The psychometric properties of the panic disorder module of the Patient Health Questionnaire (PHQ-PD) in high-risk groups in primary care.

AIMS To study the validity of detecting panic disorder (PD) using the Patient Health Questionnaire (PHQ) in a high-risk population in primary care and to test whether modified evaluation algorithms improve the operating characteristics of this questionnaire. Furthermore, the influence of psychiatric comorbidity on the test characteristics of the panic module was studied. METHODS The PHQ was administered in a primary care sample with patients at high-risk for psychiatric disorders. The total sample of 479 high-risk patients comprised 311 frequent attenders (FA), 39 patients with unexplained somatic complaints (USC) and 191 patients with mental health problems (MHP). The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID- I) was the reference standard for the presence of PD. Sensitivity, specificity, and predictive values were calculated. The conditional test characteristics were calculated based on the observed prevalence of PD in the three high-risk groups. RESULTS PD was diagnosed in 4.8% of the FAs, in 9.8% of the USCs and in 7.6% of the MHPs. The PHQ achieved moderate operating characteristics. Modified evaluation algorithms of the questionnaire led to an improvement of test characteristics, especially the screening question: sensitivity .71 and specificity .83. Psychiatric comorbidity increased sensitivity while decreasing specificity. CONCLUSION The original and modified algorithms of the PHQ-PD performed moderately in screening for panic disorder. Using only the first question of the PHQ-PD showed the best psychometric properties (sensitivity). For screening purposes requiring high sensitivity we endorse to use the screening question instead of the original algorithm.