OBJECTIVES: Methodological guidance for ITCs of RCTs typically recommend anchored analysis over unanchored analyses, since the latter relies on the assumption of all prognostic variables being balanced or adjusted. We sought to explore scenarios where unanchored analyses may be more credible than anchored counterparts

METHODS: (None)

RESULTS: Unanchored comparisons (either adjusted or unadjusted) may be more credible than anchored alternatives in three broad scenarios: Chains of evidence are long or travel through studies with variable populations; Events in the shared comparator are rare leading to unstable effect estimates and inappropriately inflated uncertainty intervals; and highly effective comparators have biological/empirical rationale to be stable across patient populations while placebo/control is highly variable. In these situations in mature disease areas where most prognostic variables are known/reported, adjusted (and sometimes unadjusted) unanchored ITCs may be expected to make weaker assumptions than their anchored counterparts.

CONCLUSIONS: Despite generally unanimous focus on anchored comparisons are methodologically preferrable, unanchored comparisons may actually make fewer or weaker assumptions in special cases. It is likely inappropriate to maintain a strict hierarchy of ITC methods.