Patterns of Use and Discontinuation of Secondary Prevention Medications After Stroke

OBJECTIVE: To investigate whether certain patient, acute care, or primary care factors are associated with medication initiation and discontinuation in the community after stroke or TIA. METHODS: This is a retrospective cohort study using prospective data on adult patients with first-ever acute stroke/TIA from the Australian Stroke Clinical Registry (April 2010 to June 2014), linked with nationwide medication dispensing and Medicare claims data. Medication users were those with ≥1 dispensing in the year postdischarge. Discontinuation was assessed among medication users and defined as having no medication supply for ≥90 days in the year postdischarge. Multivariable competing risks regression, accounting for death during the observation period, was conducted to investigate factors associated with time to medication discontinuation. RESULTS: Among 17,980 registry patients with stroke/TIA, 91.4% were linked to administrative datasets. Of these, 9,817 adults with first-ever stroke/TIA were included (45.4% female, 47.6% aged ≥75 years, and 11.4% intracerebral hemorrhage). While most patients received secondary prevention medications (79.3% antihypertensive, 81.8% antithrombotic, and 82.7% lipid-lowering medication), between one-fifth and one-third discontinued treatment over the subsequent year postdischarge (20.9% antihypertensive, 34.1% antithrombotic, and 28.5% lipid-lowering medications). Prescription at hospital discharge (sub-hazard ratio [SHR] 0.70; 95% confidence interval [CI] 0.62-0.79), quarterly contact with a primary care physician (SHR 0.62; 95% CI 0.57-0.67), and prescription by a specialist physician (SHR 0.87; 95% CI 0.77-0.98) were all inversely associated with antihypertensive discontinuation. CONCLUSIONS: Patterns of use of secondary prevention medications after stroke/TIA are not optimal, with many survivors discontinuing treatment within 1 year postdischarge. Improving postdischarge care for patients with stroke/TIA is needed to minimize unwarranted d