In a study on predictors, is it required to report both the adjusted and unadjusted odds ratio. How to interpret if there is a difference between the two.
Hi Abinash,
The odds ratio is simply the ratio of the odds in the two groups, for exampl, in two groups of treatment. We said that it is unadjusted because it does not take into account the influence, if any, of other factors, for example, age of patients, gender, ...
In some domains, it is important to take into account the effects of other factors, for example, demographics information (age, gender, carrer), living style (smoking, drinking), medical history (hypertension,...). There these information is included in a logistic model as additional covariate, beside the treatment effect.
If you think that the inclusion of the covariates is important, they should be their in your model. Normally, I would repor the adjusted odds ratio. The unadjusted odds ratio should be there for the comparison purpose.
Abinash,
I think the decision on reporting both the unadjusted and adjusted results will depend on the journal. In a medical or epidemiology journal reporting both may be expected, but for a social sciences journal just reporting the adjusted OR may be enough.
As far as interpreting the difference between the two, most statistical programs that have logistic regression will report a pseudo R-squared value (e.g., Nagelkerke or Cox and Snell). Like linear models, the larger R-squared will indicate a larger amount of variance accounted for by the model.
Mike
Dear Abinash Panda, I do not agree with the idea "the decision in reporting odds ratio depends on the interest of journals or researchers". Since the utlimate goal of researchers is to prepare self sufficient output, I recommend to report both crude (unadjusted) and adjusted odds ratio. This is because readers can easily know which variables were included in.simple and multivaribale log regression model
Dear Abinash,
I think it is better to report both adjusted and unadjusted as it gives an idea of odds of having.. before and after adjustment of confounders.
Misgan Legesse Liben I agree with you because I will be easy to readers to know which factor contributing what. and depends on researchers intention what he/she wants to report
Please tell me how to calculate unadjusted odds ratio in spss? Thanks in advance.
Md. Saiful Islam just leave covariates out of your model. Best, D. Booth
Hi Md. Saiful Islam
to get unadjusted odds ratio, simply go to logistic regression by including single independent variable (called univariate). At the end, from the output you will get unadjusted odds ratio and also get to know which variables are significantly affecting. After that, you can go for multivariable logistic regression to get adjusted odds ratio by including variables which have shown a significant effect in univariate analysis.
yes reporting boteh is good for reader and sometimes the publisher also put as a requirement.
I think this is an important question. I stop reporting it as I found little value in reporting the OR. The very plan of the manuscript was to finding out the AOR so why go the long route. I also do not conduct tests of associations when all my variables will be going into the regression model anyway. I am being asked by a reviewer to include the OR and I am struggling to address this as I see no value in it as per the aim of the study. It was stated albinitio that the finding was focused on the AOR. The OR ha no role in the study objective
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication