Predictor = frailty, outcome = cognitive impairment (Y or N)
I am using a frailty index to quantify frailty. This ranges from 0 to a theoretical maximum of 1. However, in reality a score of 1 is impossible, the upper limit is ~0.67.
Sample: n=200
Frailty index: mean =0.11, SD=0.07, range 0-0.51
Q: continuous predictor variable (frailty) has very large OR (~800) and wide CI (~9,71,000) in binary logistic regression. What is the best method to transform this variable?
I think part of this is because of the scale of the frailty index. A 1 unit increase in the frailty index is not possible in reality. When I transform the frailty index, for example log transformation, the odds ratio is much more "normal" and the association remains significant.
Is there a "best" method to transform the variable?
I am using SPSS to analyse my data and I am new to statistics (and researchgate).
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS