Ali Saad, MD

Overvad TF, Skjøth F, Lip GYH, Lane DA, Albertsen IE, Rasmussen LH, and Larsen TB. Duration of Diabetes Mellitus and Risk of Thromboembolism and Bleeding in Atrial Fibrillation: Nationwide Cohort Study. Stroke. 2015

The authors questioned the current CHA2DS2-VASc guidelines of treating all patient with afib + diabetes with an anticoagulant to prevent stroke. Using the nationwide Danish registry they followed about 130,000 patients with a diagnosis of nonvalvular afib between 2000 and 2011 for 4 years. 12.4% of these patients also had diabetes. primary outcome was thromboembolic events. They found that the longer a person had diabetes, the higher their risk of thromboembolism was: 

  • 0-4 years conveyed a risk of HR 1.11 (1.03-1.2) 
  • 15 or more years conveyed a risk of 1.48 (1.29-1.7)

On further analysis they found that the relationship was dose dependant. Duration had no effect on bleeding risk while on anticoagulation, which challenges the idea of prior stroke + diabetes as a relative exclusion criteria for IV tPA in the 4.5 hour window per the ECASS III paper. It makes sense that the longer you have a atherogenic disease such as diabetes, the more likely you are to develop thromboembolic disease. Diabetes may serve as marker of other (or the currently established) mechanisms underlying afib.

Limitations of this study include:
– when they controlled for baseline use of an antiplatelet agent, used death as an outcome measure, included patients with a diagnosis of cancer, or controlled for CHA2DS2-VASc components, there was no difference in primary outcome between short and long duration of diabetes
– when limiting the analysis to patients treated with a vitamin K antagonist, there was no difference in primary outcome
– retrospective nature
– it’s unclear how long these patients actually had diabetes, they may have just been newly diagnosed when admitted for another reason. regardless, there still appears to be a dose dependant relationship between duration of diabetes and thromboembolic risk.
– the study excluded patients with thromboembolism/TIA on the day of admission and those who died
– study is limited to the use of vitamin K antagonists and not the DAOCs
– bleeding related to a vitamin K antagonist was only captured in those who had their vitamin K antagonist started within 30 days of afib diagnosis
– it’s difficult to control for comorbidities as they would increase in prevalence as the duration of diabetes increases

How does this study change my practice?
Like any other component of metabolic syndrome, I’ll counsel patients that the longer they’ve had the disease, the more damaged their heart and vasculature are as a result. I don’t think we can conclude that diabetes independently conveys additional risk as the association disappeared with controlling for various things as mentioned above. as the authors note, the duration of diabetes likely correlates with how sick one’s cardiovascular system is due to the various associated comorbidities. I don’t currently use diabetes + prior stroke as an exclusion criteria for the 4.5 hour window for tpa.