Dear R list
I wonder please if anyone has experience they can share of the JMdesign package
that performs sample sized for (surviival/longitudinal) joint models and is
based on the paper "Sample size and power determination in joint modeling of
longitudinal and survival data " by Chen et al (Statistics in Medicine 2011) .
If so I do have a few queries please.
1. the vignette states that JMdesign can work when �the variance-covariance
matrix Sigma_theta is unknown.� This is the the covariance matrix of the
intercept, linear and quadratic effects of the longitudinal profiles of the
longitudinal (time-dependent) variable thats affecting event time.
I wonder if that refers to the input SigmaTheta to the R function powerLongSurv
in the package ? Because it looks to me as though most of the examples in the
vignette do supply a value for that matrix and it errors if its missing. I
realise section 3.2 of the Liddy Chen paper does refer to the unknown
covariance matrix situation
2. I�m also a little confused as to what is meant in the JMdesign vignette
by �Example 1� and "formula 4.6". Do these refer to the Chen paper ? I can�t
see a formula denoted 4.6.
3. The original Chen paper refers on page 3 to two possible objectives: 1)
power for testing the effect of the longitudinal trajectory and 2) power for
testing the effect of a fixed covariate, eg treatment, that affects the
trajectory (and therefore indirectly survival) but also has a direct effect on
survival.
Does the power calculation in the current JMdesign package only provide power
of the first test ? I think it does but from the paper both calculations are
possible.
4. Can anyone help with how to estimate the covariance matrix of the
intercept, linear and quadratic effects for the Sigma_theta input. I thought
theVarCorr output from a lmer mixed model analysis of the profiles might
provide the elements for that matrix
my apologies for so many questions - many thanks for any thoughts on any of them
best wishes
Bernard
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
