topic Re: Goodness of fit to poisson distribution for monthly rates in Statistical Procedures

Goodness of fit to poisson distribution for monthly rates

simkinm2 — Tue, 05 Sep 2017 17:49:49 GMT

I have monthly data for events for which I'd like to ensure a good fit for monthly rates to a poisson distribution. The data looks like this:

Month Number_of_events Patient_days Events rate per_1000_patient_days

Jan2016 125 500 250

Feb2016 150 523 286.80

etc.....

I am plotting the monthly rates on a control chart and want to ensure I am meeting the assumption that the underlying distribution is a poisson distribution, otherwise the control limits I've calculated are not valid. I believe I need to use proc genmod to model number_of_events and use the log of patient_days but want to make sure I am not missing anything. Any advice is greatly appreciated!

Re: Goodness of fit to poisson distribution for monthly rates

PGStats — Tue, 05 Sep 2017 20:41:19 GMT

What have you tried so far?

Re: Goodness of fit to poisson distribution for monthly rates

PeterClemmensen — Wed, 06 Sep 2017 06:35:59 GMT

There are plenty of articles on fitting Poisson distributions like

http://blogs.sas.com/content/iml/2012/04/04/fitting-a-poisson-distribution-to-data-in-sas.html

http://sasnrd.com/fit-discrete-distribution/

Re: Goodness of fit to poisson distribution for monthly rates

StatDave — Wed, 06 Sep 2017 14:40:03 GMT

See this note.

Re: Goodness of fit to poisson distribution for monthly rates

simkinm2 — Wed, 06 Sep 2017 16:13:02 GMT

I've tried something like this:

data a;

set a;

log_pt_days= log(patient_days);

run;

proc genmod data = a;

model number_of_events/ dist = poi offset = log_pt_days;

run;