The problem is that Stata computes confidence intervals using standard normal and chisquared distributions and it has no corrections for clustered standard error option. But, you can use alternate command. This link could be helpful: https://www.stata.com/statalist/archive/2013-01/msg00068.html