Bayesian Intervals

A simple tool to quickly perform basic Bayesian parameter estimation for a binary outcome using a beta-binomial model, using publication-level trial data.

Authors
Affiliations

Jake Barlow

Fellow, CVICU, Auckland City Hospital

David Sidebotham

Consultant, CVICU, Auckland City Hospital

Published

March 24, 2024

How to Use

  1. Input the trial data into the contingency table.
  2. Next, set a prior probability for the control and intervention groups.
    The sliders will set 95% of the probability mass between each extreme, with the maximum value approximately between them. Identical rates in both groups indicate clinical equipoise.
  3. If required, adjust the Highest Density Interval (HDI) and Region of Practical Equivalence (ROPE) to the desired values.
    • The 95% threshold for the HDI is fundamentally arbitrary, but is used by convention.
    • The Region of Practical Equivalence is the region where the effect size is weak enough to be considered clinically negligible. This is intervention-dependent: complex or expensive treatments will generally require a wider ROPE.

Risk Calculator