Author + information
- Viola Vaccarino, MD, PhDa ()
We thank Dr. Jiang and colleagues for their thoughtful comments. We were pleased to read their well-done study (1)published almost concurrently with ours (2). Whereas their study shares a number of similarities with ours, it also presents a number of differences that might explain why the association of depression with mortality was stronger in their study than in ours. First, as these investigators point out, the assessment of depression was different in the two studies.
Second, the populations of heart failure (HF) patients also differed. In the attempt to restrict HF to cases of ischemic or hypertensive etiology, we included predominantly elderly patients (50 years and older), and the mean age of our sample was 73 years. In contrast, Jiang and colleagues enrolled patients 18 years and older. As a result, their population was much younger and most likely it included a broader variety of HF etiology.
Third, while we assessed mortality at six months, Jiang et al. assessed this outcome at three months and one year. The association of depression with death may vary according to length of follow-up. These investigators, for example, observed that major depression was more strongly associated with mortality at one year than at three months. At three months, this association was not significant in their study, and the point estimate was not very different from our estimate for “severe depression” at six months. Therefore, it might be that if a longer follow-up were available in our study, we would have observed a stronger association as well.
Fourth, while Jiang and colleagues adjusted for a number of potential confounders, they did not include in their models a number of covariables that were adjusted for in our study, including race, education, previous hospitalizations for HF, diabetes, systolic blood pressure, serum creatinine and patient’s self-reported disability level (number of limitations in activities of daily living). Our study, therefore, might have obtained a tighter control for potential confounders, especially those factors related to disease severity and comorbidity. We do not believe that model overfitting was a problem in our analysis, because the relative risk estimate declined toward the null with progressive adjustment for potential confounders, and because a reduced model provided similar results.
However, in our study, depression was still associated with substantial (although nonsignificant) mortality risk. Patients with severe depression had 68% higher mortality risk compared with nondepressed patients, and there was a dose-response relationship with mortality according to severity of depressive symptoms. As we pointed out in our study, our main focus was the examination of the association of depression with functional decline. Our study was not powered to examine mortality as a separate end point, and a larger study might have found such a mortality difference to be statistically significant. We concur with Dr. Jiang and colleagues that, based on the results of both studies, depression poses serious adverse effects on the outcomes of patients with HF.
- American College of Cardiology Foundation