The Multi-Ethnic Study of Atherosclerosis is a multicenter, longitudinal cohort study of the occurrence and correlates of subclinical cardiovascular disease and its progression.29
Because an additional objective of the study was to assess ethnic differences in subclinical disease prevalence, risk of progression, and rates of clinical cardiovascular disease, participants were recruited from 4 prespecified racial/ethnic groups.29
Between July 2000 and August 2002, 6814 men and women aged 45 through 84 years who identified themselves as white, black, Hispanic, or Chinese and who were free of self-reported clinical cardiovascular disease were recruited from 6 US communities: Baltimore City and Baltimore County, Maryland; Chicago, Illinois; Forsyth County, North Carolina; Los Angeles County, California; Northern Manhattan and the Bronx, New York; and St Paul, Minnesota. Additional details about the design and objectives of the study have been previously published.29
The first examination took place between 2000 and 2002. The second examination occurred between 2002 and 2004, and the third examination occurred between 2004 and 2005. Written informed consent was obtained from participants and the study was approved by institutional review boards of each institution. Participants were notified by letter of all laboratory results, including glucose values, with a copy to their physician if they requested it. Because the Center for Epidemiologic Studies Depression Scale (CES-D) is not designed to ascertain clinical depression, participants were not notified about elevated symptoms.
Assessment of Depressive Symptoms
Depressive symptoms were assessed at visits 1 and 3 using the CES-D, a 20- item questionnaire developed to assess depressive symptoms in community populations.30
The CES-D items represent the major components of depression and include depressed mood, feelings of worthlessness, feelings of hopelessness, loss of appetite, poor concentration, and sleep disturbance. The Cronbach α for its reliability ranges between 0.84 and 0.93.31
Participants were asked to rate each item on a scale from 0 to 3 based on “how often you have felt this way during the past week.” Scores range from 0 to 60, with higher scores indicating more severe depressive symptoms. We recognize that the CES-D assesses self-reported depressive symptoms and not clinical depression. For the purpose of our analyses, elevated depressive symptoms were defined by a CES-D score of 16 or higher, consistent with mild to moderate depression or dysthymia,32
self-reported use of antidepressant medications (tricyclics, nontricyclics, and monoamine oxidate inhibitors), or both. Participants with a CES-D score of 16 or higher or who were taking antidepressant medications at visit 3 who did not meet these criteria at visit 1 were considered to have incident elevated depressive symptoms, as previously defined by our group.20
The CES-D was administered in English, Spanish, Cantonese, and Mandarin. The reliability of the CES-D is comparable with European American, African American, Mexican American, and Chinese American groups.33,34
Assessment of Diabetes Status
Impaired fasting glucose and type 2 diabetes status were determined at each visit. Participants were asked to fast for 12 hours and to avoid smoking and heavy physical activity for 2 hours before each examination. Fasting blood samples were drawn by venipuncture from an antecubital vein into vacuum tubes between 7:30 AM and 10:30 AM. Serum samples were frozen and stored at −70°C. Details of serum sampling and processing have been described previously.29
Impaired fasting glucose (100 to 125 mg/dL) and type 2 diabetes (fasting glucose, ≥126 mg/dL; or use of oral hypoglycemic medication, insulin, or both) were defined according to the 2003 American Diabetes Association criteria. 35
Those with diabetes were further subdivided into participants who were untreated or treated. Incident diabetes was defined among participants who did not have diabetes at baseline but developed diabetes at subsequent visits. The date of incident diabetes was estimated at one-half the interval between the last known date without diabetes and the examination at which it was diagnosed.
Covariates were assessed at baseline examination using standard protocols as previously described.20,29
Sex, age, race/ ethnicity, years of education, cigarette smoking history, and annual income were self-reported. Prescription and overthe- counter medications were determined by transcription of medications brought into clinic. Weight and height were measured using a balance beam scale and a stadiometer, respectively, with participants wearing light clothing and no shoes. Height was recorded to the nearest 0.5 cm and weight to the nearest 0.5 lb. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. All anthropometric measures were taken in duplicate and averaged. Blood pressure and lipid levels were measured and categorized using standard procedures and current recommendations.20,36-38
Interleukin 6 (IL-6) and C-reactive protein (CRP) were measured using standard techniques previously described. 20
A spot urine sample was collected from each participant, preferably in the early morning at the beginning of the clinic visit. Urinary creatinine and albumin were measured using previously described techniques.20
Participants' usual diet was characterized using a 120-item food frequency questionnaire, modified from the validated Insulin Resistance Atherosclerosis Study in which comparable validity was observed for non-Hispanic white, African American, and Hispanic participants.39,40
The dietary assessment was modified to include foods typically eaten by Chinese groups. We used total daily caloric intake as a summary variable of diet. Physical activity was assessed using the 28-item Typical Week Physical Activity Survey.29,41
Wesummarized physical activity as the metabolic equivalent task of minutes per week spent in moderate to vigorous household, outdoor, sporting, conditioning, and volunteer activities.
Because the 2 objectives of our study required different population samples, we describe the analysis samples and procedures separately below. However, for both sets of analyses, we began by comparing the distribution of baseline characteristics by depressive symptom status (analysis 1) or type 2 diabetes status (analysis 2) using t tests for continuous variables and χ2 tests for categorical variables. For the nonnormally distributed continuous variables, comparisons were made using the Wilcoxon rank sum test for the 2-group comparison (analysis 1) or the Kruskal-Wallis test for the 4-group comparison (analysis 2). Prior to modeling, we tested for effect modification by creating interaction terms between categorically elevated depressive symptoms and covariates of interest (ie, age, sex, race/ethnicity, BMI) or between type 2 diabetes status and covariates of interest. Although there was a suggestion that the association of depressive symptoms with incident type 2 diabetes was stronger for whites and Chinese Americans, the direction of association was similar for all ethnicities. Because no interactions were found for either analysis, pooled analyses are presented. Incidence rates for type 2 diabetes were calculated using a Poisson regression person- years approach (analysis 1). The cumulative incidence of elevated depressive symptoms is presented for analysis 2. A 2-sided P value <.05 was used to determine statistical significance. Statistical analyses were performed using Stata version 8.2 (Stata Corp, College Station, Texas).
Depressive Symptoms and Incident Diabetes (Analysis 1)
Participants were excluded if they had prevalent untreated or treated type 2 diabetes at the first visit (n=1209); were missing data on type 2 diabetes status at any of the 3 visits, or on the CES-D score at baseline (n=201), or on covariates of interest (n=203), leaving 5201 participants for analysis. Incidence rates of type 2 diabetes were calculated for participants with elevated depressive symptoms or who were taking antidepressant medication and compared with incidence rates among those without elevated depressive symptoms. The relative hazard (RH) of developing type 2 diabetes in participants with elevated depressive symptoms compared with those without was calculated using Cox proportional hazards regression models. Additionally, we modeled CES-D as a continuous variable and reported theRHper 5-point increase in CES-D score.
To explore mechanisms explaining the relation between elevated depressive symptoms and type 2 diabetes, we used the following strategies: model 1 included terms for age, race/ethnicity, sex, and examination site. Model 2 included the terms in model 1 plus baseline BMI. Models 3 through 6 included the terms in model 2 plus the following terms: model 3, metabolic variables (lipid levels, blood pressure, fasting insulin concentration); model 4, inflammatory markers (IL-6 and CRP levels); model 5, socioeconomic variables (educational achievement, annual household income); and model 6, lifestyle variables (smoking history, daily caloric intake, alcohol use, physical activity level).
Diabetes and Incident Depressive Symptoms (Analysis 2)
Participants were excluded if they had prevalent depressive symptoms (CESD≥16, antidepressant medication use, or both) at baseline (n=1022), if they were missing data on CES-D score at baseline and visit 3 or on type 2 diabetes status at baseline (n=908), or on covariates of interest (n=37), leaving 4847 who were included in this analysis. The cumulative incidence of elevated depressive symptoms was calculated for participants by baseline fasting glucose status. Because incident depressive symptoms were only assessed at 1 follow-up visit, we used logistic regression to calculate the odds ratio (OR) of developing elevated depressive symptoms in participants with impaired fasting glucose and untreated and treated type 2 diabetes compared with participants who had normal fasting glucose at baseline. We again used a series of multivariate models to investigate mechanisms explaining the relation between type 2 diabetes status and elevated depressive symptoms. The base model was the same as that described above. Subsequent models included terms for BMI, socioeconomic status, lifestyle variables, and markers of diabetes severity (dyslipidemia [triglycerides ≥200 mg/dL, high-density lipoprotein <40 mg/dL, or both], presence of hypertension [blood pressure ≥140/90 mm Hg or antihypertensive medication use], and microalbuminuria [urinary albumin:creatinine ratio ≥30 mg/g]).42
We selected these factors a priori because of their previously reported association with the development of depression in individuals with preexisting type 2 diabetes.
To convert triglycerides to millimoles per liter, multiply by 0.0113; high-density lipoprotein to millimoles per liter, multiply by 0.0259; and glucose to millimoles per liter, multiply by 0.0555.