Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: pairwise comparison of a Survival data

 From "T. Alp Ozkan" To statalist@hsphsun2.harvard.edu Subject st: pairwise comparison of a Survival data Date Tue, 17 Apr 2012 13:43:16 +0300

```Dear listers,

I'm working on a data of renal carcinoma. In 2010 TNM cancer staging
classification have been changed. According to this I want to validate
new TNM classification. In a part of this analyse I need a pairwise
survival comparisons of consecutive T stages (like T1avsT1b or T1b vs
T2a)

First I used  stcox for TNM 2010 and test each other.

. stcox i.pT2010

failure _d:  CSS
analysis time _t:  followuptimemonths

No. of subjects =         1889                     Number of obs   =      1889
No. of failures =          151
Time at risk    =  63708.43333
LR chi2(6)      =    144.47
Log likelihood  =   -949.97419                     Prob > chi2     =    0.0000

------------------------------------------------------------------------------
_t | Haz. Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
pT2010 |
11  |   2.879476   1.326301     2.30   0.022      1.16747     7.10201
21  |   5.796228   2.753477     3.70   0.000     2.284469    14.70638
22  |   10.66265   5.002194     5.04   0.000     4.251472    26.74184
30  |   14.05566   6.091828     6.10   0.000     6.010826    32.86763
31  |   48.52258   24.10974     7.81   0.000     18.32321     128.495
40  |   21.73121   9.848165     6.79   0.000     8.939966    52.82409
------------------------------------------------------------------------------

. test 11.pT2010 21.pT2010

( 1)  11.pT2010 = 0
( 2)  21.pT2010 = 0

chi2(  2) =   14.33
Prob > chi2 =    0.0008

As I read more help documents I realized another stratification with
sts test variable, strata(subvariable) command.

Would you help me to understand which one is suitable for this purpose?

Thanks Alp

. sts test pT2010, strata (T1bvsT2a)

failure _d:  CSS
analysis time _t:  followuptimemonths

Stratified log-rank test for equality of survivor functions

|   Events         Events
pT2010 |  observed    expected(*)
-------+-------------------------
T1b    |        22          28.29
T2a    |        17          10.71
-------+-------------------------
Total  |        39          39.00

(*) sum over calculations within T1bvsT2a

chi2(1) =       5.16
Pr>chi2 =     0.0231

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```