|Home | About | Journals | Submit | Contact Us | Français|
To prospectively examine the association of major depression with incidence of the metabolic syndrome in women.
Data were drawn from one of seven sites of the Study of Women’s Health Across the Nation (SWAN), a prospective cohort study of the menopausal transition. Participants were 429 (34.5% African-American) women. Major depression and comorbid diagnoses were assessed via the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition Axis I Disorders at baseline and seven annual follow-up evaluations. The metabolic syndrome was measured at baseline and each follow-up evaluation (except the second) based on National Cholesterol Education Program (NCEP) criteria.
Longitudinal generalized estimating equations (GEE) models indicated that, in women who were free of the metabolic syndrome at baseline, a lifetime major depression history or current major depressive episode at baseline was significantly associated with the onset and presence of the metabolic syndrome during the follow-up (odds ratio = 1.82; 95% Confidence Interval (CI) = 1.06–3.14). Survival analyses showed that, in women who were free of the metabolic syndrome at baseline, a lifetime major depression history or current major depressive episode at baseline predicted increased risk of developing the metabolic syndrome during the follow-up (hazard ratio = 1.66; 95% CI = 0.99–3.75). Lifetime history of alcohol abuse or dependence predicted incident metabolic syndrome and attenuated the association between depression and the metabolic syndrome in both models.
This study documents that major depression is a significant predictor of the onset of the metabolic syndrome. Intervention studies targeting depression may prevent the development of the metabolic syndrome in women.
There is substantial evidence for an association between depression and onset and progression of coronary artery disease (CAD), Type 2 diabetes, and complications from these diseases (1–3). Recent attention has turned to the metabolic syndrome, a cluster of aberrations including insulin resistance or glucose intolerance, central adiposity/obesity, dyslipidemia, and hypertension, as a major risk factor for the onset of CAD and Type 2 diabetes, and for all-cause mortality (4). Age-adjusted estimates for the metabolic syndrome range from 34% to 40% for men and 35% to 38% for women, depending on the criteria used (5). Given the putative link between psychosocial factors and the metabolic syndrome (6) and evidence for an association of depression and the metabolic syndrome with risk of CAD and Type 2 diabetes, it is important to consider major depression in the etiology of the metabolic syndrome.
Three cross-sectional studies have addressed the role of major depression in the metabolic syndrome. In a nationally representative sample of young adults (National Health and Nutrition Examination Study, NHANES III), Kinder et al. (7) demonstrated that a history of major depression was associated with approximately twice the odds of having the metabolic syndrome in women but not in men. The second study, conducted with a sample of middle-aged depressed outpatients, showed that patients with the metabolic syndrome at a follow-up visit were significantly more likely to continue to have a diagnosis of major depression at that same visit (8). In women with suspected CAD, self-reported history of treated depression was unrelated to the metabolic syndrome, unless current depressive symptoms were also taken into account (9). Eight other studies examined depressive symptoms: findings of four cross-sectional analyses were positive (10–13) and one was null (14). Two longitudinal analyses from the same sample showed a positive association between depressive symptoms and the development of the metabolic syndrome in women (15,16). The third longitudinal analysis showed increased odds of developing depressive symptoms in men and women with the metabolic syndrome (17). Although these findings together provide preliminary support for an association of depression with the metabolic syndrome, they are limited by either cross-sectional designs (7–13) or measurement of depressive symptoms only (14–17). They provide no information about whether clinical depression influences risk for incident metabolic syndrome.
Thus, the primary purpose of the current study was to test the hypothesis that a lifetime history or current major depressive episode predicts incidence of the metabolic syndrome. The study sample was composed of women enrolled in the Study of Women’s Health Across the Nation (SWAN), Pittsburgh site. SWAN is a multisite, community-based cohort investigation designed to examine prospectively the biological and psychosocial correlates of the menopausal transition (18). Lifetime history and current depression diagnoses were obtained via the Structured Clinical Interview for the DSM-IV Axis-I Disorders-Non-Patient Edition (SCID-IV/NP) (19) and data on the metabolic syndrome were collected annually at every visit for 7 years (with the exception of the second follow-up), providing a unique opportunity to investigate the longitudinal association of major depression with the metabolic syndrome in women.
The 429 participants (281 whites and 148 African Americans) in the present report were drawn from an ancillary Mental Health Study conducted at the only SWAN site that administered the SCID-IV interviews annually. At this site, women were recruited using random digit dialing and a voter’s registration list. Initial eligibility criteria included being from White or African-American ethnicity (self-identified) with the target of 1/3 minorities; ages 42 to 52 years; having an intact uterus; menstruating within the prior 3 months; and not using reproductive hormones. (After women were enrolled, becoming menopausal or using hormone therapy did not exclude them from further follow-up.) Of the 463 women who enrolled in the Pittsburgh SWAN site, 443 also participated in the Mental Health Study and did not differ in sociodemographic factors and Center for Epidemiological Studies Depression scale scores (20) from those who did not participate. Fourteen women of the 443 did not have baseline metabolic syndrome data, primarily because of the absence of a fasting blood draw, resulting in 429 participants.
The Institutional Review Board of the University of Pittsburgh approved all procedures described herein. Written informed consent was obtained at the baseline visit as part of the SWAN protocol. Data were collected at baseline and annually over 7 years of follow-up (1996–2003). The following assessments were performed at baseline and each annual follow-up visit: self-report and interviewer-administered questionnaires pertaining to medical, reproductive, and health history, lifestyle behaviors, medication use, and psychosocial factors; a fasting venipuncture for the measurement of lipids and lipoproteins, insulin, and glucose (with the exception of visit 2); anthropometric measurements; and psychiatric diagnoses. All blood samples were maintained at 4°C until spun and separated, and then frozen at −20°C and shipped on dry ice to the central laboratory (Medical Research Laboratories, Highland Heights, Kentucky) for analysis (certified by the National Heart Lung and Blood Institute, Centers for Disease Control Lipid Standardization Part III program) (21). To ensure accuracy and consistency of data collection, all phlebotomists and technicians were trained and annually certified according to a common protocol.
For each evaluation except the second (because of no lipid or glucose data), a participant was classified as having the metabolic syndrome based on the US National Cholesterol Education Program Adult Treatment Panel III (22) criteria of ≥3 of the following: a) waist circumference (WC) >88 cm; b) triglyceride >150 mg/dl; c) high density lipoprotein cholesterol (HDL-C) <50 mg/dl; d) systolic blood pressure (SBP) >130 or diastolic blood pressure (DBP) >85 mm Hg or self-reports of taking “blood pressure pills”; e) fasting glucose >110 mg/dl (or having ever been classified as diabetic).
WC was measured at the level of the natural waist, defined as the narrowest part of the torso as seen from the anterior aspect. All lipid, lipoprotein, and apolipoprotein fractions were analyzed on ethylenediami-netetraacetic acid (EDTA)-treated plasma (23,24). Total cholesterol and triglycerides were analyzed by enzymatic methods (Hitachi 747 analyzer, Boehringer Mannheim Diagnostics, Indianapolis, IN) (23), and HDL-C was isolated, using heparin-2 M manganese chloride (24). Blood pressure was measured twice with a minimum 2-minute rest period between measures, with readings taken on the right arm, with the respondent seated and feet flat on the floor for at least 5 minutes before the measurement. Respondents had not smoked or consumed any caffeinated beverage within 30 minutes of blood pressure measurement. Appropriate cuff size was determined based on arm circumference. A standard mercury sphygmomanometer was used to record SBP and DBP at the first and fifth phase Korotkoff sounds. Two sequential blood pressure values were averaged. Glucose was measured, using a hexokinase-coupled reaction (Hitachi 747 to 200 analyzer, Boehringer Mannheim Diagnostics).
The SCID-IV/NP was administered at baseline to determine lifetime history and current diagnoses of major depression and annually thereafter to diagnose major depression in the past year. The SCID-IV/NP is a semistructured diagnostic interview designed to enable trained interviewers to determine lifetime and current diagnoses of psychiatric disorders, including mood, anxiety, and substance use disorders according to the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) (25). The current study focused on the diagnosis of a lifetime history or current major depressive episode at baseline.
The SCID-IV interviewers were trained clinicians with a Master’s degree, PhD, or degree in psychiatric nursing who had been certified via a training session with the Biometrics Institute at the College of Physicians and Surgeons of Columbia University (New York, New York), conducted at least 10 practice interviews evaluated by the study’s principal investigator (J.T.B.), and had taped interviews reviewed and approved by the Biometrics Institute. The κ score for lifetime diagnoses of major depression was 0.81, indicating adequate reliability.
There is some evidence to suggest that health risk behaviors may be one mechanism linking depression with the metabolic syndrome (6). Based on these findings, smoking, total caloric intake, and physical activity were selected as covariates. Self-reported current smoking status (yes/no) was measured at each visit via a self-report questionnaire. To reduce participant burden for the SWAN parent investigation, measures of physical activity and total caloric intake were not administered at every follow-up visit. Physical activity was assessed at baseline using a modified version of the Baecke physical activity questionnaire (26), a well-validated self-report measure of typical total physical activity during the previous year. Daily total caloric intake (including alcohol) was measured at baseline, using an interviewer-administered modification of the Block food frequency questionnaire (27,28).
The χ2 analyses were conducted to examine whether lifetime history of or current major depression at baseline was associated with odds of having the metabolic syndrome and each of its individual components at baseline in the full sample of women.
To examine whether lifetime history of or current major depression at baseline was associated with odds of developing the metabolic syndrome over the 7-year follow-up period in women who were free of the metabolic syndrome at baseline (n = 341), we used generalized estimating equations (GEE) with an exchangeable correlation structure, as implemented by Stata Release 9.2 (StataCorp, College Station, Texas). GEE is a general linear model used for clustered data which accounts for multiple incidents of the metabolic syndrome, adjusts for the within-subjects correlation present among repeated observations over time, and corrects for missing data by weighting each individual’s data according to the number of available observations (29). Accordingly, GEE does not require at least one follow-up observation for all individuals. The GEE models estimated odds ratios (ORs) and 95% Confidence Intervals (CIs) for developing the metabolic syndrome associated with baseline depression (lifetime history/current; dichotomous), adjusted for age (time-varying) and race.
Survival models used the LIFEREG procedure of the SAS statistical software (30) to calculate risk of first developing the metabolic syndrome across the follow-up period. These analyses were restricted to the 320 women who were free of the metabolic syndrome at baseline and had data for at least one follow-up visit. Multivariate analyses adjusting for baseline age and race were followed by analyses adjusted for a race-by-depression interaction term to examine whether the effect of depression on risk of the metabolic syndrome differed between African Americans and whites. Visit year was the unit of analysis, and the incident (failure) date was defined as the year of the follow-up visit during which the participant was first classified as having the metabolic syndrome. Participants were censored once they developed the metabolic syndrome (n = 56); women who remained free of the metabolic syndrome were censored at the end of the study observational period. This type of analysis predicts only the initial onset of the metabolic syndrome, as opposed to the previously described GEE analyses that evaluate the metabolic syndrome as a persistent condition.
Women were categorized as pre-, peri-, and postmenopausal/surgical menopause. Analyses were run with and without women using hormone replacement therapy. Results of both analyses were similar; thus, findings with the full sample are reported herein.
To examine whether the relationship between a lifetime history of major depression and risk of developing the metabolic syndrome was affected by lifestyle factors, three separate survival models were conducted controlling for baseline age, race, and a) smoking, b) total physical activity, and c) total caloric intake.
To examine whether there were differences between women with no lifetime history/no current depression, a single depressive episode (n = 66), and recurrent depressive episodes (n = 77) at baseline in risk of developing the metabolic syndrome, survival analyses paralleling those described above were conducted with depression coded categorically as none, single, or recurrent.
There is some debate about whether certain components of the metabolic syndrome are more important that others, and whether it is useful to examine the individual components of the metabolic syndrome in addition to the metabolic syndrome as a single construct (31,32). Therefore, five separate post hoc survival analyses were conducted to explore whether a lifetime history of major depression predicted significantly increased risk of developing each of the components of the metabolic syndrome (i.e., hypertension, low HDL-C, high WC, high triglycerides, high glucose). Visit year was the unit of analysis, and the incident (failure) date was defined as the year of the follow-up visit during which the participant first reached the cutoff for the component of the metabolic syndrome. Participants were censored once they reached the cutoff for the component.
Recent evidence suggests that the metabolic syndrome may be associated with the development of depressive symptoms in men and women (17). Based on this finding, post hoc analyses were conducted to explore whether the metabolic syndrome at baseline was associated with the onset of a major depressive episode in women who had no lifetime history/no current major depressive episode at baseline. GEE and survival analyses, which paralleled the analyses described above, were implemented with baseline metabolic syndrome as the independent variable and major depression at each follow-up visit as the dependent variable, controlling for age and race. Sample size for the GEE analyses was 278; sample size for the survival analyses was 267 (total number of women who had no history of depression or current depressive episode at baseline and had SCID data for at least one follow-up visit). For the survival analyses, visit year was the unit of analysis, and the incident (failure) date was defined as the year of the follow-up visit during which the participant was first diagnosed with a major depressive episode. Participants were censored once they were diagnosed with major depression (n = 45); women who remained free of the metabolic syndrome were censored at the end of the study observational period.
A p < .05 was considered statistically significant for all analyses.
Baseline descriptive statistics for demographic characteristics, the metabolic syndrome, and depression are presented in Table 1. The mean age of the sample was 45.6 years; approximately one third of the sample was African-American. Over one third of the sample (104 whites, 47 African Americans) met the criteria for a lifetime history or current major depressive episode at baseline; an additional 17 women reported a lifetime history or current use of antidepressant medication at baseline but did not meet the criteria for a lifetime history or current major depressive episode. Eighty-eight (n = 35 African Americans) of the 429 women met the criteria for the metabolic syndrome at baseline, and 56 (n = 21 African Americans) developed the metabolic syndrome over the course of the follow-up visits. By the seventh follow-up visit, 75 women remained pre- or early perimenopausal, 41 women were late perimenopausal, and 195 women (approximately 45% of the sample at follow-up visit 7) were postmenopausal.
At baseline, 24.5% of the women with a lifetime history or current diagnosis of depression also had the metabolic syndrome, whereas 18.3% of women with no lifetime history/no current diagnosis of depression had the metabolic syndrome. This difference was not statistically significant (univariate χ2 = 2.28, p > .10). As presented in Table 1, there were no significant differences between women with and without a lifetime history/current diagnosis of depression in the baseline frequency of any of the individual components of the metabolic syndrome (all p > .05).
In women who were free of the metabolic syndrome at baseline, GEE models adjusted for age and race showed that a lifetime history or current major depressive episode at baseline was associated with 1.82 (95% CI = 1.06–3.14) times the odds of developing the metabolic syndrome during the follow-up period; this effect was statistically significant. Similarly, survival analysis demonstrated a nonsignificant trend, which showed that women with a lifetime history or current major depressive episode at baseline had a 66% greater risk of developing the metabolic syndrome (hazard ratio [HR] = 1.66; 95% CI = 0.99–3.75) compared with women with no baseline depression, adjusted for age and race (Figure 1). Findings were unchanged when menopausal status (time-varying) was added to the model (results not shown). There was no significant interaction between race and depression, indicating that race did not affect the association between depression and the metabolic syndrome.
One potential explanation for the above findings is that individuals with depression may engage in behaviors that would promote the metabolic syndrome (e.g., smoking, poor diet, physical inactivity). However, none of the findings of the GEE and survival models was accounted for by smoking or daily caloric intake. When physical activity was added to the GEE and survival models, the association of depression with the metabolic syndrome remained significant and was slightly stronger (e.g., GEE model: OR = 1.79; 95% CI = 1.04–3.09). To determine whether baseline diabetes influenced the effect of depression on risk of developing the metabolic syndrome, women with a baseline diagnosis of diabetes (n = 17) were removed from the incident GEE and survival models. The effect of depression on the metabolic syndrome remained significant and became slightly stronger (results not shown). Finally, the GEE and survival analyses were conducted with the 17 women who reported a history or current use of antidepressants (without meeting the criteria for major depression) added to the baseline depression group; results of these models were largely unchanged.
Because individuals with major depression often have an anxiety or substance use disorder, we repeated the GEE and survival analyses adjusting for lifetime history of anxiety and alcohol use disorders. The latter included both abuse and dependence. We did not examine drug use disorders separately because only 14 women reported history of drug use without alcohol use. For the GEE analyses, inclusion of any anxiety disorder (n = 61) into the model resulted in significantly increased odds of the metabolic syndrome associated with depression (OR = 1.73; 95% CI = 1.00–3.00). The effect of anxiety disorders on odds of developing the metabolic syndrome was not significant (OR = 1.65; 95% CI = 0.85–3.18). Inclusion of alcohol use disorders (n = 40) into the model resulted in a nonsignificant trend for increased odds of the metabolic syndrome associated with depression (OR = 1.61; 95% CI = 0.92–2.81), and significantly greater odds of the metabolic syndrome associated with alcohol use disorders (OR = 2.68; 95% CI = 1.33–5.37).
For the survival analyses, inclusion of any anxiety disorder (n = 56) into the model resulted in a significant HR = 1.52 (95% CI = l.94–3.49) associated with depression and a non-significant HR = 1.68 (95% CI = 0.94–4.25) for anxiety disorders. Inclusion of alcohol use disorders (n = 35) into the model resulted in a nonsignificant trend for increased risk of developing the metabolic syndrome associated with depression (HR = 1.54; 95% CI = 0.93–3.40), and significantly greater risk of developing the metabolic syndrome associated with alcohol use disorders (HR = 2.09; 95% CI = 1.06–6.06). There were no significant interactions between depression and anxiety or alcohol disorders, indicating the effects of comorbid conditions on metabolic syndrome risk were additive.
Results of the post hoc exploratory survival analyses examining differences between women with no depression, a single depressive episode, and recurrent depression at baseline in risk of developing the metabolic syndrome showed that, compared with women with no lifetime history/no current depression at baseline, women with a lifetime history of recurrent depression had a marginally significant increase in risk of developing the metabolic syndrome (HR = 1.83; 95% CI = 0.99–4.76). A single episode of depression, compared with no lifetime history/no current depression, was not associated with significantly increased risk of developing the metabolic syndrome (HR = 1.47; 95% CI = 0.79–3.96).
Results of the post hoc exploratory survival analyses examining depression as a predictor of risk for each component of the metabolic syndrome were not significant. Specifically, the HRs and 95% CIs for each component were as follows: hypertension, HR = 1.18 (0.80–2.16); low HDL-C, HR = 1.08 (0.71–2.16); high WC, HR = 1.47 (0.94–2.89); high triglycerides, HR = 1.11 (0.78–1.83); high fasting glucose, HR = 1.22 (0.75–2.92). This suggests that a lifetime history or current episode of major depression at baseline was not associated with the metabolic syndrome because of a single risk factor.
Results of the post hoc exploratory GEE and survival analyses were not significant, suggesting that the metabolic syndrome was not associated with the development of a major depressive episode in women with no lifetime history/no current major depression at baseline (data not shown).
This study is the first to show that major depression predicts increased risk for developing the metabolic syndrome during a 7-year follow-up period in a sample of White and African-American middle-aged women. Results also suggest that women with a lifetime history of recurrent depression may have the greatest risk of developing the metabolic syndrome. Our findings are consistent with previous longitudinal studies documenting an association of depressive symptoms with increased risk for developing the metabolic syndrome in women (15,16), as well as the cross-sectional evidence for an association between major depression and prevalence of the metabolic syndrome in women (7). Although each of the aforementioned studies provides indirect support for a link between depression and risk of the metabolic syndrome, conclusions drawn from these studies are tempered by cross-sectional data (7), self-reports (9), or the measurement of depressive symptoms as opposed to clinical depression (15,16). The current investigation extends prior findings by using a well-validated, standardized interview measure of clinical depression and demonstrating consistent effects in African-American and White women. In addition, the availability of multiple measurements for both depression and the metabolic syndrome permitted investigation of the direction of this association. Results of exploratory analyses suggest that the metabolic syndrome may not be a risk factor for the onset of major depression in this sample of middle-aged women. Finally, the current findings suggest that a lifetime history of depression is associated with increased risk of the metabolic syndrome as a unified construct, but not its individual components.
Several notable behavioral and pathophysiological factors may link depression with the metabolic syndrome. First, poor diet and sleep, physical inactivity, and smoking are all more common among individuals with depression (33–36), and emerging evidence suggests that these behaviors are also associated with the metabolic syndrome (37–39). In the current investigation, the association between depression and the metabolic syndrome was not attenuated by daily caloric intake, physical activity, or smoking. However, with the exception of smoking, we examined only baseline measures of these lifestyle variables as diet and physical activity are notoriously difficult to measure. Second, a history of alcohol abuse or dependence may be an important link as well. It was a significant predictor of the metabolic syndrome and attenuated the association of major depression and the metabolic syndrome in the survival model and GEE models of women with no metabolic syndrome at baseline. Consistent with data showing a positive association between a lifetime history of heavy alcohol use and the metabolic syndrome (40–42), results of the current study suggest that excessive alcohol consumption, independent of depression, may be an important factor in the development of the metabolic syndrome in women. Excessive alcohol consumption has also been linked to components of the metabolic syndrome, including hypertension (43) and increased central (44,45) and visceral (41,46) adiposity.
Plausible physiological mechanisms linking depression with the metabolic syndrome include dysregulation of the hypothalamic-pituitary adrenal (HPA) axis and autonomic nervous system (ANS) via such pathways as an accumulation of visceral adiposity and impaired insulin sensitivity (47), as well as aberrant serotonergic functioning (48). Although anxiety disorders are also associated with HPA dysregulation, they are characterized by hypocortisolemia and increased glucocorticoid receptors, as opposed to the hypercortisolemia associated with depression and the metabolic syndrome (49). Similar mechanisms may link heavy alcohol use with the metabolic syndrome, as heavy drinkers have been shown to exhibit dysregulation of the HPA axis and ANS (50).
Given our data and the recent report that depressive symptoms and the metabolic syndrome predict clinical events in women with suspected CAD (9), it is possible that the metabolic syndrome may be one pathway linking depression with disease in women. Further research is needed to better understand the underlying biobehavioral mechanisms by which depression may influence risk for the metabolic syndrome over time, and investigate whether the metabolic syndrome is a mechanism linking depression with CAD and diabetes. In addition, future studies of the effect of psychological and behavioral interventions for depression on the development or remediation of the metabolic syndrome are warranted.
The current investigation has several limitations. First, it only evaluates the role of depression in middle-aged women and not in men. Some evidence suggests that the findings for depression may be stronger in women than in men (7); thus, these findings should not be generalized to men. Second, it should be kept in mind that the sample has a higher rate of major depression than anticipated, possibly because of the close and confidential relationship between participants and interviewers. Third, because of a relatively small number of incident cases of metabolic syndrome, the current study could not fully evaluate the role of comorbid conditions.
On the other hand, this investigation is the first to show that major depression is a significant predictor of the prevalence and onset of the metabolic syndrome in middle-aged women. Depression is one of the most prevalent psychiatric disorders in women, with lifetime prevalence rates in middle-aged women estimated at approximately 22% (51). In addition to impairment in mental health and social functioning, depression has significant physical health consequences and increases a woman’s risk for CAD and Type 2 diabetes, two diseases for which the metabolic syndrome is a putative risk factor. Questions regarding a woman’s history or current experience of depression may be a useful way to help identify women who are at high risk for developing the metabolic syndrome and, subsequently, aid in early prevention or treatment of CAD and diabetes. In addition, psychological and behavioral interventions targeting depression in women may be a valuable tool for the prevention and treatment of the metabolic syndrome. A better understanding of the relationship between depression and the metabolic syndrome as well as underlying mechanisms may have substantial implications for the development of interventions designed to reduce the risk of disease in women.
The authors thank the study staff and all of the women who participated in SWAN.
The Study of Women’s Health Across the Nation (SWAN) has grant support from the National Institutes of Health (NIH), Department of Health and Human Services (DHHS), through the National Institute on Aging (NIA), the National Institute of Nursing Research (NINR) and the NIH Office of Research on Women’s Health (ORWH) (Grants NR004061, AG012505, AG012535, AG012531, AG012539, AG012546, AG012553, AG012554, and AG012495). The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the NIA, NINR, ORWH, or the NIH.