CE4 ANALYSIS OF FACTORS INFLUENCING ACCEPTANCE OF DATA FROM MATCHING-ADJUSTED INDIRECT COMPARISONS BY NICE

Abstract

The objective of this study was to identify factors related to acceptance of findings from matching-adjusted indirect comparisons (MAICs) in appraisals performed by the National Institute for Health and Care Excellence (NICE). NICE Single Technology Appraisal documents were searched as of 31 December, 2019 to identify appraisals where MAICs were performed to provide clinical or economic data as part of the submission. Publicly available appraisal consultation documents and committee papers were then reviewed to identify MAIC methodological considerations such as rationale for use of the MAIC methodology, use of unanchored/anchored analyses, covariates included and rationale for inclusion, reductions in effective sample size. The perspective of the Evidence Review Group / committee was also analyzed to determine NICE’s perspective on the MAIC results based on the methodology employed. A total of 17 STAs were identified and included as part of the analysis, with nearly all of identified MAICs were performed in the oncology setting (94%; 16/17). The majority of MAICs were performed due to the single-arm nature of the clinical evidence available (36%; 5 of 14 where information are available), though other commonly stated reasons included lack of connection to a relevant comparator (29%; 4/14), and a desire to match for inclusion/exclusion criteria and other prognostic variables (21%; 3/14). Reductions in effective sample size (when reported) ranged from 9.52% to 99%. Ultimately, data produced by most MAICs was rejected by NICE (57%; 8/14). Common reasons for rejection included lack of inclusion of all relevant prognostic variables, variable and uncertain results due to selection of different prognostic variables, and concerns about small effective sample size. Though MAICs are increasingly utilized by manufacturers in submissions to NICE, careful consideration must be taken to covariate selection and effective sample size in order to secure acceptance of findings.