topic understanding Incidence rate_proc genmode in Statistical Procedures

understanding Incidence rate_proc genmode

GreenTree1 — Mon, 20 Jan 2020 20:29:39 GMT

Hi,

I have a data that looks as follows and I want to compute incidence rate.

ID events person-days exposure

1 2 80 0

2 0 11 1

3 11 60 1

4 19 30 0

where events is the dependent variable and exposure is the disease status which is dependent and person days is the total person days contributed by each individual/id. I am using proc genmod for this and want to understand the code. I understand that I have to consider the distribution poisson/poisson-nb, zero inflated. following is the code.

proc genmode data= mydata;

class exposure;

model event= exposure/offset=logpersondays dist=poisson

link=log type3;

run;

Can someone please comment if I have the variables in correct order? Also will this code also give me the crude incidence rates?

Will appreciate any suggestions.

I am using proc genmod and want to understand the code

Re: understanding Incidence rate_proc genmode

StatDave — Wed, 22 Jan 2020 16:14:24 GMT

The code looks fine. GENMOD is a general modeling procedure, so it doesn't provide specialized summary statistics like the crude rates. But you can easily get those with simple calculations:

proc means nway data=a;
class exposure;
var events persondays;
output out=out sum=sumevents sumdays;
run;
data crude;
set out;
obsrate=sumevents/sumdays;
run;
proc print;
run;

Re: understanding Incidence rate_proc genmode

GreenTree1 — Wed, 22 Jan 2020 16:45:42 GMT

Awesome, I ran this and got the crude rates that I had calculated using a long convoluted code, being a new SAS user :). This one is so much more efficient.

I am now interested in knowing if my results are significant. I think we are already discussing this in another post started by me that you have answered, and I will keep an eye on that one.

For those who are following this, here is the link to the other post

https://communities.sas.com/t5/Statistical-Procedures/how-to-compute-Incidence-rate-and-incidence-rate-ratios-with/m-p/619162#M29810