Sample Size Determination for Clinical Trials by Using Bayesian Approach
The sample size determination is the statistical estimation of the number of patients to be enrolled in the study, which provides correct analysis, high accuracy level, and also validates the significant expected results of the study (endpoints). Today, sample size calculation has become an important issue in the pharmaceutical sector because the too large sample may waste time, resources and cost, while too small sample size may lead to incorrect results. Therefore, the prior determination of sample size is a pivotal part of the statistically well-designed clinical study, which results in the fruitful conclusion. For sample size calculation, information about two parameters is necessary, i.e., clinically meaningful difference between a test treatment and control, and population variation across the previously conducted clinical trial. Generally, these parameters are unknown, but they can be usually estimated from existing literature
What is Bayesian Approach?
There are various statistical methods for sample size calculation like a simple formula, Frequentist, classical methods, etc., but the Bayesian approach is often used because of its robust decision-making theoretical framework. Based on the Bayes’ theorem, the Bayesian approach provides a predictive probability for calculation of sample size based on existing as well as new parameters. The Bayesian method also provides power decision and precision analyses. In power decision, it provides a statistics-based decision for treating the cost or effectiveness problem as well as recovery to the missed data during the clinical study. While in precision analyses, Bayesian approach provides true parameters for effective calculation of sample size for potential outcomes. Therefore, in the past several decades, this approach for sample size determination has attracted many clinicians for research and development of the drug.
Why Bayesian Approach is considered for Sample Size Determination?
Currently, sample size estimation has become a significant challenge to the pharmaceutical sector for the success of the clinical study. But statistical approaches help the clinicians to predict the effective estimation of sample size; some statistical methods predict the sample size based on unknown parameters, which increase the possibilities of the wrong estimation. But Bayesian approach overcomes this because it estimates the sample size on the basis of actual parameters. Now, pharmaceutical companies are using the Bayesian approach in their statistically designed studies for effective estimation of sample size. The Bayesian approach provides various other benefits, which include:
- Providing median and maximum estimates of sample size with risk factors
- Giving a robust theoretical framework for the success of the clinical study
- Comprising fruitful estimation of sample size on the basis of factual parameters and not the random ones
- Recovering the missing data problems during the study
- Providing the statistical analysis for reduction of cost, resources, and time during the clinical study
- Providing potential outcomes or a better interpretation of result through a flexible parametric model
Including benefits of Bayesian approach, it also has some disadvantages like this approach provides high computational cost mainly in models which have several parameters and sometimes can show variations in the results as predicted by the Bayesian approach.