use proc surveylogistic to draw RCS

[qiqiqi]

I want to use proc surveylogistic to draw a restrictive cubic spline curve (Y-axis is the odds ratio  and x-axis is the exposure), but I don't know how to use the command.

At present, I can only draw a restrictive cubic spline curve with Y-axis as the possibility. Would like to Can someone tell me how to solve it? Thank you！

[SteveDenham]

Not sure if I understand the question correctly, but have you considered using the EFFECT statement to spline your X axis values?

SteveDenham

[qiqiqi]

Thank you very much for your answer!

I used the EFFECT statement. However, I don't know how to further export the OR and 95CI% for each exposure.

Here are the related commands I used:

proc surveylogistic data=analysis total=totals;

strata shengfen;

cluster site_ no;

weight weight_ quanzhong;

effect dmdurationsp=spline(dmduration/naturalcubic basis=tpf(noint) knotmethod=percentiles(5) details );

class DR2fen gender/param=ref;
model DR2fen (event="1")= gender dmdurationsp age /expb;
store DR2fenmodelsp;

output out=predicted_ values pred=pred lower=lower upper=upper;

run;

[SteveDenham]

The LSMEANS statement with the ODDSRATIO option should give what you want.

SteveDenham

[qiqiqi]

Thank you very much again! I have read the LSMEANS statement , but there are somethings that I still didn't quite understand how to use LSMEANS statement to get these parameters. If I want to use the model information, and draw the RCS graph with X-axis as the dmduration(the above variable) and y-axis as the corresponding OR and 95CI%(DR2fen, the above variable).