Background: Evidence has consistently shown that adherence to AM is poor in systemic lupus erythematosus (SLE) patients. However, data on the impact of adherence to AM on mortality is scarce.
Objectives: To assess the effect of AM adherence on all-cause mortality in SLE patients from the general population.
Methods: This study used administrative databases from British Columbia, Canada. We created an incident SLE cohort between January 01, 1997, and March 31, 2015, using the physician billing data and a 7-year washout period. The inclusion criteria were at least two physician visits, at least two months apart, within two years, with an ICD-9 code (710.0) or ICD-10 code (M32.1, M32.8, M32.9) for SLE. Follow-up started at the first day of having both SLE and AM, i.e., at the SLE index date (second ICD code) for those whose first AM use occurred before the SLE index date, or the date of the first AM use if otherwise. Our outcome was all-cause mortality, obtained from the vital statistics registry. In the analysis, the follow-up time was divided into 30-days windows, for a total of 293,190 person-months. For each window, a measure of adherence, the proportion of days covered (PDC), was calculated and categorized as adherent (PDC≥0.90), non-adherent (0<PDC<0.90), and discontinuer (no drug or PDC = 0). We used both Cox’s proportional hazards models and marginal structural models (MSM) to estimate the effect of AM adherence on all-cause mortality. Both analysis controlled for baseline demographics (age, sex, residence, income quintile), as well as the following baseline and time-varying covariates: immunosuppressive and other medications, hospitalizations, impatient, and other visits, and Charlson comorbidity index. To account for the possibility of a few time-varying covariates being mediators in the causal pathway from AM adherence to mortality, which may cause the Cox model to yield biased estimates of the adherence effects, we conducted the MSM analysis that can produce valid estimates as it balances the distributions of time-varying confounders among the three adherence groups via inverse probability weighting.
Results: We identified 3,385 individuals with incident SLE (mean age 47.3 years, 89% were women) who had at least one filled AM prescription. Over the mean follow-up of 6.66 years, 288 (8.5%) incident SLE patients died. The incidence rate (IR) of mortality for AM adherent, non-adherent, and discontinuer patients were 4.31, 11.86, and 19.51 per 1000 person-years, respectively. Using the Cox model, the adjusted hazard ratio (HRs) obtained for AM adherent and non-adherent SLE patients were 0.20 and 0.66, respectively, compared to discontinuer SLE patients (Table 1). Using MSM, those adjusted HRs were found as 0.18 and 0.64. Also, the adjusted HRs for adherers compared to the non-adherers were 0.30 (Cox) and 0.28 (MSM). A statistically significant linear trend in the HRs of mortality risk over the adherence levels was found (Table 1, Linear Trend).
Conclusion: SLE patients that adhere to AM therapy have a lower risk of death than patients who do not adhere or who discontinue AM (5 and 3 times, respectively) in both the MSM and Cox analysis. Our findings support the importance of AM adherence to prevent premature deaths in SLE patients.
Disclosure of Interests: None declared
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.