|Home | About | Journals | Submit | Contact Us | Français|
To the Editors:—Recently, Monahan et al. concluded that the PHQ-2 and PHQ-9 were “valid/reliable for assessing DSM-IV depressive disorders and depression severity among adults living with HIV/AIDS in Western Kenya”1. However, they did not actually use any accepted assessment tool to diagnose DSM-IV depressive disorders. Instead, they used a “diagnosis” of depression based on the self-report PHQ-9 as their criterion standard. A recent meta-analysis2 found that the PHQ-9 was acceptable for depression screening, but it was not superior to results from other depression screening tools in identifying cases of depression,3 and it was poorly sensitive in some patient groups2. Based on the authors use of the PHQ-9 as a diagnostic criterion, one might conclude that any depression questionnaire could appropriately be substituted for a gold standard structured interview. This is clearly not the case, and the PHQ-9 is not appropriate for that purpose either.
Monahan et al. arrived at an inflated estimate of the performance of the PHQ-2 by committing the error of incorporation bias, in which the PHQ-2 was validated against the PHQ-9, of which it is a subset. This method will always produce favorable, albeit unrealistic, results. QUADAS, which is a tool for assessing the quality of diagnostic accuracy studies,4 contains an item explicitly assessing whether the reference criterion was independent of the index test: this study violates that standard. Monahan et al.’s favorable assessment of the PHQ-2 contradicts a more measured conclusion from a recent meta-analysis that found that ultra-short depression screens like the PHQ-2 have high false-positive rates and should only be used as an initial method to rule out diagnoses5.
Monahan et al. minimized findings that the PHQ-9 had only moderate test-test stability, which is inconsistent with their claims it was validly assessing clinically significant DSM-IV depressive disorders. They also minimized indications from focus groups that patients were “challenged” by the response format. Overall, the authors’ conclusion that the PHQ can be implemented in understaffed Kenyan clinics to enhance care for depression is troubling and contradicts a meta-analysis that clearly indicated that introducing screening for depression does not improve depression outcomes without a substantial infusion of resources6. Based on evidence, there is no reason to believe that introducing PHQ screening in Kenya would benefit patients, and such claims retard the attraction of sufficient resources and reorganization of care systems that would be required to improve depression care.