The impact of responder analyses on trial power in patients with kidney disease and depression: A simulation study

Investigators can transform continuous data into dichotomous data by splitting it at a threshold (e.g. its mean, median, an arbitrary value or patient important threshold) and designating individuals above and below this threshold into two separate groups1. Investigators commonly perform dichotomization to simplify the analysis and interpretation of results. Other reasons include conservative hypothesis testing1 that is often a prerequisite for regulatory approval2.
Source: Journal of Clinical Epidemiology - Category: Epidemiology Authors: Source Type: research