Hello everyone and thanks in advance for helping me out with my problem (or taking the time to read my post), I am trying to calculate age-adjusted incidence rates (not ratios) for a medical procedure for each calendar year in my data set (2000-2012). See the code pasted below. For some reason when I run this, I get all zeros back for my incidence rates specified in the ESTIMATE statements. I tried exponentiating the beta coefficients myself and got the same problem (results *10^-11). Again, results included below. Now, I know I shouldn't do this, but when I add age agesq ageneg1 to the CLASS statement my estimates seem to make sense. Still - I don't trust them since I want age modeled continuously, not categorically. My data is in the form of aggregated count and person time (one observation for each year of age, for each calendar year). Please let me know if working in this aggregated format is the problem, because I can obtain more detailed data. Thanks for your time! Sorry if I'm missing something very obvious. I've been searching through the documentation for SAS procedures and nothing helped me to see the error of my ways. Here's the code I'm trying to run: PROC GENMOD DATA=work.CountData_AgeCyear CLASS CalYr; MODEL TotalCount = age agesq ageneg1 CalYr / DIST=poi LINK=log OFFSET=log_pt SCALE=deviance; ESTIMATE "IR: 2000" int 1 calyr 1 0 0 0 0 0 0 0 0 0 0 0 0; ESTIMATE "IR: 2001" int 1 calyr 0 1 0 0 0 0 0 0 0 0 0 0 0; ESTIMATE "IR: 2002" int 1 calyr 0 0 1 0 0 0 0 0 0 0 0 0 0; ESTIMATE "IR: 2003" int 1 calyr 0 0 0 1 0 0 0 0 0 0 0 0 0; ESTIMATE "IR: 2004" int 1 calyr 0 0 0 0 1 0 0 0 0 0 0 0 0; ESTIMATE "IR: 2005" int 1 calyr 0 0 0 0 0 1 0 0 0 0 0 0 0; ESTIMATE "IR: 2006" int 1 calyr 0 0 0 0 0 0 1 0 0 0 0 0 0; ESTIMATE "IR: 2007" int 1 calyr 0 0 0 0 0 0 0 1 0 0 0 0 0; ESTIMATE "IR: 2008" int 1 calyr 0 0 0 0 0 0 0 0 1 0 0 0 0; ESTIMATE "IR: 2009" int 1 calyr 0 0 0 0 0 0 0 0 0 1 0 0 0; ESTIMATE "IR: 2010" int 1 calyr 0 0 0 0 0 0 0 0 0 0 1 0 0; ESTIMATE "IR: 2011" int 1 calyr 0 0 0 0 0 0 0 0 0 0 0 1 0; ESTIMATE "IR: 2012" int 1 calyr 0 0 0 0 0 0 0 0 0 0 0 0 1; RUN; Here are the parameter estimates I get (all 0's are output from my ESTIMATE statements [not displayed]): Parameter DF Estimate Error Limits Chi-Square Pr > ChiSq Intercept 1 -24.3139 0.3298 -24.9603 -23.6675 5434.65 <.0001 AGE 1 0.3573 0.0068 0.3439 0.3707 2737.97 <.0001 agesq 1 -0.0018 0.0000 -0.0019 -0.0017 1798.56 <.0001 ageneg1 1 179.2731 4.6522 170.1550 188.3913 1484.96 <.0001 CalYr 2000 1 -0.0399 0.0567 -0.1511 0.0712 0.50 0.4813 CalYr 2001 1 -0.0313 0.0431 -0.1158 0.0533 0.52 0.4688 CalYr 2002 1 0.0528 0.0360 -0.0179 0.1235 2.15 0.1430 CalYr 2003 1 -0.0556 0.0316 -0.1174 0.0063 3.10 0.0783 CalYr 2004 1 0.0254 0.0282 -0.0299 0.0808 0.81 0.3678 CalYr 2005 1 0.1314 0.0265 0.0794 0.1833 24.53 <.0001 CalYr 2006 1 0.3067 0.0256 0.2565 0.3569 143.36 <.0001 CalYr 2007 1 0.3010 0.0254 0.2512 0.3509 140.00 <.0001 CalYr 2008 1 0.3772 0.0224 0.3333 0.4211 283.17 <.0001 CalYr 2009 1 0.3541 0.0224 0.3101 0.3980 249.30 <.0001 CalYr 2010 1 0.3451 0.0223 0.3013 0.3889 238.80 <.0001 CalYr 2011 1 0.2672 0.0219 0.2242 0.3101 148.81 <.0001 CalYr 2012 0 0.0000 0.0000 0.0000 0.0000 . . Scale 0 2.0939 0.0000 2.0939 2.0939 Some other output that may be useful: Class Level Information Class Levels Values CalYr 13 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF Deviance 1040 4559.5819 4.3842 Scaled Deviance 1040 1040.0000 1.0000 Pearson Chi-Square 1040 4674.5441 4.4948 Scaled Pearson X2 1040 1066.2219 1.0252 Log Likelihood 157709.7869 Full Log Likelihood -5447.7431 AIC (smaller is better) 10927.4862 AICC (smaller is better) 10928.0098 BIC (smaller is better) 11006.8821
... View more