The pharmaceutical industry is increasingly turning to Bayesian methods in an attempt to improve efficiency and enhance decision making. Assurance (Bayesian Power) is becoming a key feature of these Bayesian methods. This is not to replace traditional power but to complement it.
Before we continue, let us quickly examine what Bayesian Assurance is.
Assurance is the unconditional probability that the trial will yield a ‘positive outcome’. A positive outcome usually means a statistically significant result, according to (some) standard frequentist significance test. The assurance is then the prior expectation of the power, averaged over the prior distribution for the unknown true treatment effect.
In the well received paper published by O’Hagan - he argues that Assurance is an important measure of the practical utility of a proposed trial, and indeed that it will often be appropriate to choose the size of the sample (and perhaps other aspects of the design) to achieve a desired assurance, rather than to achieve a desired power conditional on an assumed treatment effect. 
Today we see a rapidly rising number of pharma companies are adopting prior elicitation and seeking to make an Assurance calculation a standard procedure in their clinical trial planning framework - with GSK the most public in publishing their reflections on it. With that in mind let us review five reasons why Pharma companies are Calculating Bayesian Assurance.
Biostatisticians are tasked with discovering the probability of success of a trial. This involves analysis and mitigating threats to the clinical trial. Using Assurance, statisticians can gain a greater insight to move towards a more informed answer to this question.
Researchers acknowledge the complications of having completely correct assumptions when calculating tradition power. By assigning a prior to where we know there is uncertainty, researchers need to focus more on the uncertainty and have a more realistic interpretation of what you actually know. By using assurance we move towards the true probability - the unconditional probability of success.
Bayesian Assurance is becoming an important tool at key milestones in the drug development process. This is due to the methodical approach required to reach an assurance calculation proved by elicitation their priors (e.g SHELF Framework) they have reviewed and identified threats and further opportunities that may not of previously been visible. This provides all parties a more informed decision at both macro and micro levels.
Calculating Bayesian Assurance also serves an important purpose at Scientific review boards. Presenting team beliefs in a probability distribution provides all involved a formalized opportunity to weight their beliefs and use this to identify gaps or opportunities to strengthen their study design and affect the later sample size.
Connected to our 5 Essential Steps to Determine Sample Size & Power, a sensitivity analysis is the usual prescription for uncertainty in planning parameters. It provides information on which assumptions have the largest influence on the sample size required. However, one of the biggest flaws with sensitivity analysis is that there are no formal rules for how a sensitivity analysis should be conducted for a sample size determination. We are of the belief that calculating Bayesian Assurance can help formalize your sample size sensitivity analysis.
nQuery Sample Size Software has a dedicated Bayesian module that allows the easy calculation of Assurance. If you are interested in trialling this yourself, click here to request a free trial.
We recently hosted a webinar Bayesian Assurance: Formalizing Sensitivity Analysis For Sample Size.
You can watch this webinar on demand by clicking the image below.
In this webinar you’ll learn about:
 O’Hagan A, Stevens JW, Campbell MJ. Assurance in clinical trial design. Pharmaceutical Statistics 2005; 4:187-201 - https://doi.org/10.1002/pst.175
These Stories on Bayesian Clinical Trials
No Comments Yet
Let us know what you think