Cochrane recommends using the diagmeta() function to do a bivariate meta-analysis of multiple cutoffs, which also provides the optimal cutoff. However, when I apply this method, the results are not as expected or clinically appropriate. This may be due to the limited number of studies that apply multiple cutoffs in my study and the differences between the applied cutoffs.

If I have two thresholds that are frequently used by the studies, can I use the likelihood ratio test via the lrtest() function to compare the fit of the cutoffs with adding the cutoff as a covariate?

Thank you!