I have a data set of about patient contacts with a provider. There are about 6000 contacts across 200 patients. The data of interest in this analysis are binary e.g. I spoke directly to the provider or not (1, 0). Patients are in one of two groups, EXP vs Control. They call in daily, and provide multiple records over two cycles of treatment. Thus patient records are nested within patient and cycle. In addition, the patients are nested within provider - they are seen by one of 11 providers and providers see both EXP and Control pts but are blinded to condition.

I'd like to test for group differences in the proportion of calls in which a patient talks directly to the provider but I am not sure how to model the analyses to indicate the three within patient variables pt_id cycle_num and provider_id and the one between subjects variable Tx_group. I could simply analyze the percent of calls in which the pts spoke to the provider and use and ANOVA (as in proc mixed) but I wonder if there might be a better way to handle the data using proc Genmod.

Thanks for any help you might be able to provide.

Bill

**********************************************************************

William N. Dudley, PhD Emma Eccels Jones Nursing Research Center

University of Utah

College of Nursing

10 South 2000 East

Salt Lake City, UT 84122-5880

http://www.nurs.utah.edu/faculty/william_dudley.htm

**********************************************************************