Abstract
The statistical analyses of oncology trials are usually performed by calculating P values, although these are poorly understood. Using P value cutoffs, such as P < 0.05, may lead to some treatments being accepted which have little benefit, and other therapies being rejected which have considerable benefit. A more intuitive and direct probability- that an experimental treatment is better than a standard treatment-can be calculated by Bayesian statistics. Here we used software to obtain the outcomes of 194,129 patients enrolled across 230 trials and then calculated probabilities of benefit. Interpretations based on P values disagreed with the probabilities of benefit in one-third of trials. This study suggests that probabilities of benefit would considerably enhance the interpretation of oncology trials.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have