Through SAS, I produced an unstructured working correlation matrix using proc genmod. However, the matrix in question contains three rows and columns all pertaining to the time interval (time 0, time 1, time 2). That said, how would I go about interpreting this matrix if the outcome of interest was participant health status, in addition to holding treatment and diagnosis constant? SAS outputs and code found below. Please if you could, provide your interpretation and how you have gotten to that conclusion as I'm a tad bit lost.

case: patient identifier

treat: 0 = standard, 1 = new drug.

diagnose: baseline depression severity , 0 = Mild, 1 = Severe

outcome: depression response variable , 1 = Normal, 0 = Abnormal

time: 0 = week 1, 1 = week 2 , 2 = week 4

SAS Code

proc genmod data = Tmp1.Depressed;

class case time;

model outcome = diagnose treat time treat*time / dist = bin link = logit;

repeated subject = case / type = UN corrw;

run;

Working Correlation Matrix Col1 Col2 Col3 Row1 1.0000 0.0732 -0.0274 Row2 0.0732 1.0000 -0.0563 Row3 -0.0274 -0.0563 1.0000 GEE Fit Criteria QIC 1174.7114 QICu 1174.7704 Parameter Intercept diagnose treat Analysis of GEE Parameter Estimates Empirical Standard Error Estimates Standard Estimate Error 95% Confidence Limits Z Pr>IZI -0.9769 0.1840 -1.3376 -0.6162 -5.31 <.0001 time treat working correlation matrix col1 col2 col3 row1 row2 row3 gee fit criteria qic qicu parameter intercept diagnose analysis of estimates empirical standard error estimate confidence limits z pr>IZI -0.9769 0.1840 -1.3376 -0.6162 -5.31 <.0001 time treat>