Mixture and Non-mixture Cure Models for HTA (Gibson et al. guidance)
This guidance explains how to use cure models—statistical tools that estimate the proportion of patients who are 'cured' by a treatment—in health technology assessment. It helps analysts, reviewers, and decision-makers understand when and how to apply mixture and non-mixture cure models, especially when long-term follow-up data are limited.
At a glance
Use when
Assessing treatments with potential for cure (e.g., cancer immunotherapies, gene therapies); limited follow-up data; need for long-term survival extrapolation
Avoid when
No biological plausibility of cure; insufficient evidence of survival plateau; high uncertainty in cure fraction estimation
Inputs
Survival data with potential for long-term survivors, treatment follow-up data, baseline covariates
Outputs
Estimated proportion of cured patients, survival curves incorporating cure fraction, extrapolated long-term outcomes
How it works
The document provides a tutorial on mixture and non-mixture cure models, describing their structure, assumptions, and implementation. It emphasizes the use of flexible parametric non-mixture cure models in HTA contexts where a cure assumption is plausible but follow-up is short. The guidance includes practical examples and Stata code for model implementation, and discusses model selection, interpretation, and validation.
- HTA domains
- Clinical Effectiveness, Costs & Economic Evaluation, Patient and Social Aspects
- Assumptions
- A subset of patients is permanently disease-free ('cured'); survival approaches a plateau; appropriate parametric form for uncured patients' survival
- Strengths
- Enables extrapolation beyond observed follow-up; accounts for potential cures; flexible parametric forms improve fit; supports cost-effectiveness analyses in curative settings
- Limitations
- Relies on strong assumptions about cure; sensitive to model specification; limited validation in real-world HTA submissions; requires careful interpretation when follow-up is short
- Also known as
- Cure Models for HTA, Gibson et al. cure models
Questions this answers
- › When are cure models appropriate in HTA?
- › What is the difference between mixture and non-mixture cure models?
- › How can cure models be implemented with limited follow-up data?
- › What are the key assumptions of cure models?
- › How should decision-makers interpret cure model results?
- › What are the advantages of flexible parametric non-mixture cure models in HTA?
References & sources
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.