I am conducting a diagnostic test accuracy on RevMan and I have different index tests for different outcomes. I am comparing each outcome for each test and then I would like to compare paired test data. However, I am struggling to understand how I can actually use subgroup analysis on RevMan.
I have Sensitivity and Specificity for each test but sometimes for different groups (e.g low-risk, high-risk). I have tried to use the covariates method but my understanding is that it is useful only to assess heterogeneity.
How can I actually enter the Sensitivity and specificity for each subgroup on each outcome?
RevMan doesnt do any statistical analysis. You can click on analysis properties however and if you have added covariates to your dataset you can arrange the Forest Plot by covariates and sort by other characteristics such as study quality. Similarly for the SROC plot click on sources of heterogeneity and then you can produce ROC plots by subgroup . I believe RevMan uses the Moses Littenberg model for producing the SROC curve so statistically its not considered very rigorous.
Thank you. I agree I need to do some meta-regression. So probably the best thing to do at the moment is to set a different covariate for each subgroup if I am right.