find enough no. of studies that reported medical/neurological comorbidities. Sixth, due to lack of clarity on their use of standardised assessments, several large samples (e.g., registry based datasets), were not represented. Seventh, we could include only clinic-based studies for meta-analysis, given the conceptual differences from community-based studies of which there were only a few with substantial numbers of individuals with OCD. The results of the meta-analysis are generalizable, therefore, to treatment-seeking clinical populations. However, clinic-based studies may suffer from Berkson's bias (104), i.e., clinical samples of OCD are more likely to be comorbid and therefore may show higher comorbidities compared to general population or community settings. Eighth, given the aims and nature of epidemiologic research, our inclusion Criteria for a sample size of ≥100 individuals with OCD would have resulted in the exclusion of a large number of community-based studies, especially those that assessed children and adolescents. Finally, while outside of the scope of the current paper, it is notable that several other factors may influence comorbidity rates, including study design, year of publication, country of origin, and socio-cultural differences.
