Re: st: Interpretation of sts list

[Steven Joel Hirsch Samuels <sjhsamuels@earthlink.net>]

To add to Svend's explanation: Your calculation 17/48 assumes that  
all 48 people were at risk for the full study period--to 39 days.   
Not all were.  For example, 2 of 48 failed between 1 and 5 days, but  
only 42 were at risk at 5 days.  That means that 4 people  (48 - 2  
-42) had observation end during the period while still alive.  These  
are referred to as 'censored'.  So, you cannot use all 48 as the  
denominator for failure.    Call Duke's Dept of Biostatistics and  
Informatics, and I am sure someone will take you through the basic  
calculations..  A couple of online resources are:

http://socserv.mcmaster.ca/jfox/Courses/soc761/survival-analysis.pdf
http://www.statsoft.com/textbook/glosc.html#Censoring

The topic is covered, as Sven states, in almost every biostatistics  
book.


On Nov 14, 2007, at 4:16 PM, Christina M Gibson-Davis wrote:


> Hello,
> I am having difficulty interpreting the "survivor function" column  
> of sts list. I had thought it reported the percentage of cases that  
> had survived during a particular time interval, and thus it would  
> be possible to tell, by looking at the last entry, how many cases  
> had survived without failing. However, by using the cancer data  
> available through Stata (webuse cancer), I see that this cannot be.
>
> If I type:
> webuse cancer
> stset study died
> stdes
>
> I see that there are 48 subjects, and 31 failed, for a survival  
> rate of 35% (48-31=17; 17/48 = 35%).
>
> But if I type "sts list, at(0 5 to 39)", I end up with a survival  
> rate of 19% (please see below).
>
>
>              Beg.                      Survivor      Std.
>    Time     Total     Fail             Function     Error     [95%  
> Conf. Int.]
> ---------------------------------------------------------------------- 
> ---------
>       1        48        2              0.9583    0.0288     0.8435  
> 0.9894
>       5        42        6              0.8333    0.0538     0.6943  
> 0.9129
>       9        32        6              0.7061    0.0661     0.5546  
> 0.8143
>      13        24        6              0.5628    0.0742     0.4060  
> 0.6931
>      17        20        3              0.4870    0.0761     0.3326  
> 0.6249
>      21        16        0              0.4870    0.0761     0.3326  
> 0.6249
>      25        10        6              0.2922    0.0767     0.1543  
> 0.4449
>      29         8        1              0.2557    0.0753     0.1247  
> 0.4093
>      33         4        1              0.1918    0.0791     0.0676  
> 0.3634
>      37         2        0              0.1918    0.0791     0.0676  
> 0.3634
> ---------------------------------------------------------------------- 
> ---------
>
> I don't understand why the survivor function doesn't end up at 35%,  
> if overall, 35% of the cases survived.
>
> Thank you very much in advance for your assistance.
>
> Christina
>
> ___________________________________________
>
> Christina Gibson-Davis
> Assistant Professor of Public Policy
> Center for Child and Family Policy
> Terry Sanford Institute
> Duke University
> Box 90245
> Durham NC 27708
> cgibson@duke.edu
> (919) 613-7364
>
> The thing that impresses me the most about America is the way  
> parents obey
> their children.
> -- King Edward VIII
> __________________________________________
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
Steven  Samuels

sjhsamuels@earthlink.net
18 Cantine's Island
Saugerties, NY 12477
Phone: 845-246-0774
EFax: 208-498-7441




*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/