How to calculate doubly robust estimators following psmatch2

[Michael Andersonnnn]

Hi everyone, any help would be greatly appreciated here. I am attempted to calculate doubly robust estimators following psmatch2

I am specifically attempting to estimate the ATT of experiencing an adverse event in hospital on the readmission rate following discharge.
My code is as follows:

psmatch2 adverse_event alter sex i.charlindex i.imddecile i.q1_score, outcome(readmission)

I have seen a post which says this can be done by using _weights from psmatch2, but does not explain how to do this.

Does anyone have any ideas?

Thank you!

[Felix Bittmann]

Use kmatch. See here, for example slide 27

https://www.stata.com/meeting/uk17/slides/uk17_Jann.pdf

[Michael Andersonnnn]

Thanks so much for replying, could you possibly provide a bit more of a explanation how kmatch is implementing doubly robust estimators in this scenario? Looking at slide 27, it seems that kmatch is being used for Multivariate-distance matching?

[Felix Bittmann]

Well I think you need to define what you mean by "double robust". In my reading, any regression adjustment is double robust. So you could use a PSM Matching with RA in kmatch "out of the box". I think you can combine the RA with any type of matching algorithm in kmatch.