05GEE_p3 - PubH8452 Longitudinal Data Analysis - Fall 2011...

Info iconThis preview shows pages 1–6. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: PubH8452 Longitudinal Data Analysis - Fall 2011 Marginal Model for Categorical Data: Case Studies Marginal Model for Categorical Data: Case Studies A 2 2 Crossover Trial A crossover trial of efficacy of two treatments on cerebrovascular deficiency. Sixty-seven subjects from one center was used in the analysis for illustration. Two treatment arms (A: active drug, B: placebo) in the trial. Thirty-four patients received the active drug (A) followed by placebo (B); another 33 patients were treated in the reverse order. Binary outcome, 1: normal electrocardiogram reading, 0: abnormal reading. 1 PubH8452 Longitudinal Data Analysis - Fall 2011 Marginal Model for Categorical Data: Case Studies Crossover design is one in which subjects are given a sequence of treatments with the objective of studying the difference between individual treatment. In crossover design, a subject can be considered as his/her own control to eliminate between subject variation, hence crossover design is more powerful than similar size parallel design. Period-by-treatment interactions may indicate carry over effect . A reasonable wash out period is needed. 2 PubH8452 Longitudinal Data Analysis - Fall 2011 Marginal Model for Categorical Data: Case Studies >xover <- read.table ("../data/xover1.data", col.names = c("id", "class", "y", "intercept", + "trt", "period", "xover", "BA")) > xover$trtA <- 1-xover$trt > xover$trtAP <- xover$trtA*xover$period > xoverw <- reshape (xover[,c("id", "y","period","BA")], + direction = "wide", v.names = "y", timevar = "period",idvar = "id") > xoverw$respat <- ifelse(xoverw$y.0==0,2,3) > xoverw$respat[(xoverw$y.0+xoverw$y.1)==2] <- 1 > xoverw$respat[(xoverw$y.0+xoverw$y.1)==0] <- 4 3 PubH8452 Longitudinal Data Analysis - Fall 2011 Marginal Model for Categorical Data: Case Studies > #Table 8.1 in DHLZ book > tab8.1 <- cbind(table(xoverw$BA,xoverw$respat),table(xoverw$BA), + table(xover$BA[xover$y==1],xover$period[xover$y==1])) > dimnames(tab8.1) <- list(c("AB","BA"),c("(1,1)","(0,1)","(1,0)","(0,0)", + "Total","Period1","Period2")) > tab8.1 (1,1) (0,1) (1,0) (0,0) Total Period1 Period2 AB 22 6 6 34 28 22 BA 18 4 2 9 33 20 22 What is the treatment effect if only period 1 data are considered? > #odds ratio comparing the chance of being normal for the active drug versus placebo + for the data at period 1 > (28*13)/(20*6) [1] 3.033333 > #standard error for the estimated log-odds ratio > sqrt(1/28+1/6+1/20+1/13) [1] 0.5738502 Any within-subject correlation? 4 PubH8452 Longitudinal Data Analysis - Fall 2011 Marginal Model for Categorical Data: Case Studies Fit a GEE marginal model for the 2 2 Crossover Trial. > library(gee) > summary (gee (y ~ trtA+period+trtAP, data = xover, cor = "exchangeable", + id = id, family = binomial, scale.fix = TRUE)) Beginning Cgee S-function, @(#) geeformula.q 4.13 98/01/27 running glm to get initial regression estimate (Intercept) trtA period trtAP 0.4307829 1.1096621 0.1753529-1.0226507 GEE:...
View Full Document

This note was uploaded on 11/21/2011 for the course PUBH 8452 taught by Professor Xianghualuo during the Fall '11 term at Minnesota.

Page1 / 37

05GEE_p3 - PubH8452 Longitudinal Data Analysis - Fall 2011...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online