All statistical analyses were performed in Stata v.11. The Poisson option was used to calculate exact confidence intervals for prevalence estimates. Gender ratios and rates of co-occurring OCD and ADHD in subjects with TS and CT were compared with unaffected controls using χ2 statistics with Yates' adjustment for small sample sizes; exact confidence intervals were also calculated. Heterogeneity analyses between different TS/CT disease definitions were performed using Cochran's Q and I2 statistics. Because these TS/CT disease definitions are nested, heterogeneity was examined by comparing subjects in the more narrowly defined group (for example, TS Narrow) to the additional subjects which comprised the broader definition (e.g., TS Intermediate subjects excluding those who met criteria for TS Narrow). All statistical tests were two-tailed.
