Member-only story
The Haybittle-Peto Boundary: Interim Checks of Experiment Design, Part 3
There are three distinct methods of checking the p-value (probability value) of an experiment at various points. Let’s explore how Haybittle-Peto deviates from the O’Brien-Fleming and Pocock boundaries.
When interim checks are used effectively, you can stop an experiment in order to save time, money and possibly lives.
The Haybittle-Peto boundary is a statistical concept that is used in the design of randomized controlled trials (RCTs). It is a threshold that is used to determine whether or not to stop a clinical trial early because of an apparent treatment effect.
In a clinical trial, an interim analysis is an evaluation of the data that has been collected up to a certain point in time, before the trial has been completed. Interim analyses are typically planned in advance and are conducted at predetermined time points during the course of the trial. They are used to monitor the progress of the trial and to assess the safety and effectiveness of the treatment being studied.
You can use various stages for an experiment, but it is important to maintain the overall experimental wide alpha level at 0.05 if that is the predetermined threshold.
The final analysis will usually tested at a significance level less than 0.05 due to the alpha spending at the interim analyses. An analyst can use various approaches in handling the multiplicity issue due to the interim analyses have been proposed. The main advantage of the Haybittle–Peto boundary is that the same threshold is used at every pre-determined interim, unlike the O’Brien–Fleming boundary, which changes at every analysis.
Also, using the Haybittle–Peto boundary means that the final analysis is performed using a 0.05 level of significance as normal, which makes it easier for investigators and readers to understand. The main argument against the Haybittle–Peto boundary is that many people believe that the Haybittle–Peto boundary is too conservative and makes it too difficult to stop a trial.
