OBJECTIVES: As the cultural adaptation process of PROs (Patient Reported Outcome instruments), every version of PROs should be validated. However, little is known whether those validated PROs are identical in terms of the test performance (operating characteristics). In this study, we investigated PHQ-9 (Patient Health Questionnaire-9 for depression) using the meta-analysis. METHODS: We searched PubMed for studies examining diagnostic accuracy of PHQ-9.  Then we extracted or retrieved 2 x 2 data from each study and meta-analysis was performed using Bivariate model (Reitsma et al. 2005). As the language versions of PHQ-9 and regions where the studies were conducted would be potential confounding factors, we performed meta-regression. Bivariate meta-analysis of the subgroup (region, language) was explored. RESULTS: 28 studies met our inclusion criteria. Univariate meta-regression showed that all the covariates assessed such as language, region significantly contributed to the heterogeneity of sensitivity (I^2=82.7) and specificity (I^2=92.5) among the studies. Meta-analysis of subgroup of the regions and languages was made using bivariate model; studies conducted in North America (6 studies), Europe (13 studies), or Asia & others (9 studies); LogitSe of North America, Europe and Asia& other were 1.66, 1.45, and 1.48, respectively. Logit Sp of these regions was 1.87, 1.90 and 2.22, respectively. CONCLUSIONS: Results suggest that operational characteristics would influenced by the regions, language version of PHQ-9. Therefore, when considering a multiregional study using PHQ-9 as screening tool, there would be a chance of some imbalance among regions. Therefore, before initiation of the clinical studies, efforts need to minimize the heterogeneity of the definite diagnosis, e.g. training to the investigators, or central diagnosis.