What are the key messages?


The BRAT frameworks alone are helpful for structuring and giving insights to the benefit-risk balance.

Evidence gathering and data preparation

The experience from this case study was that the steps of BRAT should be considered in a circular or parallel manner, especially this was clear between the steps "identify outcome" and "identify data sources" where each step was revisited several times.

When aggregating data from different sources it is important to be aware of issues such as different definitions of outcomes and different way of measuring certain effects, and also issue of bias when combining data. This led to some exercise of data transformation or even criteria customization in order to have matching criteria across the sources of evidence.


The circular nature of the benefit-risk assessment was emphasised in this case study, where the analysis of warfarin versus placebo for primary prevention of stroke in patients with atrial fibrillation was first done on the bases of RCTs which was taken as the base for the assessment of warfarin versus no treatment based on actual practice data, and then finally extended to incorporate data from newer RCTs of warfarin versus newer anti-coagulating agents.

With the help of patient level data we were able to show that although the overall benefit risk balance for a product seems acceptable, it might be very different from one patient to another. Some patients might benefit more than others, while others might have more risks. In this case study we have tried to use methods to identify these patients. A person's benefit-risk balance may be influenced by the specific combination of other risk factors, or by the fact that the usage of a product in real life is very different from a clinical trial setting. By mapping this benefit-risk profile of a medicine we might help prescribing physicians in giving the right drug to the right patient. Advantage of this method is that it can be easily adapted to different scenarios by changing the input data such as the relative rates and the weights. Therefore this method can be applied for both older drugs as new drugs.


Use of sensitivity analysis especially relating to preference proved very useful for this case study where some criteria were very broadly defined and preferences weights therefore difficult to assign.

Conclusion and dissemination

The visuals proposed in the BRAT framework (value-tree, key benefit-risk summary table, and forest plot) provided good tool for dissemination of the benefit-risk assessment, they introduce overview of criteria and effect of treatment to be compared. Although the criteria were listed according to importance,the final conclusion still required a trade-off to be made by the decision maker