Ensayos clínicos en UCI: «Más allá del valor p»

Of the frequentist methods in statistics, the “p-values” when interpreting the results do not inform about the effect size, they are based only on the 5% threshold to declare a real effect in the hypothesis test, and at the same time do not provide probabilistic information. That is why it is important to consider other alternatives or statistical methods in the current context. Bayesian inference or approach as this alternative method addresses the above limitations by having the ability to directly estimate the probability of a significant clinical benefit, and allowing researchers to quantify the influence of prior information on the interpretation of the final results, in order to provide temporary solutions to the problem of significance in the frequentist framework. As a general objective we will review the Bayesian approach or inference trying to address these limitations, some aspects such as statistical and mathematical formulas are beyond the scope of this review.