I need one info, please correct me or answer which can help me regarding regression analysis
can we dichotomized ed5d index score? if yes, on which basis we can do it? and what is the criteria for it?
if not then if we have to apply binary logistic regression, what should be appropriate method? to obtain odd ratios
replies will be appreciated.
Hello, Aqeel.
I´d work with the median. If you calculate the median of the EQ-index, you will get “EQ-index ≥ median” and “EQ-index < median”. Then you can calculate the unadjusted odds ratios and perform a multivariable logistic regression directed to the adjusted OR.
Instead of the EQ-index, you can also dichotomize the results in the dimensions of EQ-5D: (a) 1-1-1-1-1 as the optimal, and (b) any other result, suboptimal.
This may lead also to know the unadjusted ORs and adjusted OR.
Both analysis are complementary.
Regards
- Badges
- Science topic
- Similar topics
- Correction