Voluntary and non-remunerated blood donation by 2025? A comparative study of blood donor eligibility in 19 healthcare institutions in Lebanon.

Bayesian methods for analysis of clinical trials have been widely discussed in the statistical and medical literature. Bayesian statistical inference provides a formal framework for combining evidence obtained within the trial with relevant evidence that is available outside the trial and offers potential advantages over frequentist analysis. Borrowing evidence from external sources can be beneficial in settings such as rare diseases or paediatric populations, where performing large-scale trials can be very difficult, or sometimes impossible. Trial results obtained from a Bayesian analysis can be expressed using probabilistic language, for example by reporting the probability that the experimental treatment was superior to the control treatment, which may be viewed as more intuitive and easier to interpret than confidence intervals and values. Predictions are easily made from a Bayesian analysis, meaning that investigators can predict the probability of achieving a significant result in a planned trial given previous data or in an ongoing trial given interim data, and base decisions on these probabilities. Alongside the advantages of taking a Bayesian rather than frequentist approach to analysis of a trial, there are potential disadvantages to consider. Informative prior distributions based on opinion or data must be carefully chosen and justified, and may be criticised later by reviewers or readers for not reflecting the views of all relevant groups. Vague prior distributions representing lack of prior knowledge should also be chosen carefully because results can be sensitive to choice of vague prior, particularly for parameters informed by sparse data. Ideally, a Bayesian analysis should be repeated using two or more different priors, to assess the sensitivity of posterior inferences. If an informative prior is used, sensitivity can be explored by varying the weight allocated to the external data or opinion. Most trial statisticians have substantially more experience and training in the use of frequentist methods, meaning that errors in implementation may be more likely. Trial investigators may be concerned that Bayesian results are less likely to be accepted by regulators or policy makers, because of a lack of familiarity and the limited regulatory guidance available on Bayesian approaches. There is also a danger that Bayesian methods appear opaque and cannot be followed or easily reproduced by others not involved in the analysis. Bayesian analyses are now widely used in the drug development process, to inform internal ‘go/no-go’ decisions about planned studies, for example when deciding whether a drug should proceed from phase II to phase III trial. However, Bayesian analyses are still not commonly used for analysis of phase III (confirmatory) trials. The aims of this article are to review the recent use of Bayesian analyses in confirmatory trials, to explore which types of trial have chosen Bayesian methods and how they were used. We have reviewed published papers over a 6-year period from 2019 to 2024 and explored the characteristics of trials using Bayesian methods for their primary analysis, why a Bayesian approach was chosen, whether any informative priors were used and if so how they were informed. Next, we selected four trials from the review as case studies and presented their motivation for using Bayesian methods and their Bayesian analyses in more detail.