification and dosage, costly in-hospital treatments, long-term disability status with ICD-10 codes, date and cause of death from death certificates. It covers almost the entirety of the French population (67 million individuals) and is now routinely used for research purposes.18 19 We were granted full institutional access to all variables available for research purposes. SNDS is a pseudonymised database and no individual patient consent was required.ParticipantsWe included patients with incident GCA diagnosed between 1 January 2010 and 31 December 2022. Incident GCA was defined as the first hospitalisation with GCA recorded as the principal diagnosis. During the study period in France, hospitalisation was standard practice for almost all patients with new-onset GCA. Only one principal diagnostic code (ie, reason for hospitalisation) was required. The positive predictive value of GCA diag- nostic codes recorded as the principal diagnosis (ICD-10 codes M31.5 and M31.6) in the hospital discharge data- base has been reported to be 97%.20 The minimum look- back period was 3 years. To enhance specificity and ensure inclusion of incident GCA, we excluded patients with more than two GC reimbursements in the year preceding the first GCA code, as well as patients with any dispensing of tocilizumab or methotrexate prior to first GCA code (online supplemental table S1). As the study focused on GC exposure, we further required at least one GC prescription dispensed in the month following the index hospitalisation discharge. This first postdischarge dispensation defined the index date.ExposureOral GC exposure was assessed using data from the national health insurance claims database. Daily oral GC use during follow-up was estimated from the total amount of GCs dispensed per month. In case of hospitalisation, we assumed that the outpatient daily dose at the time of admission was maintained (conservative hypothesis) and provided by the hospital, thus any unconsumed medica- tion dispensed prior to hospitalisation was carried over and included in the calculation of the first postdischarge period. ### Page 3 3Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866VasculitisVasculitisVasculitisTocilizumab exposure was captured both through in-hospital intravenous infusions and subcutaneous injections dispensed in local pharmacies whereas meth- otrexate was only captured in local pharmacy dispensing (online supplemental table S1).OutcomesAdverse outcomes of interest were all-cause mortality, MACE defined as a composite outcome of myocardial infarction, stroke or death from cardiovascular cause, serious infection defined as an infection requiring hospi- talisation and major osteoporotic fracture (hip, pelvic, vertebral or humeral fracture). All ICD-10 codes and specific procedural codes (‘CCAM’) are presented in online supplemental table S1. We also evaluated time to sustained GC-free remission, which was defined as the first occurrence of a continuous 6-month period without any GC use.Statistical analysisProportions are expressed as percentages, and contin- uous variables as mean (SD) or median (Q1–Q3), as appropriate. First, we modelled daily GC dose trajectories over the 2 years following diagnosis using latent class growth mixture (LCGM) models, with daily dose modelled as a function of time treated as a continuous variable. Full details of model selection are provided in online supple- mental material, page 2. In the selected model, time was expressed as a natural spline with 4 df, and a patient-level random slope without random intercept was specified as a linear function of time. Patients’ characteristics within each trajectory were described without pairwise or global statistical comparisons to avoid highlighting statistically significant but clinically irrelevant differences given the large sample size. Second, to describe trends in GC use over time, we first calculated the cumulative GC dose within 2 years of diag- nosis, overall and according to year of GCA diagnosis. We then estimated Kaplan- Meier survival curves for time to sustained GC- free remission by calendar year of diagnosis as well as cumulative incidence of tocilizumab or metho- trexate prescription over time. Finally, to capture the burden associated with GC expo- sure in GCA, we used multivariable Fine-Gray models to estimate subdistribution HRs. For time to sustained GC-free remission, we used a time-fixed Fine- Gray model. One model included baseline variables only, and two additional models with time origins set at 3 and 6 months were fit to evaluate the effect of GC dose during the first 3 and 6 months, respectively, and included patients who were alive and still receiving GC at the corresponding time zero. Then, we used a multivariable Cox model to estimate HRs for all-cause mortality, and cause-specific Cox models for infection, MACE and major osteoporotic fractures, with two time- varying exposures of interest: (i) cumulative dose (in grams) and (ii) cumulative dura- tion of exposure to GCs (in years), within each of the four clinically relevant daily dose categories (1–5 mg/ day, 5–15 mg/day, 15–30 mg/day and over 30 mg/day). Person-time was allocated to each category according to the patient’s average daily GC dose each month. One model per outcome and cumulative exposure variable was performed. Models were adjusted for baseline vari- ables including age, sex, year of diagnosis, polymyalgia rheumatica, French Deprivation Index,21 dementia, hypertension, chronic kidney disease, atrial fibrillation, heart failure, stroke, myocardial infarction, diabetes, dyslipidaemia, cancer, proxies for tobacco use and obesity, anterior ischaemic optic neuropathy and osteo- porosis. Number of hospitalisation days for GCA, number of C reactive protein measurements per month, cumula- tive number of measurements since GCA diagnosis and prescriptions of tocilizumab and methotrexate were included as time-varying covariates. In sensitivity analyses, all analyses were repeated without excluding patients with more than two steroid prescription dispensations in the year before GCA diagnosis. We also repeated the multi- variable Fine- Gray models to look for factors associated with early sustained GC-free and DMARD-free remission. Covariate definitions are detailed in online supplemental table S1. All analyses were conducted in R V.4.3.3. LCGMs were fitted using the lcmm package.22RESULTSWe included 18 301 patients, of which 63% were female, mean age was 75.2 years (SD 8.7) and mean follow-up duration was 6.7 years (SD 3.7). A flow chart is available in online supplemental figure S1 and characteristics of patients included are shown in table 1. The mean 2-year cumulative oral GC dose following GCA diagnosis in the overall cohort was 8.9 g (SD 4.7). Sustained GC- free remission, defined as a continuous period of 6 months without any GC use, occurred in 7.7%, 35.6% and 80.6% of patients at 1, 2 and 5 years, respectively, with a median time to sustained GC-free remission from Kaplan-Meier estimator of 2.40 years (Q1–Q3: 1.74–4.07 years). Base- line use of methotrexate and tocilizumab was uncommon.Glucocorticoid daily dose trajectories and trends over timeFigure 1A presents the four identified trajectories of GC use. The different trajectories identified include: medium-dose quick taper (MDQT) (n=9397, 51.3%), low-dose quick taper (LDQT) (n=6143, 33.6%), high-dose quick taper (HDQT) (n=1880, 10.3%), high-dose slow taper (HDST) (n=881, 4.8%). The first three trajectories showed a rapid taper over about 6 months, followed by an increase in daily GC dose after 1 year of follow-up, and all trajectories converged at 2 years of follow-up (figure 1A an 1B). The proportion of patients in the HDQT and HDST decreased from 9.8% and 8.1% in 2010 to 7.3% and 1.6%, respectively, while the MDQT trajectory increased from 46.8% to 53.4% and the LDQT from 35.3% to 37.4% (figure 1C). The cumulative GC dose in the 3 and ### Page 4 4 Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866RMD OpenRMD OpenRMD Open6 months following GCA diagnosis was relatively stable across the study period with increasing calendar year, whereas the cumulative dose at 1 and 2 years of follow-up showed a modest decreasing trend (figure 1D). Patients diagnosed in 2022 had an average 2-year GC cumulative dose of 7.6 g (SD 3.1) vs 9.4 g (SD 4.8) in 2010. Patients’ characteristics across trajectories are summarised in table 1. Patients in the LDQT trajectory were older at diagnosis, were more frequently female and had a greater frequency of polymyalgia rheumatica at GCA diagnosis. In contrast, the HDQT trajectory was char- acterised by the highest proportion of males and patients with obesity, the highest proportion of anterior ischaemic optic neuropathy (AION) at diagnosis, while showing the lowest mean age at diagnosis. Patients in the HDST trajectory were least likely to be diagnosed in teaching hospitals (20%), and had the highest rate of deaths, the greatest GC cumulative dose and the longest GC treat- ment duration. GC-sparing agents were prescribed more frequently in the high-dose trajectories. The trajectories were not modified when including patients with preva- lent GC use (online supplemental figure S2).Glucocorticoid daily dose during follow-up and sparing agents use over timeThe proportion of living patients receiving GCs over time, stratified by daily dose, is shown in figure 2A. At 2 years mark, 55% of alive patients did not have GC dispensationTable 1 Patients’ characteristics overall and according to treatment trajectory All (n=18 301) Class 1: MDQT(n=9397) Class 2: LDQT(n=6143) Class 3: HDQT(n=1880)Class 4: HDST (n=881)Female 11 597 (63%) 5806 (62%) 4344 (71%) 896 (48%) 551 (63%) Age, years (mean, SD) 75.2 (8.7) 74.8 (8.5) 76.3 (8.9) 73.5 (8.7) 75.0 (9.0) Associated polymyalgia rheumatica 4596 (25%) 2126 (23%) 1852 (30%) 426 (23%) 192 (22%) Diagnosis in teaching hospital 5822 (34%) 3053 (35%) 2060 (36%) 550 (32%) 159 (20%) 2- year cumulative steroid dose, g (mean, SD) 8.9 (4.7) 9.2 (3.2) 5.6 (2.9) 13.5 (4.4) 18.3 (5.8) Baseline comorbiditiesAnterior optic ischaemic neuropathy 1190 (6.5%) 692 (7.4%) 233 (3.8%) 218 (12%) 47 (5.3%)Atrial fibrillation 2454 (13%) 1180 (13%) 898 (15%) 254 (14%) 122 (14%)Hypertension 13 339 (73%) 6776 (72%) 4465 (73%) 1443 (77%) 655 (74%)Lipid disorder 9475 (52%) 4854 (52%) 3187 (52%) 978 (52%) 456 (52%)Diabetes 4187 (23%) 2140 (23%) 1327 (22%) 524 (28%) 196 (22%)Smoker 2839 (16%) 1456 (15%) 893 (15%) 339 (18%) 151 (17%)Obesity 1167 (6.4%) 578 (6.2%) 295 (4.8%) 235 (13%) 59 (6.7%)Chronic kidney disease 980 (5.4%) 477 (5.1%) 370 (6.0%) 92 (4.9%) 41 (4.7%)Cardiac failure 1147 (6.3%) 507 (5.4%) 453 (7.4%) 124 (6.6%) 63 (7.2%)Ischaemic cardiopathy 1232 (6.7%) 622 (6.6%) 434 (7.1%) 130 (6.9%) 46 (5.2%)Stroke 649 (3.5%) 352 (3.7%) 206 (3.4%) 63 (3.4%) 28 (3.2%)Peripheral artery disease 300 (1.6%) 150 (1.6%) 96 (1.6%) 38 (2.0%) 16 (1.8%)History of osteoporotic fracture 561 (3.1%) 243 (2.6%) 237 (3.9%) 51 (2.7%) 30 (3.4%) Follow- upDeath 5442 (30%) 2552 (27%) 1982 (32%) 533 (28%) 375 (43%)Methotrexate at GCA diagnosis 150 (0.8%) 66 (0.7%) 65 (1.1%) 15 (0.8%) 4 (0.5%)Methotrexate during follow- up 2323 (13%) 1295 (14%) 595 (9.7%) 272 (14%) 161 (18%)Tocilizumab at GCA diagnosis 134 (1.3%) 52 (0.9%) 73 (2.1%) 8 (0.8%) 1 (0.3%)Tocilizumab during follow- up 1474 (14%) 810 (14%) 401 (11%) 186 (18%) 77 (21%)GC withdrawal at 2 years 6244 (34%) 2913 (31%) 2770 (45%) 493 (26%) 68 (7.7%)GC withdrawal at 5 years 13 425 (73%) 6914 (74%) 4714 (77%) 1334 (71%) 463 (53%)Time to achieve GC withdrawal* (years) (median, Q1, Q3) 2.15 (1.63, 3.06) 2.22 (1.75, 3.09) 1.87 (1.28, 2.62) 2.38 (1.84, 3.41) 3.33 (2.50, 4.71) *Withdrawal was defined as the earliest time point when a patient had no deliveries for a continuous period of 6 months. GC, glucocorticoid; GCA, giant cell arteritis; HDQT, high-dose quick taper; HDST, high-dose slow taper; LDQT, low-dose quick taper; MDQT, medium-dose quick taper. ### Page 5 5Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866VasculitisVasculitisVasculitisat this timepoint; however, at 5 years, this proportion was only 72.1%, with 16.7% being on daily doses ≤5 mg/ day and 11% on doses >5 mg/day. The median time to sustained GC- free remission decreased from 2.61 years (95% CI 2.55 to 2.65) in patients with GCA onset in the 2010–2015 period to 2.17 years (95% CI 2.13 to 2.21) among those with GCA onset in the 2020–2022 period (figure 2B). At 1 year, we observed 2388 patients in GC-free and DMARD- free remission out of 17 105 live patients (14%). While cumulative incidence of methotrexate usage over time seems to be stable since 2020, prescription of tocilizumab has sharply increased in recent years, with >25% of patients diagnosed in 2022 receiving tocili- zumab by the end of follow-up (figure 2C,D). Mean age at GCA diagnosis was 71.9 (SD 8.0) and 71.6 years (SD 8.0) for tocilizumab and methotrexate users, respectively, compared with 75.9 years (SD 8.7) among patients who did not receive either.Factors associated with time to sustained GC-free remissionThe main results from multivariable models of variables associated with time to sustained GC-free remission are presented in table 2. Factors which were associated with faster time to sustained GC-free remission were male sex, use of tocilizumab or methotrexate at baseline. Conversely, age ≥75 years, hypertension, neurocognitive disorder, chronic kidney disease, anterior optic ischaemic neuropathy and higher GC daily dose at index date were associated with slower time to GC-free remission. Higher cumulative GC dose in the first 3 and 6 months was also associated with longer time to GC-free remission as were daily GC dose above 15 mg/day at 3 months and any dose higher than 7.5 mg/day at 6 months. Cause-specific Cox models are shown in online supplemental table S2, with similar results except for neurocognitive disorder and chronic kidney disease, which had oppo- site associations in Cox and Fine-Gray models. This likely reflects the strong association between these comorbidi- ties and death. Results of Fine-Gray models were similar without excluding patients with prevalent GC (online supplemental table S4) and when considering sustained GC-free and DMARD- free remission (online supple- mental table S5).Adverse outcomes associated with GCs according to dose and cumulative duration of exposureCumulative dose of GC was significantly associated with all outcomes. The HR for all-cause mortality was 1.024 (95% CI 1.021 to 1.027) per gram of GC (table 3). HR for serious infection was 1.050 (95% CI 1.038 to 1.062), for MACE 1.015 (95% CI 1.010 to 1.020) and for majorFigure 1 Glucocorticoid use trajectories over time and cumulative glucocorticoid dose within 2 years of follow-up. (A) Daily glucocorticoid dose trajectories according to latent class growth mixture models. (B) 2-year glucocorticoid cumulative dose of 100 random individuals within each trajectory. (C) Proportion of patients per trajectory according to year of giant cell arteritis (GCA) diagnosis. (D) Cumulative glucocorticoid dose at 3, 6 months, 1 and 2 years according to year of GCA diagnosis. HDQT, high-dose quick taper; HDST, high-dose slow taper; LDQT, low-dose quick taper; MDQT, medium-dose quick taper. ### Page 6 6 Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866RMD OpenRMD OpenRMD Openosteoporotic fractures 1.013 (95% CI 1.006 to 1.019) per gram of GC. In table 3, we present HRs associated with different cumulative dose scenarios. For example, corre- sponding with a mean 2-year cumulative GC dose of 8.9 g, as observed in the whole cohort, the associated HR for all-cause mortality was 1.23 (95% CI 1.20 to 1.27). Results of multivariable models assessing the associa- tion of cumulative duration of exposure in years to GC daily dose categories with all-cause mortality, serious infections, MACE and major osteoporotic fractures are presented in figure 3. First, there seemed to be a dose-effect relationship regarding all-cause mortality (HR 1.43 (95% CI 1.30 to 1.57) at highest dose category vs 1.00 (95% CI 0.98 to 1.02) at lowest dose category) and infection risk, with HR 2.30 (95% CI 1.77 to 2.99) at highest dose category vs 1.13 (95% CI 1.001 to 1.29) at lowest dose category. Moreover, cumulative exposure to lowest dose category (≤5 mg/day) was not safe, with an increased associated risk of MACE and serious infections. Results were similar without excluding patients with prev- alent GC (online supplemental figure S3).DISCUSSIONIn this study, we identified four distinct GC usage trajec- tories. Over the past decade, the proportion of patients in trajectories characterised by high GC dose use has declined. Concurrently, the use of GC-sparing agents has steadily increased, especially tocilizumab, which is prescribed in more than a quarter of patients diagnosed in recent years. We identified that female sex, older age as well as some comorbidities such as chronic kidney disease, neurodegenerative disorders and hyperten- sion were associated with slower GC tapering. Cumula- tive duration of exposure to high doses of GC was more strongly associated with serious infections and all-cause mortality than cumulative duration of exposure to lower doses; however, exposure to low doses (≤5 mg/day) was not safe either. Although treatment strategies in GCA have been progressively shifting towards less GC use, our study shows that patients diagnosed in 2022 were still exposed to an average cumulative GC dose of 7.6 g in the 2 years following diagnosis. However, the latent class models clearly demonstrate that trajectories involving high doses, particularly those maintained over extended periods, have sharply decreased in recent years. Our findings suggest that this trend may be partly driven by the growing use of GC- sparing agents, and possibly in conjunction with recent guidelines that set explicit tapering goals.17 Studies specifically describing factors associated with duration of GC treatment in GCA are scarce and predom- inantly non-population-based, making them prone toFigure 2 Glucocorticoid (GC) daily dose and time to GC-free remission following diagnosis, trends in use of tocilizumab and methotrexate over time. (A) Proportion of patients according to daily GC dose over time. (B) Cumulative incidence of achieving GC withdrawal according to calendar year of giant cell arteritis (GCA) diagnosis. (C) Cumulative incidence of methotrexate initiation according to calendar year of GCA diagnosis. (D) Cumulative incidence of tocilizumab initiation according to year of GCA diagnosis (restricted to diagnosis from 2016 onwards). ### Page 7 7Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866VasculitisVasculitisVasculitisselection bias.23 Interestingly, male sex was associated with faster sustained GC-free remission. Moreover, consistent with recent results from Narvaez et al,24 we observed that older patients experienced a longer time to sustained GC-free remission. The same pattern was observed among patients with comorbidities such as neurocognitive disor- ders, chronic kidney disease or hypertension. This could reflect clinician caution driven by comorbidity burden or concern for adverse effects, rather than more severe underlying disease or true biological resistance. This is particularly concerning, since these vulnerable popula- tions are at higher risk of GC-related adverse effects and should be prioritised for timely GC withdrawal. Furthermore, we sought to address a complex ques- tion regarding the association of cumulative duration and cumulative dose with GC-induced morbidities. Most previous studies have focused on current daily dose and risk of adverse events, which does not account for prior and future cumulative dose or duration of exposure.10 25Table 2 Multivariable Fine-Gray model of factors associated with time to sustained GC-free remission with death as a competing eventSHR (95% CI) P valueAnalysis with T0 at index dateSex (male) 1.11 (1.07 to 1.15) <0.001Age ≥75 years 0.83 (0.80 to 0.86) <0.001Associated polymyalgia rheumatica 0.97 (0.94 to 1.01) 0.13Diagnosis in teaching hospital 1.03 (0.96 to 1.09) 0.43Tocilizumab at GCA diagnosis 1.39 (1.10 to 1.76) 0.006Methotrexate at GCA diagnosis 1.22 (1.02 to 1.45) 0.031Neurocognitive disorder 0.83 (0.75 to 0.91) <0.001Hypertension 0.90 (0.87 to 0.94) <0.001Chronic kidney disease 0.84 (0.77 to 0.93) <0.001History of cancer 0.97 (0.92 to 1.02) 0.19Diabetes 1.03 (0.99 to 1.07) 0.21Anterior optic ischaemic neuropathy 0.93 (0.86 to 1.00) 0.045History of MI or stroke 0.99 (0.93 to 1.06) 0.84History of osteoporotic fracture 0.97 (0.87 to 1.08) 0.54Initial daily steroid dose (per 10 mg/day) 0.99 (0.98 to 0.99) <0.001 Analysis with T0 at 3 months*Cumulative GC dose in the first 3 months (g) 0.94 (0.93 to 0.96) <0.001Daily GC dose at 3 months <0.001<7.5 mg/day Reference7.5–15 mg/day 0.95 (0.88 to 1.04)15–30 mg/day 0.87 (0.81 to 0.93)>30 mg/day 0.83 (0.76 to 0.90)Analysis with T0 at 6 months*Cumulative GC dose in the first 6 months (g) 0.95 (0.94 to 0.96) <0.001Daily GC dose at 6 months <0.001<7.5 mg/day Reference7.5–15 mg/day 0.79 (0.75 to 0.83)15–30 mg/day 0.66 (0.62 to 0.70)>30 mg/day 0.64 (0.59 to 0.70)Models were adjusted for age, sex, year of diagnosis, ICD-10 code for polymyalgia rheumatica, French Deprivation Index, dementia, hypertension, chronic kidney disease, atrial fibrillation, heart failure, stroke, myocardial infarction, diabetes, dyslipidaemia, cancer, proxies for tobacco use and obesity, anterior ischaemic optic neuropathy, falls and fractures. Some variables are not presented in the table for ease of presentation. *In these models, time 0 was set respectively at 3 and 6 months following index date, and included patients alive and on GC at time 0. GC, glucocorticoid; GCA, giant cell arteritis; ICD- 10, International Classification of Diseases 10; MI, myocardial infarction; SHR, subdistribution HR. ### Page 8 8 Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866RMD OpenRMD OpenRMD OpenAdverse events related to GC in GCA are frequent, as recently reported by Curtis et al in a retrospective cohort in the MarketScan database. They reported that 83% of the GCA population experienced at least one GC-related adverse event within the first year of diagnosis.26 Galeet al, using MarketScan and the UK Clinical Practice Research Datalink in patients diagnosed between 2000 and 2015, reported an increased risk of osteoporosis (HR 1.03, 95% CI 1.02 to 1.04) and serious infection (HR 1.03, 95% CI 1.02 to 1.04) per gram of GC cumulative expo- sure, with no increased risk of myocardial infarction. However, the requirement for at least 1 year of follow-up may have introduced selection bias by excluding patients who died early from cardiovascular events, potentially underestimating cardiovascular disease risk.27 In another case-control study in the UK Clinical Practice Research Datalink, Wilson et al found that being currently exposed to doses >30 mg/day was associated with higher risk of death (OR 2.1, 95% CI 1.3 to 3.5), infection (OR 3.3, 95% CI 2.2 to 5.2) and fractures (OR 2.6, 95% CI 1.6 to 4.3), although this did not take into account the duration of exposure to such dose.16 Despite methodological differences, our results are in line with these findings. We found that exposure to doses >5 mg/day is associated with mortality, with a dose-dependent trend. Furthermore, the HR for mortalityTable 3 HR per outcome associated with cumulative glucocorticoid doseAll- cause mortality HR (95% CI) Serious infection HR (95% CI) MACE HR (95% CI) Major osteoporotic fracture HR (95% CI)HR per gram of cumulative GC 1.024 (1.021 to 1.027) 1.050 (1.038 to 1.062) 1.015 (1.010 to 1.020) 1.013 (1.006 to 1.019) Predicted HR according to different mean 2-year cumulative GC dose scenarios (vs no GC)Whole cohort: 8.9 g 1.23 (1.20 to 1.27) 1.54 (1.39 to 1.70) 1.14 (1.09 to 1.20) 1.12 (1.06 to 1.18)MDQT trajectory: 9.2 g 1.24 (1.21 to 1.28) 1.56 (1.41 to 1.73) 1.15 (1.09 to 1.20) 1.12 (1.06 to 1.19)LDQT trajectory: 5.6 g 1.14 (1.12 to 1.16) 1.31 (1.23 to 1.39) 1.09 (1.05 to 1.12) 1.07 (1.03 to 1.11)HDQT trajectory: 13.5 g 1.37 (1.32 to 1.43) 1.93 (1.66 to 2.24) 1.22 (1.14 to 1.31) 1.18 (1.09 to 1.29)HDST trajectory: 18.3 g 1.54 (1.45 to 1.63) 2.43 (1.99 to 2.98) 1.31 (1.19 to 1.45) 1.26 (1.12 to 1.41) Predicted HRs are expressed relative to a theoretical reference of zero cumulative glucocorticoid exposure. Predicted HRs were derived by raising the per-gram HR to the power of the corresponding mean dose. GC, glucocorticoid; HDQT, high-dose quick taper; HDST, high-dose slow taper; LDQT, low-dose quick taper; MACE, major adverse cardiovascular events; MDQT, medium-dose quick taper.Figure 3 Multivariable cause- specific Cox models evaluating the association between glucocorticoid cumulative duration of exposure and risk of all-cause mortality, serious infections, major cardiovascular events and major osteoporotic fractures, according to daily glucocorticoid dose category. All variables are expressed as continuous variables, with cumulative time expressed in years. Infections were defined as hospitalisation for infections. Major osteoporotic fractures were defined as in-hospital code for vertebral, hip, pelvic, humeral or multiple rib fractures. Models were adjusted for age, sex, year of diagnosis, International Classification of Diseases 10 code for polymyalgia rheumatica, French Deprivation Index, dementia, hypertension, chronic kidney disease, atrial fibrillation, heart failure, stroke, myocardial infarction, diabetes, cancer, proxies for tobacco use and obesity, anterior ischaemic optic neuropathy, falls and fractures at baseline, tocilizumab or methotrexate use, number of hospitalisations for giant cell arteritis and C reactive protein laboratory measurement during follow-up. Cum time, cumulative time; MACE, major adverse cardiovascular events. ### Page 9 9Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866VasculitisVasculitisVasculitisincreased by 2.4% per gram of cumulative GC exposure. Based on our findings, HR for all-cause mortality asso- ciated with the mean cumulative dose observed in our cohort was thus 1.23, which is considerable. These results are particularly relevant given that recent randomised trials of tocilizumab and upadacitinib have demon- strated reductions of 1–2 g in cumulative GC exposure compared with placebo, yet lacked the statistical power and follow-up duration to detect the clinical benefits of such reductions.2 5 We also found that even at the lowest dose categories (≤5 mg/day), cumulative exposure was still associated with increased serious infection risk, suggesting even such low doses are not safe. Regarding cardiovascular events, our findings align with those of Pujades-Rodriguez et al, who concluded in a study including patients with multiple inflammatory conditions, including GCA, that no GC dose is entirely safe.10 Interestingly, we did not observe a dose- dependent effect regarding risk of MACE. This may reflect limited statistical power, but alternatively, GCs may act as proatherogenic agents,28 in which case cumulative exposure duration rather than dose intensity could be the more relevant risk factor. Likewise, as other studies reported in other diseases, we showed that in GCA, there is an association between GCs and increased risk of osteo- porotic fractures even at doses inferior to 7.5 mg/day.29 30 Overall, although cumulative exposure to doses below 5 mg/day was not associated with an increased risk of all-cause mortality, it was associated with major GC-related morbidities. This finding is particularly significant as we show that a non-negligible proportion of patients are exposed to GCs for longer than 2 years, even in recent years. Despite the strengths of this population-based method- ology, which limits selection bias and allows for a large sample size, and the complete capture of patients’ phar- macy medications dispensed, we acknowledge several limitations. As with most studies using claims databases, we did not have access to medical records and were therefore unable to assess clinical or biological vari- ables. Consequently, disease activity could not be fully adjusted for and was instead approximated using proxies, including initiation of steroid-sparing agents, C reactive protein measurements and hospitalisation for GCA. Moreover, in- hospital prescriptions were unavailable, including methylprednisolone pulses, which may have led to underestimation of cumulative GC exposure and precluded evaluation of its association with sustained GC-free remission. Residual confounding remains possible. Regarding the diagnostic accuracy of ICD-10 codes for GCA, previous studies have shown excellent performance.20 To ensure we included incident patients with GCA, we excluded patients with more than two GC prescription dispensation in the year before the GCA code and included only those with a GC dispensation within 1 month after the GCA code. Although it was common clinical practice during the study period to hospitalise all patients with new-onset GCA, it remains possible that we missed some patients with mild disease activity who were not hospitalised. Finally, while outcomes such as all-cause mortality and hip fractures are reliably recorded in the database, other events such as vertebral fractures are known to be under-reported. As a result, GC-related osteoporotic events are likely underestimated. We believe our work has several important implica- tions. First, it shows that despite efforts over the past decade, average GC use in GCA remains high and sustained. Second, this study should serve as a reference point for evaluating GC use in the coming years, espe- cially with the increasing adoption of GC- sparing agents. Third, although randomised trials have aimed to reduce GC use in GCA, the morbidity associated with reduced GC exposure had not been clearly established. Our find- ings highlight that the burden associated with cumulative GC exposure remains important, and that even low GC doses do not appear to be safe. Overall, our work under- scores the need for continuous and accelerated efforts to reduce GC use in GCA.CONCLUSIONIn the first 2 years following GCA diagnosis, four distinct oral GC dose trajectories were identified: MDQT, LDQT, HDQT and HDST. Although tapering has become faster and the use of GC-sparing agents has increased in recent years, cumulative GC exposure remains substantial, both in terms of dose and duration. This is particularly relevant since each additional cumulative gram of GC is associated with increased risk of mortality, serious infection, MACE and major osteoporotic fractures. A dose-dependent rela- tionship was also observed between daily dose and the risks of mortality and serious infection. These findings support continued efforts to limit GC use in the manage- ment of GCA.Author affiliations1Sorbonne Université, INSERM, Institut Sorbonne de Santé Publique, Equipe PEPITES, AP-HP, Hôpital Pitié Salpêtrière, Département de Santé Publique, Centre de Pharmacoépidémiologie (Cephepi), Paris, France 2Arthritis Research Canada, Vancouver, British Columbia, Canada 3Division of Rheumatology, Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada 4Sorbonne Université - APHP, CRI-IMIDIATE Clinical Research Network, Department of Rheumatology, Hôpital Pitié Salpêtrière, Paris, France 5Department of Rheumatology, Hôpital Bicêtre, Assistance Publique - Hôpitaux de Paris, Université Paris-Saclay, Orsay, France 6Centre de Recherche en Epidémiologie et Santé des Populations, Université Paris-Saclay, Orsay, France 7Inserm, Institut Gustave Roussy, Université Versailles Saint Quentin, Versailles, FranceContributors MB and FT contributed to the study conceptualisation and design. MB performed data extraction and statistical analysis. MB, DL, FT, BF, RS and DH contributed to data analysis and interpretation. MB drafted the original version of the manuscript. MB, DL, FT, BF, RS and DH reviewed and revised the manuscript. MB is the guarantor. During the preparation of this work, MB used OpenAI chatGPT V.5.0 to assist with manuscript writing, specifically to correct grammatical and syntactical errors, as the author is not a native English speaker. No AI tools were used for data generation or interpretation. After using this tool, the author reviewed and edited the content as needed and takes full responsibility for the content of the published article. ### Page 10 10 Beydon M, et al. RMD Open 2026;12:e006866. doi:10.1136/rmdopen-2026-006866RMD OpenRMD OpenRMD OpenFunding MB received a grant for doctoral research from the National Institute of Health and Medical Research (INSERM) and a grant from Assistance Publique des Hôpitaux de Paris (CRC).Competing interests BF received consulting fees and honoraria from AbbVie, Alfasigma, Amgen, Bristol Myers Squibb, Celltrion, Janssen, Lilly, Nordic Pharma, Novartis, Pfizer, Sobi and UCB. RS received consulting fees and honoraria from GlaxoSmithKline, Boehringer Ingelheim, Kiniksa, Janssen, Bristol Myers Squibb, AstraZeneca, Alfasigma and Novartis, and received support for attending meetings from GlaxoSmithKline, Novartis and Amgen. FT is Head of the Centre de Pharmacoépidémiologie (Cephepi) of Assistance Publique – Hôpitaux de Paris and of the Clinical Research Unit of Pitié-Salpêtrière Hospital, which have received research funding, grants and consultancy fees from a large number of pharmaceutical companies that have contributed indiscriminately to the salaries of their employees. FT has not received any personal remuneration from these companies.Patient consent for publication Not applicable.Provenance and peer review Not commissioned; externally peer reviewed.Data availability statement Data may be obtained from a third party and are not publicly available. The data used in this study are not publicly available due to confidentiality and privacy restrictions. Data are available to individuals with clearance from the Caisse Nationale de l’Assurance Maladie to access the SNDS.Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: https://creativecommons.org/licenses/by-nc/4.0/.ORCID iDsMaxime Beydon https://orcid.org/0000-0003-4924-093X Bruno Fautrel https://orcid.org/0000-0001-8845-4274 Raphaèle Seror https://orcid.org/0000-0002-5523-1856REFERENCES1 Narváez J, Domínguez-Álvaro M, Hernández-Rodríguez I, et al. Epidemiology of major and minor relapses in giant cell arteritis according to EULAR definitions: insights from the ARTESER registry.Rheumatology (Oxford) 2025;64:4985–94.2 Stone JH, Tuckwell K, Dimonaco S, et al. Trial of tocilizumab in giant- cell arteritis. N Engl J Med 2017;377:317–28.3 Castan P, Dumont A, Deshayes S, et al. Impact of glucocorticoid cumulative doses in a real-life cohort of patients affected by giant cell arteritis. J Clin Med 2022;11:1034.4 Petri H, Nevitt A, Sarsour K, et al. Incidence of giant cell arteritis and characteristics of patients: data-driven analysis of comorbidities.Arthritis Care Res (Hoboken) 2015;67:390–5.5 Blockmans D, Penn SK, Setty AR, et al. A phase 3 trial of upadacitinib for giant-cell arteritis. N Engl J Med 2025;392:2013–24.6 Felten L, Leuchten N, Aringer M. Glucocorticoid dosing and relapses in giant cell arteritis-a single centre cohort study. Rheumatology (Oxford) 2022;61:1997–2005.7 Mensch N, Hemmig AK, Aschwanden M, et al. Rapid glucocorticoid tapering regimen in patients with giant cell arteritis: a single centre cohort study. RMD Open 2023;9:e003301.8 Christ L, Seitz L, Scholz G, et al. Tocilizumab monotherapy after ultra- short glucocorticoid administration in giant cell arteritis: a single- arm, open- label, proof- of- concept study. Lancet Rheumatol2021;3:e619–26.9 Moreel L, Betrains A, Molenberghs G, et al. Epidemiology and predictors of relapse in giant cell arteritis: a systematic review and meta- analysis. Joint Bone Spine 2023;90:105494.10 Pujades- Rodriguez M, Morgan AW, Cubbon RM, et al. Dose-dependent oral glucocorticoid cardiovascular risks in people with immune- mediated inflammatory diseases: a population-based cohort study. PLoS Med 2020;17:e1003432.11 George MD, Baker JF, Winthrop K, et al. Risk for serious infection with low-dose glucocorticoids in patients with rheumatoid arthritis: a cohort study. Ann Intern Med 2020;173:870–8.12 Broder MS, Sarsour K, Chang E, et al. Corticosteroid- related adverse events in patients with giant cell arteritis: a claims-based analysis. Semin Arthritis Rheum 2016;46:246–52.13 Changes in mortality risk after stopping glucocorticosteroids—a population-based study in rheumatoid arthritis [abstract]. ACR Meeting Abstracts [Internet]. Available: https://acrabstracts.org/abstract/changes-in-mortality-risk-after-stopping-glucocorticosteroids-a-population-based-study-in-rheumatoid-arthritis/14 Avina-Zubieta JA, Abrahamowicz M, De Vera MA, et al. Immediate and past cumulative effects of oral glucocorticoids on the risk of acute myocardial infarction in rheumatoid arthritis: a population-based study. Rheumatology 2013;52:68–75.15 Aviña-Zubieta JA, Abrahamowicz M, Choi HK, et al. Risk of cerebrovascular disease associated with the use of glucocorticoids in patients with incident rheumatoid arthritis: a population-based study. Ann Rheum Dis 2011;70:990–5.16 Wilson JC, Sarsour K, Collinson N, et al. Serious adverse effects associated with glucocorticoid therapy in patients with giant cell arteritis (GCA): a nested case-control analysis. Semin Arthritis Rheum 2017;46:819–27.17 Hellmich B, Agueda A, Monti S, et al. 2018 Update of the EULAR recommendations for the management of large vessel vasculitis.Ann Rheum Dis 2020;79:19–30.18 Tuppin P, de Roquefeuil L, Weill A, et al. French national health insurance information system and the permanent beneficiaries sample. Rev Epidemiol Sante Publique 2010;58:286–90.19 Maillard O, Bun R, Laanani M, et al. Use of the French National Health Data System (SNDS) in pharmacoepidemiology: a systematic review in its maturation phase. Therapie 2024;79:659–69.20 Caudrelier L, Moulis G, Lapeyre-Mestre M, et al. Validation of giant cell arteritis diagnosis code in the French hospital electronic database. Eur J Intern Med 2019;60:e16–7.21 Rey G, Jougla E, Fouillet A, et al. Ecological association between a deprivation index and mortality in France over the period 1997 - 2001: variations with spatial scale, degree of urbanicity, age, gender and cause of death. BMC Public Health 2009;9:33.22 Proust-Lima C, Philipps V, Liquet B. Estimation of extended mixed models using latent classes and latent processes: the R package lcmm. J Stat Softw 2017.23 Moreel L, Betrains A, Molenberghs G, et al. Duration of treatment with glucocorticoids in giant cell arteritis: a systematic review and meta- analysis. J Clin Rheumatol 2023;29:291.24 Narváez J, Domínguez-Álvaro M, Galíndez-Agirregoikoa E, et al. Sustained drug- free remission in giant cell arteritis. Rheumatology (Oxford) 2025;64:2919–27.25 Listing J, Kekow J, Manger B, et al. Mortality in rheumatoid arthritis: the impact of disease activity, treatment with glucocorticoids, TNFαinhibitors and rituximab. Ann Rheum Dis 2015;74:415–21.26 Curtis JR, Kruzikas DT, Romero A, et al. Real-world analysis of patient characteristics, adverse events, and economic burden among patients with giant cell arteritis treated with glucocorticoids.Rheumatol Ther 2025;12:835–54.27 Gale S, Wilson JC, Chia J, et al. Risk associated with cumulative oral glucocorticoid use in patients with giant cell arteritis in real-world databases from the USA and UK. Rheumatol Ther 2018;5:327–40.28 Bouillet B, Gautier T, Denimal D, et al. Glucocorticoids impair HDL-mediated cholesterol efflux besides increased HDL cholesterol concentration: a proof of concept. Eur J Endocrinol2020;183:EJE- 20- 0477:297–306:.29 Van Staa TP, Leufkens HG, Abenhaim L, et al. Use of oral corticosteroids and risk of fractures. J Bone Miner Res2000;15:993–1000.30 Balasubramanian A, Wade SW, Adler RA, et al. Glucocorticoid exposure and fracture risk in a cohort of US patients with selected conditions. J Bone Miner Res 2018;33:1881–8.