Borrowing strength from `indirect' evidence: methods and policy implications for health technology assessment

Sparse relative effectiveness evidence is a frequent problem in HTA. For example, evidence on a particular comparator or randomised evidence in a specific population (e.g. paediatric) may be limited. Where evidence directly pertaining the decision problem is sparse, one could expand the evidence base to include studies relating to the decision problem only indirectly: for instance, when there is no evidence on a specific comparator, evidence from other treatments of the same molecular class could be used; similarly, a decision on children may borrow strength from evidence on adults. Usually, in HTA, such indirect evidence is either included by ignoring any differences (`lumping') or is not allowed to influence the decision (`splitting'). However, more sophisticated methods exist in the literature which, rather than lumping or splitting, borrow strength from the indirect evidence by imposing more moderate, and perhaps more appropriate, degrees of information-sharing.

This thesis commences with a systematic review that sought to identify methods to combine evidence directly and indirectly relating to a research question. A classification of Information-Sharing Methods (ISMs) according to the main assumption employed to facilitate information-sharing is proposed. Subsequently, detailed descriptions of methods’ assumptions and implementation suggestions are provided in the context of a specific synthesis problem. To aid transparency in selecting ISMs, a step-by-step approach that could be useful for HTA analysts and policy-makers is proposed. Then, all applicable methods are used to borrow strength from indirect evidence on relative effectiveness in a case-study. Findings imply that the choice of method can affect how much strength is borrowed, impact relative effectiveness estimates, and influence adoption and research recommendation decisions. Then, the strength of information-sharing imposed by the various methods is investigated using probabilistic scenarios. Finally, lessons learned throughout the thesis are distilled into a set of recommendations for HTA practice.