Background: According to the American College of Chest Physicians 2016 guidelines, the minimum recommended duration of oral anticoagulant therapy for all patients with venous thromboembolism (VTE) is three months. However, clinical equipoise regarding extending treatment exists in patients with unprovoked VTE, where the decision to extend anticoagulant treatment depends on individual patient risk of recurrence and bleeding. The objective of the dissertation was to utilize the target trial framework to evaluate the effectiveness and safety of extending treatment with rivaroxaban, the first directly acting oral anticoagulant (DOAC) to be approved by the Food and Drug Administration for secondary prevention of VTE recurrence, among patients with unprovoked VTE. Methods: We used 2011 to 2017 Medicare fee-for-service insurance claims database to emulate the target trials described. In the first two chapters of the dissertation, insurance claims data was used to emulate a rivaroxaban discontinuation trial among patients age 66 and older, with acute unprovoked VTE who completed an initial three-month course of rivaroxaban therapy. The intervention was extending versus discontinuing treatment with rivaroxaban within 14 days grace period since time zero and adhering to their treatment strategy unless clinically indicated. We estimated both the observational analogue of intention-to-treat effect (Chapter 2) and the per-protocol effect or effect of sustained treatment strategies (Chapter 3). In Chapter 4, we emulated a target trial with different durations of treatment extension with rivaroxaban among patients age 66 years and older, with unprovoked VTE who completed an initial three-month course of rivaroxaban therapy. The intervention was initiating one of five different rivaroxaban treatment extensions (0-, 3-, 6- 9- and 12-months) and we estimated the observational analogue of per-protocol effect. The outcomes of interest for all chapters were VTE recurrence, major bleeding requiring hospitalization and all-cause mortality. For all three chapters, we used a three-step process of cloning, censoring and using inverse probability weighting to account for the 14-day grace period given to patients for initiating their treatment strategies from time zero, baseline confounding and potential selection bias due to artificial censoring. Results: There were 5114 eligible patients in each of the target trial emulations. The observational analogue of the intention to treat hazard ratio (HR) for VTE recurrence comparing continuing versus discontinuing rivaroxaban treatment was 0.51 (95% confidence interval (CI), 0.34 to 0.79), while the per-protocol effect was 0.39 (95% CI, 0.24 to 0.64). In Chapter 4, we observed that HR comparing strategy of 3-months extended therapy to strategy of 0-month extended therapy was 0.34 (95% CI 0.16, 0.73) and comparing strategy of 12 months extended therapy to 0-month extended therapy was 0.28 (95% CI 0.14, 0.56). We did not observe a corresponding increase in the risk of major bleeding with extended rivaroxaban in any of the three chapters. Conclusions: We concluded that extended treatment with rivaroxaban for at least a year after an initial three-month treatment episode was better than short treatment durations in patients with unprovoked VTE. We also demonstrated that we could leverage real-world healthcare databases, such as insurance claims, to help answer research questions for which trials may not be available or feasible such as comparing different treatment durations.