Adjusting for baseline: change or percentage change?

Stat Med. 1989 Oct;8(10):1183-90. doi: 10.1002/sim.4780081002.


Clinical trials data often include baseline and response measurements on each patient. Comparisons of treatments commonly employ either change from baseline or percentage change from baseline as the analysis variable. This paper provides guidance on the choice between these two by means of plots of these variables versus baseline. I provide two examples that illustrate the problem and I discuss the impact of use of the wrong adjusted response.

Publication types

  • Comparative Study

MeSH terms

  • Clinical Trials as Topic / standards*
  • Cluster Analysis
  • Computer Simulation*
  • Data Interpretation, Statistical*
  • Decision Support Techniques*
  • Drug Therapy / statistics & numerical data
  • Models, Statistical*
  • Normal Distribution
  • Predictive Value of Tests
  • Reference Standards