Common flaws exist in published cost-effectiveness models of pharmacologic stroke prevention in atrial fibrillation.