curesurv
curesurv is a software tool that helps estimate the proportion of patients who are 'cured' of a disease, particularly in cancer studies, by analyzing survival data. It uses advanced statistical models to distinguish between patients who are cured and those who are not, based on how long they survive after diagnosis.
At a glance
Use when
Analyzing long-term survival data where a proportion of patients may be cured, such as in cancer or chronic disease registries.
Avoid when
When follow-up time is too short to observe plateauing survival, or when cure is biologically implausible.
Inputs
Individual-level survival data (time-to-event, event indicator), covariates, and population mortality rates for excess hazard modeling.
Outputs
Estimated cure fraction, time to cure, survival curves for cured/uncured subgroups, model parameters, and goodness-of-fit statistics.
How it works
curesurv implements both mixture and non-mixture parametric cure models using excess hazard modeling. It supports the mixture cure model (Phillips et al., 2002) with a Weibull distribution for uncured patient survival, and non-mixture models such as the time-to-null excess hazard model (Boussari et al., 2020). The package enables estimation of cure indicators like cure fraction and time to cure, leveraging population-based survival data.
- HTA domains
- Clinical Effectiveness, Patient and Social Aspects
- Assumptions
- The population includes both cured and uncured individuals; the survival of uncured patients follows a parametric distribution (e.g., Weibull); excess hazard diminishes over time and may reach zero.
- Strengths
- Enables flexible modeling of cure phenomena using both mixture and non-mixture frameworks; incorporates excess hazard to account for background mortality; based on peer-reviewed methodology with real-world applicability in cancer survival analysis.
- Limitations
- Relies on strong parametric assumptions; requires long-term follow-up data to reliably estimate cure indicators; may be sensitive to model specification and choice of distribution.
- Also known as
- curesurv R package
Questions this answers
- › What proportion of patients can be considered cured of a disease?
- › How long does it take for the excess hazard to reach zero (time to cure)?
- › How does the survival of uncured patients differ from the general population?
- › Which parametric model best fits the observed survival data in a cure fraction context?
- › Can we distinguish between cured and uncured patient subgroups using population-based data?
- › How does the inclusion of cure models improve long-term survival predictions?
References & sources
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.