Predictors of new onset atrial fibrillation in patients with heart failure

• Published in: International journal of cardiology 2014-08, Vol.175 (2), p.328-332
• Main Authors: Campbell, Niall G, Cantor, Emily J, Sawhney, Vinit, Duncan, Edward R, DeMartini, Chiara, Baker, Victoria, Diab, Ihab G, Dhinoja, Mehul, Earley, Mark J, Sporton, Simon, Davies, L. Ceri, Schilling, Richard J
• Format: Article
• Language: English
• Subjects: Aged; Atrial fibrillation; Atrial Fibrillation - diagnosis; Atrial Fibrillation - mortality; Biological and medical sciences; Cardiac dysrhythmias; Cardiology. Vascular system; Cardiovascular; Cohort Studies; Coronary heart disease; Female; Follow-Up Studies; Heart; Heart failure; Heart Failure - diagnosis; Heart Failure - mortality; Heart failure, cardiogenic pulmonary edema, cardiac enlargement; Humans; Implantable defibrillator; Ischaemic heart disease; Male; Medical sciences; Middle Aged; Nephrology. Urinary tract diseases; Nephropathies. Renovascular diseases. Renal failure; Predictive Value of Tests; Renal failure; Renal impairment; Retrospective Studies
• ISSN: 0167-5273; 1874-1754
• DOI: 10.1016/j.ijcard.2014.05.023

Abstract Introduction Stroke associated with atrial fibrillation (AF) is more frequent in heart failure. It is unknown what variables predict future AF in these patients and how AF might evolve over time. We investigated this in patients with implantable cardiac defibrillators (ICD) where AF detection is optimal. Methods Single centre, retrospective, observational cohort study. All ischaemic cardiomyopathy patients with dual chamber, primary prevention ICD implants between Aug 2003 and Dec 2009 were screened and included if at implant, they had no known AF history. Nine variables were analysed. AF was defined as any atrial tachyarrhythmia ≥ 180 bpm and ≥ 30 s. Multivariable, binary logistic regression models were built by adding variables significant in the univariate models. Variables were retained in the final multivariate models if p < 0.05. Results n = 197 met the inclusion criteria (85.8% male, median age: 66.8 years). After median follow-up for 2.8 years, 44.2% developed AF. After univariate analysis, the baseline variables associated with AF after implant were age, NYHA class and renal impairment (RI, defined eGFR < 60 ml/min/1.73 m2 ) (p < 0.05). After multivariable analysis, the only variable which was associated with AF was RI (HR: 2.04 (CI: 1.10–3.79)). Two baseline variables were independently associated with all-cause mortality: RI (HR: 2.42 (1.14–5.12)) and non-white ethnicity. Conclusion RI at time of implant was independently associated with both future AF and all-cause mortality during long-term follow-up. RI was a stronger predictor of AF than age. Those patients with heart failure and RI should be regularly screened for asymptomatic AF, regardless of age, to ensure that stroke prophylaxis may be initiated.