PROC CAUSALGRAPH and IPW

Epi_Stats — Sun, 19 May 2024 10:25:32 GMT

Hi everyone,

I’ve used PROC CAUSALGRAPH to identify a valid adjustment set of variables to include in my model (which will use inverse probability weighting).

In the SAS documentation, and specifically the example “Causal Model of the Effect of Persistent Perfluoroalkyl Substances on Breastfeeding Duration”, data on six variables are available: Age, BMI, Education, Employment, Parity, and PrevBF.

PROC CAUSALGRAPH recommends the following valid adjustment set of 4 variables: Education, Employment, Parity, and PrevBF.

Note, Age and BMI are not included in this adjustment set.

I understand the reasoning behind this, but my question is, if I plan to run inverse probability weighting, using PROC CAUSALTRT, these are the 4 variables I will be included in both PSMODEL and MODEL statements…. and in the output, I’ll get unweighted and weighted values which I will then include in a Table in my report.

Staying with the same example from the SAS documentation, if I do this, then I won’t have an unweighted and weighted value for age and BMI, so what do I do when reporting this in the table?

Any help/comments would be greatly appreciated,

Thank you,

topic PROC CAUSALGRAPH and IPW in Statistical Procedures

PROC CAUSALGRAPH and IPW