Biostatistics 515, Winter 2004



Biostatistics 515, Winter 2004

Homework 8 solutions/comments

2. WHAS data: there are total 481 subjects in the data.

grouped cohort alive=0 dead=1 total follow-up time(min, max)

1 58 102 160 (1, 5843)

2 82 93 175 (1, 3665)

3 92 54 146 (1, 1829)

a) The size of grouped cohort 1 is n=160. Fitted the Cox PH model is the following:

coxph(formula = Surv(lenfol, fstat) ~ age + sex + cpk + chf +

miord, data = whas, subset = (yrgrp == 1))

coef exp(coef) se(coef) z p

age 3.41e-02 1.03 0.009629 3.5410 0.0004

sex 1.26e-02 1.01 0.211653 0.0598 0.9500

cpk -2.38e-05 1.00 0.000174 -0.1370 0.8900

chf 3.80e-01 1.46 0.212086 1.7909 0.0730

miord 1.93e-01 1.21 0.203880 0.9463 0.3400

exp(coef) exp(-coef) lower .95 upper .95

age 1.03 0.966 1.015 1.05

sex 1.01 0.987 0.669 1.53

cpk 1.00 1.000 1.000 1.00

chf 1.46 0.684 0.965 2.22

miord 1.21 0.825 0.813 1.81

(b) Model 1: [pic]

H0: βi = 0 , for i = 1 … 5

H1: at least one of βi is not equal to 0

LR = -2logL(reduced model) –[-2logL(full model)] = 25.4 > 11.07= [pic]

So, we reject the null hypothesis at α = 0.05 significant level and conclude that the model is significant.

(c)There are five parameters in the above model, so we will perform five Wald tests to see if each covariate contribute to the model given other covariates are in the model.

---- Testing sex given age, cpk, chf and miord in the model:

H0: β2 = 0

H1: β2 ≠ 0

Z score is 0.0598 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download