flexsurv
flexsurv is a tool in R that helps researchers model how long it takes for health events to happen, like recovery or disease progression, using flexible statistical methods. It can predict outcomes and support decision-making in healthcare evaluations.
At a glance
Use when
Modeling time-to-event data with a need for extrapolation (e.g., in cost-effectiveness models), when standard non-parametric methods are insufficient, or when incorporating complex covariate effects or multi-state transitions
Avoid when
Non-parametric or semi-parametric methods (e.g., Kaplan-Meier, Cox models) are sufficient, or when there is insufficient data to support parametric assumptions
Inputs
Survival data in the form of 'Surv' objects (including time, event status, and optionally truncation), covariates, user- or built-in parametric distribution specifications (PDF, hazard, CDF)
Outputs
Fitted parametric survival models, parameter estimates with confidence intervals, predicted survival curves, transition probabilities in multi-state models, extrapolated time-to-event outcomes
How it works
flexsurv is an R package for fully-parametric survival analysis, supporting user-defined or built-in time-to-event distributions (e.g., generalized gamma, F, Royston-Parmar spline models). Parameters can be modeled as linear or log-linear functions of covariates. Model fitting via maximum likelihood using flexsurvreg, with syntax similar to survreg. Supports right-censoring, left-truncation via Surv objects. Provides inference and visualization for functions of parameters. Integrates with mstate for multi-state modeling.
- HTA domains
- Clinical Effectiveness, Costs & Economic Evaluation
- Assumptions
- Time-to-event data follow a specified parametric distribution; proportional or time-dependent hazards as modeled; correct specification of covariate effects; maximum likelihood assumptions (e.g., independence, correct censoring structure)
- Strengths
- High flexibility in choosing and defining survival distributions,Support for complex parametric models including spline-based and multi-state models,Seamless integration with R and other packages like mstate,User-friendly syntax modeled on standard survival package,Comprehensive inference and visualization of functions of parameters
- Limitations
- Requires correct specification of parametric distribution, which may be unknown,May be sensitive to model misspecification in extrapolation,Steeper learning curve for users unfamiliar with parametric survival modeling,Computational intensity increases with model complexity
- Also known as
- flexsurv package, flexsurv R package
Questions this answers
- › What is the expected time-to-event under different treatment scenarios?
- › How do covariates influence survival outcomes parametrically?
- › Which parametric survival distribution best fits the observed time-to-event data?
- › How can survival curves be extrapolated beyond observed follow-up times?
- › What are the uncertainty estimates (confidence intervals) for survival metrics?
- › How can multi-state disease progression be modeled parametrically?
References & sources
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.