Vascular disease and renal impairment mediate the higher thromboembolic risk in atrial fibrillation patients with heart failure with reduced ejection fraction

Background/Introduction Atrial fibrillation (AF) frequently coexists with heart failure (HF). While the CHA2DS2-VASc score guides thromboprophylaxis, the contribution of left ventricular ejection fraction (LVEF) phenotype to thromboembolic risk burden remains incompletely characterized. We hypothesised that patients with HF with reduced ejection fraction (HFrEF) carry a distinct comorbidity profile that independently elevates thrombotic risk beyond the HF component of the score. Purpose To investigate whether HF phenotype, classified by LVEF, is associated with differences in CHA2DS2-VASc score and to identify which comorbidities mediate this association in patients with AF. Methods We conducted a single-centre retrospective analysis of 431 consecutive patients with non-valvular AF. Patients were classified by LVEF as HFrEF (LVEF ≤40%), HF with mildly reduced ejection fraction (HFmrEF, LVEF 41–49%), or HF with preserved ejection fraction (HFpEF, LVEF ≥50%). CHA2DS2-VASc and HAS-BLED scores were calculated. Individual score components and comorbidities including vascular disease, renal impairment, diabetes mellitus, and hypertension were compared across groups using Kruskal-Wallis and chi-square tests. Multivariable linear regression modelled the CHA2DS2-VASc score as a continuous outcome, and sequential logistic regression assessed whether the HFrEF–thrombotic risk association was mediated by specific comorbidities. Results Of 363 classifiable patients (mean age 73.8±9.7 years, 51.4% male), 309 had HFpEF, 37 HFrEF, and 17 HFmrEF. CHA2DS2-VASc scores were significantly higher in HFrEF (4.2±2.0) and HFmrEF (4.9±1.7) compared with HFpEF (3.6±1.6; p=0.007), while HAS-BLED scores did not differ (p=0.209). Vascular disease was markedly more prevalent in HFrEF (26.7%) versus HFpEF (8.5%; p=0.003), as was renal impairment (33.3% versus 7.2%; p<0.001). In multivariable linear regression (R²=0.92), HFrEF independently added 0.56 points to the CHA2DS2-VASc score (p<0.001). However, sequential logistic regression revealed that the unadjusted association between HFrEF and high CHA2DS2-VASc (≥4) was non-significant (odds ratio 1.05, p=0.91) and was fully attenuated after adjustment for vascular disease and renal impairment (odds ratio 0.74, p=0.48), identifying these comorbidities as the principal mediators. Conclusions In patients with AF, HFrEF is associated with significantly higher thromboembolic risk scores compared with HFpEF, driven not by the reduced LVEF itself but by a clustering of vascular disease and renal impairment. These findings suggest that the elevated thrombotic burden in HFrEF reflects an ischaemic–cardiorenal phenotype rather than an independent effect of ventricular dysfunction, and support individualised risk assessment beyond LVEF classificationRisk profilesMediation Analysis and Predictors