simPH (version 0.7.2)

coxsimSpline: Simulated quantities of interest for penalised splines from coxph models.

Description

Simulates relative hazards, first differences, hazard ratios, and hazard rates for penalised splines from Cox Proportional Hazards models. These can be plotted with simGG.

Currently does not support simulating hazard rates form multiple stratified models.

Usage

coxsimSpline(obj, bspline, bdata, qi = "Relative Hazard",
    Xj = 1, Xl = 0, nsim = 1000, ci = 0.95, spin = FALSE)

Arguments

obj

a coxph fitted model object with a penalised spline.

bspline

a character string of the full pspline call used in obj. It should be exactly the same as how you entered it in coxph. You also need to ente

bdata

a numeric vector of splined variable's values.

qi

quantity of interest to simulate. Values can be "Relative Hazard", "First Difference", "Hazard Ratio", and "Hazard Rate". The default is qi = "Relative Hazard". Think carefully befor

Xj

numeric vector of values of X to simulate for.

Xl

numeric vector of values to compare Xj to. Note if qi = "Relative Hazard" or "Hazard Rate" only Xj is relevant.

nsim

the number of simulations to run per value of Xj. Default is nsim = 1000.

ci

the proportion of middle simulations to keep. The default is ci = 0.95, i.e. keep the middle 95 percent. If spin = TRUE then ci is the convidence level of the shortest probability interval. Any value from 0 t

spin

logical, whether or not to keep only the shortest proability interval rather than the middle simulations.

Value

• a simspline object

Details

coxsimSpline simulates quantities of interest from penalised splines using multivariate normal distributions.

References

Luke Keele, "Replication data for: Proportionally Difficult: Testing for Nonproportional Hazards In Cox Models", 2010, http://hdl.handle.net/1902.1/17068 V1 [Version]

King, Gary, Michael Tomz, and Jason Wittenberg. 2000. ''Making the Most of Statistical Analyses: Improving Interpretation and Presentation.'' American Journal of Political Science 44(2): 347-61.

Liu, Ying, Andrew Gelman, and Tian Zheng. 2013. ''Simulation-Efficient Shortest Probablility Intervals.'' Arvix. http://arxiv.org/pdf/1302.2142v1.pdf.

See Also

simGG, survival, strata, and coxph

Examples

## dontrun
# Load Carpenter (2002) data
# data("CarpenterFdaData")

# Load survival package
# library(survival)

# Run basic model
# From Keele (2010) replication data
# M1 <- coxph(Surv(acttime, censor) ~  prevgenx + lethal + deathrt1 +
#				acutediz + hosp01  + pspline(hospdisc, df = 4) +
#				pspline(hhosleng, df = 4) + mandiz01 + femdiz01 + peddiz01 +
#				orphdum + natreg + vandavg3 + wpnoavg3 +
#				pspline(condavg3, df = 4) + pspline(orderent, df = 4) +
#				pspline(stafcder, df = 4), data = CarpenterFdaData)

# Simulate Relative Hazards for orderent
# Sim1 <- coxsimSpline(M1, bspline = "pspline(stafcder, df = 4)",
#                    bdata = CarpenterFdaData$stafcder,
#                    qi = "Hazard Ratio",
#                    Xj = seq(1100, 1700, by = 10),
#                    Xl = seq(1099, 1699, by = 10), spin = TRUE)

## dontrun
# Simulate Hazard Rates for orderent
# Sim2 <- coxsimSpline(M1, bspline = "pspline(orderent, df = 4)",
#                       bdata = CarpenterFdaData$orderent,
#                       qi = "Hazard Rate",
#                       Xj = seq(2, 53, by = 3),
#                       nsim = 100)