survivalLandmarkSummary: Produce a summary of survival probabilities at selected time...

In RichardBirnie/survtools: A collection of useful functions for survival analysis

Description

Produce a summary of survival probabilities at selected time points from parametric survival models

Usage

survivalLandmarkSummary(survival.models, time.points = c(1, 3, 5, 10),
  time.unit = 365.25, dp = 3, ...)

Arguments

survival.models	A named list of survival models as returned by flexsurvreg. Note that names are essential. Even if models are just named 'model 1', 'model 2' etc
time.points	A vector of time points to produce summaries. Defaults are 1, 3, 5 and 10 years
time.unit	Preferred time units. This sets a multiplier to calculate landmark time points. Default is 365.25 to give time points in years. Survival at 5 years would calculated as 5 x time.unit = 5 x 365.25 = 1826.25 days. It is assumed that time in the survival model was in days. For months set this to 30.4. For days set this to 1. Only days, months or years are recognised
dp	Number of decimal places in the estimates. Default = 3
...	Additional arguments to be passed to summary.flexsurvreg NOT TESTED

Details

Note that this has only been tested for simple models with a single covariate, e.g treatment

Value

A data frame with survival probabilities (including 95

Examples

library(flexsurv)
library(survtools)

# Load data ---------------------------------------------------------------
#load some example data from the flexsurv package
#Survival times of 686 patients with primary node positive breast cancer
#censrec = censoring, 1 = dead, 0 = censored
#rectime = time of death or censoring in days
#recyrs = time of death or censoring in years
#group = prognostic group
data(bc)

# Fit parametric survival models -------------------------------------------
#fit 6 standard distributions
#this fits the distributions and collects the model fit objects in a list
#Simple model estimating survival for each prognostic group
psm.list <- list(
  exp = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'exp'),
  weibull = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'weibull'),
  gompertz = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'gompertz'),
  llog = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'llogis'),
  lnorm = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'lnorm'),
  gengamma = flexsurvreg(Surv(rectime, censrec) ~ group, data = bc, dist = 'gengamma')
)

#Get summaries an 1, 3, 5, 10 and 15 years. Note that time was modelled above in days
landmark.summ <- survivalLandmarkSummary(survival.models = psm.list, time.points = c(1, 3, 5, 10, 15),
                                         time.unit = 365.25, dp = 3)

#Simpler model estimating survival for the whole cohort
psm.single <- list(
  exp = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'exp'),
  weibull = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'weibull'),
  gompertz = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'gompertz'),
  llog = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'llogis'),
  lnorm = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'lnorm'),
  gengamma = flexsurvreg(Surv(rectime, censrec) ~ 1, data = bc, dist = 'gengamma')
)

#Get summaries an 1, 3, 5, 10 and 15 years. Note that time was modelled above in days
landmark.summ <- survivalLandmarkSummary(survival.models = psm.single, time.points = c(1, 3, 5, 10, 15),
                                         time.unit = 365.25, dp = 3)

RichardBirnie/survtools documentation built on Aug. 3, 2019, 3:20 a.m.