Psychometric analysis of the Patient Health Questionnaire in Danish patients with an implantable cardioverter defibrillator (The DEFIB-WOMEN study)

- Analysis showed disordered response thresholds in eight of nine items.
- When combining response options 2 and 3, the rescored PHQ-9 had a better fit.
- The unidimensionality and the reliability of the PHQ-9 were confirmed.
- Items did not have differential functioning across variables influencing depression.

ObjectiveTo assess the psychometric properties of the Patient Health Questionnaire (PHQ-9), a measure of depressive symptoms, in a large Danish national cohort of patients with heart disease, implanted with an implantable cardioverter defibrillator (ICD), using item response theory.MethodsA prospective cohort of patients implanted with an ICD (n = 1531; 80.4% men) completed the PHQ-9 at the time of implant. Data were analyzed using two item response theory models, the partial credit model and the generalized partial credit model.ResultsThe analysis showed disordered response thresholds in eight of nine items for the partial credit model and five of nine items for the generalized partial credit model, indicating that respondents have difficulty discriminating between response options. When collapsing response options 2 and 3, the rescored PHQ-9 had a better fit to both models. The unidimensionality and the precision of the rescored PHQ-9 were confirmed. Items did not have any differential functioning (DIF) across educational level, age, indication for ICD implantation, and severity of heart failure that influence depression outcomes in patients with an ICD. One item exhibited DIF by gender. Three items did not fit the partial credit model, but the generalized partial credit model could be fitted to the full item set.ConclusionThe unidimensionality and reliability of the Danish version of the PHQ-9 were confirmed. However, the associated consequences of the number of response options (3-point versus 4-point Likert scale) need to be further examined for the PHQ-9 both as a screening tool and outcome measure.