Hi.
I wonder if anyone can help me with this. I'm trying to analyze incidence rates in a population. For instance patients with mechanical ventilation in different time periods. I have a database with the question yes/no mechanical ventilation and 4 years: 2010, 2011, 2012 and 2013
As an 'incidence' rate is not expressed as a Percent, and should be reported as cases per time period. I would like to know how to report incidence rates correctly with 95% CIs around all incidence rates and p-values.
Hello, This may help you.
Incidence rate = no. of disease onsets / Sum of person - time @ risk
You can perform Poisson (and Negative Binomial) regression via the GENLIN procedure in SPSS, and exponentiating the coefficients (eB) yields incidence rate ratios (and a baseline incidence rate for the constant). See the UCLA web-page linked below for examples. I just tried a quick example, and GENLIN also allows you to estimate an intercept-only model, in case that is something you want. HTH.
http://www.ats.ucla.edu/stat/spss/dae/poissonreg.htm
Hi Ignacio,
I usually calculate incidence rates using excell. The most correct and standarised way to show this kind of parameters is incidence density (as you said quite well it is not a percent) and for this you'll need, as the first comment said, number of events and time in risk for this event of every case. If you need futher information please send me mail or wahtsapp ;)
Hi Ignacio,
The incidence density is calculated as the first comment said.
IR= Number of new cases of disease or injury during/specified period Size of population at start of period*1000.
Negative binomial regression with SPSS produce IRR (incidence rate ratio). is there any difference between IR and IRR? Can anybody explain pls
Sounds very interesting while I have been engaged in people with mechanical ventilation. If you want to read it, I attach the paper:
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS