Background There is an evidence of poor adherence (ranging from 25% to 57%) to antimalarial (AM) medications in systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA). We examine the association between AM adherence and cardiovascular (CVD) events and mortality among incident SLE and RA patients, and assess if the association in SLE differs from that in RA.

Methods All patients with incident SLE or RA and incident AM use in British Columbia, Canada, between January 1997 and March 2015 were identified using provincial administrative databases. The outcomes were incident CVD events and mortality attributed to myocardial infarction, stroke or venous thromboembolism. We used marginal structural models (MSM) to estimate the effect of AM adherence on incident CVD events and mortality, accounting for potential confounders and competing events due to death unrelated to CVD. In the analysis, for each 90-days window of follow-up time, the proportion of days covered (PDC), a measure of adherence, was calculated and categorized as adherence (PDC≥0.90), partial adherence (0<PDC<0.90), and non-taking (PDC=0) for AM use. The analyses were controlled for baseline demographics as well as the following sets of baseline and time-varying covariates: medication use, health resource utilization, comorbidities, and Romano adaptation of Charlson comorbidity index. Also, SLE and RA patients were analyzed separately.

Results We identified 21,114 individuals with incident SLE or RA (3,496 SLE; 17,618 RA patients, mean age 55.8 years, 75.9% female) with ≥1 filled AM prescription. Over the mean follow-up of 8.8 and 8.6 years, 2759 (14.3%) and 900 (4.3%) patients experienced incident CVD events and CVD mortality, respectively. The incidence rates of CVD mortality for AM adherence, partial adherence, and non-taking were 3.10, 4.68, and 5.65 per 1000 person-years. Using MSM, the adjusted hazard ratios (aHRs) of CVD mortality obtained for AM partial adherence and adherence in SLE or RA patients were 1.08 (95% CI: 0.91-1.31) and 0.51 (95% CI: 0.42-0.62), respectively, relative to non-taking (table 1). Also, the aHR for adherence compared to partial adherence was 0.47 (95% CI: 0.37-0.60). Findings were similar for CVD events (table 1). There was no significant difference in risk estimates between SLE and RA patients (Wald test p-values, table 1) though SLE patients had nonsignificant results, possibly due to the small sample size, especially for mortality.

The multivariable models were adjusted for baseline covariates including demographic variables (age, sex, location of residence, neighborhood income quintile), indicator of having RA or SLE (not considered for individual RA and SLE patients cases), health resource utilization (hospital visits, physician and specialist visits including rheumatologist, nephrologist, and psychiatrist visits), medication usage (statins, other cardiovascular drugs, hormone replacement therapy, glucocorticoids, anticoagulant therapy, Cox-2 inhibitors, immunosuppressive drugs), comorbidities (hypertension, chronic obstructive pulmonary disease, angina), and the Romano adaptation of the Charlson comorbidity index for administrative data. Also, the time-varying variables of health resource utilization, medication usage, Charlson comorbidity index, and comorbidities were used to calculate weights in marginal structural model.

Conclusions SLE and RA patients’ adherent to AM therapy had 53% and 30% lower risk of CVD mortality and incident CVD events, respectively, than partially adherent patients.