Testing for baseline balance in clinical trials

Stat Med. 1994 Sep 15;13(17):1715-26. doi: 10.1002/sim.4780131703.


Once the data from a clinical trial are available for analysis it is common practice to carry out 'tests of baseline homogeneity' on prognostic covariates before proceeding to analyse the effects of treatment on outcome variables. It is argued that this practice is philosophically unsound, of no practical value and potentially misleading. Instead it is recommended that prognostic variables be identified in the trial-plan and fitted in an analysis of covariance regardless of their baseline distribution (statistical significance).

MeSH terms

  • Algorithms
  • Analysis of Variance
  • Data Interpretation, Statistical*
  • Humans
  • Multivariate Analysis*
  • Randomized Controlled Trials as Topic / statistics & numerical data*
  • Research Design*
  • Sample Size
  • Selection Bias*
  • Treatment Outcome