The beta error and sample size determination in clinical trials in emergency medicine

Excerpts

Failure to achieve statistical significance between interventions does not prove the absence of any difference. Proper planning of a clinical trial with attention to beta error and sample size determination allows the critical investigator to acknowledge the probability of a type II error and therefore the probability of detecting a clinically meaningful difference if one exists. The reader and clinician must be aware that negative trials may in fact be falsely negative, and should look for specific reporting of alpha, beta, and $\Delta_c$ error to provide the details of the reliability of such conclusions. Type II error should be as essential a requirement for publication, and as rigorously analyzed, as the traditional and far more common type I (alpha) error.

Reference

Charles G Brown, Gabor D Kelen, James J Ashton, Howard A Werman “The beta error and sample size determination in clinical trials in emergency medicine” (1987) DOI: 10.1016/s0196-0644(87)80013-6

@Article{brown1987,
  title = {The beta error and sample size determination in clinical trials in emergency medicine},
  volume = {16},
  issn = {0196-0644},
  url = {http://dx.doi.org/10.1016/S0196-0644(87)80013-6},
  doi = {10.1016/s0196-0644(87)80013-6},
  number = {2},
  journal = {Annals of Emergency Medicine},
  publisher = {Elsevier BV},
  author = {Brown, Charles G and Kelen, Gabor D and Ashton, James J and Werman, Howard A},
  year = {1987},
  month = {feb},
  pages = {183–187}
}