Back

Ivabradine and Atrial Fibrillation Incidence: A Nested Matching Study

KIM, K.; Keur, J.; Anderson, H.; Radwanski, P.; Munger, M. A.

2025-01-12 pharmacology and therapeutics
10.1101/2025.01.10.25320367 medRxiv
Show abstract

ObjectiveThe objective of this real-world study was to quantify and project the cumulative incidence of atrial fibrillation (AF) associated with exposure to ivabradine (IVB) in naive-AF heart failure (HF) patients. MethodsA retrospective observational comparative study was performed using a health plan claims database. The new diagnosis of AF was compared between adult HF patients receiving IVB versus no IVB controls (CTRs) over 180-day follow-up period. Eligible IVB subjects were free from AF diagnosis before the IVB index date, which should be longer than 6 months. Incidence-density sampling was performed to select matched CTRs based on clinical characteristics, AF naive status and time to be selected as a CTRs corresponding to the time to receive IVB from the initial HF diagnosis. The measure of IVB-AF association was tested using Cox Proportional Hazards Regression model. ResultsOf the 153 IVB and 4,494,305 CTRs meeting the AF-naive HF status, the analytic cohort of 107 IVB and 321 matched controls were created. The groups were well matched for age: 52.9{+/-}11.3 (IVB) vs. 53.5{+/-}11.8, gender: 57% male, HFrEF diagnosis: 66.4%, co-morbidities: hypertension 81.3% and coronary artery disease event 38.3%, and goal-directed medical therapy: {beta}-adrenergic blockers 89.7%, ACEi/ARB/sacubitril/valsartan 79.4%, mineral corticoid antagonists 39.3%. The adjusted HR [95% CI] estimates was 7.293 [4.985-10.668]. The only variable that predicted AF incidence was the IVB prescription. ConclusionsIn HF AF-naive patients, analysis of healthcare claims point to a significant risk of AF receiving IVB.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.