PS 4-30
EXAMINING THE IMPACT OF INTRACRANIAL HEMORRHAGE RELATIVE TO ISCHEMIC STROKE IN ANTICOAGULATED PATIENTS WITH NONVALVULAR ATRIAL FIBRILLATION
Decision analyses regarding anticoagulation must weight hemorrhagic and ischemic outcomes. In recognition of the generally worse outcomes of intracranial hemorrhage compared with ischemic stroke, many analyses use a fixed weight of 1.5, with sensitivity analyses from 1.0 to 2.0. But the use of a uniform weight across different ages and risk factors may mask patient heterogeneity. It may therefore be appropriate to use different weights for different subpopulations, any of which may differ meaningfully from the range in current use.
To better inform future decision-analytic models, we derived a ratio of quality-adjusted-life-years (QALYs) lost to intracranial hemorrhage compared with ischemic stroke among patients with nonvalvular atrial fibrillation.
Method:
We created a Monte Carlo simulation predicting the QALYs lost to intracranial hemorrhage compared to ischemic stroke. We began with a synthetic population intended to mirror the atrial fibrillation population of the United States. Each hypothetical patient was simulated in a stroke condition and an intracranial hemorrhage condition, drawing from a variety of datasets to predict downstream morbidity and mortality. The QALY loss in each condition, and the ratio of the two conditions, were calculated. We then created a regression model of the simulation results (a “meta-model”) to demonstrate the sensitivity of this ratio to each of the input variables, and predicted the marginal QALY loss ratio at various ages.
Result:
The ratio of QALY loss from intracranial hemorrhage relative to ischemic stroke is skewed (mean of 1.56, median 1.02, IQR 0.70-1.82). Patients with longer life expectancies have, on average, a higher ratio of QALY loss from intracranial hemorrhage compared with ischemic stroke. For example, the marginal predicted ratio of QALY loss from intracranial hemorrhage, relative to ischemic stroke, is 2.82 in a 40-year-old patient, and 0.89 in a 90-year-old patient (see table).
Conclusion:
The use of a fixed weight of intracranial hemorrhages relative to ischemic stroke biases decision-analytic models in favor of anticoagulation in patients with long remaining life expectancy, and away from anticoagulation in patients with short remaining life expectancy. Use of a dynamic weighting ratio that varies with age can improve future decision-analytic models in anticoagulation.
Table. Mean marginal weighting ratio at selected ages.
Age |
Mean marginal weighting ratio |
40 |
2.82 |
50 |
2.43 |
60 |
2.05 |
70 |
1.66 |
80 |
1.28 |
90 |
0.89 |