Discussion
We employed a large-sized claims database to identify an association between higher medication copayments and lower adherence in individuals with SLE. For hydroxychloroquine, azathioprine and mycophenolate mofetil, the ORs of adherence were 0.61 (95% CI 0.55 to 0.68), 0.44 (95% CI 0.30 to 0.66) and 0.69 (95% CI 0.49 to 0.96), respectively, for copayments above $10 compared with lower copayments. For methotrexate, the association did not reach statistical significance, possibly due to a smaller sample size of patients prescribed this medication (only 4.4% of our cohort). In our analysis of more than 10 000 individuals with SLE, the differences in odds of adherence per high copayments persisted even after adjustments for demographics, several comorbidities and social factors. These results underscore the urgency of ameliorating our patients’ access to affordable medications, especially considering that in this study, we examined medications that are the backbone of SLE treatment (hydroxychloroquine) and some that have been scarcely studied in regard to their cost-related adherence (azathioprine, mycophenolate mofetil and methotrexate).
Our research highlights and confirms the high burden of medication costs among individuals with rheumatic diseases. Out-of-pocket costs may pose significant financial strains on persons with RA who need biological disease-modifying antirheumatic drugs (DMARDs).25–29 In a study of Medicare beneficiaries, coinsurance (percentage paid by patient) for biological DMARDs averaged 29.6%, with a mean out-of-pocket cost of $835/month across all medications examined.25 A systematic review of six studies evaluating the association between out-of-pocket costs and adherence in RA found that costs negatively affect treatment adherence.27 Notably, one of the studies showed that increments of $5.5 in weekly out-of-pocket costs were enough to impact adherence.28 A retrospective cohort of individuals with RA from a national Medicare Advantage and Prescription Drug plan assessed the relationship between out-of-pocket costs and initial prescription, fill and refill of biological DMARDs.29 This study revealed that 18.2% of the therapy initiators abandoned their prescription after a reversed (non-paid) claim for a biological agent in the 180-day follow-up period. The rate of prescription abandonment further increased with the increase in out-of-pocket costs.29 In patients with non-rheumatological chronic illnesses, the burden of copayments has been shown to affect treatment adherence and worsen disease outcomes.30–32 On the other hand, decreasing or eliminating copayments has been associated with increased adherence, especially among low-income individuals.33
In a case–control study that assessed sociodemographic and prescription data from an SLE cohort, individuals with a new diagnosis were matched by age, sex, race and geographical residence to those with other chronic diseases. Cases were twice as likely as controls to report cost-related non-adherence and more likely to skip medication doses, take less medications and delay filling their prescriptions due to costs.34 In our study, even seemingly small increments in the costs of SLE medications had a notable negative association with adherence. A review of SLE adherence studies reiterated fear of side effects, low economic status and medication costs as factors associated with medication non-adherence.35 Despite the overall adherence in our study being higher than that of many studies in the literature, azathioprine and mycophenolate mofetil presented the lowest adherence (69.5% and 76.3% of individuals on these medications had ≥80% PDC, respectively). A previous Medicaid claims data study reported adherence to azathioprine and mycophenolate mofetil to be as low as <22%,36 while a small retrospective study of connective tissue diseases (SLE, mixed connective tissue disease, myositis, vasculitis and Sjogren’s syndrome were included) found an overall adherence to mycophenolate mofetil of 68.3%.37 Interestingly, in our study, azathioprine had the largest magnitude of decrease in odds of adherence for high copayments (56%), and mycophenolate mofetil had the highest absolute copayment (median of $10), as shown in table 2.
Non-adherence to treatment in SLE has been linked to increased emergency department visits and hospitalisation, even after consideration of sociodemographic factors and comorbidities.38 In a patient interview-based prospective study (n=982), patients who reported forgetting medications ‘all the time’ were more likely to have emergency room visits than those forgetting ‘some’ or ‘most of the time’.39 In a US paediatric cohort study, claims data-derived adherence to hydroxychloroquine and immunosuppressants was predictive of future hospitalisation. This study also used PDC ≥80% as measure for adherence.40 A longitudinal cohort with 332 individuals with SLE evaluated the association between patient medication cost concerns (encompassing difficulty affording medications, delaying refills, requesting lower-cost alternatives, etc) and patient-reported outcomes. In this study, the group of patients that expressed medication cost concerns had worse disease activity, depression symptoms, pain, fatigue and poorer sleep compared with the group without medication cost concerns, and this difference persisted over time.41 In Canada, adherence to antimalarials in SLE was shown to decrease the risk of death by 83% and 42% compared with those with limited adherence.42 Notably, this study used a stricter definition of adherence than ours (PDC >90%).
Our study identified that increased copayments are associated with lower odds of adherence. Mechanisms that may explain our findings pertain to both patients’ financial insecurity and physicians’ orientation toward the financial obstacles to treatment. Individuals with SLE suffer from a high comorbidity burden,43 and often are affected by polypharmacy and its associated costs.44 The burden of polypharmacy which frequently accompanies SLE treatment may exacerbate financial strain due to prescription costs. Concurrently, physicians have limited ability to estimate their patients’ drug costs. In a survey of physicians practising in New York, 80% of respondents were unaware of the costs of medications, and 55% inaccurately estimated prices for commonly used drugs.45 In another survey that included 131 randomly selected rheumatologists, participants had difficulty estimating out-of-pocket costs for patients even when given adequate information about their patients’ insurance plans.46 On the provider level, there is room for improvement in awareness of drug costs, with more engagement in patient–doctor discussions regarding medication cost concerns.
Our study contributes to the body of scholarship which has demonstrated the financial costs of SLE. Prior literature has reported the significant financial toll on the healthcare system.13 47–49 We now demonstrate how patient-level costs may modify medication adherence, fundamental for maintaining treatment and reducing disease morbidity and adversity. Importantly, studies have focused on the costs of newer medications such as biologics; however, in our study, we show the impact of costs on the usage of non-biological immunosuppressive medications for SLE, which are usually considered first line of the disease treatment. Targeted healthcare policies to lower patient cost-sharing are essential and may translate to increased medication adherence and improved health outcomes.30 Despite the available data demonstrating the effects of adherence on healthcare utilisation, costs and possibly patient outcomes, guidelines do not yet consider medication prices as a component of treatment choice.2
The strengths of our study include using a large and diverse cohort of individuals with SLE, including hydroxychloroquine, azathioprine, methotrexate and mycophenolate mofetil, along with several relevant covariates. As was shown by another claims data study, concordance between adherence to hydroxychloroquine and adherence to immunosuppressants is only moderate, demonstrating the importance of considering other SLE therapies individually.40
Our study has some important limitations. Foremost is the potential for misclassification that is inherent to using claims data. To reduce potential misclassification, we employed ICD-9/10 codes specific to SLE with the additional stipulation that individuals are prescribed medications for SLE. Second, while our cohort is large sized with geographical viability, the generalisability is limited by including only individuals with insurance. We expect challenges to medication access would be all the more challenging in the USA for those with chronic rheumatological diseases and who lack health insurance. Third, we used PDC to quantify adherence. We recognise that PDC accounts for claims from filling prescriptions but does not measure actual medication usage and ingestion. Observed therapy is both impractical and not feasible when using administrative claims data. Fourth, we observed small sample sizes for analyses of azathioprine, methotrexate and mycophenolate mofetil individually. Finally, we are unable to exclude residual confounding from unmeasured factors that may contribute towards the association between medication copayment and adherence, such as disease severity or other barriers not measured by administrative data.
In conclusion, we used a large, administrative health claims database to determine that increased copayments are associated with reduced adherence to hydroxychloroquine and immunosuppressive medications in individuals with SLE. Incorporating awareness of the burden of copayments and its consequences into medication adherence is essential for improving medication access in individuals with SLE, and has the potential to improve outcomes and decrease healthcare utilisation in a complex and socially costly chronic disease.