0
$\begingroup$

I would like to compare cumulative incidence/ event count across two different time periods. As an example, I am looking at patients presenting with stroke in a one month period in 2015 and again in 2017 (to eliminate the remote possibility of seasonal variation).

There is a small possibility that these patient groups may overlap, but assuming that they are completely independent (the events are independent, at the very least, even if the the patients may not be), is there a statistical test that can compare events occurring in these time periods directly?

I wondered about the chi square test (assuming that the events are entirely independent) if that was valid when different time periods are taken into account.

I am a relative novice at this and would be very grateful for any advice. I have tagged the question as SPSS solely because that is the only statistical package I am familiar with at present.

$\endgroup$
0
$\begingroup$

A Poisson regression model is commonly used for count data. SPSS will fit such models in the GENLIN procedure for either independent counts (Analyze>Generalized Linear Models>Generalized Linear Models) or repeated counts (Analyze>Generalized Linear Models>Generalized Estimating Equations), and the GENLINMIXED procedure will also handle both kinds (Analyze>Mixed Models>Generalized Linear).

For starters I would suggest the basic approach using GENLIN with Analyze>Generalized Linear Models>Generalized Linear Models, applied to data with a binary variable indicating the time period to use as a predictor variable and a variable with the count as the dependent variable. Poisson loglinear is one of the options for Type of Model on the initial tab.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.