Despite the wide availability of Direct oral anticoagulation (DOAC), warfarin remains an important oral anticoagulant1 especially in patients with mechanical valve replacement or chronic renal failure where the evidence of DOAC is limited. However, the dosing of warfarin is challenging as it has a narrow therapeutic index and is highly influenced by dietary vitamin K intake and drugs that interacts with cytochrome (CYP) P4501. The time outside therapeutic range will carry the risk of adverse events such as thrombosis and bleeding. Numerous algorithms have been proposed that utilized either clinical or pharmacogenetics factors. The Clinical Pharmacogenetics Implementation Consortium (CPIC) has recommended the use of 4 dosing algorithms. However, these algorithms require the input of genetic information such as CYP2C9 and VKORC1 type which are not widely available locally and are less relevant in maintenance phase. Furthermore, these algorithms target mainly at predicting the initiation and the maintenance dose of warfarin. This has provided us with the opportunities to explore the Prediction model for on-treatment warfarin dose titration for outside therapeutic range.

Despite the wide availability of Direct oral anticoagulation (DOAC), warfarin remains an important oral anticoagulant1 especially in patients with mechanical valve replacement or chronic renal failure where the evidence of DOAC is limited. However, the dosing of warfarin is challenging as it has a narrow therapeutic index and is highly influenced by dietary vitamin K intake and drugs that interacts with cytochrome (CYP) P4501. The time outside therapeutic range will carry the risk of adverse events such as thrombosis and bleeding. Numerous algorithms have been proposed that utilized either clinical or pharmacogenetics factors. The Clinical Pharmacogenetics Implementation Consortium (CPIC) has recommended the use of 4 dosing algorithms. However, these algorithms require the input of genetic information such as CYP2C9 and VKORC1 type which are not widely available locally and are less relevant in maintenance phase. Furthermore, these algorithms target mainly at predicting the initiation and the maintenance dose of warfarin. This has provided us with the opportunities to explore the Prediction model for on-treatment warfarin dose titration for outside therapeutic range.

Inclusion Criteria:

1. Patients who have been on maintenance warfarin treatment
2. Patients whose INR have fallen outside therapeutic range will be identified using CDARS enquiry

Exclusion Criteria:

1. No exclusion criteria

This is based on the primary objective, previous data in Hong Kong and calculation