Presentation (CTI)

OBJECTIVES: In France, the CEESP issues cost-effectiveness opinions, yet no cost-effectiveness threshold exists, leaving the incremental cost-effectiveness ratio (ICER) without an interpretive scale. We test whether the efficiency verdict on 2025 CEESP opinions depends on the reference value used, and also we quantify the effort behind them.
METHODS: ICERs were extracted from the 2025 CEESP activity report and confronted with three French anchors: the statistical QALY value (VSQ; Téhard et al. 2020), willingness-to-pay per QALY (WTP-Q; Téhard et al. 2024), and an emerging national league table. We estimated a bottom-up production cost per opinion across all stakeholders.
RESULTS: Of 20 opinions transmitted to the pricing committee (CEPS), 18 had an interpretable ICER; 17 were cost per QALY, 2 of them dominant. Across the remaining 15, the median ICER was 101,688 EUR/QALY (mean 151,691; range 17,880 to 461,452; Q1 42,968; Q3 219,144). The median fell below the lower VSQ bound (147,093 EUR/QALY), suggesting efficiency, yet sat atop the WTP-Q range (27,847 to 112,586 EUR/QALY), suggesting borderline acceptability; Q3 and the maximum exceeded every anchor. The same opinion can thus be judged efficient or not. Estimated production cost was about 470,000 EUR per opinion (range 400,000 to 550,000), roughly 10 to 13 million EUR yearly across 24 eligible files. CEPS reportedly negotiates around 300 reimbursement files annually while receiving about 20 opinions, with no framework to interpret them.
CONCLUSIONS: Without a reference value, a published ICER is not interpretable, and the efficiency verdict shifts with the chosen anchor. We propose a triangulated reading rather than a single threshold: VSQ as a breakeven ceiling, WTP-Q as a preference benchmark, league tables as an empirical anchor, while acknowledging that public ICERs based on list prices diverge from confidential net prices. This would turn a costly but underused analytical effort into an explicit input to price setting.