Disentangling health complexity among internal medicine, hospitalized patients

factorial structure in Germany. Method: The PHQ-15 was used to gather the somatic symptoms from a group of 491 Chinese study participants and a group of 2,517 German patients. The confirmatory factor analyses were used for the comparison between both groups and examination of the questionnaire for the Rasch scalability. Results: Descriptive analyses showed significant differences in the total score of PHQ-15. The factorial structure of the instrument could be confirmed in both groups (Chinese sample: CFI = 0.954, TLI = 0.944; German sample: CFI = 0.929; TLI = 0.914). However, it was not possible to find an acceptable model across both groups. For further analyses to identify latent classes across both groups, items had to be excluded which could not be integrated into the Rasch model. A 4-group model was found as best-fitting model. It showed also significant differences in distributions of relative frequencies in both samples. Conclusion: There are differences in the factorial structure between a Chinese sample and a German sample. It must be assumed that if the PHQ-15 is used in different cultural groups, the items have different meanings. A direct comparison of the total value does not seem to be meaningful.