1\name{svykm} 2\alias{svykm} 3\alias{plot.svykm} 4\alias{plot.svykmlist} 5\alias{lines.svykm} 6\alias{quantile.svykm} 7\alias{confint.svykm} 8%- Also NEED an '\alias' for EACH other topic documented here. 9\title{Estimate survival function. } 10\description{ 11Estimates the survival function using a weighted Kaplan-Meier 12estimator. 13} 14\usage{ 15svykm(formula, design,se=FALSE, ...) 16\method{plot}{svykm}(x,xlab="time",ylab="Proportion surviving", 17 ylim=c(0,1),ci=NULL,lty=1,...) 18\method{lines}{svykm}(x,xlab="time",type="s",ci=FALSE,lty=1,...) 19\method{plot}{svykmlist}(x, pars=NULL, ci=FALSE,...) 20\method{quantile}{svykm}(x, probs=c(0.75,0.5,0.25),ci=FALSE,level=0.95,...) 21\method{confint}{svykm}(object,parm,level=0.95,...) 22} 23%- maybe also 'usage' for other objects documented here. 24\arguments{ 25 \item{formula}{Two-sided formula. The response variable should be a right-censored 26 \code{Surv} object} 27 \item{design}{survey design object} 28 \item{se}{Compute standard errors? This is slow for moderate to large 29 data sets} 30 \item{\dots}{in \code{plot} and \code{lines} methods, graphical 31parameters } 32 \item{x}{a \code{svykm} or \code{svykmlist} object} 33 \item{xlab,ylab,ylim,type}{as for \code{plot}} 34 \item{lty}{Line type, see \code{\link{par}}} 35 \item{ci}{Plot (or return, for\code{quantile}) the confidence interval} 36 \item{pars}{A list of vectors of graphical parameters for the 37 separate curves in a \code{svykmlist} object} 38 \item{object}{A \code{svykm} object} 39 \item{parm}{vector of times to report confidence intervals} 40 \item{level}{confidence level} 41 \item{probs}{survival probabilities for computing survival quantiles 42 (note that these are the complement of the usual 43 \code{\link{quantile}} input, so 0.9 means 90\% surviving, not 90\% dead)} 44} 45\value{ 46 For \code{svykm}, an object of class \code{svykm} for a single curve or \code{svykmlist} 47 for multiple curves. 48} 49\details{ 50 When standard errors are computed, the survival curve is 51 actually the Aalen (hazard-based) estimator rather than the 52 Kaplan-Meier estimator. 53 54 The standard error computations use memory proportional to the sample 55 size times the square of the number of events. This can be a lot. 56 57 In the case of equal-probability cluster sampling without replacement 58 the computations are essentially the same as those of Williams (1995), 59 and the same linearization strategy is used for other designs. 60 61 Confidence intervals are computed on the log(survival) scale, 62 following the default in \code{survival} package, which was based on 63 simulations by Link(1984). 64 65 Confidence intervals for quantiles use Woodruff's method: the interval 66 is the intersection of the horizontal line at the specified quantile 67 with the pointwise confidence band around the survival curve. 68} 69\references{ 70Link, C. L. (1984). Confidence intervals for the survival function using 71Cox's proportional hazards model with covariates. Biometrics 40, 72601-610. 73 74Williams RL (1995) "Product-Limit Survival Functions with Correlated 75Survival Times" Lifetime Data Analysis 1: 171--186 76 77Woodruff RS (1952) Confidence intervals for medians and other 78position measures. JASA 57, 622-627. 79 } 80 \seealso{\code{\link{predict.svycoxph}} for survival curves from a Cox model 81 } 82\examples{ 83data(pbc, package="survival") 84pbc$randomized <- with(pbc, !is.na(trt) & trt>0) 85biasmodel<-glm(randomized~age*edema,data=pbc) 86pbc$randprob<-fitted(biasmodel) 87 88dpbc<-svydesign(id=~1, prob=~randprob, strata=~edema, data=subset(pbc,randomized)) 89 90s1<-svykm(Surv(time,status>0)~1, design=dpbc) 91s2<-svykm(Surv(time,status>0)~I(bili>6), design=dpbc) 92 93plot(s1) 94plot(s2) 95plot(s2, lwd=2, pars=list(lty=c(1,2),col=c("purple","forestgreen"))) 96 97quantile(s1, probs=c(0.9,0.75,0.5,0.25,0.1)) 98 99s3<-svykm(Surv(time,status>0)~I(bili>6), design=dpbc,se=TRUE) 100plot(s3[[2]],col="purple") 101 102confint(s3[[2]], parm=365*(1:5)) 103quantile(s3[[1]], ci=TRUE) 104 105} 106% Add one or more standard keywords, see file 'KEYWORDS' in the 107% R documentation directory. 108\keyword{survey} 109\keyword{survival}% __ONLY ONE__ keyword per line 110