Classifying information-sharing methods

Abstract Background Sparse relative effectiveness evidence is a frequent problem in Health Technology Assessment (HTA). Where directly pertaining to the decision sparse, it may be feasible expand evidence-base include studies that relate only indirectly: for instance, when there no on comparator, other treatments of same molecular class could used; similarly, children borrow-strength from adults. Usually, HTA, such indirect either included by ignoring any differences (‘lumping’) or not at all (‘splitting’). However, range more sophisticated methods exists, primarily biostatistics literature. The objective this study identify and classify breadth available information-sharing methods. Methods Forwards backwards citation-mining techniques were used set seminal papers topic information-sharing. Papers if they specified (network) meta-analytic combining information distinct populations, interventions, outcomes study-designs. Results Overall, 89 included. A plethora synthesis have been Most ( n =79) described shared treatment effects. Amongst these, was strong emphasis across multiple =42) =25), with fewer focusing study-designs =23) populations =8). We categorise discuss under four ’core’ relationships information-sharing: functional, exchangeability-based, prior-based multivariate relationships, explain assumptions made within each these core approaches. Conclusions This highlights available. These often impose moderate than lumping splitting. Hence, degree potentially considered appropriate. Our identification ‘core’ allows an improved understanding underpinning different Further research required understand how differ terms strength sharing implications health care decisions.