ABSTRACT
BACKGROUND: Depression is a frequent comorbidity of COPD and leads to worse clinical COPD-outcomes. PHQ-9 and PHQ-2 are two widely used brief instruments to assess depression. However, psychometric properties in COPD patients are unknown. This study examines factorial validity, measurement invariance and composite reliability (CR) of PHQ-9/PHQ-2, respectively, and concordance between both tools. METHODS: This is a secondary analysis of N = 561 COPD patients who filled out the PHQ-9 at the begin (T0), the end (T1) and 3/6/9/12 (T2/T3/T4/T5) months after pulmonary inpatient rehabilitation. Structural equation modeling was used to examine factorial validity and measurement invariance between gender, GOLD disease severity groups and over time. Concordance was assessed using Cohen's Kappa, Yules Y, positive and negative agreement. RESULTS: A one-factor model (with one freed residual covariance) showed best model fit. At least partial scalar invariance could be established. Concordance between both instruments was substantial. 31.7% (26.2%) COPD patients showed clinically relevant depression according to PHQ-9 (PHQ-2) at T0. At T0-T2, PHQ-9 classified more patients as depressed than did PHQ-2. According to both measures, depression rates declined after rehabilitation. Reliability was high for both PHQ-9 (CR = 0.94) and PHQ-2 (CR = 0.89). LIMITATIONS: No gold-standard (clinical interview) to assess depression was used. Therefore, diagnostic accuracy for PHQ-9/PHQ-2 remains unclear. CONCLUSIONS: PHQ-9 and PHQ-2 fulfill important psychometric criteria (factorial validity, invariance, reliability) for measuring depression in COPD. The results support their use in clinical practice to assess severity of depression. Diagnostic accuracy to identify major/minor depression of both instruments should be examined in future studies.
