Date:         Thu, 4 Feb 2010 00:20:22 -0800
Reply-To:     sousimou <elidrissi.m@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         sousimou <elidrissi.m@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Re: proc lifetest and survival analysis on sas
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

Hello,
thank you for interesting to my problem.

in ordre to estimate vaccine efficacy on clinical trial (experimental
drug versus active controle), I will use the time to event using cox
model.
I have 1000 subjects which should be followed during 2 period (period
1 = 01Mars to 31 mars and period 2 = 01 june to 30june) in order to
collect un event (death)
time= duration between start date of a period and date of event if
arise within that period, the event=1
time= duration between start date of a period and end date of that
period if no event, the event=0
group= 1 for experimental grug and 0 for active controle.

for each each period (1 or 2), the following procedure will be used :
PROC PHREG DATA=table;
             MODEL time*event(0)=group  / RL ALPHA=0.05;
RUN;


for analysis on overall study period both period. the following model
will be used :

PROC PHREG DATA=table;
           MODEL (start, stop)*event(0)=group period / RL ALPHA=0.05;
RUN;

period=1 if period 2 and 0 if period 1
start = 0 days for start date of period 1 and 92 days for period 2
stop = 31 days if no event and number of day between start date and
day of event in case of death for period 1.
       = 122 days if no event and number of day between start date of
period 1 and day of event in case of death in period 2.


the cox model assume proportional hazards throughout the follow up
period and within each period in case of the period in the model.
tp check this assumption I will use Log-log Kaplan Meier survival
curve.

for each period (1 or 2) : I will generte the graphic using :
PROC LIFETEST DATA=table   PLOTS=(LLS) ;
                      TIME time**event(0);
                       STRATA group;
RUN;


but check the assumption for the second model, I don't know what will
be the methodology to generate this graph ?

Could you have any suggestion :


Thank you




On 3 fév, 19:42, rh...@UNMC.EDU (Robin R High) wrote:
> It might be of value if we knew more about the context of your problem and
> what objective you are seeking.
>
> Robin High
> UNMC
>
> From:
> sousimou <elidriss...@GMAIL.COM>
> To:
> SA...@LISTSERV.UGA.EDU
> Date:
> 02/03/2010 08:35 AM
> Subject:
> proc lifetest and survival analysis on sas
> Sent by:
> "SAS(r) Discussion" <SA...@LISTSERV.UGA.EDU>
>
> hello every body,
> may be someone have solution for my problem :
>
> time to event analysis we use usually this :
>
> PROC PHREG DATA=table;
>             MODEL time*censor(0)=group  / RL ALPHA=0.05;
> RUN;
> the corresponding kaplan meier graph, we use this procedure :
> PROC LIFETEST DATA=table   PLOTS=(SURVIVAL) ;
>                       TIME time**censor(0);
>                       STRATA group;
>                     RUN;
>
> question for  this model how to complete the proc lifetest procédure
> with start and stop
> PROC PHREG DATA=table;
>             MODEL (start, stop)*censor(0)=group  / RL ALPHA=0.05;
>           RUN;
>
> thanks you