Zhiyu Yan

Background Atrial fibrillation (AF) is a potent risk factor for stroke. The presence of competing etiologies can modify disease outcomes and demand different treatment strategies. Objectives The primary purpose of the study was to examine the differences in outcomes for patients with AF admitted with a recurrent stroke, stratified according to the presumed etiology of the stroke. Methods We analyzed AF patients admitted for a recurrent ischemic stroke in an academic comprehensive stroke center. Recurrent strokes were categorized as “Cardioembolic”, meaning AF without any competing mechanism, versus “Undetermined” etiology due to competing mechanisms. We used logistic regression to test the association between recurrent stroke etiology and favorable outcome (discharge home), after accounting for important covariates. Results We included 230 patients, with a mean age 76.9 (SD ± 11.3), 52.2% male, median National Institute of Health Stroke Scale (NIHSS) score of 7 (IQR 2–16). Patients with cardioembolic stroke (65.2%) had higher median NIHSS 8.5 (3–18) versus 3 (1–8) and were more likely to be treated with reperfusion therapies. The favorable outcome was reached by 64 patients (27.8%), and in-hospital mortality was 15.2% overall. After adjustment, there was no difference in outcome between patients with cardioembolic versus undetermined stroke etiology (odds ratio for discharge home: 1.41; 95% CI: 0.65–3.15). Conclusions In this single-center sample of AF patients with history of stroke, there was no difference in discharge outcomes between those with cardioembolic and those with undetermined stroke etiology. This question warrants examination in larger samples to better understand the importance of the stroke mechanism and secondary prophylaxis.

Purpose: To examine indications for, duration of use, and rate of adverse drug events (ADE) attributable to anticonvulsant initiation, as adjudicated by expert review of electronic health records (EHR) of older adults.Methods: We identified a cohort of community dwelling Medicare beneficiaries with linked EHR (aged 65+, continuously enrolled with a large health system/until death between 2012 and 2014, n = 20 945) and drew a stratified EHR review sample (n = 1534). An expert reviewed all records to adjudicate anticonvulsant use, years of use, indication for use, and evidence of ADEs attributable to anticonvulsant initiation. After excluding patients with insufficient EHR data (n = 37; 2%), we reconstructed the cohort using inverse probability weights to resemble the original cohort of eligible beneficiaries (n = 20 380). Among incident users of a single anticonvulsant, we estimated the rate of ADEs and described the type and severity of ADEs.Results: Overall, 12% (n = 2469) of eligible beneficiaries used at least one anticonvulsant in the 2012 to 2014 period (4% [n = 757] incident users, 8% [n = 1712] prevalent users). Incident users were most frequently prescribed gabapentin (n = 461/757, 61%), benzodiazepines (n = 122/757, 16%), and levetiracetam (n = 74/757, 10%); the most common indication was pain relief (n = 214; 28%) followed by epilepsy (n = 53; 7%). Among incident users, the overall ADE rate was 10/100 person‐years (95% CI 4‐20/100 person‐years), of which 29% (n = 28/97) were life threatening (eg, somnolence). Most ADEs among incident monotherapy users were nervous system related (68%, n = 66/97).Conclusion: Many older adult community dwelling traditional Medicare beneficiaries had clinically significant ADEs likely attributable to the initiation of anticonvulsant therapy, which was begun for a range of indications.