Concept Architecture
Concept
Theoretically, the Non-Mixture Cure Model is a parametric survival model that estimates long-term survival by allowing the survival function to converge asymptotically to a cure fraction rather than explicitly modelling separate cured and susceptible populations. It is founded on cure modelling theory and bounded cumulative hazard models. The model exists to represent situations in which a proportion of patients become effectively cured while maintaining a single continuous survival function.
Mathematically, the non-mixture cure model incorporates the cure fraction directly into the cumulative hazard or survival function. As time approaches infinity, the cumulative hazard reaches a finite limit, causing the survival function to converge to the estimated cure fraction. Model parameters, including the cure fraction and survival distribution parameters, are estimated simultaneously using maximum likelihood estimation.
In practice, non-mixture cure models are fitted to individual patient survival data using specialised survival modelling software. Model fit is evaluated using likelihood-based criteria, graphical diagnostics and comparison with alternative cure models. In health economics, non-mixture cure models are frequently used in oncology and advanced therapies where long-term survival plateaus are observed and realistic lifetime extrapolation is required for cost-effectiveness analysis.
Purpose
Used to estimate long-term survival when a proportion of patients are effectively cured, providing smooth survival extrapolation without explicitly separating patients into cured and susceptible subpopulations.
Mathematical Formulae
Primary Formula
S(t) = �^(1 ? F?(t))
where:
� = cure fraction
F?(t) = cumulative distribution function of the underlying survival distribution
Supporting Formulae
F?(t) = 1 ? S?(t)
H(t) = ?ln(S(t))
L = ?S(t?)???? ? h(t?)??
where:
H(t) = cumulative hazard
L = likelihood function
�? = event indicator
Related Mathematical Methods
- Maximum Likelihood Estimation
- Cure Fraction Model
- Mixture Cure Model
- Parametric Survival Analysis
- Weibull Model
- Log-Normal Model
- Generalised Gamma Model
Example
A study evaluates an immunotherapy for advanced melanoma using a non-mixture cure model. The estimated cure fraction is � = 0.28. At 10 years, the underlying cumulative distribution is estimated as F?(10) = 0.85.
The predicted overall survival is:
S(10) = 0.28^(1 ? 0.85)
S(10) = 0.826
The model predicts an 82.6% probability of survival at 10 years while converging smoothly towards the long-term cure fraction.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| POWER | =POWER($B$1,1-C2) | Calculate survival from the estimated cure fraction and cumulative distribution. |
| LN | =-LN(D2) | Calculate cumulative hazard from survival. |
| EXP | =EXP(-E2) | Convert cumulative hazard to survival where required. |
| Solver | Maximum likelihood optimisation | Estimate cure fraction and survival model parameters. |
VBA (Optional)
Automate estimation of non-mixture cure models, compare alternative cure models and generate lifetime survival projections for economic evaluation.
Sources
- Yakovlev AY, Tsodikov AD. Stochastic Models of Tumour Latency and Their Biostatistical Applications.
- Sy JP, Taylor JMG. Estimation in a Cox Proportional Hazards Cure Model. Biometrics. 2000.
- Maller RA, Zhou X. Survival Analysis with Long-Term Survivors.
- Othus M, Barlogie B, LeBlanc ML, Crowley JJ. Cure Models as a Useful Statistical Tool for Analyzing Survival. Clinical Cancer Research. 2012.
- NICE. Health Technology Evaluation Manual.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
Related Concepts (2)
Library
Publications
1
NICE DSU Technical Support Document 15: Cost-effectiveness modelling using patient-level simulation — Davis, Stevenson, Tappenden & Wailoo, TSD 15 ed., 2014 (NICE Decision Support Unit (University of Sheffield))
Guidance on individual patient-level (microsimulation) cost-effectiveness modelling — when to use it in preference to cohort models, how to structure it, and how to handle the associated computational and uncertainty challenges.
Frequently Asked Questions (7)
What is a non mixture cure model?
An alternative cure model structure formulating survival directly in terms of the cured fraction without explicitly splitting the population into two subgroups.
Source: Boag 1949
What is a non-mixture cure model?
A non-mixture cure model is an alternative cure model structure that formulates survival directly in terms of the cured fraction without explicitly splitting the population into cured and uncured subgroups. Instead of modelling survival as a mixture of two groups, it uses a mathematical form that incorporates the cure fraction into the overall survival function, still producing a plateau at the cured proportion. It offers a different derivation of a cure model, related to Boag's work, with properties that some find advantageous for estimation.
Source: Boag 1949
What survival model allowing for cure is a non-mixture cure model?
A non-mixture cure model is a survival model that allows for a proportion of patients who will never experience the event of interest, treating them as effectively cured. Rather than splitting the population explicitly into cured and uncured groups, it builds the cure fraction into the overall survival function through a particular mathematical form. As follow-up lengthens, the modelled survival flattens towards the cured proportion. This structure suits diseases where long-term survivors are plausible, such as some cancers. It provides an alternative to the mixture formulation of cure.
Source: Lambert et al. 2007
How does a non-mixture cure model work?
A non-mixture cure model works by defining overall survival through a function that includes the cure fraction directly, typically formulated so that as time increases, survival approaches the cured proportion, without separately specifying cured and uncured subgroup survival as a mixture. It arises from a biologically motivated formulation, such as the distribution of the time for surviving damaged cells to cause the event. This yields a survival curve with a plateau at the cure fraction, like the mixture model, but through a different structure.
Source: Boag 1949
How does a non-mixture cure model differ from a mixture cure model?
A non-mixture cure model formulates survival directly in terms of the cure fraction without explicitly dividing the population into cured and uncured subgroups, whereas a mixture cure model models overall survival as a weighted mixture of the two subgroups' survival. Both yield a survival plateau at the cured proportion and represent cure, but they differ in mathematical structure and interpretation. The mixture form is more intuitive, directly representing the subgroups, while the non-mixture form arises from a different, often biologically motivated, derivation.
Source: Latimer 2013
When is a non-mixture cure model used?
A non-mixture cure model is used to represent cure in survival data, as an alternative to the mixture cure model, particularly where its formulation is considered more suitable, for instance for its statistical properties or its biological motivation. Like the mixture model, it is appropriate when a cured fraction is present and a plateau is seen. The choice between mixture and non-mixture formulations may depend on estimation behaviour and interpretation, so the non-mixture model is one option when modelling a cured fraction.
Source: Latimer 2013
What are the advantages of the non-mixture formulation?
The non-mixture cure model's formulation is sometimes preferred for its statistical properties, such as more stable estimation in some settings, and for a biological interpretation in terms of an underlying process, such as the failure of surviving damaged cells. It avoids explicitly specifying the uncured subgroup's distribution as in a mixture, which some find advantageous. However, like the mixture model, it requires sufficient follow-up to estimate the cure fraction, so its advantages concern formulation and estimation rather than removing the fundamental challenges of cure modelling.
Source: Boag 1949
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 22 Oct 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-EM-SM-057
Stable URI · Machine-readable · Resolvable · CC BY 4.0