Background: It is critical to evaluate levels of depressive symptoms in patients with heart failure (HF) because they have a substantial impact on outcomes. Because there are shared physical symptoms between HF and depression, depressive symptom instruments that include physical symptoms of depression might artificially inflate scores in HF patients. One way to assess the accuracy of measurement of depressive symptoms is to explore the iimpact of physical symptom on outcomes using a depression instrument with physical symptoms, and then to use it again with the physical symptoms removed. The purpose of this study was to compare the predictive ability for event-free survival between the full Patient Health Questionnaire-9 (PHQ-9) and the PHQ-9 with the physical symptom items removed. Hypothesis: The full PHQ-9 will account for a greater amount of variance in event-free survival of HF patients than the PHQ-9 with the physical symptom items removed after controlling for relevant variables. Methods: Data from 214 patients with HF (mean age: 61 ± 12, female: 33%, NYHA III/IV: 60%) were collected. The PHQ-9 was used to assess levels of depressive symptoms. Event-free survival (mortality, rehospitalization, or emergency department visit for cardiac reasons) was collected for a median follow-up of 335 days. To examine predictive ability on event-free survival after adjusting for relevant variables, of the two versions of the PHQ-9, the scores of each of the two instruments were entered in two separate hierarchical Cox proportional hazard regression analysis. Because the scores of PHQ-9 were skewed, the scores were transformed. Results: After controlling for age, gender, ethnicity, body mass index, comorbidity scores, and NYHA class, scores from the full PHQ-9 predicted event-free survival (HR=1.32, 95% CI=1.05–1.66). Subsequent analysis was done to evaluate the impact of physical symptoms in the PHQ-9 by excluding 3 physical symptoms. Scores of the PHQ-9 without physical symptoms were an independent predictor of first cardiac event (HR=1.12, 95% CI=1.02–1.23). Conclusion: Although the hazard ratio is lower for the prediction of event-free survival, removal of physical symptom items from the PHQ-9 does not eliminate the association of depressive symptoms with event-free survival. It is possible to obtain an accurate assessment of depressive symptoms in HF patients using an instrument that includes physical symptoms.
FULL TEXT