Mediation analysis on multiple imputed data sets: How to get standard errors from the "medeff" command.

Hi everyone. I want to conduct mediation analysis (using "medeff") with continuous treatment, mediator and outcome variables on 10 imputed data sets but "medeff" is not supported by the command "mi: estimate". Alternatively, I can run "medeff" on each imputed data set and then applies Rubin´s formulas to calculate both point estimates and standard errors; however, "medeff" only saves point estimates and confidence intervals which are asymmetrical for all kind of effects. Is there any way to access to the variance-covariance matrix generated by "medeff" or to derive standard errors from the confidence intervals provided by the command?

Thanks in advance.