Abstract
Introduction
Patients with rheumatoid arthritis (RA) often fail to achieve treatment targets with first-line treatment because of poor efficacy or tolerability. However, comparative long-term adherence and switching data among patients initiating a second-line therapy after a first-line tumor necrosis factor inhibitor (TNFi) remain limited.
Methods
Retrospective data were from the Merative MarketScan® claims database, August 2018–October 2024. Eligible patients were aged ≥ 18 years; diagnosed with RA; initiated upadacitinib (UPA), tofacitinib (TOF), adalimumab (ADA), etanercept (ETA), abatacept (ABA), or tocilizumab (TOC); and had discontinued a first-line TNFi within 12 months prior to index. Treatment adherence and switching outcomes were reported at 1- and 3-year follow-up.
Results
Overall, 3782 and 1182 patients were included for the 1- and 3-year analyses, respectively. The proportion of patients with adherence ≥ 80% was highest with UPA at 1-year (49.5%) and 3-year follow-up (35.7%) versus other treatments. Compared with UPA through 3 years, odds of being adherent were significantly lower for TOF (adjusted odds ratio [95% confidence interval]: 0.45 [0.29–0.71]), ADA (0.50 [0.33–0.75]), ETA (0.37 [0.24–0.58]), ABA (0.43 [0.26–0.73]), and TOC (0.46 [0.22–0.99]; p < 0.05). Switch rates were lowest for UPA versus other treatments at 1 year (28.6%) and remained lowest through 3 years (47.6%), while rates for all other treatments increased to approximately 60%. Through 3 years, patients were also significantly (p < 0.05) more likely to switch treatment when receiving TOF (adjusted hazard ratios [95% confidence interval]: 1.40 [1.07–1.84]), ADA (1.44 [1.12–1.86]), ETA (1.57 [1.21–2.03]), ABA (1.56 [1.16–2.10]), and TOC (1.54 [1.03–2.32]) compared with UPA. Median 3-year time to switch was not reached for UPA and was 528–660 days for all other treatments.
Conclusion
Among patients with RA who discontinued a first-line TNFi, second-line UPA was associated with significantly greater long-term adherence and lower switching rates than other advanced therapies.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s40744-026-00872-z.
Keywords: Adherence, JAK inhibitor, Persistence, Real-world evidence, Rheumatoid arthritis, TNF inhibitor, Treatment sequencing, Upadacitinib
Key Summary Points
| Why carry out this study? |
| Poor treatment adherence and frequent treatment switching are associated with poorer clinical outcomes and higher costs in patients with rheumatoid arthritis (RA), and limited evidence exists that compares outcomes for second-line treatments after a first-line tumor necrosis factor inhibitor (TNFi) |
| This study investigated the 1- and 3-year treatment adherence and switching patterns among patients with RA initiating second-line treatment with upadacitinib compared with other advanced therapies following a first-line TNFi |
| What was learned from this study? |
| At 1- and 3-year follow up, patients receiving upadacitinib after a first-line TNFi demonstrated significantly higher treatment adherence and lower switching rates compared with patients receiving other advanced therapies |
| These findings suggest that upadacitinib may offer favorable long-term treatment continuity in real-world practice, which may help inform treatment decisions following discontinuation of a first-line TNFi |
| Results of this analysis should be interpreted with consideration of the limitations inherent to insurance claims data, including the lack of disease activity information, and the current study’s methodology; future studies investigating long-term clinical comparative effectiveness are needed to further contextualize these findings |
Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune disease that causes inflammation of the joints, leading to pain, stiffness, swelling, and fatigue [1]. Over time, this chronic inflammation causes loss of cartilage and bone erosion, leading to joint destruction and irreversible disability [2–4].
The primary therapeutic target for all patients with RA should be sustained remission or low disease activity, according to the most recent guidelines from the European Alliance of Associations for Rheumatology (EULAR) and the American College of Rheumatology (ACR) [5, 6]. EULAR and ACR guidelines also recommend that if no improvement is seen within 3 months of initiating therapy, patients should switch to an alternative therapy [5, 6].
Current pharmacologic treatments available for RA include corticosteroids, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs; such as methotrexate), biologic DMARDs (bDMARDs; including tumor necrosis factor inhibitors [TNFi] and interleukin [IL]-6 inhibitors), and targeted synthetic DMARDs (tsDMARD; including Janus kinase inhibitors [JAKi]) [1, 7]. The most initiated advanced therapies for first-line treatment are TNFis; however, many patients do not reach the treatment targets and either switch to an alternative TNFi or change to a therapy with a different mode of action [8]. In cases where patients fail to achieve target disease control with a bDMARD or tsDMARD, ACR conditionally recommends switching to a DMARD from a different class rather than cycling within the same class [5].
A recent real-world study in patients with moderate to severe RA in the US reported that among patients who received first-line TNFis, 30% discontinued within 6 months, rising to 64% between 12 and 24 months [9]. This study also found that among those who discontinued initial TNFi treatment, 15% later restarted the same TNFi, 46% initiated another TNFi, 28% initiated a non-TNFi biologic, and 12% initiated a JAKi [9]. This process of TNFi cycling has been shown to incur higher healthcare costs than switching to a drug with another mechanism of action (MoA) [10].
Comparisons between second-line treatments are important to inform clinical decisions following treatment failure with a first-line TNFi. One retrospective claims analysis reported that although more patients cycled to another TNFi after first-line TNFi failure, patients who started a second-line non-TNFi had lower risk of switching [11]. Furthermore, a meta-analysis found that switching to a drug with a new MoA, including JAKis, IL-1/6 receptor inhibitors, and CD80/86/20 inhibitors, was more effective and was associated with lower withdrawal rates than cycling to a different TNFi after first-line TNFi failure [12]. Similarly, another retrospective claims analysis found that patients switching to a drug with a new MoA, including abatacept, tocilizumab, and tofacitinib, had higher persistence, lower odds of switching, higher odds of being effectively treated, and lower drug and healthcare costs compared with patients cycling to a second-line TNFi [13].
However, despite evidence suggesting worse clinical outcomes with TNFi cycling compared with switching to a new MOA [11–13], TNFi cycling by trial and error continues to occur frequently in clinical practice [9]. As such, real-world second-line effectiveness studies comparing newer treatments, including the JAKi upadacitinib (UPA), to more established therapies are needed to provide improved treatment insight.
Furthermore, data that may allow comparisons of long-term clinical effectiveness between individual treatments, such as UPA and TNFis, are still accumulating in real-world data sources. In lieu of these data, treatment adherence and switching can serve as potential indicators of treatment effectiveness and tolerability. This study examined 1- and 3-year treatment adherence and switching to UPA compared with other advanced therapies among patients with RA who initiated second-line treatment after receiving a first-line TNFi.
Methods
Study Design
This retrospective cohort study used data from the Merative MarketScan® commercial and Medicare Supplemental Research Databases (Ann Arbor, MI) from August 16, 2018 to October 31, 2024, to evaluate patients who initiated a second-line RA advanced therapy after switching from a first-line TNFi.
Eligible patients were aged ≥ 18 years with ≥ 2 RA diagnoses during the 12-month baseline period and had a first pharmacy or medical claim (index date) for UPA, tofacitinib (TOF), adalimumab (ADA), etanercept (ETA), abatacept (ABA), or tocilizumab (TOC) within 12 months of discontinuing a first-line TNFi (i.e., ≥ 1 claim for a TNFi within 12 months before the index date). Patients must also have no record of receiving the index medication or any other listed advanced therapy at any time prior to the index date. Continuous insurance enrollment was required for ≥ 12 months pre-index and either ≥ 12 months post-index for analysis of 1-year outcomes or ≥ 36 months post-index for analysis of 3-year outcomes.
Outcomes
All adherence and switching outcomes were assessed at 1 year among patients with ≥ 1 year of follow-up and at 3 years in a separate cohort of patients with ≥ 3 years of follow-up.
Treatment adherence was assessed using the proportion of days covered (PDC) method. PDC represents the percentage of days during the observation period that the patient had access to the index medication [14, 15]. For this study, the observation periods were 1 year (365 days) and 3 years (1095 days). The number of days with medication access was calculated via actual days’ supply from pharmacy claims or approximated days’ supply from medical claims based on US Food and Drug Administration-approved labeling for the medication. If a subsequent prescription from a pharmacy claim was filled before the end of the preceding prescription, we considered the subsequent prescription an “early refill.” In this case, the start date of the subsequent prescription was adjusted to the day after the end of the preceding prescription to avoid counting the overlap between the two prescription fills. Adherence measures reported here included median PDC and the percentage of patients adherent to treatment (PDC ≥ 80%), which is considered the threshold for the likelihood of clinical benefit [14].
The percentage of patients switching to a new treatment during follow-up and median time to treatment switch were reported. Patients were considered to have switched to another advanced therapy if a pharmacy or medical claim for a different advanced therapy from the index treatment was present during the 1- or 3-year follow-up period. Switching from an originator medication to a biosimilar, or switching between biosimilar medications, was not included in the switching definition in this study and was not counted as switching from a first- to second-line treatment. The switching date was defined as the first date of the non-index advanced therapy.
Statistical Analysis
Baseline demographics and characteristics were reported descriptively in the 12-month baseline period. Categorical outcomes were reported as the number and percentage of patients, and continuous outcomes were reported as the median and interquartile range (IQR) or the mean and standard deviation.
Logistic regression was used to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the likelihood of treatment adherence (PDC ≥ 80%) among UPA initiators compared with other advanced therapy initiators. Risk of treatment switching among UPA users compared with users of other advanced therapies was estimated using Cox proportional hazards model; adjusted hazard ratios (HRs) and 95% CIs were reported. Models were adjusted for age, sex, and Charlson Comorbidity Index. Kaplan-Meier curves were constructed to visualize treatment switching over time within each medication-initiation cohort.
Subgroup analyses evaluated outcomes at 1 year in the cohort with 3 years of follow-up to assess potential differences between the 1- and 3-year follow-up cohorts.
For all analyses, p < 0.05 was considered statistically significant. Instant Health Data software (Panalgo, Boston, MA) and R version 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria) were used for all analyses.
Ethics
The authors had a data license contract that granted them permission to access and conduct analyses using data from the Merative MarketScan® Research Databases and publish the results. This was a retrospective database analysis that did not collect, use, or transmit patient-identifiable data, so informed consent was not needed. Ethics committee or Institutional Review Board approval was not needed. This study complied with the data security requirements of the Health Insurance Portability and Accountability Act of 1996.
Results
Study Population
In total, 3782 patients were included in the 1-year analyses (UPA, N = 675; TOF, N = 520; ADA, N = 926; ETA, N = 957; ABA, N = 529; TOC, N = 175). For 3-year analyses, 1182 patients were included (UPA, N = 185; TOF, N = 217; ADA, N = 316; ETA, N = 270; ABA, N = 142; TOC, N = 52). Approximately 80% of patients were female, with mean age ranging from 47.9 to 52.4 years across treatment groups in the 1- and 3-year cohorts (Table 1).
Table 1.
Baseline demographics and clinical characteristics among patients with 1 and 3 years of follow-up
| Upadacitinib | Tofacitinib | Adalimumab | Etanercept | Abatacept | Tocilizumab | P valuea | |
|---|---|---|---|---|---|---|---|
| Patients with 1 year of follow-up | |||||||
| N | 675 | 520 | 926 | 957 | 529 | 175 | |
| Age (years), mean ± SD | 50.8 ± 9.7 | 50.9 ± 9.7 | 50.3 ± 10.6 | 49.4 ± 10.7 | 52.4 ± 10.1 | 50.4 ± 10.4 | < 0.0001 |
| Female, n (%) | 532 (78.8) | 414 (79.6) | 733 (79.2) | 759 (79.3) | 430 (81.3) | 145 (82.9) | 0.78 |
| CCI, mean ± SD | 1.5 ± 0.9 | 1.5 ± 0.9 | 1.5 ± 1.0 | 1.5 ± 1.0 | 1.7 ± 1.2 | 1.6 ± 1.1 | < 0.0001 |
| US region, n (%) | < 0.0001 | ||||||
| South | 361 (61.1) | 213 (47.3) | 370 (46.7) | 415 (52.6) | 221 (52.3) | 70 (50.0) | |
| Midwest | 88 (14.9) | 103 (22.9) | 168 (21.2) | 160 (20.3) | 117 (27.7) | 27 (19.3) | |
| West | 73 (12.4) | 59 (13.1) | 142 (17.9) | 111 (14.1) | 39 (9.2) | 21 (15.0) | |
| Northeast | 69 (11.7) | 75 (16.7) | 112 (14.1) | 103 (13.1) | 46 (10.9) | 22 (15.7) | |
| Patients with 3 years of follow-up | |||||||
| N | 185 | 217 | 316 | 270 | 142 | 52 | |
| Age (years), mean ± SD | 51.2 ± 8.2 | 50.7 ± 8.3 | 49.5 ± 9.2 | 48.8 ± 9.2 | 50.0 ± 8.8 | 47.9 ± 7.9 | 0.02 |
| Female, n (%) | 139 (75.1) | 175 (80.7) | 258 (81.7) | 221 (81.9) | 120 (84.5) | 45 (86.5) | 0.26 |
| CCI, mean ± SD | 1.4 ± 0.7 | 1.4 ± 0.9 | 1.5 ± 1.0 | 1.4 ± 0.9 | 1.6 ± 1.1 | 1.5 ± 0.9 | 0.05 |
| US region, n (%) | 0.02 | ||||||
| South | 98 (57.7) | 95 (49.7) | 136 (48.8) | 132 (56.4) | 65 (54.6) | 27 (58.7) | |
| Midwest | 31 (18.2) | 46 (24.1) | 59 (21.2) | 58 (24.8) | 36 (30.1) | 4 (8.7) | |
| West | 24 (13.5) | 26 (13.6) | 46 (16.5) | 22 (9.4) | 7 (5.9) | 10 (21.7) | |
| Northeast | 18 (10.6) | 24 (12.6) | 38 (13.6) | 23 (9.4) | 11 (9.2) | 5 (11.0) | |
ANOVA one-way analysis of variance, CCI Charleston Comorbidity Index, SD standard deviation
aChi-square tests were used to compare categorical variables across groups; one-way ANOVA was used to compare continuous variables across groups
Treatment Adherence
At 1-year follow-up, the percentage of patients with PDC ≥ 80% was highest in patients receiving UPA (49.5%), followed by ABA (48.2%) and TOF (47.7%) (Fig. 1a). The median IQR PDC was 0.80 (0.41–0.95) for UPA, 0.74 (0.33–0.93) for TOF, 0.69 (0.38–0.95) for ADA, 0.55 (0.23–0.92) for ETA, 0.77 (0.38–0.94) for ABA, and 0.61 (0.23–0.89) for TOC. Through 1 year of follow-up, the odds of being adherent to treatment were numerically lower in all treatment groups compared with UPA and significantly lower in patients initiating ETA (adjusted OR [95% CI] 0.56 [0.46–0.69]; p < 0.05) and TOC (0.57 [0.40–0.80]; p < 0.05; Fig. 2a).
Fig. 1.
Proportion of patients adherent to treatment through (a) 1-year follow-up and (b) 3-year follow-up. Treatment adherence was defined as PDC ≥ 80%. ABA abatacept, ADA adalimumab, ETA etanercept, PDC proportion of days covered, TOC tocilizumab, TOF tofacitinib, UPA upadacitinib
Fig. 2.
Odds of adherence to treatment relative to UPA at (a) 1-year follow-up and (b) 3-year follow-up. *p < 0.05. Treatment adherence was defined as PDC ≥ 80%. aOR with 95% CI for each treatment is shown relative to UPA (reference, aOR = 1). ABA abatacept, ADA adalimumab, aOR adjusted odds ratio, CI confidence interval, ETA etanercept, PDC proportion of days covered, TOC tocilizumab, TOF tofacitinib, UPA upadacitinib
At 3-year follow-up, the UPA-treated population had the highest percentage of patients with PDC ≥ 80% (35.7%). Percentage of patients with PDC ≥ 80% ranged from 16.3 to 21.2% for all other treatment groups (Fig. 1b). The median (IQR) PDC was 0.57 (0.16–0.90) for UPA; 0.30 (0.14–0.71) for TOF, 0.28 (0.13–0.69) for ADA, 0.20 (0.08–0.66) for ETA, 0.26 (0.13–0.76) for ABA, and 0.29 (0.13–0.71) for TOC. In adjusted analyses, patients receiving other advanced therapies were less than half as likely to be adherent compared with patients receiving UPA: TOF (adjusted OR [95% CI] 0.45 [0.29–0.71]), ADA (0.50 [0.33–0.75]), ETA (0.37 [0.24–0.58]), ABA (0.43 [0.26–0.73]) and TOC (0.46 [0.22–0.99]) (p < 0.05; Fig. 2b).
Subgroup analyses of 1-year outcomes among the cohort of patients with 3 years of follow-up showed similar results to the primary analyses (Fig. S1a), with patients receiving ADA also being statistically less likely to be treatment adherent compared with patients receiving UPA (OR [95% CI] 0.66 [0.45–0.95]; p < 0.05; Fig. S1b).
Treatment Switch Rates
Through 1 year of follow-up, patients receiving UPA had the lowest switch rates (28.6%), followed by patients receiving ABA (33.3%) and TOF (36.4%; Fig. 3a). Kaplan-Meier analysis showed that patients receiving UPA also had the slowest switch rate, with the UPA curve diverging from other treatments at approximately Day 150 and continuing to separate over time (Fig. 4). Compared with UPA, patients were significantly (p < 0.05) more likely to switch treatment when receiving TOF (adjusted HR [95% CI] 1.32 [1.08–1.61]), ADA (1.49 [1.25–1.78]), TOC (1.65 [1.26–2.15]), and ETA (1.71 [1.44–2.02]; Fig. 5a).
Fig. 3.
Treatment switch rates through (a) 1-year follow-up and (b) 3-year follow-up. ABA abatacept, ADA adalimumab, ETA etanercept, TOC tocilizumab, TOF tofacitinib, UPA upadacitinib
Fig. 4.
Kaplan-Meier curves for treatment switching through 1 year of follow-up. Kaplan-Meier curves showing the proportion of TNFi-experienced second-line initiating patients remaining on index treatment over 1 year. ABA abatacept, ADA adalimumab, ETA etanercept, TNFi tumor necrosis factor inhibitor, TOC tocilizumab, TOF tofacitinib, UPA upadacitinib
Fig. 5.
Risk of treatment switching relative to UPA at (a) 1-year follow-up and (b) 3-year follow-up. *p < 0.05. aHR with 95% CI for each treatment are shown relative to UPA (reference, aHR = 1). ABA abatacept, ADA adalimumab, aHR adjusted hazard ratio, CI confidence interval, ETA etanercept, TOC tocilizumab, TOF tofacitinib, UPA upadacitinib
During 3 years of follow-up, switch rates increased, but patients receiving UPA still had the lowest percentage of switching (47.6%). Switch rates for all other treatment groups increased to approximately 60% (Fig. 3b). Kaplan-Meier analysis of the 3-year cohort showed that the UPA curve started to deviate from other treatments at approximately 175 days post-initiation, and this separation increased throughout the 3-year follow-up (Fig. 6). Median time to switch at 3 years was not reached for UPA. For other treatments, median (IQR) time to switch in days was 638 (253–1096) for TOF, 660 (211–1097) for ADA, 555 (207–1096) for ETA, 528 (196–1096) for ABA, and 536 (188–1096) for TOC. Through 3 years follow-up, patients receiving TOF (adjusted HR [95% CI] 1.40 [1.07–1.84]), ADA (1.44 [1.12–1.86]), ETA (1.57 [1.21–2.03]), ABA (1.56 [1.16–2.10]), and TOC (1.54 [1.03–2.32]) were all significantly more likely to switch treatments than patients receiving UPA (p < 0.05; Fig. 5b).
Fig. 6.
Kaplan-Meier curves for treatment switching through 3 years of follow-up. Kaplan-Meier curves showing the proportion of TNFi-experienced second-line initiating patients remaining on index treatment over 3 years. ABA abatacept, ADA adalimumab, ETA etanercept, TOC tocilizumab, TNFi tumor necrosis factor inhibitor, TOF tofacitinib, UPA upadacitinib
Results from subgroup analyses were similar to the primary analyses, with the exception of the rate of switching in the ABA group, which was higher in this subgroup analysis compared with the main analysis (Figs. S2a and S2b).
Discussion
This study is, to our knowledge, the first to compare long-term treatment adherence and switching between UPA and other advanced therapies in second-line therapy after discontinuation of a first-line TNFi in a real-world setting. The study leverages a large cohort of patients, followed for up to 3 years after switching from a TNFi, providing valuable insights into post-switch treatment durability over time. Furthermore, the use of recent administrative claims data supports this study’s relevance to current clinical practice. Overall, second-line treatment adherence was highest with UPA compared with other treatments at 1-year follow-up, and this effect became more pronounced through 3 years of follow-up. Switch rates followed a similar pattern, with patients on UPA having the lowest rates at both 1 and 3 years compared with all other treatments. Over the course of the study, switch rates for all treatments began to diverge from UPA at approximately Day 150, with the difference in switching rates between UPA and all other treatments accelerating after Day 300.
Higher treatment adherence and persistence are important indicators of real-world treatment continuity and may suggest adequate symptom management, especially since improved adherence in RA has been associated with lower disease activity [16]. Although these outcomes are not a direct measure of clinical efficacy, improved persistence and lower rates of switching may serve as proxies for additional outcomes including effectiveness, tolerability, patient and physician preference, and affordability, although further studies investigating these associations are warranted. Nonadherence and poor persistence have previously been associated with increased healthcare costs, worse health outcomes, lower quality of life, and disease progression [17, 18]. Moreover, patients who do not comply with therapy are more likely to experience polypharmacy, dose escalations, and more specialized outpatient care [19, 20]. Therefore, by maintaining consistent disease control, improved adherence and persistence may translate into better long-term outcomes for both patients and healthcare systems.
The topic of optimal sequencing after inadequate response to first-line advanced therapy in RA remains an area of discussion. Current ACR guidelines conditionally recommend switching to a therapy with a different MoA rather than cycling within a class [5]. Several studies have provided evidence for better treatment persistence, which can ultimately lead to lower healthcare costs when switching MoA rather than cycling within the TNFi class [10–13]. Our results support these guidelines, as patients treated with UPA showed greater adherence and lower rates of switching than those receiving ADA or ETA after prior TNFi failure. However, it is not clear from the current study whether other non-TNFi therapies provide better outcomes than cycling with ADA or ETA because differences in adherence and persistence outcomes were not consistent. However, a previous study using a real-world prospective Korean cohort showed that treatment continuation and remission or low disease activity at 1 year were higher for patients with RA receiving second-line non-TNFi biologics (TOC, ABA, and rituximab) than for patients receiving either second-line TNFis or JAKis (TOF and baricitinib) [21]. Furthermore, another prospective study found that at 24 months after switching to a second-line non-TNFi, patients were three times more likely to achieve a long-term clinical response than patients receiving second-line TNFis [22]. Additionally, a previous study found that patients with secondary failure with a first-line TNFi had higher odds of achieving Clinical Disease Activity Index ≤ 10 than patients with first-line TNFi failure, regardless of the choice of second agent [23]. Taken together, these results support switching to an alternative therapy after first-line TNFi treatment, though longer-term analyses are warranted.
In addition to corroborating prior results, the present study is also the first to provide a long-term snapshot (up to 3 years) of the adherence and switching outcomes for individual RA advanced therapies, including UPA, after a first-line TNFi. The higher adherence and lower switching over 3 years observed in our study for UPA compared with other RA advanced therapies is consistent with earlier results from a 1-year follow-up analysis showing that patients receiving UPA were more likely to be treatment-adherent and persistent compared with those receiving ADA, TOF, or baricitinib, regardless of prior TNFi use [17]. Similarly, our results also align with a recent Australian retrospective, real-world analysis, which showed that propensity-matched patients receiving UPA as monotherapy or in combination with csDMARDs had longer persistence and greater symptom improvements than those receiving other JAKis or biologics [24]. Additionally, a Swiss cohort study in patients with RA showed significantly higher drug maintenance for the JAKi baricitinib compared with TNFis [25]. In terms of clinical outcomes, a recent real-world study demonstrated favorable clinical remission rates and physician-reported outcomes for patients who switched from first-line TNFi therapy to UPA compared with those who switched to a second TNFi or to a treatment with another MoA [26]. Results from the SELECT-SWITCH trial, which enrolled patients with an inadequate response to a first TNFi, showed that at Week 12, significantly more patients receiving UPA achieved low disease activity (Disease Activity Score in 28 joints-C-reactive protein [DAS28-CRP] ≤ 3.2) and remission (DAS28-CRP < 2.6) compared with those receiving ADA (43.3% vs. 22.4% and 28.4% vs. 14.5%, respectively; p < 0.001 for all) [27]. Taken together, these results suggest that second-line treatment with UPA may offer a viable long-term treatment option for patients with RA following a first-line TNFi, resulting in higher adherence rates and lower switch rates compared with other advanced therapies.
This study is subject to several limitations. Some factors, such as drug-related adverse events, treatment response (e.g., clinical effectiveness and patient-reported outcomes), physician preferences, and other patient characteristics, may influence medication adherence and switching. Due to a lack of available data in the claims database, statistical adjustments for these covariates (e.g., disease activity, such as DAS28 and Clinical Disease Activity Index) could not be made in this study, so how these factors contribute to the findings cannot be determined. Additionally, a significant proportion of the study period occurred during the COVID-19 pandemic, so typical clinical care patterns, such as in-office versus at-home treatment administration, may have been disrupted, affecting treatment adherence and switching patterns. In addition, these results may not be generalizable to other patient populations, such as individuals with noncommercial health insurance such as government-sponsored programs or no health insurance. These data are subject to limitations inherent to insurance claims data, such as data entry errors or coding limitations. Furthermore, it is difficult to determine the reason for treatment discontinuation based on the available claims data, and it remains unknown whether patients discontinued treatment because of side effects, lack of efficacy, provider or patient preference, or insurance restrictions.
Conclusion
Among patients with RA who discontinued a first-line TNFi, treatment adherence was higher and treatment switching was lower in patients receiving second-line UPA than in those receiving other advanced RA therapies at both 1- and 3-year follow-up, with the differences favoring UPA becoming more pronounced over time. These results highlight the potential long-term benefit of UPA in maintaining treatment continuity and may help inform treatment decisions following discontinuation of a first-line TNFi.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We gratefully acknowledge the individuals included in the Merative MarketScan® Research Databases.
Medical Writing, Editorial, and Other Assistance
Medical writing support was provided by Kay Hallam of Avalere Health Ltd and was funded by AbbVie.
Author Contributions
All authors meet the ICMJE requirements for authorship. Study conceptualization and design: Patrick M Zueger, Yi Peng, Siran Fang, Richard Thielen, and Gregory Markley; Acquisition, analysis, or interpretation of data: all authors; Article drafting or critical revision for important intellectual content: Christina Charles-Schoeman, Kurt Oelke, Patrick M Zueger, Yi Peng, Gregory Markley, Richard Thielen, Siran Fang, Martin Bergman; Approval of the final article to be published and agreement to be accountable for all aspects of the work: all authors.
Funding
This study was funded by AbbVie. AbbVie funded this study and participated in the study design, research, analysis, data collection, interpretation of data, reviewing, and approval of the publication. All authors had access to relevant data and participated in the drafting, review, and approval of this publication. No honoraria or payments were made for authorship. The journal’s Rapid Service Fees were funded by AbbVie.
Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Conflict of Interest
Christina Charles-Schoeman has received consultancy fees from AbbVie, BMS, Boehringer Ingelheim, Octapharma, Pfizer, Recludix and Sana Biotechnology; and grant/research support from AbbVie, Alexion, BMS, CSL Behring, Janssen, Octapharma, Pfizer and Priovant. Kurt Oelke has been a member of the speakers’ bureau for AbbVie, Amgen, Bristol Myers Squibb, Crescendo Biosciences, GSK, Pfizer Inc, and UCB. Patrick M Zueger, Yi Peng, Gregory Markley, Richard Thielen, and Siran Fang are employees of AbbVie and hold shares and/or stock options in the company. Martin Bergman is a shareholder of Johnson and Johnson (parent company of Janssen).
Ethical Approval
The authors had a data license contract that granted them permission to access and conduct analyses using data from the Merative MarketScan® Research Databases and publish the results. This was a retrospective database analysis that did not collect, use, or transmit patient-identifiable data, so informed consent was not needed. Ethics committee or Institutional Review Board approval was not needed. This study was compliant with data security requirements of the Health Insurance Portability and Accountability Act of 1996.
References
- 1.Brown P, Pratt AG, Hyrich KL. Therapeutic advances in rheumatoid arthritis. BMJ. 2024;384:e070856. [DOI] [PubMed] [Google Scholar]
- 2.Johnson KJ, Sanchez HN, Schoenbrunner N. Defining response to TNF-inhibitors in rheumatoid arthritis: the negative impact of anti-TNF cycling and the need for a personalized medicine approach to identify primary non-responders. Clin Rheumatol. 2019;38(11):2967–76. [DOI] [PubMed] [Google Scholar]
- 3.Ajeganova S, Forslind K, Hafström I. The equivocal opposite effects of lower and higher body mass index at diagnosis on radiographic joint damage progression in early rheumatoid arthritis: an inception cohort study over 15 years. RMD Open. 2025;11(3):e005784. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ten Klooster PM, Versteeg LGA, Oude Voshaar MAH, et al. Radiographic progression can still occur in individual patients with low or moderate disease activity in the current treat-to-target paradigm: real-world data from the Dutch rheumatoid arthritis monitoring (DREAM) registry. Arthritis Res Ther. 2019;21(1):237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Fraenkel L, Bathon JM, England BR, et al. 2021 American college of rheumatology guideline for the treatment of rheumatoid arthritis. Arthritis Rheumatol. 2021;73(7):1108–23. [DOI] [PubMed] [Google Scholar]
- 6.Smolen JS, Landewé RBM, Bergstra SA, et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2022 update. Ann Rheum Dis. 2023;82(1):3–18. [DOI] [PubMed] [Google Scholar]
- 7.Kanda R, Miyazaki Y, Nakayamada S, et al. Effective second-line b/tsDMARDs for patients with rheumatoid arthritis unresponsive to first-line b/tsDMARDs from the first registry. Rheumatol Ther. 2025;12(2):353–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bergman MJ, Kivitz AJ, Pappas DA, et al. Clinical utility and cost savings in predicting inadequate response to anti-TNF therapies in rheumatoid arthritis. Rheumatol Ther. 2020;7(4):775–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Edgerton C, Frick A, Helfgott S, et al. Real-world treatment and care patterns in patients with rheumatoid arthritis initiating first-line tumor necrosis factor inhibitor therapy in the United States. ACR Open Rheumatol. 2024;6(4):179–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chastek B, Chen CI, Proudfoot C, et al. Treatment persistence and healthcare costs among patients with rheumatoid arthritis changing biologics in the USA. Adv Ther. 2017;34(11):2422–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Karpes Matusevich AR, Duan Z, Zhao H, et al. Treatment sequences after discontinuing a tumor necrosis factor inhibitor in patients with rheumatoid arthritis: a comparison of cycling versus swapping strategies. Arthritis Care Res (Hoboken). 2021;73(10):1461–9. [DOI] [PubMed] [Google Scholar]
- 12.Migliore A, Pompilio G, Integlia D, Zhuo J, Alemao E. Cycling of tumor necrosis factor inhibitors versus switching to different mechanism of action therapy in rheumatoid arthritis patients with inadequate response to tumor necrosis factor inhibitors: a Bayesian network meta-analysis. Ther Adv Musculoskelet Dis. 2021;13:1759720x211002682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chastek B, Becker LK, Chen CI, Mahajan P, Curtis JR. Outcomes of tumor necrosis factor inhibitor cycling versus switching to a disease-modifying anti-rheumatic drug with a new mechanism of action among patients with rheumatoid arthritis. J Med Econ. 2017;20(5):464–73. [DOI] [PubMed] [Google Scholar]
- 14.Pharmacy quality alliance adherence measures. https://www.pqaalliance.org/adherence-measures. Accessed 26 Feb 2026.
- 15.Dalli LL, Kilkenny MF, Arnet I, et al. Towards better reporting of the proportion of days covered method in cardiovascular medication adherence: a scoping review and new tool ten-spiders. Br J Clin Pharmacol. 2022;88(10):4427–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Li L, Cui Y, Yin R, et al. Medication adherence has an impact on disease activity in rheumatoid arthritis: a systematic review and meta-analysis. Patient Prefer Adherence. 2017;11:1343–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bergman M, Chen N, Thielen R, Zueger P. One-year medication adherence and persistence in rheumatoid arthritis in clinical practice: a retrospective analysis of upadacitinib, adalimumab, baricitinib, and tofacitinib. Adv Ther. 2023;40(10):4493–503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Menditto E, Cahir C, Malo S, et al. Persistence as a robust indicator of medication adherence-related quality and performance. Int J Environ Res Public Health. 2021;18(9):4872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Aksoy N, Ozturk N, Agh T, Kardas P. Adherence to the antirheumatic drugs: a systematic review and meta-analysis. Front Med (Lausanne). 2024;11:1456251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Carballo N, Garcia-Alzórriz E, Ferrández O, et al. Impact of non-persistence on healthcare resource utilization and costs in patients with immune-mediated rheumatic diseases initiating subcutaneous TNF-alpha inhibitors: a before-and-after study. Front Pharmacol. 2021;12:752879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Park DJ, Choi SE, Kang JH, et al. Comparison of the efficacy and risk of discontinuation between non-TNF-targeted treatment and a second TNF inhibitor in patients with rheumatoid arthritis after first TNF inhibitor failure. Ther Adv Musculoskelet Dis. 2022;14:1759720x221091450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Bogas P, Plasencia-Rodriguez C, Navarro-Compán V, et al. Comparison of long-term efficacy between biological agents following tumor necrosis factor inhibitor failure in patients with rheumatoid arthritis: a prospective cohort study. Ther Adv Musculoskelet Dis. 2021;13:1759720x211060910. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Bessette L, Movahedi M, Reed G, et al. Does the type of failure and the choice of the second biologic influence response and persistence on medication in rheumatoid arthritis? J Clin Rheumatol. 2023;29(7):332–40. [DOI] [PubMed] [Google Scholar]
- 24.Youssef P, Ciciriello S, Tahir T, et al. Real-world persistence and effectiveness of upadacitinib versus other janus kinase inhibitors and tumor necrosis factor inhibitors in Australian patients with rheumatoid arthritis. Rheumatol Ther. 2025;12(1):173–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gilbert BTP, Mongin D, Aymon R, et al. Comparative effectiveness of baricitinib and alternative biological DMARDs in a Swiss cohort study of patients with RA. BMJ Open. 2024;14(3):e072300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Caporali R, Kadakia A, Howell O, et al. A real-world comparison of clinical effectiveness in patients with rheumatoid arthritis treated with upadacitinib, tumor necrosis factor inhibitors, and other advanced therapies after switching from an initial tumor necrosis factor inhibitor. Adv Ther. 2024;41(9):3706–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mysler E, Sidiropoulos PI, Balanescu A, et al. Upadacitinib vs adalimumab in patients with rheumatoid arthritis and a prior inadequate response or intolerance to a tumour necrosis factor inhibitor: 12-week results from the randomised, double-blind, SELECT-SWITCH study. Ann Rheum Dis. 2026. 10.1016/j.ard.2026.03.013. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.