Dear colleagues,
I am looking for a way to obtain absolute risk reduction, relative risks, relative risk reduction and NNT from logistic regression models based on the paper of Austin (2010 - see attached) using (preferably) a mix of Excel, SPSS and/or Mplus. I have problems applying the manuscripts content to my data. I guess I can calculate the ARR, RR, RRR and NNT with Excel; however, calculating the confidence intervals via bootstrap would be problematic. Any help with the application of the manuscript would be very welcome!
http://www.ncbi.nlm.nih.gov/pubmed/19230611
If you have SPSS available, then you can obtain these by using the Generalized Linear Model procedure. With a binary family, using different links will get you the following:
logit link odds ratio
log link rate ratio
identity link risk difference
and of course, the NNT is 1 / risk difference
Dear Adrian,
Thanks a lot for the info, I will check it later today and keep you posted!
Took me some time, but I figured it out step by step. Is there any way to check how SPSS calculates the adjusted RRs and RDs i.e. which approach SPSS uses?
Please contact Dr. Diego Rojas Gualdrón, CES University. E-mail: [email protected]
Be careful using SPSS for regression. It doesn't seem to let you know if the algorithm converged and I haven't seen it make much feedback on collinearity, even when it produces a result. Thus you can get a meaningless result.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS