Programming the statistical procedures from SAS

Proc Phreg command and Interpretation of the Results

Reply
Contributor
Posts: 23

Proc Phreg command and Interpretation of the Results

[ Edited ]

Hello,

 

I am trying to get the unadjusted Hazard Ratio with the 95% Confidence Interval with the cancer patient data by using health claim data, and the last day of the data set is 2015-06-30. My original data looks like the following:

 

Variables:

Patient ID=Patient Identification Number

Chemo_Type=Type of Chemotherapy (1=type A, 2=type B, 3=typeC)

Chemo_First_Date=First Date when Chemotherapy Started

Death = Death

 

Chemotherapy Type A (Chemo_type=1) is the Study Group, so I want to compare Chemotherapy Type A vs. Chemotherapy Type B

and Chemotherapy A vs. Chemotherapy C. (No comparison between B and C) 

 

------------------------------------------------------------------------------------------------------------------------------

Patient_ID    Chemo_Type   Chemo_First_Date       Death              Death Date

10012                      1                     2012-01-01              1                  2013-05-15

10013                      2                     2013-04-08              0                       .

10014                      3                     2010-11-15              0                       .    

10015                      2                     2011-12-05              1                  2014-01-02

10016                      1                     2012-07-04              0                        .

10017                      2                     2011-02-15              1                 2013-04-18

10018                      2                     2012-05-17              0                        .  

10019                      3                     2014-01-12              1                  2015-05-07

.

.

.

(cont'-about 2000 patients..)

------------------------------------------------------------------------------------------------------------------------------

 

To run cox proportional hazard model by using the proc phreg command in SAS, I have entered the observation finish date and

calculated the difference from the first date of chemotherapy.

 

 

 

--------------------------------------------------------------------------------------------------------------------------------------------------------------------

Patient_ID    Chemo_Type   Chemo_First_Date       Death      Observation_Finish_Date        Survival Weeks

10012                      1                     2012-01-01              1                    2015-06-30                           182.3

10013                      2                     2013-04-08              0                    2015-06-30                           116.1        

10014                      3                     2010-11-15              0                    2015-06-30                           241.1

10015                      2                     2011-12-05              1                    2015-06-30                           186.1

10016                      1                     2012-07-04              0                    2015-06-30                            155.9

10017                      2                     2011-02-15              1                    2015-06-30                             228

10018                      2                     2012-05-17              0                    2015-06-30                            162.7

10019                      3                     2014-01-12              1                     2015-06-30                           76.2 

.

.

.

(cont'-about 2000 patients..)

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------

 

 

PROC PHREG DATA=w.cancer; Class Chemo_Type (PARAM=REFERENCE REF="1"); MODEL FU_WEEK*DEATH(0) = Chemo_Type/RL; RUN;

 

 

I am getting the following results

-------------------------------------------------------------------------------------------------------------------------------

Parameter                 Pr>Chisq       Hazard Ratio       95% Hazard Ratio Confidence Limits

Chemo_type   2        <0.0001               0.190                          0.153       0.237

Chemo_type   3         <0.0001              0.161                          0.133       0.196

------------------------------------------------------------------------------------------------------------------------------

 

My questions is

 

#1. Did I make the data set correctly to run Cox Proportional Hazard Model? Should I only enter the Survival Time (difference between Chemo_First_Date and Observation Finish Date) for the patients who survived?  

 

#2. If I set up and ran the Cox Proportional Hazard Model correctly, does it mean that the crude Hazard Ratio for Chemo_Type 2 (0.190, 95% CI: 0.153-0.237) and Chemo_Type 3 (0.161, 95% CI: 0.133-0.196) are lower than Cemo_type 1 1.00 (reference)?  

(I did not adjust for age, sex, and other covariates yet)

 

I would really like to know if I set up the Cox Proportional Hazard Model correctly and the interpretation of the results.....

 

Any advice or help would be appreciated.

 

Thank you.

 

P.S sorry, I cannot upload the complete data set or commands due to the privacy issues...

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