Rheumatoid arthritis (RA) is a chronic inflammatory disease that may result in significant disability, morbidity and increased mortality [1]. In recent years, earlier aggressive treatment with disease-modifying drugs (DMARDs) has been shown to play a major role in improving patient outcomes. Those benefits will be achieved only if patients follow prescribed treatment regimens reasonably closely.

Both adherence and compliance are words commonly used to describe how patients take their medication. Traditionally, adherence to therapy has been defined as 'the number of patients continuing treatment with a particular drug prescribed by their health care provider, regardless of the clinical response' [2, 3]. Guidance regarding the meaning of compliance (adherence) and persistence has been proposed recently [4]. Accordingly, medication compliance (synonym adherence) is defined as 'the extent to which a patient acts according to the prescribe interval and dose of a dosing regimen' and medication persistence is defined as 'the duration of time from initiation to discontinuation of therapy' [4].

Compliance to and persistence on medication may be monitored through different methods, direct and indirect, each one with particular advantages and disadvantages [2, 5]. No method is considered the gold standard. Patient self-reports are simple, inexpensive and the most useful method in the clinical setting [6, 7], although higher rates of adherence/persistence have been reported when patient self-reports are used for evaluation [5].

Adherence rates are typically higher among patients with acute conditions as compared with those patients with chronic conditions. Persistence of adherence among patients with chronic conditions is disappointedly low and drops dramatically after the first 6 months of therapy [8–10]. Efforts had been made to develop a questionnaire to investigate patient compliance and persistence with antirheumatic drugs [11–13], but few studies have examined the topic in chronic inflammatory rheumatic conditions – the majority of these studies focused on RA treatment. In those studies, different definitions of adherence and persistence have been used and different populations and medications evaluated, thus limiting any conclusions. The current literature, however, suggests that nonadherence to and nonpersistence on DMARD therapy is a substantial problem, ranging in occurrence from 20% to 70% [10, 14–23].

Many factors have been related to patient's medication behavior in RA patients, including younger age [16, 21, 23], male sex [20, 21], belonging to an ethnic minority [17], lower education [17], side effects [17], availability of financial resources and social support [18, 23], medication-taking behavior and beliefs [19], increased disability [20], better perceived health status at the beginning [10], poor quality of contact with health professionals [20], poor personal knowledge about the disease and its treatment [20], comorbidity [23] and the class of DMARDs [10, 15, 22].

Compliance and persistence with prescribed medication regimens (and placebo regimens) predict better outcomes. Collecting adherence and persistence data from patients is considered an essential part of clinical trials.

By contrast, poor compliance and nonpersistence with medication contribute to substantial worsening of disease and death, and increase healthcare costs [2].

We report our experience with persistence on DMARDs and corticosteroids evaluated through patient–physician structured interviews, during 2 years of follow-up of an early RA cohort of patients. The aims of the study were to determine persistence on DMARDs and corticosteroids over 13 consecutive visits each 2 months apart in a cohort of early RA patients, to investigate the relationship between persistence on therapy and disease activity, to investigate the relationship between persistence on therapy and both disability and structural damage, and to identify whether any baseline factors are associated with nonpersistence on medication.

The terminology, definitions and methods to determine adherence and persistence differ greatly in the published literature [27]. A need for improvement in the quality and consistency of medication compliance and persistence research is mandatory. In 2006 the International Society of Pharmaeconomics and Outcomes Research published a consensus document that was intended to improve the consistency and quality of analysis regarding this topic and to understand the impact of compliance and adherence on health outcomes [28]. Two years later, Cramer and colleagues provided specific definitions for compliance and persistence, and encouraged their adoption by health outcome researchers [4]. In the present study we have adopted those definitions and analyzed persistence on DMARDs and corticosteroids in early RA patients. Nonetheless, poor adherence may be related to therapy discontinuation (that is, nonpersistence according to the recent proposal) and most of the published studies do not include persistence as a separate construct from compliance, which makes difficult a comprehensive discussion on the topic.

The present study highlights the impact of therapy persistence on different RA outcomes. After 6 months of follow-up, persistent patients showed lower scores and greater improvements in disease activity and disability and had more frequent and earlier sustained remission than nonpersistent patients. It may be argued that it is not clear whether better outcomes were a cause of or an effect of persistence on therapy. Nonetheless, at initial evaluations both groups of patients had similar clinical status, demography (except age), comorbidities, and disease characteristics. Treatment at baseline and during follow-up did not differ between them, suggesting that greater clinical improvement in persistent patients was probably related to persistence on therapy. To the best of our knowledge there is only one study that has evaluated compliance and RA outcomes. Viller and colleagues showed more frequent improvement in disability (no relationship with disease activity was investigated) in consistently compliant European RA patients over 3 years of follow-up than in those who changed behavior [21]. They included 556 patients with < 5 years of disease duration, and compliance with drug dosages and dosing times was assessed yearly using a questionnaire. We also found that more nonpersistent patients developed erosive disease than persistent patients, although the difference was not statistically significant – we were probably limited by the number of patients with erosive disease. Interestingly, most nonpersistent patients (56%) omitted treatment once during follow-up; among them, the number of therapy-discontinuation days was low (4% of the indicated days of therapy) but impacted negatively on the disease prognosis.

Among potential candidates, older age was the only predictor to be associated with nonpersistence on DMARD and corticosteroid therapy. Kristensen and colleagues identified high age (in addition to low C-reactive protein serum level, elevated HAQ score, and higher previous number of DMARDs) as a predictor of premature treatment termination (which may be considered a synonym for nonpersistence) with etarnecept and infliximab in 1,161 patients with active RA [29]. By contrast, Viller and colleagues found older age (in addition to female sex, decreased disability, very satisfactory contacts with health professionals, and more personal knowledge about the disease and its treatment) significantly associated with good compliance in their early RA cohort [20]. Self-reporting of current medication use with DMARDs (methotrexate, sulphasalazine, and corticosteroids) has been considered to have good to excellent agreement with information obtained from the medical charts [30]. In our study, the time between two visits was 2 months so the treatment recorded could be considered current treatment; meanwhile, compliance with drugs was assessed annually in the study of Viller and colleagues. Tuncay and colleagues performed three assessments for drug compliance over 1 year of follow-up in 100 RA patients [16]. Consistently compliant patients were older than consistently noncompliant patients, although no regression analysis was performed. In both studies, patients had longer disease duration than patients from our study and the concept of medication persistence as a different construct from medication adherence was not defined. Finally, Curkendall and colleagues showed that persistence on (and adherence with) anti-TNFα was better among older patients, in an inception cohort of anti-TNFα-naïve-RA patients [23]; worse persistence was also associated with greater out-of-pocket costs, with higher Charlson's comorbidity score and with previous prescription for a narcotic analgesic. Adherence and persistence were measured using claims data on a particular RA population of unknown disease duration, which may have accounted for the different result.

Persistence on therapy decreased in our cohort of early RA patients up to 34% at 2 years of follow-up. Curkendall and colleagues showed that persistence on (and adherence with) anti-TNFα decreased over 1 year of follow-up in their inception cohort of RA patients, up to 32% in those patients with out-of-pockets costs above $50/week [23]. Similar results were found regarding a decline of compliance by de Klerk and colleagues in 127 outpatients, 81 of whom had RA [10]; by Viller and colleagues, who identified 35.7% of consistently compliant patients after three years of follow-up [20]; and by Tunkay and colleagues, who found that 30.2% of 100 RA patients were consistently compliant over 1 year of follow-up [16]. Two other studies performed in 66 patients and 108 RA patients, respectively, showed higher rates of compliance and persistence with therapy, ranging from 61% to 73% [15, 31]. Numerous randomized clinical trials performed in RA patients had also assessed drug continuation (persistence) over different follow-up periods [32–38], although the probability of discontinuation for reasons other than adverse events and lack of efficacy was very low.

The present study has several limitations. We did not use a well-validated questionnaire scale to assess compliance/persistence [11, 13, 39]. We assessed persistence through the simplest and most used way in clinical practice, which is to ask the patient whether he/she is still taking the medication as prescribed [7, 40–43]. We used a partially arbitrary and recently proposed definition to define nonpersistence. We choose a lag time of 1 week, as 89% of our patients were taken methotrexate and the drug is indicated weekly. We analyzed neither the whole spectrum of nonadherence/nonpersistence nor its repercussion on disease activity and damage – clinical outcomes of treatment are affected not only by how long patients take their medication but also by how well; accordingly, adherence should have been defined and measured to characterize medication-taking behavior comprehensively. In terms of guaranteeing a better prognosis, how to define a reasonable adherence/persistence is still unknown. Finally, the outcome is not the result of compliance and persistence alone but can be influenced by many other factors [44]. In that sense, the present study was carried out in an inception cohort of early RA patients, with particular sociodemographic characteristics, ethnicity, treatment and health system, and our results may not be generalized to RA populations with different characteristics [45].

In 2006, the American College of Rheumatology endorsed a starter set of quality indicators for several rheumatic diseases, among them RA [46, 47]. Two measures pertinent to treatment were included and strategies proposed to be used when there is increased disease activity or damage progression over a 6-month period. The recommendations add potential adverse events, increase costs, and are based on the assumption that patients are fully (or partially) compliant with therapy and persistent on medication. The present study reinforces the necessity to consistently investigate compliance and persistence with treatment before regimens are modified. Potential benefits are better prognosis, cost reductions and patient safety.

Nonpersistence on DMARDs was frequent and progressive over the first 2 years of follow-up in a cohort of early RA patients, a time frame that has been proposed as crucial for disease control/remission. Persistent patients were younger at baseline evaluation and had lower disease activity and disability during their follow-up than nonpersistent patients. Persistent patients reached sustained remission more frequently and earlier than nonpersistent patients. Persistence and adherence with medication should routinely be evaluated during RA follow-ups, especially at the beginning of the disease when adequate treatment has a major impact on disease outcomes.