AF anticoagulation

Both scores give C (heart failure) 1, H (hypertension) 1, A₂ (age ≥75) 2, D (diabetes) 1, S₂ (prior stroke/TIA/thromboembolism) 2, V (vascular disease) 1 and A (age 65–74) 1. CHA₂DS₂-VASc adds Sc (female sex) 1 (max 9); CHA₂DS₂-VA drops it (max 8).

ACC/AHA/ACCP/HRS 2023 keeps CHA₂DS₂-VASc and sets thresholds by annual risk: OAC recommended at ≥2%/yr, e.g. score ≥2 in men or ≥3 in women (Class 1, LOE A); reasonable at ≥1 to <2%/yr, i.e. 1 in men or 2 in women (Class 2a, LOE A). For intermediate scores, other risk modifiers can help: AF burden, persistent or permanent AF, obesity, HCM, poorly controlled BP, eGFR <45, proteinuria and LA enlargement.

ESC 2024 uses CHA₂DS₂-VA. OAC is recommended at a score of ≥2 (Class I, LOE C) and should be considered at 1 (IIa, LOE C). Female sex acts as an age-dependent risk modifier rather than an independent factor. Removing it means men and women share the same thresholds.

Both guidelines: aspirin, alone or with clopidogrel, is not an alternative to OAC for stroke prevention. Re-evaluate stroke and bleeding risk and dosing at intervals.

Swedish AF cohort: 90 490 patients never on warfarin during follow-up (some took aspirin). These are unadjusted event rates per 100 patient-years, from Table 2. The paper also gives rates adjusted for aspirin, which are about 15% higher. Absolute rates vary a lot between cohorts, so read this table as an estimate.

H uncontrolled hypertension (SBP >160) · A abnormal renal function (dialysis, transplant, creatinine ≥200 µmol/L) and/or abnormal liver function (cirrhosis, or bilirubin >2× ULN with AST/ALT/ALP >3× ULN), 1 each · S prior stroke · B bleeding history or predisposition · L labile INR (TTR <60%, warfarin only) · E elderly (>65) · D antiplatelet/NSAID and/or alcohol ≥8 drinks/week, 1 each. Maximum 9; ≥3 is high risk.

Both guidelines say a bleeding score should not be the reason to withhold OAC. ACC/AHA 2023 states that bleeding risk scores should not be used in isolation to determine OAC eligibility; they are for finding and modifying bleeding risk factors (3: No Benefit). ESC 2024 does not endorse a specific score. Many items also raise stroke risk, so a high HAS-BLED usually means the net benefit of OAC is still large. Fix what can be fixed: blood pressure, unnecessary antiplatelets or NSAIDs, alcohol, labile INR (switch to a DOAC).

CrCl: Cockcroft–Gault with actual body weight, as the rivaroxaban and edoxaban labels specify. This tool uses the CrCl rounded to the nearest whole number against the label cut-offs.

• eGFR is not CrCl. Label cut-offs are Cockcroft–Gault CrCl in mL/min. eGFR (CKD-EPI, mL/min/1.73 m²) uses a different equation and is indexed to body size. It can move an older or low-weight patient into the wrong dose band.
• Inappropriate under-dosing is common. Off-label low DOAC doses are associated with more stroke or systemic embolism (OR 1.21) without less bleeding, and with higher all-cause mortality (HR 1.24) (ACC/AHA 2023 §6.3). Use the full dose unless the label criteria are met.
• Apixaban needs 2 of 3 criteria. Age ≥80 alone, or weight ≤60 kg alone, is not a reason for 2.5 mg. The criterion is serum creatinine ≥1.5 mg/dL, not CrCl.
• Edoxaban's weight and P-gp reductions are for VTE only. In AF the only dose cut is for CrCl 15–50.
• Valves: with a mechanical valve or moderate–severe (rheumatic) mitral stenosis, use warfarin, not a DOAC. A bioprosthetic valve or other native valve disease is not an exclusion under the 2023 guideline. Check each label: dabigatran's label does not recommend it with a bioprosthetic valve.
• Re-check renal function at least yearly and more often in CKD or acute illness, and re-dose as CrCl changes.