survextrap
survextrap is an R package that helps estimate long-term survival from short-term clinical trial data by combining it with other data sources like disease registries or population statistics. It uses advanced statistical modeling to make predictions more reliable and transparent, especially when projecting survival over long periods.
At a glance
Use when
Estimating long-term survival in health technology assessments when only short-term trial data are available; combining clinical trial data with real-world evidence; modeling complex hazard patterns or non-proportional treatment effects.
Avoid when
When sufficient long-term individual-level data are already available; when external data sources are unreliable or not comparable; when users lack familiarity with Bayesian modeling or R programming.
Inputs
Individual-level right-censored survival data, optionally combined with summary survival data from external sources (e.g., disease registries, population mortality); prior distributions in Bayesian framework; treatment or covariate data.
Outputs
Posterior distributions of survival curves, hazard functions, and model parameters; long-term survival extrapolations with uncertainty intervals; estimates of treatment effects over time.
How it works
survextrap implements a Bayesian parametric survival model using M-spline functions to flexibly model the hazard function, allowing changes in hazard trajectories over time. It integrates individual-level right-censored data with aggregate survival data from external sources within a unified Bayesian framework. The package supports proportional and non-proportional hazards, cure models, additive hazards with background mortality, and waning treatment effects. Models are fitted using standard R survival syntax, enabling accessible and transparent survival extrapolation for health technology assessment.
- HTA domains
- Clinical Effectiveness, Costs & Economic Evaluation, Safety
- Categories
- Cost-effectiveness ModellingData Exploration & VisualizationEvidence SynthesisModel Validation
- Assumptions
- The hazard function can be represented by an M-spline; external data are relevant and combinable with trial data; Bayesian priors are appropriately specified; proportional or specified non-proportional hazards structures reflect underlying biology.
- Strengths
- Flexible hazard modeling using M-splines; principled incorporation of multiple data sources; transparent quantification of uncertainty through Bayesian inference; supports complex survival mechanisms like cure models and waning effects; integrates with standard R survival workflows.
- Limitations
- Requires specification of priors and model structure; may be sensitive to choice of external data and their relevance; computational complexity increases with model flexibility; assumes correct specification of spline and model components.
- Also known as
- survextrap R package
Questions this answers
- › How can long-term survival be estimated from short-term clinical trial data?
- › How can external survival data (e.g., from registries) be combined with trial data in survival analysis?
- › How can uncertainty in survival extrapolation be appropriately quantified?
- › How can non-proportional hazards or time-varying treatment effects be modeled?
- › How can cure fractions or waning treatment effects be incorporated into survival models?
- › How can transparent and flexible survival extrapolation be implemented in HTA?
References & sources
Related methods
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.