Fit an exposure-response time-to-event model based on survreg()

Fit an exposure-response time-to-event model based on survreg()

Usage

ertte_model(formula, data, dist = "weibull", ...)

Arguments

formula: Model formula, with a survival::Surv() object as the response, e.g. Surv(time, event) ~ exposure.
data: Data set
dist: The AFT distribution to fit, as for survival::survreg(). Defaults to "weibull". Tested and officially supported for "exponential", "weibull", "lognormal", and "loglogistic" – see ertte_select_distribution() for choosing among them by AIC.
...: Other arguments passed to survival::survreg().

Value

A survreg object with an extra ertte_model class

Details

The returned object has class c("ertte_model", "survreg"): it is a survreg object, with a little extra metadata attached. This means all of the usual survreg methods work unchanged, without needing an ertte-specific equivalent – e.g. summary(), coef(), vcov(), confint(), predict(), AIC(), BIC(), logLik(), and anova() for comparing nested models. ertte_predict() is a separate, ertte-specific alternative to predict() that returns survival probabilities with confidence intervals in a tidy data frame; the two are complementary, not competing.

All four supported distributions are log-location-scale AFT models: log(T) = mu + scale * W, where mu is the linear predictor (intercept + covariates) and W follows a distribution that depends only on dist (extreme-value for "exponential"/"weibull", standard normal for "lognormal", standard logistic for "loglogistic") – see ertte_predict() and .ertte_dist_info().

Examples

mod <- ertte_model(survival::Surv(time, event) ~ aucss, ertte_data)
mod
#> Call:
#> survival::survreg(formula = formula, data = data, dist = dist)
#> 
#> Coefficients:
#>   (Intercept)         aucss 
#>  4.8563375087 -0.0006407913 
#> 
#> Scale= 0.7164032 
#> 
#> Loglik(model)= -1207.3   Loglik(intercept only)= -1263.9
#> 	Chisq= 113.25 on 1 degrees of freedom, p= <2e-16 
#> n= 300 

# other AFT distributions are also supported
mod_ln <- ertte_model(survival::Surv(time, event) ~ aucss, ertte_data, dist = "lognormal")
mod_ln
#> Call:
#> survival::survreg(formula = formula, data = data, dist = dist)
#> 
#> Coefficients:
#>   (Intercept)         aucss 
#>  4.4908037548 -0.0006546191 
#> 
#> Scale= 0.978441 
#> 
#> Loglik(model)= -1222.3   Loglik(intercept only)= -1271.4
#> 	Chisq= 98.21 on 1 degrees of freedom, p= <2e-16 
#> n= 300