Wald p-values and CIs for stcox

[Sabina Murphy]

Can Stata produce P-values and confidence intervals for the hazard ratio based on the Wald statistic from a Cox model rather than likelihood ratio? This can be done in SAS using 'Risk Limits' but I cannot locate a parallel command in Stata. Thanks in advance.

[Clyde Schechter]

The -test- command produces Wald statistics after -stcox- (and any other Stata estimation command).

[Sabina Murphy]

Thanks. That gives the p-value but not CIs. Any other suggestions?

[Clyde Schechter]

-lincom-

[Marcos Almeida]

With stcox, try the option vce(robust)

[Marcos Almeida]

Below, using the dataset "drugtr" (available in Stata), an example of what happens with each command.

First, the LR "ominbus" test with the default command:

. stcox drug age

         failure _d:  died
   analysis time _t:  studytime

Iteration 0:   log likelihood = -99.911448
Iteration 1:   log likelihood = -83.551879
Iteration 2:   log likelihood = -83.324009
Iteration 3:   log likelihood = -83.323546
Refining estimates:
Iteration 0:   log likelihood = -83.323546

Cox regression -- Breslow method for ties

No. of subjects =           48                     Number of obs   =        48
No. of failures =           31
Time at risk    =          744
                                                   LR chi2(2)      =     33.18
Log likelihood  =   -83.323546                     Prob > chi2     =    0.0000

------------------------------------------------------------------------------
          _t | Haz. Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        drug |   .1048772   .0477017    -4.96   0.000     .0430057    .2557622
         age |   1.120325   .0417711     3.05   0.002     1.041375     1.20526
------------------------------------------------------------------------------

Then, with "vce(robust)", and now having the Wald "omnibus" test:

. stcox drug age, vce(robust)

         failure _d:  died
   analysis time _t:  studytime

Iteration 0:   log pseudolikelihood = -99.911448
Iteration 1:   log pseudolikelihood = -83.551879
Iteration 2:   log pseudolikelihood = -83.324009
Iteration 3:   log pseudolikelihood = -83.323546
Refining estimates:
Iteration 0:   log pseudolikelihood = -83.323546

Cox regression -- Breslow method for ties

No. of subjects      =           48                Number of obs   =        48
No. of failures      =           31
Time at risk         =          744
                                                   Wald chi2(2)    =     34.46
Log pseudolikelihood =   -83.323546                Prob > chi2     =    0.0000

------------------------------------------------------------------------------
             |               Robust
          _t | Haz. Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        drug |   .1048772   .0434523    -5.44   0.000     .0465601     .236237
         age |   1.120325   .0366286     3.48   0.001     1.050786    1.194465
------------------------------------------------------------------------------

[Marcos Almeida]

A last note, but directly related to the subject: if I understood correctly, no matter the type of omnibus test (Wald or LR), the specific p-values for the predictors (here, drug and age) come from the Wald statistic.