Kcentra is a plasma-based CSL-Behring drug approved for VKA reversal due to major bleeding. (VKA = Vitamin K Antagonists like Warfarin). FDA is concerned about TEE (thromboembolytic events) as a side effect. CSL sponsored a prospective observational study in a real-world setting with a health provider (Kaiser Permanente), with TEE occurrence and survival as outcomes of interest. The epidemiologists at CSL soon noticed evidence of "channeling bias": Patients with intracranial bleeds were preferentially "channeled" to Kcentra; patients with gastro-intestinal bleeds were channeled to an older treatment. Intracranial bleeds are considered a more serious condition than gastro-intestinal bleeds.

Questions:

1. Would the observed channeling bias favor or disfavor Kcentra?
2. Is the confounder of interest observed or unobserved?
3. How would you address the issue in your planned analysis?