Figure 4.

Two-way sensitivity analysis examining the effect of varying the annual risk of embolic events (x-axis) and the relative hazard of major bleeding associated with culprit alleles of CYP2C9 during the maintenance phase of anticoagulant therapy (y-axis). Hypothetical patients can therefore be assigned risk estimates for each of these two parameters. The area on the graph where they fall then allows assignment of a strategy either of (1) “Do not anticoagulate,” (2) genetic testing followed by aspirin for those individuals found to possess CYP2C9*2 and/or CYP2C9*3 culprit alleles, or (3) anticoagulation without genetic testing. Patients whose risk for thromboembolism is judged to be high are placed in the region to the right. For these patients, if the relative hazard for bleeding conferred by CYP2C9 is low, anticoagulation without prior testing is favored. For patients in whom the rate of embolic events is judged to be low (but not less than 1% per year) and the relative hazard conferred by CYP2C9 is high, genetic testing followed by aspirin in those found to possess the culprit genetic variants is best. For our base case 69-year-old man with nonvalvular AF, estimates for these two parameters place him just within the genetic testing region.