Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
3438216 | American Journal of Obstetrics and Gynecology | 2009 | 4 Pages |
Abstract
Randomized controlled trials may provide erroneous conclusions when the null hypothesis is not rejected because of insufficient analysis statistical power. The authors dispute the conclusion of a randomized controlled trial that compared chronic pain relief rates following laparoscopic adhesiolysis and diagnostic laparoscopy and recommended abandoning laparoscopic adhesiolysis. In the trial, the observed difference between pain rates (15%) was inferior to that expected (35%). On the basis of this result, we calculated the 90% confidence interval of the true difference, whose limits of -1% and 31% were found to fall outside the predetermined equivalency interval (-10% to 10%). The trial should therefore not have concluded that the 2 surgical procedures were equivalent. By doing so, it is likely that numerous surgeons have abandoned laparoscopic adhesiolysis on the basis of this statement. In any randomized trial, a calculation of statistical power is required each time that the null hypothesis cannot be rejected.
Related Topics
Health Sciences
Medicine and Dentistry
Medicine and Dentistry (General)
Authors
Horace MD, PhD, Tara F. PhD, RN, CNE, Loïc MD, PhD, Thomas C. ScD, MSPH,