Capturing Real-World Evidence Alongside Trial Evidence in HTA-Specified Systematic Literature Review: Optimizing Search Strategies across the Totality of Evidence Required

OBJECTIVES: There is increasing use of real-world evidence (RWE) in HTA with treatment effect estimates from real-world, non-randomized studies potentially considered in relative efficacy assessments. This is changing the way we search for evidence, with systematic literature review (SLR) extending from randomized controlled trials (and single-arm trials) to RWE with similar methodological rigour. SLRs are a time-consuming aspect of HTA and there is no guidance, to our knowledge, about how to combine searches for clinical evidence, including trial and RWE, alongside other evidence requirements. Other HTA requirements include quality of life and health-state utility values, which can overlap clinical RWE, alongside cost, resource use and economic evaluations. The stringency for this additional evidence can be lower than for clinical (e.g. single full-text review) and these searches are often run separately. We aimed to test overall efficiency of search strategies for HTA.

METHODS: We tested how different combinations affected the total number of citations retrieved from Embase by constructing pairs of search strings. These were across oncology indications with identical overall PICOS terms and limitations. Each pair included the totality of evidence required for HTA and RWE treatment effects but differed in how the evidence was partitioned: separate clinical (clinical trials + all other) and combined clinical (clinical RWE/clinical trials + all other). Differences in citations retrieved were due to duplication between searches.

RESULTS: Total numbers of citations retrieved ranged from 2086 to 13 541. There were consistently fewer overall citations, showing lower duplication, for the combined clinical approach vs separate clinical: mean 10.9% fewer citations (SD, 0.10%).

CONCLUSIONS: Our results show that when RWE of treatment effects may be considered alongside trial evidence it is more efficient to combine the searches across study type than to keep separate. These RWE studies then follow the same review process as trials.