OBJECTIVES: Population-adjusted indirect comparisons (PAIC) are statistical methods frequently employed to compare the effectiveness of treatments across trials where direct head-to-head comparisons are not available. In 2020, Phillippo et al published a new PAIC method, multi-level network meta-regression (ML-NMR), which is an extension to Bayesian network meta-analysis that allows for population adjustments and incorporation of patient-level data for some trials and aggregate data for others. The method overcomes issues with existing approaches in that estimates can be made in target populations of interest, and can leverage the entire evidence network, not just a single indirect comparison. The aim was to identify any use of ML-NMR in NICE submissions to date, as well as challenges with other PAIC methods that may be improved upon with ML-NMR.

METHODS: The NICE website was searched to identify drug submissions from January 2021 onwards. Terminated appraisals were excluded. Each appraisal was assessed by a reviewer and quality checked by a second reviewer. The use of any indirect comparison was extracted from the company submission; amongst submissions with a PAIC, the evidence review group report was reviewed to determine if the selected approach was considered appropriate.

RESULTS: The search, conducted on May 1^st, 2024, yielded 110 submissions; eight were excluded as they either duplicates, withdrawn or relied on older company submissions. Of the 102 remaining, 40 (39%) included at least one population-adjusted method but none employed an ML-NMR. Matching-adjusted indirect comparison (MAIC) was the most employed method (n =29, 75%). A criticism of MAICs was inconsistency in covariate selection across comparisons; a single model conducted using ML-NMR could address this concern.

CONCLUSIONS: To date, no submissions have utilized the ML-NMR approach despite its potential to address limitations of other PAIC approaches. Given its appealing features, it is expected that its utilization will increase in the following years.