Presentation (CTI)

OBJECTIVES: Medical devices (MDs), unlike treatments, experience rapid technological updates, varied usage across healthcare settings, and differing real-world outcomes, that may pose problems in their economic evaluations (EEs). This study aims to review the modelling approaches and key critiques reported for EEs of MDs in NICE’s Medical Technologies Guidance (MTG) and suggests potential strategies to address the identified challenges.
METHODS: We reviewed MTG documents published on the NICE website from January 1, 2020, to December 2, 2024. For each included MTG, we extracted information on the type of economic analysis, model structure, and the critiques raised by NICE or external reviewers.
RESULTS: Of the 41 MTGs identified, 37 were included for analysis (4 excluded due to lack of evidence reported). The most frequently used EE methods were cost-consequence analysis (9/37, 24.3%) and cost analysis (8/37, 21.6%), followed by cost-effectiveness analysis (6/37, 16.2%), cost-minimization analysis (4/37, 10.8%), cost comparison (2/37, 5.4%), cost-benefit analysis, and budget-impact analysis (each 1/37, 2.7%). Six MTGs (16.2%) did not specify EE method. The most used model structure was decision-tree (20/37, 54.1%), followed by Markov models (11/37, 29.7%), hybrid of decision-tree and Markov model (3/37, 8.1%). Three MTGs (8.1%) used simple calculations without any economic model. Key critiques were uncertainty in clinical evidence (n=11), non-reliable data sources (n=9), unrealistic assumptions (n=7), inadequate or overly complexed model structure (n=5), and inadequate time horizon (n=3). Six MTGs reported robust models, with no major critiques, whereas two were critiqued for focusing only on cost comparison and neglecting the health outcomes, and one faced question on model validity.
CONCLUSIONS: The findings highlight persistent challenges in modeling the economic impact of MDs. Refining the models to reflect real-world complexities, using flexible time horizons, improving data quality, and considering differences in MDs use across settings is recommended to produce robust results in the upcoming evaluations.