BackgroundCardioembolic stroke (CS) due to atrial fibrillation (AF) bears a high risk of unfavorable outcome. Treatment with a non-vitamin K antagonist oral anticoagulant (NOAC) reduces this risk. NOAC dosage occurs on a thin line during the acute phase of the stroke unit when the patient is threatened by both recurrent CS and a hemorrhagic stroke. It is often adapted to renal function—usually glomerular filtration rate (GFR)—to prevent both under- and overdosing. This study investigates the hypothetical risk of incorrect NOAC dosage after acute stroke when relying on plasma creatinine alone in comparison to a more exact renal function assessment including urine collection.MethodsIn a cohort study on consecutive 481 patients treated in a stroke unit with acute stroke and AF, the GFR estimated from plasma creatinine (eGFR) was compared to concurrent creatinine clearance measurement (CrCl) from urine collection regarding the hypothetically derived NOAC dosage.ResultsThe risk of incorrect dosage (mean, 95% confidence interval) was 6.9% (4.8–9.5), 26% (23–31), 38% (33–42), and 20% (16–23) for apixaban, dabigatran, edoxaban, and rivaroxaban, respectively. The overall risk for incorrect dosage of any NOAC was 23% (21–25). Thresholds for age and admission eGFR were optimized to achieve an overall risk below 5% by additional CrCl measurements in selected patients (apixaban <36 ml/min and any age, dabigatran <75 ml/min and >70 y, edoxaban >36 ml/min and >58 y, rivaroxaban <76 ml/min and >75 y, any NOAC <81 ml/min and >54 y). The resulting portion of patients requiring an additional CrCl measurement was 10, 60, 80, 55, and 65% for apixaban, dabigatran, edoxaban, rivaroxaban, and any NOAC, respectively.ConclusionsThere is a considerable risk of incorrect NOAC dosage in patients with acute CS treated in a stroke unit that can be lowered by targeted CrCl measurements in selected patients.