When it is better to estimate a slope with only one point

QJM. 2008 Oct;101(10):821-4. doi: 10.1093/qjmed/hcn099. Epub 2008 Aug 8.


When investigating the change in a biomarker, it is often believed that at least two measurements are needed from each participant, and that those with only one measurement should be excluded. In this short note, we explain why this could lead to imprecise and biased estimates. Furthermore, we discuss a standard statistical method that handles such issues.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Bias*
  • Biomarkers*
  • Computer Simulation / standards*
  • Data Interpretation, Statistical
  • Humans
  • Models, Statistical*


  • Biomarkers