[R] Parametric survival models with left truncated, right censored data

Göran Broström goran.brostrom at gmail.com
Sun Jul 20 13:04:50 CEST 2008


I uploaded eha-1.2-0 to CRAN a day ago. This version can handle
accelerated failure time and proportional hazards parametric models
for right censored, left truncated, and stratified data. Included
distributions are: Weibull, lognormal, loglogistic, Gompertz, and
extreme value. The new inner structure of eha makes it very simple
(for me) to add families of distributions of a shape-scale type (i.e.
the distribution of the log of the "baseline" survival time is a
location-scale family). Suggestions are welcome. The Gompertz and
Gompertz-Makeham families of distributions are unfortunately not of
this type. While the Gompertz case could be solved by a "trick"
(always include an intercept in the regression model), the
Gompertz-Makeham case is harder. Needs special treatment; hints are
welcome.

With the possibility of keeping the shape parameter(s) fixed, special
cases of the abovementioned distributions are also possible to fit,
e.g., the exponential (Weibull with shape = 1).

Göran

On Wed, Jan 23, 2008 at 5:46 PM, Beth Ireland <Beth.Ireland at hymans.co.uk> wrote:
> Dear All,
>
> I would like to fit some parametric survival models using left
> truncated, right censored data in R. However I am having problems
> finding a function to fit parametric survival models which can handle
> left truncated data.
>
> I have tested both the survreg function in package survival:
>
> fit1 <- survreg(Surv(start, stop, status) ~ X + Y + Z, data=data1)
>
> and the psm function in package Design:
>
> fit2 <- psm(Surv(start, stop, status) ~ X + Y + Z, data=data1)
>
>
> But neither function appears to work with left truncated data. The error
> message "Invalid survival type" is received for both functions when left
> truncated data is specified.
>
> The function weibreg in the eha package fits Weibull survival models and
> works with left truncated data. This function is useful, but I need to
> fit parametric models other than Weibull.
>
> Any suggestions of functions that can fit parametric survival models
> other than Weibull on left truncated, right censored data would be
> greatly appreciated.
>
> Kind Regards,
>
> Beth
> --------------------
> Hymans Robertson LLP is a limited liability partnership registered in
> England and Wales with registered number OC310282. A list of members
> of Hymans Robertson LLP is available for inspection at One London
> Wall, London, EC2Y 5EA, the firm's registered office. Hymans
> Robertson LLP is authorised and regulated by the Financial Services
> Authority. Certain regulated services are provided by Hymans
> Robertson Financial Services LLP, a subsidiary of Hymans Robertson
> LLP. Hymans Robertson Financial Services LLP is a limited liability
> partnership registered in England and Wales with registered number
> OC310836. A list of members of Hymans Robertson Financial Services
> LLP is available for inspection at One London Wall, London, EC2Y 5EA,
> the firm's registered office.  Hymans Robertson Financial Services
> LLP is authorised and regulated by the Financial Services Authority.
> This e-mail and any attachments are confidential. If it ...{{dropped:9}}
>
> ______________________________________________
> R-help at r-project.org mailing list
> 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.
>



-- 
Göran Broström


More information about the R-help mailing list