OBJECTIVES: Population-adjusted indirect comparison (PAIC) has emerged as a frequently employed statistical method in oncology submissions to NICE due to high study/population heterogeneity and use of single-arm trials. However, the utilisation of PAICs in non-oncological indications remains relatively less studied. We assessed the trends of PAIC use in NICE submissions for non-oncological indications.

METHODS: NICE website was thoroughly searched to identify all non-oncology drug submissions from 2019 to June 2023 (excluding terminated appraisals). For each appraisal, use of PAIC methods, Evidence Review Group’s critique, and final recommendations were assessed.

RESULTS: Of the 115 non-oncology-related submissions identified, only 10 submissions (8.6%) included a PAIC; 90% included a matching-adjusted indirect comparison (MAIC) and 10% included both MAIC and simulated treatment comparison (STC). Of these 10 submissions, two submissions each were in nervous system-related, blood-related, and skin-related conditions. Sixty percent of the submitted PAICs were anchored (conducted due to study/population heterogeneity), while the remaining 40% utilised an unanchored approach (due to lack of comparative studies). There has been an increase in overall proportion of submissions including a PAIC over recent years; from 5% in 2019 to 11% in 2022 with no submission using PAIC in 2023 till June. Results of the PAICs were accepted for decision-making in 30% submissions, accepted with caution due to high uncertainty in 30% submissions, and were judged not robust for decision-making in 40% submissions (results from alternative approaches like Bucher ITC or NMA were preferred). The main critique for the submitted PAICs was insufficient matching for covariates (60% submissions).

CONCLUSIONS: PAIC use in non-oncology submissions to NICE is limited and a cautious approach was adopted by NICE committee in accepting the results. Despite, a trend was observed regarding increasing use of PAICs (mainly MAIC) in non-oncological submissions to NICE. Advanced methods like STC or multi-level network meta-regression were rarely used.