Persistence Among Biologic/JAKi-Naive Medicare Beneficiaries with RA Initiating Synthetic DMARDs

Abstract

Introduction

Persistence to therapy to treat rheumatoid arthritis (RA) is an indirect measure of tolerability and effectiveness in the real-world setting. Previous clinical trials suggested that seropositive patients with RA may particularly benefit from abatacept. The risk of non-persistence after initiating abatacept, tumor necrosis factor inhibitors (TNFi), and Janus kinase inhibitors (JAKi) as first-line biologic or targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs) was examined.

Methods

We utilized the 100% Medicare Fee-For-Service sample linked claims with Prognos laboratory data from 2012 to 2019. Newly-initiating, anti-cyclic citrullinated peptide (anti-CCP)-positive and rheumatoid factor-(RF) positive patients were identified. Index date was initiation of abatacept, TNFi, or JAKi as first-line b/tsDMARD treatment. Persistence (measured at 12 months post-index) was defined as the absence of a treatment gap ≥ 60 days, > 90 days off-treatment, or switch in therapy. Cox regression was used to investigate risk of non-persistence between the groups.

Results

Of 3105 patients identified, 487 received abatacept (16%), 2330 TNFi (75%), and 288 JAKi (9%). Abatacept, TNFi, and JAKi twelve-month persistence was 48%, 33%, and 39%, respectively. Beneficiaries initiating with abatacept were more likely to be persistent than those initiating with TNFi or JAKi (hazard ratio [HR] for TNFi, 1.45, 95% CI 1.27–1.64; HR for JAKi, 1.43, 95% CI 1.19–1.72).

Conclusion

These real-world findings suggest that abatacept as a 1L treatment has the potential to improve persistence in patients with anti-CCP+ and RF+RA compared to the most commonly used 1L alternatives.

Key Summary Points

Why carry out this study?

In patients with early rheumatoid arthritis (RA), treatment persistence predicts better outcomes.

This study used Medicare claims data to examine real-world rates of treatment persistence in patients with anti-cyclic citrullinated peptide (anti-CCP)+, rheumatic factor (RF)+ RA initiating first-line abatacept, tumor necrosis factor inhibitors (TNFi), and Janus kinase inhibitors (JAKi).

What was learned from this study?

The rates of treatment persistence at 12 months were 48%, 33%, and 39% with abatacept, TNFi, and JAKi, respectively.

Clinicians should consider a patient’s treatment experience, as well as their RF and anti-CCP test results, when considering durable options for the treatment of RA.

Introduction

Rheumatoid arthritis (RA) is the most common type of autoimmune arthritis and affects more than 1.3 million people in the United States, approximately 75% of whom are women [1]. Blood tests used to inform disease diagnosis, prognosis, and management of RA include rheumatoid factor (RF) and anti-cyclic citrullinated peptide (anti-CCP) – positive in approximately 60–70% of patients with RA [1]. Patients with seropositive RA (defined as a positive test result for RF and/or anti-CCP) have been shown to have increased radiographic progression and joint damage [2, 3], both of which have potential implications for symptom management and treatment effectiveness [4].

Treatment options for RA have expanded over the past 30 years. Currently available therapies include conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), biologic DMARDs (bDMARDs), targeted synthetic DMARDs (tsDMARDs) as well as non-steroidal anti-inflammatory drugs (NSAIDs) and glucocorticoids used for symptoms and bridging therapy [5, 6]. The newest addition to RA treatments is Janus kinase inhibitors (JAKi), a class of oral drugs which have been shown to be comparable or superior to biologic agents despite some safety concerns and utilization restrictions in the USA [7, 8].

Real-world research has shown that persistence to therapy predicts better outcomes for patients with early RA [9]. Persistence to RA treatment is important, as persistent patients avoid gaps in treatment that can affect the efficacy of treatment and is an indirect measure of acceptable tolerability and effectiveness. Persistence is necessary to achieve clinical improvement and benefit from therapy, specifically disease remission or low disease activity [10–12].

In 2005, abatacept, a bDMARD for the treatment of moderate to severe adult RA, was approved by the US FDA. The Early AMPLE head-to-head clinical trial [13] has shown that abatacept performs numerically better than adalimumab across several endpoints in patients with biologic-naive moderate to severe RA who were anti-citrullinated protein antibody- (ACPA) positive and RF-positive. However, treatment persistence in patients with anti-CCP antibodies and RF+RA has yet to be studied in the real world. Therefore, the aim of the present study was to examine real-world treatment persistence (measured by discontinuation and switching) and treatment patterns among Medicare Fee-For-Service (FFS) beneficiaries with anti-CCP antibodies and RF+RA who are biologic- and JAKi-naive and initiate first-line (1L) b/tsDMARD treatment on abatacept, tumor necrosis factor-inhibitors (TNFi), or JAKi.

Methods

Study Design and Data Sources

This retrospective cohort study utilized the 100% sample of Medicare FFS enrollment and Parts A/B claims data and Part D Prescription Drug Event (PDE) data linked to Prognos laboratory data from January 1, 2012, to December 31, 2019. The 100% Medicare FFS Research Identifiable Files include all Part A/B medical encounters for Medicare FFS beneficiaries, and the PDE data contain all pharmacy encounters for this population. The Medicare Master Beneficiary Summary File was used to determine beneficiary monthly eligibility for Part A/B services, patient demographics, and where applicable death. Prognos laboratory data were collected from national and regional labs and are standardized for research. Data linkage was performed using a HIPAA-compliant de-identification and matching algorithm.

Study Population

The study population included Medicare FFS beneficiaries with RA who initiated biologic or JAKi treatment in the first line and had both RF-positive and anti-CCP–positive lab results. As this was a secondary data analysis of statistically de-identified data, obtaining ethics approval and patient consent was not necessary. Approval to access the Medicare data is governed by an established research oriented data use agreement with Centers for Medicare & Medicaid Services. Approval to access the Prognos laboratory data was governed by a data use agreement with Prognos.

Adult beneficiaries were required to have at least two claims separated by ≥ 7 days with an International Classification of Diseases [ICD] diagnosis code for RA (ICD-9-CM]: 714.0; ICD-10-C: M054x-M055x, M057x-M058x, M060x, M062x-M064x, M068x-M069). Individuals were required to have at least 1 claim for an RA treatment of interest during the index period of January 1, 2013, to December 31, 2018, within 60 days after RA diagnosis (Fig. 1). The 60-day switching threshold and allowable treatment gap were designed to reflect real-world clinical practice and adherence patterns while aligning with established RA persistence methodology to ensure meaningful comparisons across treatment groups. The index date was defined as the date of the first RA treatment claim. The index therapies of interest included abatacept, etanercept, adalimumab, infliximab (and biosimilars), certolizumab pegol, golimumab, tofacitinib, and baricitinib. Infusion medications were also captured in the dataset. Note that upadacitinib was not available during the study period. Patients on other treatments at index date were excluded from the study. However, patients were allowed to be on concomitant csDMARDs. Beneficiaries were required to have positive test results for RF and anti-CCP tests prior to index date. Patients must have had at least 6 months of pre-index and 12 months of post-index continuous enrollment in Medicare FFS Parts A/B/D. In order to identify biologic/JAKi-naive patients, patients were required to have no claims for a biologic or JAKi for 6 months prior to index date.

Fig. 1.

Study schematic. JAKi Janus kinase inhibitors; RA rheumatoid arthritis, TNFi tumor necrosis factor inhibitor

Patients were excluded if they had claims for multiple biologics on the same day, were pregnant within 30 days pre- or post-index date or had ≥ 2 claims for specific cancer types on different days prior to treatment initiation in which biologic or targeted therapies may be utilized.

Treatment groups with sample sizes too small to detect a significant difference (patients initiating on interleukin-6 receptor inhibitors with n = 105) were excluded. Finally, patients initiating rituximab were excluded as it is typically indicated for use in patients who have failed other biologics (ie, not b/tsDMARD-naive) or in patients with serious comorbidities such as cancer or interstitial lung disease.

Exposure Variables: Index RA Therapy

Patients were placed into three treatment groups and outcomes were evaluated based on their index RA therapy: abatacept, TNFi (etanercept, adalimumab, infliximab [and biosimilars], certolizumab pegol, golimumab), and JAKi (tofacitinib, baricitinib).

Outcomes

The probability of persistence at 12 months was the primary study outcome. Treatment persistence at 12 months post-index was also measured descriptively. Individuals were considered persistent if they did not discontinue the index treatment or switch to an alternative treatment during follow-up. Treatment discontinuation was defined as the presence of a gap in treatment ≥ 60 days or total time off treatment  > 90 days over a year. Treatment switching was defined as the initiation of a second RA treatment group within 60 days of the discontinuation date of the index therapy (defined by the date of the last day’s supply of therapy) during follow-up.

The duration of persistence was calculated as: (a) from the index date until the first treatment gap (ie, end of last prescription) ≥ 60 days or switch in therapy for non-persistent beneficiaries, whichever appeared earlier, or (b) from the index date until the end of the 12-month period for persistent beneficiaries. Consistent with past research, this analysis allowed for a < 60 days gap in days of supply across the entire year [14]. Given the allowable treatment gap (< 60 days) in between prescriptions, patients would be considered persistent through the year if they had ≥ 305 days of treatment.

Covariates

Baseline characteristics at index date included age, sex, race, census region, Deyo–Charlson Comorbidity Index, the original reason for Medicare entitlement, dual-eligibility in Medicare/Medicaid status, eligibility for a low-income subsidy, glucocorticoid use, serious infection, and csDMARD use.

Statistical Analysis

A descriptive analysis was conducted to summarize the baseline characteristics. All results were reported by the index treatment. Categorical variables were reported as frequencies and relative frequencies and the measures of central tendency (ie, mean, standard deviation [SD], median, and interquartile range [IQR]) were reported for continuous variables.

Kaplan–Meier (KM) curves were constructed to present the unadjusted persistence through the 12 months follow-up by treatment group. Treatment discontinuation, switching prior to discontinuation, or reaching the end of the 12-month follow-up period were considered events marking the end of the persistent period.

The risk of non-persistence (discontinuation or switch) over 12 months by treatment group was estimated using a Cox proportional hazards model, adjusting for the baseline demographic and clinical covariates, as outlined above. Patients were censored at treatment discontinuation, switching, or at the end of the 12-month follow-up period, whichever occurred first. Individual variables were evaluated during the univariate analyses to determine which variables were included in the multivariable Cox regression models. A two-sided p-value of < 0.05 was considered statistically significant.

Results

Sample Characteristics

A total of 3105 b/tsDMARD-naive patients with anti-CCP antibodies and RF+RA were included in the final analytic study sample (Fig. 2). This included 487 patients with an index therapy of abatacept (16%), 2330 patients initiating on TNFi (75%), and 288 patients initiating on JAKi (9%).

Fig. 2.

Sample selection of bDMARD/JAKi-naive, anti-CCP+ and RF+ beneficiaries with RA within the 100% sample of Medicare FFS claims data from 2012–2019. bDMARD biologic disease-modifying antirheumatic drug; CCP cyclic citrullinated peptide; FFS Fee-For-Service; ICD International Classification of Diseases; JAKi Janus kinase inhibitor; RA rheumatoid arthritis, RF rheumatoid factor; TNFi tumor necrosis factor inhibitor. ^aICD-9-CM: 714.0; ICD-10-CM: M054x-M055x, M057x-M058x, M060x, M062x-M064x, M068x-M069x. ^bAged <18 years, pregnant within 30 days pre- or post-index date, 2 claims for specific cancer types on different days prior to treatment in which biologic or targeted therapies of interest may be used.

All baseline characteristics are described in Table 1. There was a higher proportion of female patients identified with JAKi (82%) compared with patients identified on abatacept or TNFi (both 77%). Patients initiating treatment on abatacept had a higher mean age (68.8 years, SD 8.6) compared to patients identified on TNFi (66.2 years, SD 10.0) and JAKi (63.8 years, SD 11.0). A higher proportion of patients identified on abatacept were white (77%) compared to those initiating on TNFi (74%) and JAKi (63%). The highest proportion of Black patients was in the JAKi cohort (26%). Similarly, patients initiating with JAKi also had the highest proportion of full or partial dual eligibility in Medicare and Medicaid (60%) and Part D low-income subsidy eligibility (72%) compared to patients initiating with abatacept (25% and 30%, respectively) and those on TNFi (41% and 48%, respectively).

Table 1.

Baseline characteristics at initiation of abatacept, TNFi, or JAKi for biologic/JAKi-naive, anti-CCP+ and RF+ patients with RA within the 100% sample of Medicare FFS claims data in 2012–2019

Variable	Abatacept (n = 487)	TNFi (n = 2330)	JAKi (n = 288)
Female, n (%)	377 (77)	1797 (77)	235 (82)
Mean age, years (SD)	68.8 (8.6)	66.2 (10.0)	63.8 (11.0)
Race, n (%)
White	375 (77)	1734 (74)	181 (63)
Black	69 (14)	375 (16)	74 (26)
Othera	43 (9)	221 (10)	33 (11)
Census region, n (%)
Northeast	67 (14)	303 (13)	39 (14)
South	44 (9)	208 (9)	18 (6)
Midwest	308 (63)	1470 (63)	193 (67)
West/Unknown	68 (14)	349 (15)	38 (13)
Original reason for entitlement code, n (%)
Old age and survivor's insurance	321 (66)	1273 (55)	107 (37)
End-stage renal disease, disability, or both	166 (44)	1057 (45)	181 (63)
Dual eligibility, full or partial, n (%)	122 (25)	960 (41)	173 (60)
Part D low-income subsidy eligibility, full or partial, n (%)	146 (30)	1112 (48)	208 (72)
Mean Charlson Comorbidity Index score (SD)	3.1 (2.1)	2.9 (2.0)	3.0 (2.1)
Glucocorticoid use, n (%)	311 (68)	1581 (68)	190 (66)
Serious infection, n (%)	157 (32)	673 (29)	81 (28)
Concomitant csDMARD use, n (%)	401 (82)	2007 (86)	240 (83)

^a“Other” includes Asian and other categories. Categories combined to avoid presenting cell sizes < 11

CCP cyclic citrullinated peptide; csDMARD, conventional synthetic disease-modifying antirheumatic drug; JAKi Janus kinase inhibitor; RA rheumatoid arthritis, RF rheumatoid factor; SD standard deviation; TNFi, tumor necrosis factor inhibitor

Unadjusted Analysis

Across the abatacept, TNFi, and JAKi groups, 48%, 33%, and 39% of patients were persistent at 12 months of follow-up as the primary outcome of the study. The abatacept group also had a longer median (IQR) duration of persistence of 276 days (119–365) compared to patients indexing on TNFi (202 days [105–365]) and JAKi (197 days [58–365]) (Table 2). Additionally, the KM curves (Fig. 3) also show that a larger proportion of patients in the abatacept group were persistent at 12 months (48% vs 33% of TNFi cohort and 39% of JAKi cohort) compared to the other two treatment groups.

Table 2.

Treatment patterns of biologic/JAKi-naive, anti-CCP+ and RF+ patients with RA within the 100% sample of Medicare FFS claims data in 2012–2019, by treatment group

Outcome over 12 months	Abatacept (n = 487)	TNFi (n = 2330)	JAKi (n = 288)
Treatment persistence at 12 months, n (%)	233 (48)	778 (33)	112 (39)
Median (IQR) days of duration of persistencea	276 (119, 365)	202 (105, 365)	197 (58, 365)
Treatment discontinuation,b n (%)	195 (40)	1,404 (60)	159 (55)
Treatment switching, switching prior to discontinuation,c n (%)	59 (12)	148 (6)	17 (6)

The sum of beneficiaries who discontinue and switch totals the number of beneficiaries who are not persistent over 12 months

^aFrom treatment initiation to discontinuation of therapy, switch in therapy, or end of follow-up. This analysis of persistence allowed for a gap in therapy < 60 days. Time off therapy was limited to no more than 90 days across the entire year

^bTreatment discontinuation was defined as the presence of a gap in therapy ≥ 60 days. Time off therapy was limited to no more than 90 days across the entire year

^cTreatment switching was defined as initiation of a second RA treatment group within 60 days of the discontinuation date of the index therapy (defined by the date of the last day’s supply of therapy) during the specified follow-up time

CCP cyclic citrullinated peptide; FFS Fee-For-Service; JAKi Janus kinase inhibitor; RA rheumatoid arthritis, RF rheumatoid factor; TNFi tumor necrosis factor inhibitor

Fig. 3.

Unadjusted time to non-persistence for bDMARD/JAKi-naive, anti-CCP+ and RF+ beneficiaries with RA receiving abatacept, TNFi, or JAKi. bDMARD, biologic disease-modifying antirheumatic drug; CCP cyclic citrullinated peptide; JAKi Janus kinase inhibitor; RA rheumatoid arthritis, RF rheumatoid factor; TNFi tumor necrosis factor inhibitor

Specifically, a smaller percentage of patients identified with abatacept discontinued (40%) by 12 months compared to the TNFi cohort (60%) and JAKi cohort (55%). However, a higher proportion of patients in the abatacept group had switched during the follow-up period (12%) compared to the other two cohorts (both 6%).

Multivariable Analysis

As shown in Fig. 4, after adjusting for baseline demographic and clinical characteristics, beneficiaries initiating with abatacept were more likely to be persistent than those initiating with TNFi or JAKi (hazard ratio [HR] for TNFi 1.45, 95% CI 1.27–1.64; HR for JAKi 1.43, 95% CI 1.19–1.72). Other than TNFi and JAKi use, qualifying for a low-income subsidy (p = 0.0097), baseline Charlson Comorbidity Index (p = 0.0022), and having a serious infection prior to baseline (p = 0.0188) were significant predictors of treatment discontinuation or switching.

Fig. 4.

Forest plot of adjusted hazard ratios for non-persistence for bDMARD/JAKi-naive, anti-CCP+ and RF+ beneficiaries with RA receiving abatacept relative to TNFi and JAKi. Covariates include age, sex, race/ethnicity, census region, dual Medicare/Medicaid eligibility, low-income subsidy status, baseline Charlson Comorbidity Index score, baseline glucocorticoid use, serious infection at baseline, and prior conventional synthetic DMARD use at baseline. bDMARD biologic disease-modifying antirheumatic drug; CCP cyclic citrullinated peptide; CI confidence interval; HR hazard ratio; JAKi Janus kinase inhibitor; RA rheumatoid arthritis, RF rheumatoid factor; TNFi tumor necrosis factor inhibitor

Discussion

In this nationwide study of older adults with anti-CCP antibodies and RF+RA initiating b/tsDMARDs, we found that abatacept was associated with higher rates of persistence than TNFi or JAKi. Past research describes the importance of persistence to treatment for patients with RA given persistent patients are more likely to achieve a clinical benefit if they stay on therapy without gaps or switching, specifically clinical remission or low disease activity [10–12]. In this real-world study, a higher proportion of patients initiating abatacept were persistent to their index treatment at 12 months and had a longer persistent period compared to patients on TNFi or JAKi.

Prior studies showed mixed findings: the Canadian Rhumadata^® registry for patients with RA found no difference between abatacept and TNFi in first-line treatment, but higher persistence with abatacept in second-line [15], while a Swedish registry study found 55% of biologic-naive patients with RA remained on abatacept at 12 months [16]. Ours is one of the first studies to evaluate treatment persistence after first-line b/tsDMARD initiation for RA in the USA, while also focusing on patients with anti-CCP antibodies and RF+RA. Patients on abatacept have shown differential treatment response given their anti-CCP status [17]. With this in mind, the prior persistence study using clinical registry data [15] did not delineate based on serostatus, suggesting that use of abatacept as first-line b/tsDMARD for anti-CCP antibodies and RF+ patients may yield significantly higher treatment persistence compared to other treatment options, as shown by our study results.

Prior research in the CorEvitas RA registry found that patients with anti-CCP+RA initiating abatacept or TNFi showed similar clinical improvements in disease activity after 6 months [17], establishing comparable short-term efficacy. Our findings of superior persistence with abatacept in patients with anti-CCP antibodies and RF+RA suggest potential advantages in long-term treatment durability for this serologically defined population after a careful shared decision process that also considers safety, efficacy, contraindications, and likelihood of persistence, among other personal preferences.

Despite the higher switching rate, abatacept demonstrated superior 12-month persistence compared with TNFi and JAKi, suggesting that switches among abatacept initiators may occur later in the treatment course rather than representing early discontinuation. This pattern may reflect real-world use of abatacept in patients, where early therapeutic adjustments are followed by longer-term treatment stability once an effective regimen is established.

The three treatment cohorts differed at baseline, with JAKi initiators exhibiting higher rates of dual Medicare Medicaid eligibility, while abatacept initiators had modestly higher comorbidity burden and serious infection rates. These differences likely reflect real-world treatment sequencing and prescribing considerations, with abatacept more frequently used in patients with longer disease duration, and socioeconomic factors influencing initial treatment selection. Importantly, after adjustment for measured baseline characteristics including age, comorbidities, dual eligibility, glucocorticoid use, and serious infection, abatacept remained associated with superior 12-month persistence, suggesting that the observed persistence advantage is not fully explained by these baseline differences.

Our findings are consistent with recent large cohort studies examining biologic retention in RA. The ANSWER cohort in Japan [18] reported that abatacept demonstrated higher retention than TNFi and anti-interleukin-6 receptor antibodies due to safety. Similarly, Takami et al. [19] found tocilizumab and abatacept had superior retention compared to TNFi in Japanese patients. Our study provides unique contributions by examining the U.S. Medicare FFS population with linked Prognos laboratory data, allowing assessment of seropositivity impact on retention. The Medicare population represents a large cohort of older patients with higher comorbidity burden than typically studied cohorts.

Our study has some strengths that included the nationwide coverage of this claims-based dataset and the linked laboratory data. The database comprised individuals distributed across all geographic regions of the USA and the healthcare claims data included 100% of patients covered by traditional Medicare FFS. This study also provided an opportunity to examine patients within claims data while also including biomarker data from Prognos that are not typically available in the 100% Medicare FFS database.

Limitations

However, results must be taken with consideration for the limitations which limit causal inference. This study did not consider all classes of b/tsDMARDs due to the sample size and clinical considerations described above. A critical limitation is the exclusion of interleukin-6 receptor inhibitors, which represents a substantial portion of the biologic DMARD market for RA; for instance, the sample of n = 105 patients initiating on interleukin-6 receptor inhibitors was not large enough to analyze in the present study. Also of note, baricitinib received FDA approval during 2018, the final year of the index period, and upadacitinib, another JAKi, was not approved until August 2019, following the study period, so it was not captured in the JAKi cohort. The study follow-up also occurred before safety concerns in the ORAL-Surveillance study were reported [20–22]. Thus, the JAKi group may not be generalizable to contemporary use of this drug class. Additionally, seronegative RA and RA with single seropositivity (RF+ or anti-CCP+, but not both) were not investigated, and research into these other RA subgroups may yield variations in outcomes. Although switch rates in patients prior to any discontinuation were evaluated in the study, we did not compare the totality of switch rates that includes the post-discontinuation period.

This study did not account for reasons for discontinuation and switching, as that information is not available in claims data without linked EMR or clinical registry data. Abatacept may have a longer time to efficacy than JAKi. Therefore, HCPs may have continued abatacept for longer periods while waiting for treatment effect than the other classes of drugs, particularly JAKi. As claims data exist mainly for billing and reimbursement purposes, there are chances that some of the data that capture medical conditions and outcomes are not documented correctly. This can lead to patient misclassification either due to miscoding or misdiagnosis. For this study, identification of patients with RA relied heavily on available diagnosis codes, potentially introducing misclassification bias that could have affected cohort identification. However, also including anti-CCP antibodies and RF+ likely enhanced the validity of our RA definition. RA disease activity is not measured in claims data and differences in this could explain findings. However, all patients were new users initiating b/tsDMARD, which may limit this potential for confounding. Comorbid conditions and treatment-related outcomes may have also been underreported in claims data. Furthermore, patients who died during minimum follow-up were excluded from the study given 12 months of continuous enrollment is required to capture outcomes. This could result in survival bias. This observational study cannot determine causation, as claims data only capture those disease entities and variables that have their own specific billing codes. As such, temporality cannot be truly established with the use of claims data. The Prognos laboratory data are a convenience sample and all patients with RA in the 100% Medicare FFS data may not have available lab tests in the Prognos laboratory data, which may impact the results. Finally, results from this study cannot be generalized to all populations as patients who have less access to the healthcare system or who are uninsured, as they are less likely to be captured in the data.

Conclusions

In conclusion, we found that abatacept had higher likelihood of persistence than TNFi or JAKi among patients with anti-CCP antibodies and RF+RA initiating first b/tsDMARD. While persistence reflects a composite measure of other measures including efficacy, tolerability, and treatment satisfaction rather than direct clinical response, these findings suggest that treatment with abatacept, compared with TNFi and JAKi, may be an appropriate choice for some to extend time to treatment discontinuation or switch for some anti-CCP antibodies and RF+, bDMARD/JAKi-naive patients with RA. Clinicians should consider a patient’s treatment experience, as well as their RF and anti-CCP test results, when considering durable options for the treatment of RA. Utilization of biomarkers, such as RF and anti-CCP, to inform provider treatment decision-making may be considered for patients with RA.

Author Contributions

Taylor T Schwartz, Hanke Zheng, Laetitia N’Dri, Sang Hee Park, Kristofer Norris, Vardhaman A Patel, Scott B Robinson, Keith Wittstock, Vadim Khaychuk, Alison R Silverstein, and Jeffrey A Sparks made substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data; or the creation of new software used in the work; drafted the work or revised it critically for important intellectual content; approved the version to be published; and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy of integrity of any part of the work are appropriately investigated and resolved.

Funding

This study and the journal’s rapid service fee was funded by Bristol Myers Squibb.

Data Availability

The datasets generated during and/or analyzed during the current study are not publicly available to protect patient privacy. Data sharing requests for non-identifiable patient-level and study-level clinical data, clinical study reports, and protocols will be considered on a case-by-case basis. Bristol Myers Squibb’s policy on data sharing may be found at https://www.bms.com/researchers-and-partners/clinical-trials-and-research/disclosure-commitment.html.

Declarations

Conflict of interest

Taylor T Schwartz, Kristofer Norris: employee of Avalere, a company which received funding from Bristol Myers Squibb to perform the study and provides advisory services to the life sciences industry. Hanke Zheng: Fellowship with Bristol Myers Squibb. Laetitia N’Dri, Sang Hee Park, Vardhaman A Patel, Keith Wittstock, Vadim Khaychuk: employees and stockholders of Bristol Myers Squibb. Scott B Robinson, Alison R Silverstein: employees of Inovalon, a company which received funding from Bristol Myers Squibb to perform the study and provides advisory services to the life sciences industry. Jeffrey A Sparks: research funding: Arthritis Foundation, Boehringer Ingelheim, Bristol Myers Squibb, Johnson & Johnson, Llura Gund Award of the Gordon and Llura Gund Foundation, National Heart, Lung, and Blood Institute (R01 HL155522), National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 AR080659, R01 AR077607, P30 AR070253, and P30 AR072577), R. Bruce and Joan M. Mickey Research Scholar Fund, Rheumatology Research Foundation, and Sonoma Biotherapeutics; consulting fees: AbbVie, Amgen, Anaptys, AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Fresenius Kabi, Gilead, Inova Diagnostics, Invivyd, Johnson & Johnson/Janssen, Merck, MustangBio, Novartis, Optum, Pfizer, ReCor, Sana, Sobi, and UCB. The content is solely the responsibility of the authors and does not necessarily represent the official views of Harvard University, its affiliated academic health care centers, or the National Institutes of Health. The following authors have updated affiliations since the time of the analysis: Laetitia N’Dri (Merck), Vardhaman A Patel (Genmab), Scott B Robinson (ZS), Vadim Khaychuk (Biogen), Alison R Silverstein (EntityRisk).

Ethical Approval

As this was a secondary data analysis of statistically de-identified data, obtaining ethics approval and patient consent was not necessary. Approval to access the Medicare data is governed by an established research-oriented data use agreement with Centers for Medicare & Medicaid Services. Approval to access the Prognos laboratory data was governed by a data use agreement with Prognos.