Observational Patient Evidence for Regulatory Approval and uNderstanding Disease (OPERAND)

Study type

• Study type:
  Non-interventional study
• Scope of the study:
  Effectiveness study (incl. comparative)

Non-interventional study

• Non-interventional study design:
  Cohort

Study drug and medical condition

• Anatomical Therapeutic Chemical (ATC) code:
  (B01AA03) warfarin

  warfarin

  (B01AF01) rivaroxaban

  rivaroxaban

  (A10BJ02) liraglutide

  liraglutide

  (A10BB12) glimepiride

  glimepiride
• Medical condition to be studied:
  Type 2 diabetes mellitus

  Atrial fibrillation

Population studied

• Age groups:
  Adults (18 to < 46 years)

  Adults (46 to < 65 years)

  Adults (65 to < 75 years)

  Adults (75 to < 85 years)

  Adults (85 years and over)
• Estimated number of subjects:
  9999

Study design details

• Main study objective:
  To replicate 2 previously published RCTs of pharmacological products used as the basis of marketing approval by the FDA. For each trial, we will first mimic the inclusion/exclusion criteria, endpoint definitions, exposure windows, and other study design features and then use state-of-the-art causal inference methods to estimate ATE and compare to those reported in the original publications
• Outcomes:
  For the first study, primary outcome is a composite of the occurrence of stroke (ischemic or hemorrhagic) and systemic embolism. For the second study, primary outcome is change in hemoglobin A1C at the end of the study compared with baseline.
• Data analysis plan:
  In this first study, to mimic the primary analysis in the ROCKET AF trial, we will conduct analyses to obtain the analog of intention-to-treat and per-protocol treatment effect estimates. Cox proportional hazards regression model will be used to estimate hazard ratios for treatment effects. We will use different analytical approaches that control for baseline confounding. The 95% confidence interval (CI) for estimates in all approaches will be obtained via nonparametric bootstrap. In the second study, to mimic the primary analysis in the LEAD-2 trial, we will conduct analyses to obtain the analog of intention-to-treat treatment effect estimates. We will use a linear regression model to estimate the mean difference in change in A1C from baseline. We will use different analytical approaches that control for confounding due to baseline characteristics. The 95% CI for estimates in all approaches will be obtained via nonparametric bootstrap.