October 2011
Reviewer: Jeff C. Huffman, MD
Depressive symptoms, health behaviors, and subsequent inflammation in patients with coronary heart disease: prospective findings from the heart and soul study
Duivis HE, de Jonge P, Pennix BW, Na BY, Cohen BE, Whooley MA
Am J Psychiatry 2011; 168(9):913-920
Background: Depression has been associated with increased inflammation in patients with cardiac illness; this is an important finding given that elevated inflammatory markers have been independently associated with morbidity and mortality in this population. However, there have been conflicting studies regarding the directionality of this association: does depression subsequently lead to increased inflammation in cardiac patients, or vice versa?
Methods: This was a follow-up study involving subjects who initially participated in the Heart and Soul Study, a large epidemiologic study of outpatients with stable coronary heart disease. This follow-up substudy recruited subjects from Sept 2005-Dec 2007, 5 years after initial participation. Enrolling subjects underwent assessment for depression symptoms using the Patient Health Questionnaire-9 (PHQ-9) yearly for 6 years. In addition, the following markers of inflammation were collected from subjects at baseline and 5 years: high sensitivity C-reactive protein (hsCRP), interleukin-6(IL-6), and fibrinogen. Finally, at these timepoints, the study team collected data on potentially confounding variables (age, gender, education, race, prior MI, history of DM, congestive heart failure, and aspirin use) and on health-related behaviors that can impact inflammation (smoking, physical activity, and body mass index).
Data regarding the connection between depression and inflammation was assessed in several ways. Data analysis compared levels of inflammatory markers between three groups (depressed at zero follow-ups, one follow-up, or two or more follow-ups, with PHQ-9≥10 used as the depression cutoff), with additional analyses adjusting for relevant covariates and health behaviors. The depression-inflammation connection was assessed using absolute levels of the inflammatory markers at 5 years (analysis #1), as well as using change between baseline and 5-year levels (analysis #2).
In addition, depression symptoms were assessed as a continuous variable (i.e., using the average of PHQ-9 scores) to assess associations between these scores and 5-year inflammatory markers (analysis #3). Finally, markers of inflammation were split into categorical variables (split into quartiles), and the authors assessed rates of having the highest quartile of each marker in each of the three depression categories (analysis #4).
Similar analyses (analysis #5) were performed to assess the converse association (i.e., to assess whether higher baseline levels of inflammatory markers were associated with subsequent depression, as measured via the three categories defined above).
Results: A total of 667 (80% of the surviving original Heart and Soul study cohort) subjects enrolled in this substudy. 138 (21%) had depression at 2+ interviews, 86 (13%) had depression at 1 interview, and 443 (66%) had no depression.
Analysis #1: On unadjusted models, increasing depression frequency was generally associated with elevations in all three levels of inflammatory markers at 5 years, though the association was subthreshold (p=.06) for fibrinogen; after adjustment for medical/sociodemographic variables, depression was independently associated only with elevated IL-6 (p=.01); depression-fibrinogen remained a trend (p=.08). Finally, when health behaviors were added to the model, there were no significant associations (or trends).
Analysis #2: Regarding the relationship between depression and change in inflammatory markers, depression was only associated with fibrinogen on unadjusted analysis (p=.05), and this became a trend-level association after adjustment for medical/sociodemographic variables and health behaviors (p=.07 for both analyses).
Analyses #3/#4/#5: Using PHQ-9 scores as a continuous variable, increasing depression scores were associated with increased hsCRP (p=.002) and IL-6 (p=.009), though not fibrinogen; these associations disappeared with addition of health behaviors to the model. When inflammatory markers were split into quartiles, depression status was significantly associated with a greater likelihood of having upper quartile levels of all three markers. Finally, baseline levels of inflammatory markers were not associated with subsequent depression for any of the three markers.
Commentary: Overall, depression was generally associated with subsequent increases in the three inflammatory markers, though this was not highly powerful or entirely consistent across methods of analysis. In addition, when ‘health behaviors’ were added to statistical models, these associations between depression and inflammation appeared to essentially disappear in nearly all cases, suggesting a vital role of health behavior in moderating this association. In contrast, baseline inflammation did not appear to be associated with development of depression, though this association was less comprehensively analyzed.
If we agree with the authors’ relatively blanket interpretation that depression was associated with greater inflammation in this cohort of cardiac patients, and that health behavior may play a key role in this connection, it suggests that addressing health behavior in depressed cardiac patients may be of vital prognostic importance, given the risks associated with increased inflammation. How to address impaired health behavior in depressed cardiac patients is a complicated issue: do we treat depression first before addressing behaviors like exercise and smoking cessation to give patients a better shot at changing long-standing behavior, do we prioritize/address the behaviors first since this may help depression and will certainly help cardiac risk, or do we try to simultaneously act on both? And how do we help patients, for example, exercise? Changing behavior—especially in depressed patients—can be extremely challenging and simply recommending change can help a little but not a lot. These are important questions on an individual patient level and for psychosomatic medicine research.
This work had some substantial strengths—such as a large cohort that was followed for many years, in a systematic manner, using well-accepted methods of measuring depressive symptoms and inflammation. There are some important limitations/considerations as well. First, the classification and measurement of ‘health behaviors’ was somewhat limited. Physical activity and smoking status were measured only by self-report; there are more objective ways of measuring these behaviors. Furthermore, classifying BMI as a ‘health behavior’ is somewhat unusual—it may be a surrogate for everything from diet, physical activity, and/or poor genetics. In addition, using the PHQ-9—and especially a score cutoff—for depression can be complicated. Patients can rack up high scores on the PHQ-9 when they are generally ill (and thus have low energy, poor sleep, psychomotor slowing, and/or poor concentration) without having depressed mood, anhedonia, guilt, or thoughts of death, yet such patients would be classified as ‘depressed’. It is not clear that the covariates used would fully account for physical symptoms/health status, and perhaps it is more true that physically sicker patients had more inflammation. Supporting the importance of this is one prior study that found inflammation to predict psychological symptoms of depression (and not the other way around).
Where are we left? The upshot seems to be that depression and health behaviors are important in cardiac patients and that poor adherence to health behaviors may be an important mediating/moderating step in the depression-inflammation-mortality connections in cardiac patients. We should focus on treating both depression and health behaviors—in some order or together when possible—in this vulnerable population, and future research should continue to both study these connections and identify optimal ways to address depressed cardiac patients who have impaired health behaviors.