Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Cox survival-patients who die at entry


From   Adam Olszewski <adam.olszewski@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Cox survival-patients who die at entry
Date   Tue, 23 Apr 2013 22:22:42 -0400

Thank you very much for this comment Steve! I actually did recognize
this issue while using stpm2, I forgot about that. Your explanation
explains it for me finally.
AO

On Tue, Apr 23, 2013 at 10:18 PM, Steve Samuels <sjsamuels@gmail.com> wrote:
>
> To expand on Adam's comment:
>
> Estimated coefficients from -stcox- can be affected by the exact value
> of the added fraction (e.g. 0.1  vs. 0.5) only if the predictors include
> interactions with time. In -stcox-, these are set by the tvc() and
> texp() options. In the absence of such interactions, times enter into
> the partial likelihood equation only through their rank order.
>
> On the other hand, the contributed (SSC) flexible regression commands -stpm-,
> -stpm2- and -stgenreg- are parametric models in time. The estimated
> coefficients from these commands can be affected by the exact value of
> the added fraction even if the predictors include no interactions with
> time.
>
>
> Steve
>
>
> On Apr 22, 2013, at 3:15 PM, Adam Olszewski wrote:
>
> If the patient has survival time zero, then the patient contributes
> zero (1 person * 0 days) to time at risk. Those patients will be
> excluded from the survival analysis at the outset, when you -stset-
> your data.
> I agree with the comment to add a fraction of the day to the data,
> although in some settings I found that the Cox regression coefficients
> can change depending on how much you add (0.1 vs. 0.5 etc.) and how
> many patients are affected.
> AO
>
> On Mon, Apr 22, 2013 at 1:37 PM, Steve Samuels <sjsamuels@gmail.com> wrote:
>> I assume you have dates in days:
>>
>> . replace sdate = sdate + 0.5 if (died on day of enrollment)
>> . stset stdate
>>
>> Use more exact fractions if you have them.
>>
>> In some circumstances, a lot of ties can distort the analysis. If you have frequent ties for  days of early deaths, then perhaps you should ascertain approximate times of death for all patients with early deaths, not just those who died on day 1.
>>
>> Steve
>> sjsamuels@gmail.com
>>
>>
>> On Apr 22, 2013, at 1:11 PM, Po-Lin Chan wrote:
>>
>> Hi Statalisters,
>>
>> I am running a Cox survival but have 5 patients who die on the day of
>> enrolment (ie entry).
>> Does Stata automatically account for the contribution of these 5 patients
>> for the total patient time?
>> Do I need to add additional time to their date of death in stata, and how
>> do I do this?
>>
>> Thank you
>>
>> Polin Chan
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index