Presentation (CTI)

OBJECTIVES: Since first developing cost-effectiveness analyses (CEA) for drug therapies in 2014, the Institute for Clinic and Economic Review (ICER) has positioned itself as a champion of value based price, while critics argue that ICER’s reports are conducted prematurely, are dependent on major assumptions, and undervalue the benefits of drugs. The aim of this project is to perform an assessment between ICER and pharma-sponsored CEAs for the same drug intervention by comparing the model structures, inputs, and outputs.
METHODS: ICER reports published between January 2021 to December 2024 were obtained from the ICER website; drugs of interest were identified and corresponding pharma-sponsored CEAs were obtained from a literature review of PubMed, Google Scholar, Tuft’s CEA Registry, and US economic journals (eg, JMCP, Value in Health) for eligible studies published until December 2024. The search was conducted using the drug of interest and “cost-effectiveness analysis” to retrieve potentially relevant publications. Economic evaluation quality and congruence in model structure and inputs were performed using the Quality of Health Economics Studies (QHES) instrument.
RESULTS: Outcomes for the ICER-sponsored and pharma-sponsored report pairs (n=15) varied widely with differences in the incremental cost-effectiveness ratio ranging from <1 up to 4,268%. Both ICER-sponsored and pharma-sponsored reports graded similarly on the QHES instrument with median scores of 90 (range, 82-96) and 90 (range, 75-100), respectively. A majority of the matched reports were incongruent when it came to model structure (n=12), cost inputs (n=12), drug of interest price assumptions (n=7), clinical inputs (n=14), and health utilities inputs (n=10).
CONCLUSIONS: Despite having wide variances in results for the same drug intervention, ICER-sponsored and pharma-sponsored models graded similarly on a quality-level and were generally of high quality according to the QHES instrument. However, differences in model results were mainly related to differences in model structure, model assumptions, and inputs.