Results have changed somewhat with 850 answers. Now correlation for the whole population is 0.34 and for only those that got a non-zero score it is 0.38. Seems like the correlation is similar regardless of selection criteria after all.
And yes, the DCI is not really a very good test as is claimed by the authors, but then I don't think Tourette is a very well defined condition either. Especially the fact that stims and tics are related is quite problematic.