|Home | About | Journals | Submit | Contact Us | Français|
Depression is common in people with cardiovascular diseases (CVD) and those with HIV, and is a risk factor for CVD-related mortality. However, little is known about whether HIV influences the relationship between depression and cardiovascular risk. 526 HIV-infected and 132 uninfected women from the Women’s Interagency HIV Study were included in an analysis of women who completed twice-yearly study visits over 9.5 years. CVD risk was calculated at baseline and approximately 9.5 years later using the Framingham Risk Score (FRS). Chronic depressive symptoms were defined as Center for Epidemiologic Studies Depression Scale scores of 16 or greater at ≥75% of study visits. Over the follow-up period, 22.8% of HIV-infected women and 15.9% of HIV-uninfected women had chronic depressive symptoms (p=0.08). Baseline FRS were similar between HIV infected and uninfected women (M=−5.70±SE=0.30 vs. M=−6.90± SE=0.60, p=0.07) as was follow-up FRS (M=0.82±SE=0.30 vs. M=−0.44± SE=0.73, p=0.11). Among HIV-infected and uninfected women, together, follow-up FRS were higher among women with chronic depressive symptoms as compared to those without (M=1.3± SE=0.6 vs. M=−0.3± SE=0.40, p<0.01), after adjusting for baseline FRS and other covariates. HIV status did not modify the relationship between chronic depressive symptoms and FRS. Chronic depressive symptoms accelerated CVD risk scores to a similar extent in both HIV infected and uninfected women. This implies that the diagnosis and treatment of depression may be an important consideration in CV risk reduction in the setting of HIV-infection. The determination of factors that mediate the depression/CVD relationship merits further study.
Across multiple large-scale studies, depression and depressive symptoms have generally been significantly associated with cardiovascular disease (CVD) risk (Ferketich, Schwartzbaum, Frid, & Moeschberger, 2000; Hemmingway & Marmot, 1999; Mendes de Leon et al., 1998; Sesso, Kawachi, Volonas, & Sparrow, 1998; Wassertheil-Smoller et al., 1996; Wulsin & Singal, 2003). Depression confers a two-fold greater risk for coronary artery disease among both men and women (Rugulies, 2002). Among patients with coronary disease, coexisting depression/depressive symptoms have also been associated with poorer prognosis (Hemmingway & Marmot, 2000; Kuper, Marmot, & Hemingway, 2002; Wuslsin et al., 2003). In a cohort study of patients with coronary heart disease, those with elevated depressive symptoms were at a 31% increased risk for adverse coronary events even after controlling for comorbid conditions and disease severity (Whooley et al., 2008).
Potential mechanisms explaining the associations between depression and CVD include the increased inflammation associated with depression such as increased levels of the inflammatory markers C-reactive protein and IL-6, as well as increased platelet activity and hypercoagulable states that are thought to contribute to atherosclerosis (Elovainio et al., 2009; Frasure-Smith, Lespérance, Irwin, Talajic, & Pollock, 2009; von Kanel, Bellingrath, & Kudielka, 2009). One of the first studies to test the potential pathways that link depression and CVD found support for the mediating role of norepinephrine, C-reactive protein, and omega-three fatty acid levels as well as for behavioral factors including medication non-adherence, smoking, and physical inactivity (Elovainio et al., 2009).
Little research has been conducted on this relationship between depressive symptoms and CVD risk in the context of HIV infection. CVD is a growing concern among HIV-infected patients as they live into older age (Krik & Goetz, 2009). There are also concerns regarding potential cardiotoxic side-effects of some antiretroviral medications (Friis-Møller et al., 2003; Friis-Møller et al., 2004; Hsue et al., 2004; Mooser, 2003; Vittecoq et al., 2003). Some studies have found higher CVD event rates in HIV-infected individuals compared to uninfected individuals (Triant, Lee, Hadigan, & Grinspoon, 2007). Markers of subclinical atherosclerosis such as carotid artery plaque, stiffness, and calcification are more prevalent and are present at younger ages in HIV-infected patients as compared to uninfected comparison subjects (Depairon et al., 2001; Hsue et al., 2004; Jerico et al., 2006; Johnsen et al., 2006; Kingsley et al., 2008; Maggi et al., 2000; Seaberg et al., 2010).
Finally, it is important to investigate the relationship between depression and CVD among HIV-infected women in particular, given the over two-fold increased risk of depression among women compared to men (Centers for Disease Control and Prevention, 2006). In addition, ethnic minority women, who have a higher prevalence of self-reported depressive symptoms than white women(Centers for Disease Control and Prevention, 2004), make up one of the fastest growing subgroups of individuals infected with HIV in the US (Hader, Smith, Moore, & Holmberg, 2001). Prevalence rates of major depression among HIV-infected women have ranged from 1.9% to 35% in clinical samples and from 30% to 60% in community samples (Ciesla & Roberts, 2001; Hader et al., 2001; Goggin, Engelson, Rabkin, & Kotler, 1998; McDaniel, Fowlie, Summerville, Farber, & Cohen-Cole, 1995; Moore et al., 1999; Taylor, Amodei, & Mangos, 1996). Women with a history of recurrent major depression have a greater than 2-fold prevalence of coronary calcification than those without depression (Agatisa et al., 2005). In addition, depression predicted coronary artery disease events over an 8-year follow-up in older adult women(Mendes de Leon et al., 1998) and over a 10-year follow-up in a younger female cohort (Wilson et al., 1998). While depression and depressive symptoms are common in the setting of HIV infection, their role in CVD risk remains unknown among women with HIV. The objective of this study was to determine the effect of depressive symptoms on estimators of CVD risk, computed by the Framingham coronary risk score, and to determine whether this association varies by HIV status. The current study is important as it is one of the first to investigate the impact of depressive symptom chronicity as opposed to assessing depressive symptoms at one time point.
Participants were a subgroup of women enrolled in the Women’s Interagency HIV Study (WIHS). The WIHS is a longitudinal multicenter cohort study that enrolled HIV-infected and demographically matched HIV-uninfected women in 1994–5 and 2001–2 in six US cities. Women attended study visits every six months in which they completed an interviewer-assisted questionnaire in English or Spanish, underwent a physical and gynecological exam, and provided biological specimens. The study’s methodology, training, quality assurance activities, and cohort characteristics have been reported elsewhere (Bacon et al., 2005; Barkan et al., 1998). Study protocols were approved by the institutional review boards and informed consent was obtained from the participants.
For the present study, chronic depressive symptoms were measured using data from all visits from baseline to Visit 20 (approximately 9.5 years of follow-up). Participants were included if they had depressive symptom scores for at least 15 of the possible 20 visits between study enrollment and Visit 20. Women enrolled during the second wave of WIHS recruitment in 2001–2002 (n=1143) were therefore excluded. Of the 2623 women from the first wave of recruitment, 2372 had enough data to calculate Framingham Risk Scores (FRS) at baseline. Of those, 772 had depressive symptom data for at least 15 visits. Of the 772, only 658 had all components required to calculate FRS at Visit 20, representing the final included participants. We were unable to use shorter follow-up in order to include a greater number of subjects because measures required to calculate Framingham risk scores were only available from the baseline visit as well as from visits occurring 9–10 years later. Hence 1714 women from the baseline WIHS enrollment were excluded from the present analysis.
The key predictor variable in this analysis was chronic depressive symptoms. Depressive symptoms were measured using the Center for Epidemiological Studies-Depression Scale (CES-D; Radloff, 1977). The CES-D has excellent reliability, validity, and factor structure among numerous subgroups CES-D; Radloff, 1977), and it is commonly used in studies of HIV populations, including women (Low-Beer et al., 2000; Cook et al 2002). A standard cut-off score of 16 or higher was considered indicative of clinically relevant symptomatology (Boyd, Weissman, Thompson, & Myers, 1982; Cook et al., 2002). Women were classified as having chronic depressive symptoms if CES-D was 16 or higher for at least 75% of visits (Cook et al., 2002; Ickovics et al., 2001).
The primary outcome variable was the Framingham Risk Score (FRS). FRS was calculated at baseline and at Visit 20 (Wilson et al., 1998). Age, total cholesterol, HDL cholesterol, systolic blood pressure, diastolic blood pressure, diabetes, and current smoking status are the variables that are used in the Framingham algorithm. Systolic and diastolic blood pressure scores were the average of two measurements. Diabetes was determined at follow-up using fasting glucose levels equal to or greater than 126 mg/dL, self-report of taking a diabetes medication and/or self-report of diagnosed diabetes. As fasting glucose measurements were not available at baseline, only diabetes medication usage or self-reported diabetes diagnosis treated with medications was used. Total cholesterol and HDL cholesterol were measured at baseline and Visit 20. Random blood draws without regard to fasting status were conducted at baseline whereas all Visit 20 samples were fasting. Current cigarette smoking (yes/no) was determined by self report. The raw Framingham risk scores are used in the current analysis which yields means that can extend anywhere from less than or equal to −2 (10-year CHD risk correlate of 1%) to greater than or equal to 17 (10-year CHD risk correlate of greater than 32%) (Wilson et al., 1998).
Income and education were collected at Visit 20. Low income was defined as annual household income less than $12,000, and education was stratified by whether the participant had completed high school or an equivalent (e.g., GED) or not. Body mass index was based on measured weight and height recorded at Visit 20 and calculated as weight divided by meters squared. Women were classified as underweight (<18.5 kg/m2), normal (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (≥30.0 kg/m2). As the number of women who were underweight was very low (n=16), they were grouped with the normal group for analysis. Illicit drug use and alcohol use were assessed by self-report at Visit 20, and analyzed as any illicit drug use in the past 6 months (e.g., marijuana, cocaine, crack, heroin) and consumption of three or more drinks per week, respectively. Self-reported antidepressant use at Visit 20 was dichotomized as those who had taken antidepressant medications in the past 6 months and those who had not.
HIV status was determined at all visits for the HIV-negative women. CD4 T-lymphocyte counts were measured for all subjects at each visit and were averaged over the total number of visits. Similarly, plasma HIV-1 RNA levels were determined at each visit for the HIV-infected women and were averaged for the total number of visits. History of AIDS defining illness at Visit 20 was defined as a prior report of conditions consistent with CDC’s 1993 surveillance definition, but excluding immunologic criterion of CD4+ <200 cells/μl. Use of highly active antiretroviral therapy (HAART) was defined as self-reported therapy at Visit 20. HAART was defined using the standard WIHS definition, adapted from the Department of Health and Human Services & Kaiser Panel guidelines (Department of Health & Human Services, 2008).
Continuous variables were reported as means ± standard error. Student’s t tests were used to test for statistically significant differences in means between HIV-infected and uninfected women. Pearson’s X2 tests were used to assess differences in proportions for dichotomous and categorical variables; if the numbers within a cell were small (i.e. n<10), Fisher’s exact test was used. Univariate and multivariate one-way analyses of covariance were performed to examine the relationship between the independent variable (i.e., dichotomous chronic depressive symptoms) and the dependent variable (i.e., Framingham risk score at Visit 20), controlling for covariates. Baseline FRS was also included as a covariate to control for baseline cardiovascular risk. Selected interaction effects between HIV status, chronic depressive symptoms and other covariates were tested in the multivariate model. Estimated means for FRS were generated for each covariate. Mediational models to determine whether behavioral variables (i.e., alcohol/drug use, HAART use) explained any association between depressive symptoms and FRS were tested using analysis of variance and following the rules for testing mediation as outlined by Baron and Kenny (1986). To evaluate the relationship between chronic depressive symptoms and FRS in HIV-infected women, HIV-specific variables were included. A p-value <0.05 was used to evaluate statistical significance. Analyses were performed using SPSS version 16.0 (Chicago, Illinois).
Characteristics of the HIV-infected (N= 526) and uninfected (N = 132) women included in the study are presented in Table 1. Overall, the mean age of the women at baseline was 35.6 years (SD=7.82) and 57.8% self-identified as Black and 26.1% as Hispanic. At the baseline visit, there was a non-significant difference in FRS with a higher mean in the HIV-infected group (−5.7±0.30 vs. −6.9±0.60; p=0.07). At Visit 20, mean FRS was similar between the two groups (0.82 ± 0.30 vs. −0.44 ± 0.73; p=0.11). Although chronic depressive symptoms were more prevalent in the HIV-infected group, differences were not statistically significant (22.8% vs 15.9%, p=0.08). Self-reported illicit drug use (31.1% vs. 18.9%; p<0.01) and alcohol use (20.5% vs. 9.1%; p<0.01) were greater in the uninfected compared to the infected women. HIV-uninfected women also had higher rates of overweight/obesity compared to the infected women (p< .01). Antidepressant use was more common among the HIV-infected women as compared to the uninfected women (27.6% vs. 18.9%, p=0.04).
In unadjusted analyses, FRS at Visit 20 was positively associated with presence of chronic depressive symptoms, as well as with black race, lower household income, antidepressant use, and minimal alcohol use (all p < .05; Table 2). In multivariate analysis, controlling for baseline FRS, chronic depressive symptoms and absence of significant alcohol use were the only significant independent predictors of FRS at follow-up. Participants with chronic depressive symptoms had a higher mean FRS at visit 20 than those without chronic depressive symptoms (1.3±0.60 vs. −0.3±0.40, p<0.01). There were no interaction effects of other covariates on this relationship. Analyses regarding the mediating role of HAART use, drug/alcohol use, and the individual factors that comprise the FRS were not significant.
In univariate analyses limited to HIV-infected women (Table 3), FRS was significantly associated with chronic depressive symptoms in addition to Black race, low household income, antidepressant use, absence of significant alcohol use and a history of clinical AIDS. The effects of race, income, antidepressant use, and clinical AIDS were no longer significant in adjusted analyses. FRS remained independently associated with chronic depressive symptoms and absence of significant alcohol use (both P<0.05). There was a trend towards higher FRS among women not on HAART (0.5±0.5 vs. 1.8±0.8, p=0.08).
In this analysis, chronic depressive symptoms were associated with higher Framingham risk scores at 9.5-year follow-up even after controlling for known risk factors including baseline Framingham risk score. Chronic depressive symptoms increased CVD risk in both HIV-infected and uninfected women. These findings are important for a number of reasons. First, depression is a modifiable risk factor that could be targeted for intervention possibly reducing CVD risk. Second, in the HAART era, CVD death is a leading cause of mortality in HIV-infected persons (French et al., 2009).
The fact that there was not a statistically significant difference in chronic depressive symptoms between the HIV-infected and uninfected women was surprising although this difference (i.e., 23% HIV-infected vs. 16% uninfected) bordered on significance and may have been an issue of statistical power. Twenty-three percent of HIV-infected participants experiencing chronic depressive symptoms is somewhat lower than prior studies, but this may be due to differences in timing in data collection relative to the HAART era (Ickovics et al., 2001). It is also likely that women with chronic depressive symptoms were underrepresented because we excluded those who did not provide CES-D data for at least 15 of the 20 visits and, in general, individuals suffering from depression are less likely to be adherent to medical appointments and exclusion analyses indicated higher baseline CES-D scores among excluded participants (Holzemer et al., 1999; Gonzalez et al., 2008). Despite this possible limitation, a strength of the current study is that it is one of the first to examine the impact of depressive symptom chronicity on a CVD risk score as opposed to cross-sectionally only (Evans et al., 1995; Kaplan, Marks, & Mertens, 1997; Kilbourne et al., 2002).
In the current study, univariate analyses indicated that higher follow-up FRS was also associated with Black race and lower household income. These findings are consistent with prior cross-sectional studies showing CVD was associated with demographic factors in the HIV and non-HIV settings (Kaplan et al., 2007). However, these demographic factors did not maintain significance in the multivariate model with chronic depressive symptoms. Given the higher prevalence rates of depression among lower income and Black individuals in the US (Centers for Disease Control & Prevention, 2004), it is not surprising that the impact of these factors on FRS appeared to be attenuated by depressive symptoms.
The only consistently significant variable associated with increased FRS besides chronic depressive symptoms was a lack of significant alcohol use. It is possible that this is because participants who were HIV-uninfected were more likely to be significant drinkers than HIV-infected participants and the FRS score was somewhat, although not significantly, lower (i.e., better) among HIV-uninfected participants. However, a test of the interaction between HIV-status and alcohol use was not significant. It is also possible that our definition of significant alcohol use (i.e., >3 drinks per week) is actually protective at least when the number of drinks per week is around 3 or 4 given research indicating that moderate alcohol use can increase HDL levels (Foerster et al., 2009). A final possibility is that some of the non-drinkers in the current study are abstaining from alcohol use due to having a history of prior abuse, so it’s possible that their increased FRS is due to prior excess drinking history.
There are a number of limitations of the current study. First, we used FRS as a raw score and not as the calculated risk in order to have a large enough distribution for analysis which yields means that extend to negative numbers. Adjusted scores of 2.4 and 0.1 translate to approximate 10 year event rates of 3% and 2% (Wilson et al., 1998). Although this difference is small, the low FRSs are largely determined by young age and female gender. The women may well have been too young to demonstrate larger differences and these differences may increase with time. Second the CES-D is a measure of depressive symptomatology and not a diagnostic measure of major depressive disorder. It is possible that use of a clinical diagnostic tool may have somewhat altered results although this has not been the case in other studies that have used such measures (Devanand et al., 2004; Fraguas et al., 2007; Sesso et al., 1998). Third, the study participants were all female as well as racially and ethnically diverse and it is possible that FRS may underestimate true CV risk in female and minority populations (Cappuccio, Oakeshott, Strazzullo, & Kerry, 2002; Michos et al., 2006). However, among the various CV risk prediction equations, those derived from the Framingham Heart Study are most commonly recommended for use in the United States (Wilson et al., 1998). Fourth, women included in the study were of lower cardiovascular risk and had less prevalent depression than the large number of baseline enrollees that were not be included in the present analysis due to various reasons as outlined in the methods section. Comparison of women included in the present study to those excluded, the included women (658) were a mean of 1 year younger (35.6±7.8 years vs 36.6±8.1 years, p<0.01, had lower systolic blood pressure (115±15mmHg vs 117±16mmHg, p<0.01), higher HDL (49±13mg/dL vs 47±15mg/dL, p<0.01, and were less likely to have hypertension (14.0% vs 17.7%, p=0.03. Accordingly, their mean Framingham risk score was somewhat lower than those not included (−6.2 ± 7.0 vs−4.9 ± 7.6, p<0.01). In addition, the prevalence of depression at baseline was lower in the included group (53.2% vs.58.1%, p=0.03). Among patients not included in the study, 36.5% had died prior to the follow-up visit and hence could not be included in the present analysis, indicating the presence of possible selection bias. Fifth, survivor bias is likely present especially given the large sample size decrease after baseline among the seropositive participants. Although the mortality risk of the uninfected group may be lower than the infected group, the analysis was limited to survivors who remained active participants in the longitudinal study. Finally, because the women were relatively young and about half were African-American, the results may not generalize to other HIV-infected or at-risk populations.
Despite these limitations, the results suggest that therapeutic interventions for depression should be evaluated with respect to decreasing CV risk. In the non-HIV setting, the relationship between depression and cardiovascular disease has led to efforts to use antidepressants and cognitive behavioral therapy to improve cardiovascular prognosis (Carney, Freedland, & Veith, 2005; ENRICHD Investigators, 2003; ENRICHD Investigators, 2004). To date, such studies have been disappointing as neither antidepressants nor cognitive behavioral therapy have been shown to lower CV event rates, however, there were some inherent design flaws in those early intervention trials (Carney, Freedland, & Veith, 2005; ENRICHD Investigators, 2003; ENRICHD Investigators, 2004). The present study findings suggest that depressive symptoms may worsen CV outcomes by deleteriously influencing CV risk factors and imply that researchers need to think critically about how best to design and evaluate interventions to target depression among both HIV-infected and uninfected women with or at risk for CVD. Given the increasing life expectancy among HIV-infected individuals due to the advent of HAART, interventions focused on the reduction of depressive symptoms may result in benefits on FRS as these women age. Further studies are needed to confirm data from the current study and to determine the specific mediators of the relationship between chronic depressive symptoms and cardiovascular risk among HIV-infected women.
Funding Source: Data in this manuscript were collected by the Women’s Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Stephen Gange). The WIHS is funded by the National Institute of Allergy and Infectious Diseases (UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI-42590) and by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (UO1-HD-32632). The study is co- funded by the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute on Deafness and Other Communication Disorders. Funding is also provided by the National Center for Research Resources (UCSF-CTSI Grant Number UL1 RR024131). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.