Users' Guides to the Medical Literature
Guyatt G, Rennie D, Meade MO, Cook DJ
Part B Therapy
Chapter 9.3. Randomized Trials Stopped Early for Benefit
Victor Montori, P. J. Devereaux, Holger Schünemann, Maureen O. Meade, Deborah J. Cook, Gordon Guyatt
Although statistical simulation can readily demonstrate how truncated RCTs will overestimate...
randomized controlled trials, randomized controlled trials, truncated, stopping rules, treatment effect, truncated studies
"Taking the point estimate of the treatment effect at face value
will mislead if the decision to stop the trial resulted from catching
the apparent benefit of treatment at a random high. Consider
a hypothetical set of RCTs testing a treatment with a true,
but modest, underlying benefit. Even early in their conduct, results
will cluster around the true effect (Figure 9.3-1).
Even so, half these trials will, by chance, overestimate the true effect
and half will underestimate the true effect (Figure
9.3-1). In some, the overestimates and underestimates will
be large. The smaller the number of events, the greater the risk
that the play of chance will result in apparent effects far from
the truth (Figure 9.3-1)...."
Log in to read the full chapter:
Get full access to JAMAevidence two ways:
Subscribe to JAMAevidence
JAMAevidence is a subscription-
based website dedicated to the learning, teaching, and practicing of evidence-based medicine.
Pay Per View
Timed access to all of JAMAevidence
24 hours for $34.95
48 hours for $54.95