topic Account for time-varying covariates in conditional logistic regression analysis in SAS in Statistical Procedures

Account for time-varying covariates in conditional logistic regression analysis in SAS

zinojeng — Thu, 05 Dec 2019 14:25:06 GMT

I meet a problem by doing a nest-case control study. Since both cases and controls are from a cohort and have exposures information in time-dependent manner, just wonder how to account for these time-dependent covariates in the conditional logistic regression analysis.

Is the way in conditional regression similar as COX regression?

PROC PHREG DATA = count;
MODEL (start, stop)*censor(0) = sex marital age weight height;
STRATA id;
RUN;

Thank you.

Sincerely.
Yelson, Tseng, MD

Re: Account for time-varying covariates in conditional logistic regression analysis in SAS

StatDave — Thu, 05 Dec 2019 20:03:08 GMT

You can also fit the conditional model by using a STRATA statement in PROC LOGISTIC. See the example in the LOGISTIC documentation. The predictors do not have to be fixed within a stratum, so you can simply include the time-varying predictors along with the others in your model.

Re: Account for time-varying covariates in conditional logistic regression analysis in SAS

Ksharp — Fri, 06 Dec 2019 12:36:59 GMT

I think you should use PROC GLIMMIX , put ID and TIME variables into random / repeated statement.