Immune-mediated rheumatic diseases (IMRD) including rheumatoid arthritis (RA), psoriatic arthritis (PsA) and ankylosing spondylitis (AS) are chronic progressive diseases that impair and deteriorate joint function and structure (Sangha, 2000; Verstappen and Carmona, 2018). IMRD not only adversely affect patients in terms of diminished quality of life, pain, disability, or work-related productivity, but also represents a significant burden to society and healthcare systems due to high direct- and indirect-related healthcare costs (Li et al., 2006; Jacobs et al., 2011; Hsieh et al., 2020). The advent of biologics, particularly, subcutaneous tumor necrosis factor-alpha inhibitors (SC-TNFis) has marked a paradigm shift, revolutionizing the management of patients with IMRD (Kaneko and Takeuchi, 2014). However, taking the medication as prescribed and for a sufficient period of time is crucial for the success of treatment.

Poor adherence to long-term therapies severely, compromises effectiveness and safety outcomes, making patient adherence a critical issue from the perspective of the patient to achieve the lowest possible level of disease activity, optimize functional status and improve quality of life, as well as from the perspective of health economics (Chastek et al., 2017; Usherwood, 2017). Adherence to biological medication in RA has shown variability between 30 and 80%, mainly due to absence of a reference standard measure of adherence and differences in definition and terminology (Borah et al., 2009; Li et al., 2010). In the particular case of golimumab, a retrospective database analysis of 353 patients with rheumatic diseases receiving biological drugs, the probability of retention of golimumab at 1, 2, 3, 4, and 5 years was 85.9, 73.7, 68.5, 60.6, and 57.1%, respectively, with similar percentages across all indications and significantly greater when used as first biological agent compared with later therapies (Hernandez et al., 2019). In a systematic review of real-world treatment persistence of golimumab in the management of IMRD in Europe based on 27 studies, persistence at 24 months was about 50%, with lower persistence among axial AS (43%), but significantly better or equal persistence to other TNFis (Luttropp et al., 2019).

A number of studies have compared costs associated with persistence and non-persistence in patients with IMRD treated with SC-TNFis, including TNFi cycling, first-line and second-line TNFis prescription or switching to medications with a different mechanism of action (Dalén et al., 2016; Bhoi et al., 2017; Dalén et al., 2017; Bonafede et al., 2018; Lee et al., 2018; Sruamsiri et al., 2018; Ziegelbauer et al., 2018). Despite methodological differences, results of these studies provide evidence of non-persistence being more expensive, with higher costs to the healthcare systems of patients who discontinued their treatment compared to patients who were persistent.

All SC-TNFis, including adalimumab, etanercept, certolizumab pegol and golimumab are available in Spain and are approved for the treatment of RA, PsA and AS. A cost management control database established in our hospital provides an opportunity to evaluate costs associated with persistence versus non-persistence based on follow-up data of patients with IMRD treated with biological drugs in a real-world setting. The objective of the current study was to assess the impact of persistence compared with non-persistence to first-line SC-TNFis on healthcare resource utilization (HCRU) costs in IMRD.

At the time of data extraction, 110 patients (49 men and 61 women) with a mean (SD) age of 47.6 (14.8) years, fulfilled the inclusion criteria and were included in the study. They were divided into the groups of persistence and non-persistence, with 85 and 25 patients, respectively. The characteristics of these patients are shown in Table 1. There were no significant differences between the persistent and non-persistent groups in the distribution of patients according to IMRD disorder, with RA accounting for 43.6% of the cases followed by spondyloarthritis in 25.4%, and PsA in 11.8%. Forty-two patients (38.2%) were treated with etanercept, 29 (26.4%) with golimumab, 26 (23.6%) with adalimumab and 13 (11.8%) with certolizumab pegol.

In the Kaplan-Meier analysis, the overall rate of retention of SC-TNFis was 77.3% (95% CI 68.8–84.3) (Figure 1). As shown in Figure 2, the retention rates of SC-TNFis were 84.6% (95% CI 51.2–95.9) for PsA, 81.6% (95% CI 67.7%–90%) for spondylarthritis including AS and other spondyloarthropathies, and 70.8% (95% CI 55.8%–81.6%) for RA (log-rank 0.322). Reasons of non-persistence were lack of efficacy in 16 (64%) patients, adverse events in 7 (28%), and poor adherence in 2 (8%). Reported adverse events were injection site reactions, upper respiratory tract infection, hepatotoxicity, adverse cutaneous reactions, and flu-like syndrome.

Details of HCRU costs before and after initiation of treatment with SC-TNFis are shown in Table 2. There were statistically significant differences in the overall costs, costs associated with outpatient specialized rheumatology care, and costs of other non-biological drugs in favor of the persistence group. Mean (SD) differences in HCRU costs after and before the 12-months study periods for patients in the persistent vs. non-persistent groups were €20.80 (129.59) and €110.90 (234.56) (p = 0.023) for outpatient rheumatology care, −€241.3 (217.88) and −€193.99 (195.88) (p = 0.025) for laboratory testing, −€10.90 (157.42) and €3849.03 (4046.14) (p < 0.001) for treatment with other non-biological drugs, and −€334.67 (905.44) and €3268.90 (4821.55) (p < 0.001) for total costs.

In the treatment of IMRD, biologic agents are recommended for patients who have experienced an inadequate response to conventional DMARDs. Among the biologics currently approved for treatment of naïve patients with IMRD, we have selected adalimumab, etanercept, golimumab and certolizumab pegol that can be administered through subcutaneous injection as these TNFis are usually the first-line options when initiating a patient with IMRD on biologic therapy (Williams and Edwards, 2006; Chilton and Collett, 2008). This single-center before-and-after study of patients with IMRD starting treatment with SC-TNFis for the first time shows that persistence with SC-TNFis treatment using a 12-months period framework in comparison with the non-persistent cohort was associated with significant reductions in total HCRU costs as well as costs of outpatient rheumatology visits, laboratory testing, and the use of other non-biological drugs. The rate of non-persistence of 22.7% (25/110) found in our study is similar to data reported in other studies. In a total of 1,005 patients with AS treated with TNFis collected from the Korean College of Rheumatology Biologics Registry, discontinuation of TNFis occurred in 24.2% after a median follow-up of 14 months (Kim et al., 2021). In a study that reported drug retention rates in 24,915 biologic-naïve patients with axial SpA initiating TNFis treatment from 12 registries in the EuroSpa collaboration, the 12-months non-retention rate was 20% (Ørnbjerg et al., 2019). In addition, persistence rates can vary notably according to the country, characteristics of the healthcare services, or type of specific drug being evaluated. In our study, however, the comparison of the persistent and non-persistent cohorts before initiation of SC-TNFis therapy did not reveal significant differences, leading us to believe that persistence may not be associated with SC-TNFis costs offsets in patients with IMRD.

Other studies have also shown the positive impact of persistence on costs in patients with different rheumatic diseases treated with SC-TNFis. In a retrospective cohort study based on the German statutory health insurance database and using a 2-years time horizon, after 1:1 matching 678 persistent and 678 non-persistent patients, the cost of office-based visits, hospitalizations, co-medications, and sick leave were higher in patients who discontinued SC-TNFis after an average of 9 months compared with patients with at least 24 months of therapy (Ziegelbauer et al., 2018). In a study in which data were extracted from a Japanese claims database, of a total of 6,153 naïve patients treated for the first time with biological DMARDs for RA, the non-persistent group had a larger increase in outpatient visits, with persistence associated with a reduction of total healthcare costs of US$760 (Sruamsiri et al., 2018). The reduction in medication costs in non-persistent patients was offset by higher hospitalization costs, making non-persistence more expensive (Sruamsiri et al., 2018). In a large study population of 4,903 patients treated for the first time with SC-TNFis and 845 with their second SC-TNFi identified from the Swedish Prescribed Drug Register between May 2010 and December 2012, patients treated with the second SC-TNFi had significantly lower persistence and incurred in higher costs (Svedbom et al., 2017).

As far as we are aware, there is only one study that included a cost analysis of HCRU costs prior to initiation of SC-TNFis and 12 months post-initiation in 1,793 persistent and 1,326 non-persistent patients (Dalén et al., 2016), which is similar to our before-and-after design. They found significant differences in HCRU costs post- and prior initiation in specialized outpatient care, in-patient care, non-DMARD medication in the group of persistent patients, and in costs of in-patient care only in the group of non-persistence (Dalén et al., 2016). Although in this study between-group comparisons between the persistent and non-persistent groups are not presented, differences in the persistent group were higher than in the non-persistent group, which is consistent with results of our study. Other studies have shown higher persistence and substantially lower cost per persistent patient for those who switched from a TNFi to another drug with an alternative mechanism of action, rather than in patients who cycled between TNFis (Bonafede et al., 2018). This aspect, however, was not analyzed in the present study as only newly treated patients with SC-TNFis were included. On the other hand, costs of biologic drugs per effectively treated patient according to claims-based algorithm have been reported, with etanercept having the lowest cost (Curtis et al., 2015).

The present findings, however, should be interpreted considering the limitations of the study including the single-center design, the small number of patients included in the non-persistent group and the fact that differences in HCRU costs stratified by biologic drugs were not determined. Moreover, the total number of patients treated with different SC-TNFis for different diseases (RA, PsA, AS and other spondyloarthropathies) is relatively small. Also, as this study was conducted on a hospital-based administrative management database, data of which were integrated in the healthcare system of Catalonia, Spain, the results could not be generalizable to other settings. However, the present findings have value in daily clinical practice. The present results were evaluated collectively, and it would be interesting to analyze HCRU separately and to compare HCRU for the different disorders in further studies. The causes of non-persistence were not assessed, although lack of efficacy was the main reason (64% of patients) followed by adverse events and poor adherence.

In conclusion, in patients with IMRD initiating first-line SC-TNFis, persistence in treatment over a study period of 12 months was associated with decreased HCRU costs in outpatient rheumatological care, laboratory services, use of other non-biological therapies and total costs. Considering the lower costs for persistent patients, adherence to SC-TNFis continues to be a crucial factor given its impact on patients and payers.