BookmarkSubscribeRSS Feed
mconover
Quartz | Level 8

I am trying to conduct a propensity score analysis using inverse probability of treatment weighting (IPTW) using PROC PHREG.  I am modeling "P_OUTCOME"against the explanatory variable "INDEX_HS".  I have already coded my standardized weights as "sw", and trimmed the dataset for extreme weights.

My tables have an observation for each person as follows:

Obs    INDEX_HS    P_OUTCOME       ps       id        ExpPrev        sw


I have included my PROC PHREG coding below along with the output I get.  The program returns ratio measures (and confidence intervals) that are closely centered around one.  I also noticed in the output that the total is equal to the number of events, which I expected to roughly simulate what I see in my raw dataset.  I also have a very high value for the log likelihood.


What am I missing?  What might cause this issue?  For all I know this is running correctly and I'm just misinterpreting the output but if someone could help me I'd really appreciate i

  proc phreg data = iptw_trimmed covs(aggregate);  ** Robust variance **;

   id id;

   model P_OUTCOME = INDEX_HS / rl;

   freq sw / notruncate; run;

                                       The PHREG Procedure

                                       Model Information

                           Data Set                           WORK.IPTW_TRIMMED

                           Dependent Variable            P_OUTCOME

                           Frequency Variable            sw

                           Ties Handling                    BRESLOW

                             Number of Observations Read       65634

                             Number of Observations Used       65634

                             Sum of Frequencies Read        63536.25

                             Sum of Frequencies Used        63536.25

                       Summary of the Number of Event and Censored Values

                                                               Percent

                             Total       Event    Censored    Censored

                          63536.3     63536.3           0        0.00

                                       Convergence Status

                         Convergence criterion (GCONV=1E-8) satisfied.

                                      Model Fit Statistics

                                              Without           With

                             Criterion     Covariates     Covariates

                             -2 LOG L       1403039.2      1403039.2

                             AIC               1403039.2      1403041.2

                             SBC              1403039.2      1403050.2

                             Testing Global Null Hypothesis: BETA=0

                   Test                    Chi-Square       DF     Pr > ChiSq

                   Likelihood Ratio            0.0000        1         1.0000

                   Score (Model-Based)         0.0000        1         0.9959

                   Score (Sandwich)            0.0070        1         0.9332

                   Wald (Model-Based)          0.0000        1         1.0000

                   Wald (Sandwich)             0.0000        1         1.0000

2 REPLIES 2
mconover
Quartz | Level 8

Is it possible that I'm modeling the risk of NOT having the outcome?  Would that explain these results?  My outcome is extremely rare. 

SteveDenham
Jade | Level 19

That could be.  I wonder if you could use one of the other procedures, such as surveyfreq, for this analysis.  It appears that there is no censoring, so an analysis that requires all of the data would be equivalent.  At least those would allow you to specify the outcome one way or the other.

Steve Denham

sas-innovate-2024.png

Join us for SAS Innovate April 16-19 at the Aria in Las Vegas. Bring the team and save big with our group pricing for a limited time only.

Pre-conference courses and tutorials are filling up fast and are always a sellout. Register today to reserve your seat.

 

Register now!

What is Bayesian Analysis?

Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.

Find more tutorials on the SAS Users YouTube channel.

Click image to register for webinarClick image to register for webinar

Classroom Training Available!

Select SAS Training centers are offering in-person courses. View upcoming courses for:

View all other training opportunities.

Discussion stats
  • 2 replies
  • 2826 views
  • 0 likes
  • 2 in conversation