Stringent selection criteria substantially reduced the risk of bias. For the meta-analysis, we included only studies that reported on a clinical cohort of patients with OCD, who were not selected (included or excluded) on the basis of any comorbidities. A diagnosis of OCD and comorbid disorders was made using standardised assessment instruments. In our estimation of pooled prevalence rates, we have considered several covariates by moderator analyses, as discussed earlier. Sensitivity analyses were conducted with Baujat plots, influence plots and the leave-one-out method. A considerable impact on the pooled prevalence rate was found only for OCPD. The outlier study, in this case, was removed in the final pooled prevalence estimation. For all other comorbidities, outliers did not significantly change the pooled prevalence estimates.
