Detecting outliers in multivariate laboratory data

J Biopharm Stat. 2008;18(6):1178-83. doi: 10.1080/10543400802369046.

Abstract

Laboratory data collected in clinical trials often include outliers, and these are often the observations of most interest. In high dimensional settings, outliers can be difficult to detect and can be masked by classical statistical methods. A method of plotting robustly scaled data in such a way as to expose outliers is described and an application is presented.

MeSH terms

  • Clinical Laboratory Techniques / statistics & numerical data*
  • Clinical Trials as Topic / statistics & numerical data*
  • Data Interpretation, Statistical*
  • Drug-Related Side Effects and Adverse Reactions
  • Humans
  • Multivariate Analysis*
  • Treatment Outcome