Controlling Confounding in a Study of Oral Anticoagulants: Comparing Disease Risk Scores Developed Using Different Follow-Up Approaches