Rate Ratio of Prescription rates compares to a reference year/or previous year using proc genmod

[reemm225]

Hi,

I am new to programming in SAS. I have count data for prescriptions in each year of the study. Each row has the number of prescirpitons in that uear, the person-time, the rate and the ln(person time). I want to use proc genmod to calculate the rate ratio:

1) Between each year of the study compared to the first year of the study (1998) as a reference

2) Between each year of the study and the previous year (i.e. 1999 to 1998, 2000 to 1999 etc) 

I am having trouble to set my estimate statement in the code so I will get the rate ration for each year. 

This is my data structure: 

Yearrx	RX	PT	rate	lnPT
1998	100	135	0.740741	4.905275
1999	200	480	0.416667	6.173786
2000	250	950	0.263158	6.856462
2001	300	1100	0.272727	7.003065
2002	350	2100	0.166667	7.649693
2003	450	2500	0.18	7.824046

Hoe should my code look like?

	proc genmod data=h2only  ;
Class Yearrx (ref="1998");
model rx=Yearrx /dist=negbin  scale=deviance offset=ln;
estimate '1998' Yearrx 1 -1;
estimate '1999' Yearrx 1 -1;
estimate '2000' Yearrx 1 -2;
run;y 

[reemm225]

Is this code more correct?

proc genmod data=h2only ;
Class Yearrx (ref="1998");
model rx=Yearrx /dist=negbin scale=deviance offset=ln;
estimate '1999 vs. 1998' Yearrx 1 -1;
estimate '2000 vs. 1998' Yearrx 2 -2;
estimate '2001 vs. 1998' Yearrx 3 -3;
run;

[PaigeMiller]

Don't use ESTIMATE statement to compare levels. Use the LSMEANS command.

lsmeans yearrx / cl lines linestable;

You may not want all of those options, but that is where I would start.

[reemm225]

Thank you! The lsmeans gives a table of Analysis Of Maximum Likelihood Parameter Estimates, where I get the rate ratio for each year compared to 1998- that is what I am looking for...however is there a way to get the estimates (rate ratios) not in logarithmic number but the exponential? I can do it for the rates but not for the parameter estimates. Thank you!