Objectives: Depressive disorders are overrepresented among the patients admitted to non-psychiatric units in general hospitals. However, the majority of depressed patients fail to be identified within this care setting. Since a self-administered questionnaire (Patient Health Questionnaire, PHQ-9) has given encouraging results in English and Spanish, new research should test its criterion validity in a French-speaking environment.
Design: The study included 292 patients admitted to the internal medicine units of the University Hospitals of Geneva. Each patient filled the PHQ-9; 212 patients also underwent a blinded DSM-IV diagnostic assessment by a psychiatrist.
Methods: In order to assess the validity of PHQ-9 against the gold standard of the psychiatrist's DSM-IV diagnosis, we calculated overall accuracy, sensitivity, specificity, positive predictive value and Cohen kappa coefficients. We also studied the relationship between the PHQ-9 diagnostic and the severity of depression. Finally, analysis focused on the presence of a diagnosis of depression.
Results: Within the framework of the study, PHQ-9 showed an acceptable level of specificity. However, its sensitivity in detecting major depression was low (about 50% of false-negative results). As regards the overall presence of depressive disorders, this instrument performed hardly any better (35% of false negatives). Other characteristics of the population under investigation may have affected the data.
Conclusions: The French version of PHQ-9 demonstrated low sensitivity as compared with psychiatrist-established diagnosis of DSM-IV A criterion and major depressive episode.