topic Re: Incidence rates between populations in Statistical Procedures

Incidence rates between populations

lauramb — Tue, 17 Oct 2017 21:51:40 GMT

Hi, I'm working with the following data involving incidence of a disease.

My variables are YEAR, COUNTRY and INCIDENCE RATE (calculated on the basis of the total population of the country that year.

So far I did a Cochran Armitage trend test to see if there is a significant trend over the years.

The other thing I want to check is if I have a difference between countries.

The thing is I don't have cases, just incidence rates. Basically I have the cases for the disease, but not for the not disease (if that makes any sense).

Anyone knows how to work with this kind of data? i.e. only one value (freq) per variable (country)?

Thanks!

Any input will be welcomed!

Re: Incidence rates between populations

StatDave — Thu, 19 Oct 2017 17:50:44 GMT

When you say you "have the cases for the disease, but not for the not disease," that suggests you have a count of the number of cases with the disease and the total population for each country-year combination. The ratio of those is actually a proportion, which you are calling a rate**. If that is true, then you can fit a logistic model and then make country comparisons. For example,

proc logistic; class country; model Ndis / Total = country year; run;

If you particularly need to measure a difference of proportions (rather than the more natural odds ratio with PROC LOGISTIC will automatically provide), then see this note.

**This assumes that each person can only either have or not have the disease. If a person can have it more than once, then a different method is needed.

Re: Incidence rates between populations

lauramb — Wed, 25 Oct 2017 02:26:45 GMT

Thanks for your help!