Programming the statistical procedures from SAS

Proc PHREG, different HR

Reply
Super Contributor
Posts: 301

Proc PHREG, different HR


Hi Everyone,

I have got the next dataset, and I would like to understand the different HR using different models using proc PHREG.

data exposed;

input Days  Status Treatment Sex $ @@;

datalines;

   179   1   1   F   378   0   1   M 

   256   1   1   F   355   1   1   M

   262   1   1   M   319   1   1   M 

   256   1   1   F   256   1   1   M

   255   1   1   M   171   1   1   F 

   224   0   1   F   325   1   1   M

   225   1   1   F   325   1   1   M 

   287   1   1   M   217   1   1   F

   319   1   1   M   255   1   1   F 

   264   1   1   M   256   1   1   F

   237   0   0   F   291   1   0   M 

   156   1   0   F   323   1   0   M

   270   1   0   M   253   1   0   M 

   257   1   0   M   206   1   0   F

   242   1   0   M   206   1   0   F 

   157   1   0   F   237   1   0   M

   249   1   0   M   211   1   0   F 

   180   1   0   F   229   1   0   F

   226   1   0   F   234   1   0   F 

   268   0   0   M   209   1   0   F

;

run;

1) HR using treatment;

proc phreg data=exposed;

treatment (ref='0');

days*status(0)=treatment/RISKLIMITS TIES = EFRON; ;

run;

HR=0.449 CI (0.222,0.907) P=0.0257 (significant different)

If I am correct:

Then, it shows me the risk of death in treatment 1 is 0.449 than in treatment 0 from one time point (alive) to the next time point.

2) HR adding sex:

proc phreg data=exposed;

treatment (ref='0') sex (ref='M');

days*status(0)= treatment sex/RISKLIMITS TIES = EFRON; ;

run;

Treatment: HR= 0.315 CI (0.145,0.686) p=0.0036; (Significant different)

Sex: HR= 12.780 CI (4.451, 36.698) p< 0.001; (Significan different)

3) HR only sex:

proc phreg data=exposed;

sex (ref='M');

days*status(0)=sex/RISKLIMITS TIES = EFRON; ;


HR=9.304 CI (3.503, 24.714) P<0.001 (significant different)

Can Explain me the differences between the three different HR, and which model should I use?

thanks.

Ask a Question
Discussion stats
  • 0 replies
  • 164 views
  • 0 likes
  • 1 in conversation