Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: count-time data


From   "Jose Ignacio Pijoan Zubizarreta" <jipijoan@hcru.osakidetza.net>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: count-time data
Date   Fri, 3 Sep 2004 07:37:46 +0200

Dear stata users

I apologize for this rather basic (and long) question but it is really
getting me into despair.

I want to perform very simple survival analysis from some data provided in
"life table format" (i.e. count data):

It is about fertility rates of two groups in an artificial insemination
program.

35 patients enter the first insemination cycle in group 0 (not having some
health problem) and 16 in group 1 (having it).

There are failures (embarazos) and censoring (not getting the "failure" and
exiting the program at some time after the entrance) in both groups.

After reading ST manual´s ctst and cttost commands I built the following
"life table style" dataset:

 list

     +---------------------------------------------+
     | ciclo   embara~s   censura   entran   grupo |
     |--------------------------------------------------|
  1. |     0          0         0       35       0 |
  2. |     0          0         0       16       1 |
  3. |     1          3         0        0       0 |
  4. |     1          0         0        0       1 |
  5. |     2          3         0        0       0 |
     |---------------------------------------------|
  6. |     2          1         1        0       1 |
  7. |     3          2         0        0       0 |
  8. |     3          1         1        0       1 |
  9. |     4          3         3        0       0 |
 10. |     4          0         1        0       1 |
     |---------------------------------------------|
 11. |     5          3         0        0       0 |
 12. |     5          0         0        0       1 |
 13. |     6          1         1        0       0 |
 14. |     6          0         1        0       1 |
     +---------------------------------------------+


Then I tried to declare this data as count-time stata data:

ctset ciclo embarazos censura entran, by( grupo)

No claim from stata, with the following log:



 dataset name:  E:\Trabajos\......\tablas4.dta
            time:  ciclo
        no. fail:  embarazos
        no. lost:  censura
       no. enter:  entran
              by:  grupo

Thereafter I proceed and issue the cttost command with the following
results:

 cttost
(data are now st)

     failure event:  embarazos != 0 & embarazos < .
obs. time interval:  (0, ciclo]
 exit on or before:  failure
            weight:  [fweight=w]

----------------------------------------------------------------------------
--
       14  total obs.
        0  exclusions
----------------------------------------------------------------------------
--
       14  physical obs. remaining, equal to
       25  weighted obs., representing
       17  failures in single record/single failure data
       86  total analysis time at risk, at risk from t =         0
                             earliest observed entry t =         0
                                  last observed exit t =         6



I cannot see that there are 51 patients at the entrance of the programme
(first cycle). Similarly if I issue:

stdes, w

I get:

         failure _d:  embarazos
   analysis time _t:  ciclo
             weight:  [fweight=w]

                                   |-------------- per
subject --------------|
                         weighted     weighted              weighted
Category                   total        mean         min     median
max
----------------------------------------------------------------------------
--
no. of subjects               25
no. of records                25           1           1          1
1

(first) entry time                         0           0          0
0
(final) exit time                       3.44           1          4
6

subjects with gap              0
time on gap if gap             0
time at risk                  86        3.44           1          4
6

failures                      17         .68           0          1
1
----------------------------------------------------------------------------
--




And after:

sts list , by(grupo)



The following real (but inaccurate) life tables turn up:

    failure _d:  embarazos
   analysis time _t:  ciclo
             weight:  [fweight=w]

           Beg.          Net            Survivor      Std.
  Time    Total   Fail   Lost           Function     Error     [95% Conf.
Int.]
----------------------------------------------------------------------------
---
grupo=0
     1       19      3      0             0.8421    0.0837     0.5865
0.9462
     2       16      3      0             0.6842    0.1066     0.4279
0.8439
     3       13      2      0             0.5789    0.1133     0.3321
0.7626
     4       11      3      3             0.4211    0.1133     0.2037
0.6249
     5        5      3      0             0.1684    0.1028     0.0306
0.4024
     6        2      1      1             0.0842    0.0787     0.0056
0.3070
grupo=1
     2        6      1      1             0.8333    0.1521     0.2731
0.9747
     3        4      1      1             0.6250    0.2135     0.1419
0.8931
     4        2      0      1             0.6250    0.2135     0.1419
0.8931
     6        1      0      1             0.6250    0.2135     0.1419
0.8931
----------------------------------------------------------------------------
---



All I can see is that stata uses only 25 cases instead of 51. A careful
reading of the examples in the manual  does not enable me to quite
understand the right way to proceed although I have tried it in several
different ways

Any clue will be very much welcomed

Thanks in advance

Jose I Pijoán
Unidad de Epidemiología Clínica
Hospital de Cruces
Tfno: 94 600 6452
Fax:   94 600 6451
email: jipijoan@hcru.osakidetza.net




*
*   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/



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