Objective: To evaluate whether items of the Patient Health Questionnaire 9 (PHQ-9) function differently in persons with traumatic brain injury (TBI) than in persons from a primary care sample.
Design: This study was a retrospective analysis of responses to the PHQ-9 collected in 2 previous studies. Responses to the PHQ-9 were modeled using item response theory, and the presence of DIF was evaluated using ordinal logistic regression.
Setting: Eight primary care sites and a single trauma center in Washington state.
Participants: Participants (N=3365) were persons from 8 primary care sites (n=3000) and a consecutive sample of persons with complicated mild to severe TBI from a trauma center who were 1 year postinjury (n=365).
Interventions: Not applicable.
Main outcome measure: PHQ-9.
Results: No PHQ-9 item demonstrated statistically significant or meaningful DIF attributable to TBI. A sensitivity analysis failed to show that the cumulative effects of nonsignificant DIF resulted in a systematic inflation of PHQ-9 total scores. Therefore, the results also do not support the hypothesis that cumulative DIF for PHQ-9 items spuriously inflates the numbers of persons with TBI screened as potentially having major depressive disorder.
Conclusions: The PHQ-9 is a valid screener of major depressive disorder in people with complicated mild to severe TBI, and all symptoms can be counted toward the diagnosis of major depressive disorder without special concern about overdiagnosis or unnecessary treatment.
Copyright © 2011 American Congress of Rehabilitation Medicine. Published by Elsevier Inc. All rights reserved.