The decision regarding therapy for reducing stroke risk in NVAF involves trade-offs among bleeding, stroke, and inconveniences associated with three options: warfarin, aspirin and no therapy. A recent meta-analysis of 29 RCTs (mean age of 71 years, 35% women) found a greater absolute reduction in stroke risk and a small incremental risk of major hemorrhage associated with warfarin compared to aspirin.7
As a result, guidelines recommend the use of warfarin for patients at moderate or high risk for stroke who do not have an absolute contraindication to warfarin.8
In contrast, a number of observational studies have demonstrated that the risk of bleeding associated with warfarin is not small.9–12
In a population-based study utilizing Medicare data, the rate of bleeding resulting in hospitalization ranged from 1.9 to 12.3 per 100 patient-years.9
Several observational studies have demonstrated that both the risk of stroke and the risk of bleeding vary according to patients’ comorbidities.9–11, 13
Because of this variability in risk, there is a large range in the incremental risks and benefits associated with warfarin, aspirin and, no treatment. The provides examples of the 5-year risks of stroke and bleed, converted from annualized outcome rates,14
associated with each option for two 70-year old men with different comorbidities. The baseline stroke risk and risk of bleeding with warfarin were based on validated risk calculators derived from observational, population-based data.9, 15
The risks of stroke with warfarin and aspirin were derived by applying the 67% and 21% reduction in stroke risk associated with these two therapies published in meta-analysis of RCT data.7
Risk calculators are not available for risk of bleeding with no treatment and with aspirin; therefore, these risks were taken from a systematic review.16
Five-year risk of stroke and bleeding associated with no treatment, aspirin, or warfarin in the treatment of NVAF according to comorbid conditions:
As can be seen in the , expected outcomes calculated using the best available data vary markedly according to the individual’s comorbidities. Moreover, specific comorbidities differentially affect the risk of stroke and bleed. For example, in the case of a 70 year old man with well-controlled hypertension, heart failure, diabetes mellitus, and non-ulcer related abdominal pain, the first three comorbid conditions increase his baseline risk of stroke but not bleeding, and the last comorbid condition modestly increases his risk of bleeding with aspirin but not warfarin. In contrast, for a 70 year old man with poorly controlled hypertension, renal disease, and a history of a fall, only the first comorbid condition increases his baseline stroke risk while all the comorbidities increase his bleeding risk.
Despite the availability of a large number of studies and formal meta-analyses, the best available calculations of individualized benefit and harm in NVAF still depend upon a number of assumptions because of the absence of data needed to provide individualized estimates. The following paragraphs outline how the absence of these data affects decision making at the individual patient level.