Accounting for Cured Patients in Cost-Effectiveness Analysis

Economic evaluations often measure an intervention effect with mean overall survival (OS). Emerging types of cancer treatments offer the possibility of being "cured" in that patients can become long-term survivors whose risk of death is the same as that of a disease-free person. Describing cured and noncured patients with one shared mean value may provide a biased assessment of a therapy with a cured proportion.

The purpose of this article is to explain how to incorporate the heterogeneity from cured patients into health economic evaluation.

We analyzed clinical trial data from patients with advanced melanoma treated with ipilimumab (Ipi; n = 137) versus glycoprotein 100 (gp100; n = 136) with statistical methodology for mixture cure models. Both cured and noncured patients were subject to background mortality not related to cancer.

When ignoring cured proportions, we found that patients treated with Ipi had an estimated mean OS that was 8 months longer than that of patients treated with gp100. Cure model analysis showed that the cured proportion drove this difference, with 21% cured on Ipi versus 6% cured on gp100. The mean OS among the noncured cohort patients was 10 and 9 months with Ipi and gp100, respectively. The mean OS among cured patients was 26 years on both arms. When ignoring cured proportions, we found that the incremental cost-effectiveness ratio (ICER) when comparing Ipi with gp100 was $324,000/quality-adjusted life-year (QALY) (95% confidence interval $254,000-$600,000). With a mixture cure model, the ICER when comparing Ipi with gp100 was $113,000/QALY (95% confidence interval $101,000-$154,000).

This analysis supports using cure modeling in health economic evaluation in advanced melanoma. When a proportion of patients may be long-term survivors, using cure models may reduce bias in OS estimates and provide more accurate estimates of health economic measures, including QALYs and ICERs.