topic Statistical test for difference between two proportions in Statistical Procedures

Statistical test for difference between two proportions

iuri_leite — Wed, 11 May 2022 14:35:34 GMT

We observe are carrying out a longitudinal study event in different moments of time. But in one specific analysis we observe an event (identified by a medical exam) at enrollment (baseline) and 52 weeks later (week 52). We would like to see if there is difference between the proportions observed at baseline and week 52. It is important to emphasize that we will not observe the event for all the participants, either because the medical exam was not realized or because the participants did not have enough time to have the medical exam carried out in the week 52.

To evaluate the difference in proportions we used a logistic GEE model considering time (dummy variable indicating baseline and week 52) as a explanatory variable.

So, I would like to know if this is the best way to implement this analysis.

Thank you in advance.

Regards,

Iuri Leite

Re: Statistical test for difference between two proportions

SAS_Rob — Wed, 11 May 2022 14:38:41 GMT

It might be appropriate because of the dropouts to use a Weighted GEE model. These types of models can be fit in Proc GEE. Although it is a little more complex than your particular model, the example below should get you started.

SAS Help Center: Example 47.3 Weighted GEE for Longitudinal Data That Have Missing Values

Re: Statistical test for difference between two proportions

iuri_leite — Mon, 16 May 2022 12:49:59 GMT

Dear Rob,

thanks a lot.

regards,

Iuri