Abstract
Objective. Medication adherence in patients with rheumatoid arthritis (RA) is typically suboptimal. Nonadherence has been associated with symptom worsening and increased disability. We systematically reviewed published clinical studies to evaluate methotrexate (MTX) adherence and persistence, factors associated with MTX adherence and persistence, and the effect of MTX nonadherence on clinical outcomes in RA.
Methods. MEDLINE and Embase were systematically searched (inception to February 2016) using relevant keywords. Observational or interventional clinical studies in patients with RA that specifically reported adherence to or persistence with MTX were included. Data were extracted using a predesigned, standardized template that included study design, patient demographics, and relevant outcomes. Main outcomes were MTX adherence and persistence rates in patients with RA treated with MTX and factors associated with MTX adherence and persistence.
Results. Of 365 references screened, 31 articles met inclusion criteria and another 10 were identified from searching reference lists. Estimates of MTX adherence varied from study to study because of heterogeneity in patient populations, duration of followup, definitions of adherence, and methods of assessment. Rates of MTX persistence ranged from 50% to 94% at 1 year and 25% to 79% at 5 years. No clear trends were identified in factors that influence MTX adherence and persistence. Two studies suggested that MTX adherence was associated with superior clinical outcomes.
Conclusion. MTX adherence and persistence are highly variable in patients with RA. Research is necessary to determine the effect of nonadherence on health outcomes and to identify independent predictors of nonadherence to inform evidence-based interventions.
Highly effective pharmacotherapeutic options are available for treating patients with rheumatoid arthritis (RA). Nonetheless, many patients fail to achieve remission1,2, increasing the likelihood of functional deterioration1,3 and additional clinic visits, laboratory tests, and treatment modifications.
Although reasons for not achieving or for losing disease control are complex, one factor that may influence outcome is medication adherence, defined as the degree to which a patient takes medication in accordance with clinician instructions4. Persistence is considered an aspect of adherence and may be defined as the time from therapy initiation to discontinuation5,6. According to the World Health Organization, risk factors/predictors for nonadherence may be categorized based on 5 interacting dimensions: socioeconomic, healthcare system-related, disease-related, therapy-related, and patient-related4.
In RA, medication adherence is highly variable and typically suboptimal, with reports of adherence to conventional disease-modifying antirheumatic drugs (DMARD) ranging from 22% (underuse) to 107% (overuse)5,7,8. Nonadherence to DMARD has been associated with symptom worsening and increased disability9,10, and high adherence has been linked to lower disease activity11,12. A previous narrative review found no clear pattern in factors that influence medication adherence in patients with RA5, whereas a systematic review of methotrexate (MTX) use in RA suggested improved adherence was associated with belief in the necessity and efficacy of MTX, absence of low mood, mildness of disease, and monotherapy with MTX13.
MTX is recommended as a first-line treatment in patients with active RA14,15, and because of its additive efficacy benefits and its reductions in immunogenicity, is typically used in combination with biologics16. Although several systematic or comprehensive reviews have evaluated adherence to DMARD in general7,17,18, only 1 systematic study, to our knowledge, has specifically evaluated MTX adherence (but not persistence) in patients with RA13. Given the prevalence of MTX use and its involvement in combination treatment with biologics, it is important to understand factors that may cause patients to be nonadherent or to discontinue treatment.
Our aims were to systematically review published clinical studies to evaluate MTX adherence and persistence in patients with RA treated with MTX alone or in combination with nonbiologic or biologic DMARD, to identify factors that influence adherence and persistence, and to determine the effect of nonadherence on clinical outcomes.
MATERIALS AND METHODS
Literature search
A systematic literature search was conducted according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses statement. The MEDLINE (through PubMed) and Embase databases were searched using combined Medical Subject Headings (MeSH) and free text terms related to “methotrexate,” “rheumatoid arthritis,” “adherence,” “compliance,” and “persistence” (see Supplementary Table 1 for the full search strategy, available online at jrheum.org). Studies were limited to those published in English. No publication date or study type limits were applied. The original search was performed on November 1, 2013, and was updated on February 5, 2016. Additional articles were obtained by manually searching reference lists of relevant articles.
Study selection
Eligible publications reported results of an observational or interventional study that included adults with RA treated with MTX as monotherapy or in combination with other DMARD or biologics. Studies were included only if they reported adherence to or persistence with MTX in patients with RA as an outcome. No further exclusion criteria were applied.
All authors independently screened the titles and abstracts of all studies identified in the literature search for relevance. The relevance of each article was then discussed, and the full text of each potentially relevant article was obtained and screened.
Data extraction and quality assessment
For each study, the following were extracted using a predesigned, standardized template: study design, funding sources, patient recruitment method, setting, data collection dates, inclusion/exclusion criteria, sample size, demographics and characteristics [age, sex, RA definition, RA duration, disease activity, comorbidities, and concomitant therapies, including folate supplementation (folic or folinic acid)], followup duration, MTX dosage and administration route, MTX line of therapy, method of measuring adherence/persistence, MTX adherence/persistence rate, factors that influenced adherence/persistence (if available), effect of nonadherence/non-persistence on clinical outcomes (if available), and statistical analyses undertaken. If studies included both univariate and multivariate analyses of influencing factors or clinical outcomes, only multivariate results were extracted.
Given that no single tool is recommended for assessing quality or susceptibility to bias in observational studies, we used a simple methodology checklist that assessed the risk for bias within each study. This included the representativeness of the study sample, whether an RA diagnosis had been clinically confirmed, how loss to followup was accounted for, appropriateness of the outcome measure, risk for missing/erroneous information, and appropriate accounting for confounders (where applicable). Data extraction and quality assessment were performed by 1 person and verified by another.
The primary outcomes were MTX adherence and persistence rates in adults with RA treated with MTX. Secondary outcomes were factors associated with MTX adherence and persistence and the effect of nonadherence/non-persistence on clinical outcomes.
RESULTS
The search of MEDLINE and Embase identified 365 citations after duplicates were removed; another 10 studies were identified from manual searching of reference lists. Forty-one studies met all criteria and were included (Figure 1). For each study, data were extracted and quality was assessed. Results are summarized in Table 19,11,12,19,20,21,22,23,24,25,26,27, Table 2, Table 328–37,38–47,48,49,50,51,52,53,54,55,56, and Table 4, as well as Supplementary Table 2 and Supplementary Table 3 (available online at jrheum.org).
Flow chart of study selection process. MTX: methotrexate; RA: rheumatoid arthritis.
Adherence to MTX and associated factors in patients with RA.
Bias assessment of studies that evaluated adherence to methotrexate in patients with RA.
Persistence and factors associated with persistence to MTX in patients with RA.
Bias assessment of studies that evaluated persistence with methotrexate in patients with RA.
MTX adherence
Twelve studies were identified that evaluated MTX adherence in patients with RA (Table 1)9,11,12,19,20,21,22,23,24,25,26,27. Most patients were enrolled in rheumatology outpatient clinics and were receiving MTX for the first time (Supplementary Table 2, available online at jrheum.org).
Assessment of study quality found that most studies had notable susceptibility to bias (Table 2). Five studies included populations unlikely to be representative of the general population (enrollment in a social healthcare program for persons with low income and limited resources21, RA duration < 1 yr9, predominantly men11, sample from an outpatient clinic in India24, and economically disadvantaged12). Accounting for loss of followup was insufficient in many studies.
Adherence was evaluated using a Medication Event Monitoring System (MEMS; electronic chip that records medicine bottle opening) in 4 studies12,19,25,26. In the largest study of 129 patients receiving MTX, 58% were completely adherent and 91% were at least 80% adherent over the 16-week study period26. In another study of 23 patients taking MTX for the first time, the mean percentage of prescribed doses taken was 107% over 6 months (overuse)19. When adherence was assessed as the mean percentage of doses taken at the correct time (a more stringent measure), the rate decreased to 83% over 6 months. A third study used MEMS to evaluate MTX adherence in an ethnically diverse, economically disadvantaged US population (n = 76)12. Compared with patients who did not agree to monitoring, those monitored were more likely to be younger, female, and Hispanic. No differences were seen in disease characteristics or self-reported adherence. The mean percentage of correctly taken MTX doses over 2 years was 63%. In the final MEMS study, the proportion that was at least 80% adherent to MTX (alone or with another DMARD) decreased from 91.2% at 3 months to 69.3% at 12 months25.
Claims data were used to assess adherence in 5 studies11,20,21,22,27. In each, patients were followed from first MTX use (or prescription gap ≥ 180 days). Harley, et al20 calculated the medication possession ratio (MPR; number of prescribed days of MTX during a course divided by total course duration) for MTX in commercial and Medicare enrollees in a large US health plan (n = 1668). Throughout 1 year, 64% were adherent to MTX (MPR ≥ 80%). In a Medicaid-managed care program, median MPR over 180 days was 59%21. In an analysis of a German sickness fund, MPR was 60% at 2 years if the entire period was considered, irrespective of whether patients were actively prescribed MTX; however, if only periods when patients were prescribed MTX were considered, MPR increased to 95%27. One study22 used prescription data to perform a 10-year longitudinal study among Danish patients who were first-time MTX users. Adherence was assessed using a continuous measure of medication gaps (number of days when medication was unavailable divided by total course duration) to give a mean of 12.3%, corresponding to 10.5 months/year of MTX coverage if patients were followed for 1 year. Finally, in a retrospective analysis of a US Veterans Affairs Rheumatoid Arthritis registry11, 84% were considered adherent over the first course of MTX (MPR ≥ 80%). One of the study’s main limitations was that 92% of the population was men, making it non-representative of the general RA population.
The remaining studies used various methods of patient reporting to evaluate adherence9,23,24. In a 9-month prospective Danish study that evaluated beliefs about medication, patients who were first-time MTX users were identified through a prescription database and completed the Compliance Questionnaire Rheumatology (CQR) at baseline and 9 months23. The median CQR score was 70.1 [interquartile range (IQR) 36.8–93.0] at baseline and 70.6 (IQR 42.1–91.2) at 9 months. However, 23% did not complete the 9-month evaluation, baseline CQR scores were higher in completers, and completers believed more strongly in the necessity of MTX and were less concerned about side effects and potential longterm consequences than those who dropped out. In a 6-month prospective study in patients with early arthritis in Mexico, patients with RA durations < 1 year completed a standardized drug record registry that recorded doses, timing, and frequency of DMARD use in the 7 days before the interview9. Adherence, defined as percentage of doses taken correctly ≥ 80% of the time, was 78% for patients treated with monotherapy and 14–49% for patients treated with MTX in combination with other DMARD. A patient report was also used to evaluate MTX adherence in patients with RA at a rheumatology outpatient clinic in India24. Of the patients who returned for the 3-month visit (n = 50), 92% stated that they had taken MTX as prescribed. One study26 used 3 self-report questionnaires — CQR, Medication Adherence Rate Score, and a visual analog scale (VAS) — in addition to the MEMS to assess MTX adherence. VAS correlated best with the MEMS results (r = 0.552, p < 0.001), giving a mean adherence rate of 94% over the 16-week study26.
Persistence with MTX
Twenty-nine studies evaluated persistence with MTX (Table 3)28–37,38–47,48,49,50,51,52,53,54,55,56. Most were in the United States and Europe, and 1 each in Israel, India, and Japan. Mean/median doses of MTX ranged from 3.8 mg/week to 20 mg/week (Supplementary Table 3, available online at jrheum.org).
Different definitions of MTX discontinuation were used across studies, including MTX withdrawal, adding another treatment to MTX, or MTX interruption. Some studies included patients treated with first-line MTX, and others included patients who had previously received DMARD or MTX by a different route. This made it difficult to directly compare persistence duration. Persistence rates ranged from 50% to 94% at 1 year and 25% to 79% at 5 years. Lack of tolerability was the main reason for withdrawal (23%–79% of withdrawals). The highest withdrawal rates owing to tolerability occurred in a Japanese study45. Inefficacy was the other primary reason for withdrawal (6%–72% of withdrawals).
Assessment of bias susceptibility found that earlier studies tended to include patients from 1 center only, and in general, reports did not include details about how data were verified for accuracy or completeness (Table 4). Further, few accounted for different thresholds for MTX discontinuation between physicians or documented key potential modifiers of persistence such as folate supplementation.
Factors that influence MTX adherence and persistence
Clinical, demographic, and biologic factors that may influence MTX adherence were assessed in several studies (Table 1 and Table 5), with no clear patterns observed. A multiple regression analysis adjusted for age, sex, comorbidities, duration of RA, C-reactive protein (CRP), hemoglobin, concomitant therapies, and year of inclusion found ulcer/mild liver disease independently predicted worse adherence (adjusted estimate 0.04, 95% CI 0.004–0.084), whereas CRP > 32 ng/l predicted better adherence than CRP < 8 ng/l (adjusted estimate −0.04, 95% CI −0.070 to −0.015)22. In another study9, higher erythrocyte sedimentation rate (ESR) remained independently associated with nonadherence (OR 1.03, 95% CI 1.01–1.05, p = 0.003), as did > 3 concomitant DMARD versus no concomitant DMARD (OR 31.5, 95% CI 2.3–433.3, p = 0.009) in a multi-regression analysis that included age, years of education, followup at clinic, Health Assessment Questionnaire (HAQ) score, sex, socioeconomic status, rheumatoid factor, comorbidities, and corticosteroid use.
Summary of factors associated and not associated with MTX adherence and persistence.
A direct (non-adjusted) comparison of clinical variables between patients whose MPR were < 80% and those whose MPR were ≥ 80% found significantly more white patients (p = 0.05) and fewer African American patients (p = 0.01) in the adherent group11. At enrollment, adherent patients had better clinical measures [mean Disease Activity Score at 28 joints (DAS28), p = 0.02; tender joint count (TJC), p = 0.02; swollen joint count (SJC), p < 0.01] and lower ESR (p = 0.02) than nonadherent patients. There were no differences in age, disease duration, smoking status, autoantibody status, or concurrent therapy use in the adherent versus the nonadherent group. One study23 compared clinical and demographic variables in patients who had CQR scores in the bottom quartile with those who had CQR scores in the upper 3 quartiles. After 9 months of treatment, no significant differences in MTX dose or HAQ score were reported between groups. Another study26 used multivariate analysis to show that better mental health status, presence of comorbidities, and not living alone predicted 30% of variance in MTX adherence.
Several studies evaluated factors that influence adherence to rheumatology medications in general rather than with specific regard to MTX. A multivariate regression analysis found that disease-controlling drugs (compared with symptom-modifying drugs), less frequent dosing, female sex, better coping, and perception of poorer health together explained 66.6% of adherence variation19. A stepwise regression analysis using the percentage of correct DMARD doses taken as the dependent variable showed adherence was associated with being married/having a significant other/not being widowed or separated (p < 0.01), having lower disease activity (p < 0.05), and having better mental health (p < 0.05)12.
Factors that influenced MTX persistence were reported in several studies (Table 3 and Table 5). In a Cox regression analysis that included disease severity, age, and race, Alarcón et al28,29 found a higher risk for MTX discontinuation in patients who experienced a toxic event (p = 0.028) or began MTX treatment at an earlier year (p = 0.001). One study37 reported older age (≥ 65 yrs) increased the likelihood of discontinuation because of toxicity (p < 0.001) and noted that treatment termination rates varied between rheumatologists, particularly because of differences in withdrawal attributed to toxicity. Cox regression analysis on baseline covariates and length of the MTX therapeutic segment found that these factors were associated with shorter MTX persistence: longer disease duration (p = 0.002), fewer previous nonsteroidal antiinflammatory drugs (p = 0.003), and fewer previous DMARD (p = 0.034). These factors did not have significant effect: age at disease onset, pain score, disability index, global assessment, and education level39. In patients treated with MTX between 1993 and 200142, multivariate analysis found folate supplementation [relative risk (RR) related to MTX discontinuation 0.25, 95% CI 0.20–0.33, p < 0.001], prednisolone use (RR 0.70, 95% CI 0.54–0.90, p = 0.005), and attending rheumatologist (p = 0.002) were related to MTX retention. No independent association was found with current sulfasalazine use, number of previous DMARD, or age. Analysis of a large Canadian database of patients ≥ 65 years with newly diagnosed RA showed age > 70 years [adjusted hazard ratio (HR) 1.4, 95% CI 1.0–2.0], comorbidity score (adjusted HR 1.1, 95% CI 1.0–1.2), and intramuscular versus oral MTX (adjusted HR 2.4, 95% CI 1.1–5.7) were all independently associated with MTX discontinuation50. No significant association was found among sex, folic acid use, and RA disease severity.
MTX dose may also have an effect on persistence. In 1 study, differences in persistence between patients treated with higher doses (≥ 12.5 mg/week) and lower doses (≤ 10 mg/week) of MTX significantly favored higher dose treatment for median retention time (p < 0.05) and 5-year persistence rate (p < 0.05). However, factors such as disease severity, RA duration, and MTX administration era might have confounded these results40. In a study in Japan45, where MTX doses are generally lower than in Western countries, multi-regression analysis showed that no previous DMARD use (relative to previous DMARD use ≥ 3) was an independent predictor of MTX survival (adjusted RR 2.07, 95% CI 1.09–3.95, p = 0.027), whereas MTX dose, folate supplementation, disease duration, and age were not significant.
Effect of MTX nonadherence or non-persistence on clinical outcomes
Only 2 articles specifically evaluated the effect of MTX nonadherence on clinical outcomes11,25. A retrospective study in patients at US Veterans Affairs medical centers found patients with MTX MPR ≥ 80% had better DAS28 scores (p = 0.02), ESR (p = 0.05), CRP (p = 0.03), TJC (p < 0.01), and SJC (p = 0.02) than those with MTX MPR < 80% (unadjusted analysis)11. No association was found between adherence and HAQ score, patient’s global score, or pain. A multivariate analysis adjusted for RA classification criteria, medical history, smoking status, education, ethnicity, age, sex, RA duration, previous/current DMARD use, seropositivity, and comorbidities showed a significant independent association between the DAS28 score and MTX MPR ≥ 80% (−0.37, 95% CI −0.67 to −0.07, p < 0.05). No adjustment for baseline DAS28 was possible because some patients were already established in receiving MTX at study initiation. When the analysis was limited to the incident MTX subpopulation and baseline DAS28 score was included in the covariates, the association between MTX adherence and DAS28 lost statistical significance (−0.40, 95% CI −1.11 to 0.30). Nonadherence to MTX (alone or with DMARD) has been independently associated with poorer DAS28 score for up to 6 months after treatment initiation; however, this effect was lost after 6 months25.
DISCUSSION
Patients with RA are often required to take a cocktail of medications by different routes and at varying times, which may lead to issues with medication adherence. MTX remains the initial drug of choice for most patients and is typically used in combination with biologics. Therefore, adequate adherence to MTX remains a major therapeutic goal.
In our systematic review, estimates of MTX adherence were shown to vary considerably because of differences in patient populations, followup durations, adherence definitions, and assessment methods. Wide variation in persistence rates was also seen and may be attributed to factors such as time of study, lines of therapy, MTX dosage, and differences in rheumatologists’ thresholds for MTX termination. In most studies, more patients discontinued because of tolerability than inefficacy. Interestingly, a claims database analysis of patients with RA initiating MTX found that although about one-half discontinued MTX at 1 year, more than one-third subsequently restarted MTX52. One study found patients who received MTX subcutaneously were less likely to add or switch to another nonbiologic DMARD or to add a biologic DMARD than patients treated with oral MTX57.
MTX adherence or persistence may be superior to that seen with other conventional DMARD19,21,35,36,40,41,48, but not with biologics20,48. Direct comparisons are difficult to make, however, because of inherent biases in nonrandomized studies possibly related to weekly dosing, side effects, or differences in clinical characteristics in patients prescribed MTX compared with other DMARD. In addition, the side effects of MTX can be at least partially mitigated by use of concomitant folate, which is not the case for other DMARD.
Most of the identified factors that influence MTX adherence are inherent to the disease, treatment, or patient and are difficult to manipulate to improve adherence or persistence. Some reasons associated with poor adherence may be modifiable (e.g., MTX dose, higher folate doses, or switch to folinic acid/leucovorin). Although most studies investigating MTX dose as a factor that influenced adherence did not find an independent association23,31,45, 1 study found that a higher dose was associated with better persistence40. Reasons for withdrawal were not reported, however, making this finding difficult to interpret. Folic/folinic acid use has been shown to reduce overall withdrawals from MTX58, but a general lack of reporting of folate supplementation in the studies in the review makes commenting on the potential effect of this variable on persistence problematic. In the few studies that included folate supplementation as a covariate in regression analysis, 1 found a positive association between supplementation and adherence42; other reports, including 1 in which patients received low-dose MTX (mean dosage 5.5 mg/week), found no relationship45,50.
The effects of concomitant biologics on MTX adherence or persistence are also difficult to gauge. Two studies indicated no association between biologic use and MTX adherence12,55, whereas among patients initiating MTX therapy in a Veterans Affairs registry, a higher proportion of nonadherent patients than adherent patients were treated with concomitant tumor necrosis factor antagonists (p = 0.04)11. However, patient numbers were relatively small.
Few studies have examined interventions to improve adherence in patients with RA. One study59 did not find that targeting beliefs about medication had any effect on DMARD adherence. Another study60 found that supplying a nonadherent patient’s rheumatologist with a report about medication use and adherence did not change adherence or the patient’s beliefs about medicine.
Several studies9,12,61,62 have shown that patients with better adherence generally have less disease activity. Only 2 studies11,25 specifically evaluated the effect of MTX adherence on clinical outcomes. One study11 suggested that high rates of adherence were associated with improved DAS28 and that disease activity measures were superior at baseline in adherent users already established in receiving MTX, but not in incident users, likely reflecting that patients with better adherence already had better clinical outcomes. Nevertheless, the non-representative design of the population (veterans, 92% men) and the lack of ability to adjust for baseline DAS28 scores make definitive conclusions difficult. In another study25, the effects of MTX adherence on DAS28 were lost over time, possibly because a step-up in therapy might have occurred in patients with poor outcomes. Of course, as in any cohort study, it is difficult to disentangle causal relationships; it may be that poor disease control is the driving factor of poor adherence rather than the reverse. Similarly, adherent patients often have better outcomes, irrespective of the underlying therapy, because their attention to other health-promoting strategies is often superior to that of nonadherent patients.
Our systematic review has a number of limitations, largely because of the heterogeneity of available data. Some studies had populations that made it difficult to generalize results to the typical RA population, and others were performed several decades ago and may include outdated practices, such as using low doses of MTX, which might have influenced persistence rates because of inefficacy, or no folate supplementation, which might have contributed to increased toxicity-related withdrawals. Changes in MTX prescribing have occurred over time. Current practice is to prescribe doses of about 15 mg/week; many patients are able to tolerate doses of ≥ 20 mg/week, and folate supplementation is routine. In addition, all studies included in our review were nonrandomized. Many were susceptible to bias, mainly because of lack of clarity around how patients who were lost to followup were accounted for, data accuracy, and representativeness of the sample population.
MTX adherence and persistence are suboptimal in patients with RA. Measuring adherence is not standardized, but should be part of routine clinical practice. No consistent variables have been identified regarding MTX adherence or persistence. Research is necessary to determine the effect of MTX nonadherence on health outcomes in patients with RA and to identify independent predictors of nonadherence to inform the development of evidence-based interventions.
ONLINE SUPPLEMENT
Supplementary data for this article are available online at jrheum.org.
Acknowledgment
We thank Juliette Allport and ApotheCom, who provided writing services on behalf of F. Hoffmann-La Roche Ltd.
Footnotes
Full Release Article. For details see Reprints/Permissions at jrheum.org.
Funding for manuscript preparation was provided by F. Hoffmann-La Roche Ltd. J.R.C. has received fees from F. Hoffmann-La Roche, UCB, Janssen, Corrona, and Amgen. M.A. is an employee of F. Hoffmann-La Roche Ltd. M.S. has received fees from AbbVie, Bristol-Myers Squibb, and F. Hoffmann-La Roche.
- Accepted for publication July 9, 2016.
Free online via JRheum Full Release option
