How does this new bit of information change what we believe?
According to Hartung, Knapp and Sinha:
“Most statistical methods of Meta-analysis focus on deriving and studying properties of a common estimated effect, which is supposed to exist across all studies. However, when heterogeneity across studies is believed to exist, a meta-analyst ought to estimate the extent and sources of heterogeneity among studies. So, whether a fixed-effects model or a random-effects model, Bayesian idea of approach considers all parameters (population effects sizes for fixed effects models, in particular) as random and coming from a super population with its own parameters.”
To put it simply, Bayesian meta-analysis is the use of external evidence in the design, monitoring, analysis, interpretation and reporting of a health technology assessment. The biggest advantage of this kind of approach is that it is more flexible and ethical than the traditional methods. It also handles multiple sub studies elegantly. Other advantages of this method include allowance for parameter uncertainty in the model and the ability to include relevant information that would otherwise be excluded.
However, for the field of medical technology this method should be used cautiously. The researcher should make proper use of guidelines, software and evaluate the data critically.
One major drawback that is noticed in this approach is that it is largely concerned with the prior source and interpretation of the conclusions. This means that this method can only be used when the conclusion of a study is robust, which may limit the options for article selection in research.
For formal decision analysis, this method can combine probability distribution with quantitative measures, that makes the decision making process easier. This skill has been applied to many fields such as technology, monitoring, development of clinical recommendations and analysis in the field of environment.
In a nutshell, along with medicine, this method is also widely used in the fields of engineering, processing, expert systems, decision making, financial analysis and other complex models. Although there are still certain flaws in Bayesian meta-analysis method, as it is in the evolving stage. It is still one of the most widely used methods because it is able to accommodate complex and uncertain models.