Cluster analysis reveals “considerable heterogeneity” of atrial fibrillation


A cluster analysis of the data from the ORBIT-AF observational registry study has been published by Taku Inohara (Duke University Medical Center, Durham, USA) et al in the Journal of the American Medical Association (JAMA). The cluster analysis of the registry data identified four phenotypes of atrial fibrillation (AF) associated with distinct clinical outcomes.

The ORBIT-AF study collected data from 9,749 patients admitted for AF at 174 participating centres in the USA, with the purpose of investigating the use and outcomes of bridging during periods of interrupted oral anticoagulants. Investigator Benjamin Steinberg (Duke University Medical Center, Durham, USA) et al published the outcomes in Circulation in 2015.

The cluster analysis was undertaken by Inohara et al and performed between January and October, 2017. Using 60 baseline clinical characteristics, patients classified into four statistically driven clusters which were not defined by conventional classifications, such as AF subtype or left atrial size.

These four patient groups made up the clusters in the data analysis:

  1. 4,673 patients with considerably lower risk factor and comorbidity rates.
  2. 963 patients with AF diagnosed at a younger age and/or comorbid behavioural disorders.
  3. 1,651 patients with similarities to tachycardia-bradycardia patients and a device implantation for treatment of sinus node dysfunction.
  4. 2,462 patients with prior coronary artery disease, myocardial infarction, and/or atherosclerotic comorbidities.

Adjusted risks of major adverse cardiovascular or neurological events were significantly lower in the low-comorbidity cluster number one compared to the other three (cluster two: hazard ratio, 1.49; 95% CI, 1.10-2.00; cluster three: hazard ratio, 1.39; 95% CI, 1.15-1.68; and cluster four: hazard ratio, 1.59; 95% CI, 1.31-1.92).

Higher adjusted risk rates for major bleeding were also found for clusters two (hazard ratio, 1.35; 95% CI, 1.05-1.73), three (hazard ratio, 1.24; 95% CI, 1.05-1.47), and four (hazard ratio, 1.13; 95% CI, 0.96-1.33).

Inohara et al underline the implications of the cluster analysis, explaining that the heterogeneity of AF as well as the importance of comorbidities and substrates are highlighted by the distinct associations with clinical outcomes for each of the four clinically relevant phenotypes of AF identified by this analysis.


Please enter your comment!
Please enter your name here