Psoriatic arthritis (PsA) is a chronic seronegative inflammatory arthritis occurring in approximately a third of the patients with psoriasis [1]. Men and women are equally affected and for the majority of patients rheumatic manifestations appear 7–8 years after the initial onset of psoriasis [2].

The prevalence of PsA varies widely by geographical region and is estimated to range from 0.05 to 0.67%. A study performed in Israel by Eder et al. during 2006–2015 reported a crude prevalence of 0.153% [3].

Once thought to be a relatively benign condition, it is now recognized that PsA is a systemic disease, potentially debilitating, erosive disease, that can negatively impact many aspects of daily life [4]. Repeated joint pain, chronic fatigue, decreased occupational function, as well as psychological distress and social morbidity may result in decreased quality of life (QoL) [5]. The advent of new treatments for PsA in the past decades, mainly in the form of biological disease modifying antirheumatic drugs (bDMARDs), and targeted synthetic disease modifying anti-rheumatic drugs (tsDMARDs), along with their relative early implementation in the treatment paradigm, have significantly improved patients with PsA care, quality of life and prevented disability [6, 7]. According to numerous studies, biologic agents and apremilast, have shown to ameliorate disease activity, inhibit radiographic progression and improve quality of life to a similar extent [7, 8]. In addition, their safety profile are clearly acceptable [9].

While the variety of bDMARDs and tsDMARDs available for patients with PsA has proved to be efficacious and with a good safety profile in randomized clinical trials [10], these studies included mostly relatively young and treatment-naïve patients [9, 11]. In the real-world, these drugs are used for both treatment-naïve and treatment-experienced patients and naturally encompassing all age groups. However, data on the safety and efficacy among the latter is sparse [12, 13] and novel data are lacking [14].

Drug adherence, defined as the extent to which a patient takes a drug as prescribed by the health care provider [15] and drug survival, defined as the time from initiation to discontinuation (whether due to stop or switch) of a specific therapy [16], are commonly used, indirect measures to evaluate drug efficacy and safety [17] in real-world settings. In PsA, such studies are present [18,19,20,21,22,23]; however, they are limited and often have conflicting results due to differences in drug availability originating from regional subsidization and local health insurance policies and coverage. As treatment choices continue to expand and join the growing armamentarium in PsA, there is a growing importance to understand the benefits and potential drawbacks of these different therapies.

The aim of this study was to compare the adherence and drug survival of different bDMARDs and tsDMARDs in the treatment of PsA in a real-world setting by utilizing the electronic database of a large Israeli health state-mandated health provider.

Methods

This retrospective cohort study was conducted using the computerized databases of Maccabi Healthcare Services (MHS). MHS is the second largest state-mandated health provider operating in Israel, providing healthcare services for over 2.6 million members (25% of the Israeli population), with 1.1–1.5% leaving annually. [24] The MHS databases integrate data from the MHS central laboratory, medication prescriptions and purchases throughout the MHS pharmacy network, primary care, expert consultations, hospitalizations, procedures, and socio-demographic data. Physician diagnoses are coded using the International Classification of Disease, 9th Edition (ICD-9-CM) codes as well as internal MHS codes for subclassification. In addition, MHS has developed and validated computerized registries of patients who are suffering from major chronic conditions such as cardiovascular diseases and diabetes [1, 25, 26]

The study was conducted in accordance with the protocol, applicable regulations, and guidelines governing clinical study conduct and the ethical principles that have their origin in the Declaration of Helsinki. The MHS Institutional Review Board (IRB), approved the study protocol and related documents. The MHS’s IRB waived the requirement to obtain any informed consent for this secondary analysis of existing data (approval number 0108-18-BBL, 18 December 2018).

According to the Israeli regulatory guidelines, patients with PsA are eligible for bDMARDs or tsDMARDs after failing to achieve control over their disease with two conventional synthetic (cs) DMARDs. Prior to procurement, patients are required to receive an authorization from MHS drugs authorization center, confirming they comply with the guidelines. Patients’ co-payment is the same, regardless of which bDMARD or tsDMARD they are using.

The study population included patients who, between 1 January 2015 and 31 December 2017, first purchased one or more of the following drugs for the indication of PsA: TNF-α inhibitors [adalimumab (ADA), infliximab (IFX), golimumab (GLM), etanercept (ETN), certolizumab pegol (CTZ)]; IL-12/23 inhibitor [ustekinumab (UST)]; IL-17 inhibitor [secukinumab (SEC)]; CTLA4-Ig [abatacept (ABA)]; PDE-4 inhibitor [premilast (APR)]. All patients were prescribed with these drugs for the indication of PsA only, according to the MHS drug authorization center. The first purchase was defined as the index date for the study. Included patients were adults (age ≥ 18 years) and MHS members for ≥ 12 months before and after the index date. During the study follow-up period, all the aforementioned drugs were available as first line treatment after utilizing ≥ 2 different conventional synthetic DMARDs (i.e., methotrexate, leflunomide, sulfasalazine).

Patients were followed until the earliest of the following dates: death, leaving MHS, or the end of the follow-up period (31 December 2018).

We included treatment-naïve patients (those with no previous purchase of any of the drugs included in the study before 1 January 2015), as well as treatment-experienced patients (those treated with any of the drugs included in the study before 1 January 2015 who switched and started treatment with any of the other drugs included in the study, between 1 January 2015 and 31 December 2017).

Initiation of a new treatment line was defined by the first purchase of the drug. For all patients, we followed all lines of therapy used during the study follow-up period and numbered them. For those defined as treatment-experienced when entering the study, we numbered the lines of therapy used during the study follow-up period, using data on bDMARDs and tsDMARD dispensed up to 10 years before entering the study.

Adherence to medical treatment was evaluated by using the proportion of days covered (PDC) method. PDC reflects the number of days covered by the dispensed drug divided by the total follow-up time for a specific line of therapy. PDC was categorized as followed: nonadherent (PDC < 40%), moderately adherent (40% ≤ PDC < 80%), or highly adherent (PDC ≥ 80%).

Drug survival was measured from the initiation of treatment (i.e., the index date) until treatment discontinuation, which was defined as the first gap of 120 days or more after the last supply date. Patients who discontinued their current drug were further classified as: switching (starting a new treatment, with any of the drugs included in the study), restarting (continued the same treatment after the ≥ 120 day’s gap), stopping (neither switching nor restarting treatment with any of the drugs included in the study).

Additional data retrieved from the database included: sociodemographic factors [age, sex, residential area socioeconomic status (SES), and smoking status]; baseline comorbidities according to MHS registries [cardiovascular disease [25], diabetes [26], hypertension [27], obesity (latest body mass index (BMI) record ≥ 30), and osteoporosis [28]]. Depression and anxiety were defined according to antidepressants and benzodiazepines dispensed 180 days before the index date. The Charlson Comorbidity Index was calculated at baseline. Additional data at baseline included disease duration, visits to a primary care physician (PCP), and being hospitalized at least once in the 180 days before the index date.

Baseline characteristics, adherence rates, discontinuation rate, type, and time to discontinuation are presented using descriptive statistics, namely n (%), mean ± standard deviation (SD), or median and interquartile range (IQR), as appropriate. To assess differences in adherence and drug survival between older and younger patients, we used age of 60 years as the cutoff and patients were categorized by their age at treatment initiation (older, age < 60 years; younger, age ≥ 60 years) [29, 30]. Baseline characteristics are presented by age group and statistical differences between the two age groups were assessed.

Adherence was assessed by line of therapy, bio-experience status and mode of action (MoA), and bio-experience status and individual drug. Additionally, adherence rates are presented by bio-experience status and age group and differences were assessed using the Chi-squared test. Time to treatment discontinuation was analyzed and plotted using Kaplan–Meier estimates and the log rank test was used to evaluate statistical significance between MoA and between age group in treatment-naive and in treatment-experienced patients.

The risk of treatment discontinuation of each MoA was estimated separately for the treatment-naïve and the treatment-experienced patients using a multivariable Cox proportional hazards regression model. The model was adjusted for age group, sex, SES, smoking status, disease duration, various comorbidities (diabetes, hypertension, obesity, osteoporosis, depression, and anxiety), visits to the PCP, MoA, and line of therapy (only in the treatment-experienced).

Two-sided P < 0.05 was considered statistically significant. All statistical analyses were performed with IBM-SPSS V.25.0 standard statistical software for Windows and R V.3.5.