When the likelihood is maximized, all gradient values should be zero.  Since we're dealing with an iterative optimization algorithm, exact zero is obviously not possible, so you're looking for the gradients to be close to zero.  Of course, how close is "close" is anybody's guess.  GLIMMIX sets a bar at 0.001, though there's nothing magical about it.  So, as it says, the convergence in this is indeterminate since the biggest gradient value is bigger than that.  That's not to say that the solution presented is bad.  It just alerts you that you'll want to assess the resulting model. ... View more

One additional observation here.  See this note on computing relative risks.  You can use the log-binomial approach with augmented data as shown there to get an estimate and a confidence interval for the relative risk.     data a; do g=0,1; do i=1 to 10; y=1; f=10000*10/100010; output; y=0; f=1*10/100010; output; end; end; run; proc genmod; model y(event="1")=g / dist=bin link=log; estimate 'rr' g 1 ; run; ... View more