Social Science & Medicine 138 (2015) 74e81

Contents lists available at ScienceDirect

Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed

Depressive symptoms and self-reported adherence to medical recommendations to prevent cardiovascular disease: NHANES 2005e2010 Jessica Berntson a, Kendra Ray Stewart b, Elizabeth Vrany a, Tasneem Khambaty a, Jesse C. Stewart a, * a b

Department of Psychology, Indiana UniversityePurdue University Indianapolis, Indianapolis, IN, USA OurHealth, Indianapolis, IN, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 30 May 2015

This study's aim was to examine the relationships between depressive symptom severity and adherence to medication and lifestyle recommendations intended to prevent cardiovascular disease (CVD) in a large, diverse sample of men and women representative of the U.S. population. Participants were adults from the National Health and Nutrition Examination Survey (NHANES) 2005e2010 with a self-reported history of hypertension and/or hypercholesterolemia, but no CVD. The Patient Health Questionnaire-9 (PHQ-9) was used to assess depressive symptoms, and the Blood Pressure and Cholesterol interview was used to assess self-reported adherence to five medical recommendations: take antihypertensive medication (n ¼ 3313), eat fewer high fat/cholesterol foods (n ¼ 2924), control/lose weight (n ¼ 2177), increase physical activity (n ¼ 2540), and take cholesterol medication (n ¼ 2266). Logistic regression models (adjusted for demographics, diabetes, body mass index, smoking, and alcohol intake) revealed that a 1-SD increase in PHQ-9 score was associated with a 14% lower odds of adherence to the control/ lose weight recommendation (OR ¼ 0.86, 95% CI: 0.75e0.98, p ¼ .02) and a 25% lower odds of adherence to the increase physical activity recommendation (OR ¼ 0.75, 95% CI: 0.65e0.86, p < .001). PHQ-9 score, however, was not related to the odds of adherence to the take antihypertensive medication (p ¼ .21), eat fewer high fat/cholesterol foods (p ¼ .40), or take cholesterol medication (p ¼ .90) recommendations. Our findings suggest that poor adherence to provider recommendations to control/lose weight and to increase physical activity may partially explain the excess risk of CVD among depressed persons. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Depressive symptoms Adherence Cardiovascular disease NHANES

It is well established that adults with depressive disorders or elevated depressive symptoms are at increased risk of atherosclerotic cardiovascular disease (CVD), including coronary artery disease and cerebrovascular disease (Van der Kooy et al., 2007). Although the mechanisms underlying this relationship have yet to be elucidated, multiple candidate mediators have been proposed, including increased smoking, autonomic dysfunction, systemic inflammation, and platelet hyperactivation (Carney et al., 2005; Joynt et al., 2003; Kop and Gottdiener, 2005; Nemeroff and Musselman, 2000; Patten et al., 2009). Another potential mechanism that has received somewhat less attention is nonadherence to

* Corresponding author. Department of Psychology, Indiana UniversityePurdue University Indianapolis, 402 N. Blackford St., LD 100E, Indianapolis, IN 46202, USA. E-mail address: [email protected] (J.C. Stewart). http://dx.doi.org/10.1016/j.socscimed.2015.05.041 0277-9536/© 2015 Elsevier Ltd. All rights reserved.

medication and lifestyle recommendations intended to prevent CVD (McConnell et al., 2005). Examples of these medical recommendations include taking antihypertensive or cholesterol medications, increasing physical activity, improving diet, and losing weight (Go et al., 2013). A meta-analysis by DiMatteo and colleagues (Dimatteo et al., 2000) found that, across multiple diseases, the likelihood of nonadherence to medical recommendations is three times higher for depressed versus nondepressed patients. A more recent metaanalysis by Grenard and colleagues that specifically examined medication adherence in patients with chronic diseases found that depressed adults had a 76% greater odds of being non-adherent than nondepressed adults (Grenard et al., 2011). Among studies pertaining to CVD, Ziegelstein and colleagues (Ziegelstein et al., 2000) found an association between depression after a myocardial infarction and reduced adherence to multiple

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

recommendations, including taking prescribed medications, increasing exercise, improving diet, reducing stress, and increasing social support. In a study of heart failure patients, those with more depressive symptoms had lower overall compliance scores and were less compliant with exercise recommendations than patients with fewer symptoms (van der Wal et al., 2006). Similar relationships have been observed between depressive symptoms and medication nonadherence in samples of cardiovascular inpatients (Dempe et al., 2013; Kronish et al., 2013) and outpatients (Bane et al., 2006). In addition to CVD patients, the relationship between depression and adherence has been examined in adults with diabetes or obesity e two conditions that have a high comorbidity and increase risk of CVD. Meta-analyses of the diabetes literature indicate that depression is related to poorer adherence to diabetes treatment regimens, including diet, physical activity, medication, and other self-care recommendations (Gonzalez et al., 2008; Sumlin et al., 2015). Regarding adult obesity treatment programs, evidence suggests depression predicts higher attrition rates (Moroshko et al., 2011) but not weight loss (Teixeira et al., 2005). More recently, attention has also been given to investigating the depressionemedical adherence relationship in adults at elevated risk but free of clinical CVD. There is a current need for research in this area, as (a) evidence indicates that adherence to primary prevention efforts is poorer than to secondary prevention efforts (Naderi et al., 2012), and (b) findings from this line of research could suggest avenues (e.g., adherence interventions) for reducing the excess CVD risk of depressed individuals. In Grenard et al.'s (2011) meta-analysis, it was found that, across eight studies, depressed adults with hypertension or hyperlipidemia have a 73% greater odds of exhibiting medication nonadherence than nondepressed adults with hypertension or hyperlipidemia. A study that examined emotional well-being, rather than depressive symptoms, yielded mixed results; lower emotional well-being was associated with poorer adherence to dietary and exercise recommendations, but not medication recommendations, among hypertensive patients (Trivedi et al., 2008). Notably, most studies in this literature utilized moderately sized samples and have examined adherence to one medical recommendation only. Consequently, it is unknown whether the depressionemedical adherence relationship is present in the general U.S. population and whether depressive symptoms are associated with nonadherence to some medication and lifestyle recommendations but not others. The objective of the present study was to examine the relationships between depressive symptom severity and selfreported adherence to multiple medication and lifestyle recommendations intended to prevent CVD in a large, diverse sample of American men and women with self-reported hypertension and/or hypercholesterolemia. Data from the 2005e2010 waves of the National Health and Nutrition Examination Survey (NHANES), a survey of a large probability sample representative of the U.S. population, were analyzed. The NHANES data provided a good opportunity to examine these relationships, given that the survey includes a validated depressive symptom scale and interview questions assessing adherence to five medication and lifestyle recommendations for the management of hypertension and hypercholesterolemia, both of which are major risk factors for CVD (Go et al., 2013). 1. Methods 1.1. Study design and sample Data for this study were obtained from the publicly available data files for 2005e2010 NHANES survey years. The NHANES

75

program of studies is conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention to assess the health and nutritional status of adults and children in the U.S. The survey employs a stratified, multistage, probability sampling design to obtain a nationally representative sample of the civilian, non-institutionalized U.S. population. Detailed descriptions of the survey design and procedures are available at the study website (www.cdc.gov/nchs/nhanes.htm). Briefly, about 5000 people were recruited each survey year, and non-Hispanic Blacks and Hispanics were oversampled to ensure accurate estimates. Individuals who were selected and agreed to participate first completed a computer-assisted interview conducted by trained personnel in their homes. Approximately 1e2 weeks after the household interview, participants were invited to attend a session at a Mobile Examination Center (MEC) to undergo examinations and laboratory assessments and to complete additional interviews. This archival study of the NHANES data was approved by the Indiana University institutional review board. Fig. 1 presents a flowchart illustrating the establishment of the five overlapping subsamples. From the total sample for the 2005e2010 survey years (N ¼ 31,034), respondents aged 18 years and older were selected (n ¼ 18,318), of whom 15,830 answered 8 of the Patient Health Questionnaire-9 (PHQ-9) items and had NHANES Blood Pressure and Cholesterol interview data. Because our focus is the association of depressive symptoms with adherence to medical recommendations intended to prevent CVD, the 1619 adults who reported a history of CVD (coronary heart disease, angina, myocardial infarction, stroke, or congestive heart failure) were also excluded. We also excluded the 1044 respondents with missing data for smoking, leaving a sample of 13,167 adults. From this sample, we created the following overlapping subsamples (see Self-Reported Medical Adherence section): antihypertensive medication (HTN Med; n ¼ 3313), cholesterol medication (Chol Med; n ¼ 2266), Diet (n ¼ 2924), Weight (n ¼ 2177), and Activity (n ¼ 2540). All five subsamples combined consisted of 5066 unique patients. Briefly, the HTN Med subsample consisted of respondents with a history of hypertension and of being told to take antihypertensive medication, and the other subsamples consisted of respondents with a history of hypercholesterolemia and of being told to take cholesterol medication, manage their weight, change their diet, or increase their activity level. Respondent characteristics for these subsamples appear in Table 1. 1.2. Measures 1.2.1. Depressive symptoms To assess depressive symptom severity, the PHQ-9 (Kroenke et al., 2001) was administered during the MEC examination. Using a 0e3 scale, respondents reported the frequency with which they had experienced the nine symptoms of major depressive disorder during the past two weeks. Total scores range from 0 to 27, with scores 10 being indicative of clinically significant depressive symptoms (Kroenke and Spitzet, 2002). The PHQ-9 is a reliable and valid questionnaire in community samples, as evidenced by its high internal consistency and good sensitivity and specificity for identifying cases of major depressive disorder (Kroenke and Spitzet, 2002; Kroenke et al., 2001; Manea et al., 2012; Patten and Schopflocher, 2009; Wittkampf et al., 2007). 1.2.2. Self-reported medical adherence Data from the NHANES blood pressure and cholesterol interview (National Center for Health Statistics, 2009) administered during the household interview were used to compute an adherence variable for each of the five overlapping subsamples. During the blood pressure (BP) segment (see Fig. 1, Path A), respondents

76

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

Fig. 1. Selection method used to establish five overlapping subsamples. BPQ ¼ Blood Pressure Questionnaire. PHQ-9 ¼ Patient Health Questionnaire-9. BP-Q ¼ blood pressure question. Chol-Q ¼ cholesterol question. HTN Meds ¼ antihypertensive medication. Chol Meds ¼ cholesterol medication.

were asked a series of questions: (BP-Q1) “Have you ever been told by a doctor or other health professional that you had hypertension, also called high blood pressure?”, (BP-Q2) “Because of your hypertension/high blood pressure, have you ever been told to take prescribed medication?”, and (BP-Q3) “Are you now taking a

prescribed medicine?” From the original sample of 13,167 adults, we excluded respondents who did not provide a “Yes” response to BP-Q1 (“No”: n ¼ 9219, “Don't Know”: n ¼ 21) and BP-Q2 (“No”: n ¼ 610, “Don't Know”: n ¼ 3) because those without “Yes” responses to both questions were not asked BP-Q3. We also excluded

Table 1 Characteristics of respondents in the five overlapping subsamples. Characteristic

HTN meds subsample (n ¼ 3313)

Diet subsample (n ¼ 2924)

Weight subsample (n ¼ 2177)

Activity subsample (n ¼ 2540)

Chol meds subsample (n ¼ 2266)

Age, years Female, % Race/Ethnicity, % Non-Hispanic White Non-Hispanic Black Mexican American Other Education Level, % Less than 9th grade 9the12th grade (no diploma) High School Graduate/GED Some College or Associates Degree College Graduate or Above Diabetes, % Body Mass Index, kg/m2 Current smoker, % Number of drinks per day PHQ-9 total (possible range: 0e27) PHQ-9 total  10, % Adhering to medical recommendation, %

60.7 (13.8) 56

56.5 (14.4) 54

55.8 (13.9) 54

56.1 (14.1) 54

60.9 (12.9) 53

50 27 13 10

51 20 16 13

48 21 17 14

49 21 17 13

53 19 16 12

13 17 25 26 19 23 31.3 (7.3) 16 0.1 (0.2) 3.3 (4.4) 10 87

12 14 24 27 23 20 30.2 (6.3) 17 0.1 (0.2) 3.2 (4.2) 9 83

12 15 23 27 23 24 32.0 (6.4) 16 0.1 (0.2) 3.5 (4.5) 10 84

12 14 23 28 23 22 31.0 (6.4) 16 0.1 (0.2) 3.3 (4.3) 9 77

14 15 26 25 20 27 30.5 (6.4) 16 0.1 (0.2) 3.2 (4.4) 10 81

Note. Continuous variables are presented as mean (standard deviation), and categorical variables are presented as percentage. HTN Meds ¼ antihypertensive medication. Chol Meds ¼ cholesterol medication. PHQ-9 ¼ Patient Health Questionnaire-9.

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

the one respondent who gave a “Don't Know” response to BP-Q3, leaving 3313 adults in the HTN Med subsample. BP-Q3 served as our measure of antihypertensive medication adherence, with respondents who answered “Yes” coded as adherent (1) and those who answered “No” coded as non-adherent (0). During the cholesterol interview segment (see Fig. 1, Path B), respondents were asked a series of questions: (Chol-Q1) “Have you ever had your blood cholesterol checked?”, (Chol-Q2) “Have you ever been told by a doctor or other health professional that your blood cholesterol level was high?”, (Chol-Q3) “To lower your blood cholesterol, have you ever been told by a doctor or other health professional: (a) to eat fewer high fat or high cholesterol foods? (b) to control your weight or lose weight? (c) to increase your physical activity or exercise? or (d) to take prescribed medicine?”, and (Chol-Q4) “Are you now following this advice to [selected aed recommendation]?” From the original sample of 13,167 adults, we excluded respondents who did not provide a “Yes” response to Chol-Q1 (“No”: n ¼ 3,844, “Refused”: n ¼ 2, “Don't Know”: n ¼ 408) and Chol-Q2 (“No”: n ¼ 5,220, “Don't Know”: n ¼ 52) because those without “Yes” responses to both questions were not asked Chol-Q3 and Chol-Q4. Of the remaining 3641 respondents, those who did not provide a “Yes” response to Chol-Q3 part a (“No”: n ¼ 705, “Don't Know”: n ¼ 12), b (“No”: n ¼ 1,453, “Don't Know”: n ¼ 11), c (“No”: n ¼ 1,094, “Don't Know”: n ¼ 7), or d (“No”: n ¼ 1,369, “Don't Know”: n ¼ 4) were not asked further cholesterol questions and were excluded. Finally, we excluded the two respondents who gave a “Don't Know” response to Chol-Q 4-d, leaving 2924 adults in the Diet subsample, 2177 adults in the Weight subsample, 2540 in the Activity subsample, and 2266 in the Chol Med subsample. Chol-Q4 parts a, b, c and d served as our measures of diet, weight, physical activity, and cholesterol medication adherence, respectively, with respondents who answered “Yes” coded as adherent (1) and those who answered “No” coded as non-adherent (0). To establish concurrent validity, we ran logistic regression models (adjusted for age, sex, race/ethnicity and education) examining associations between our adherence variables and appropriate health-related measures. Respondents coded as adherent to the antihypertensive medication recommendation had lower systolic (131 vs. 138 mmHg, p < .001) and diastolic (71 vs. 73 mmHg, p < .001) BP than those coded as non-adherent. Similarly, respondents coded as adherent to cholesterol medication recommendation had lower total cholesterol than those coded as non-adherent (193 vs. 232 mg/dL, p < .001). Because lifestyle changes tend to yield smaller cholesterol reductions than medications (National Cholesterol Education Program (NCEP) 2002), we examined associations of adherence to weight, diet, and activity recommendations with total cholesterol and BMI, a more proximal outcome for the lifestyle changes. Respondents coded as adherent to weight and activity recommendations did not have lower total cholesterol than those coded as non-adherent (Weight subsample: 207 vs 211 mg/dL, p ¼ .11; Activity subsample: 209 vs. 208 mg/dL, p ¼ .74); however, they did have lower BMI (Weight subsample: 31.7 vs. 33.1 kg/m2, p < .001; Activity subsample: 30.8 vs. 31.7 kg/ m2, p < .001). For adherence to diet recommendations, differences between adherent and non-adherent respondents in total cholesterol (209 vs. 213 mg/dL, p ¼ .069) and BMI (30.2 vs. 30.7 kg/m2, p ¼ .089) were in the expected direction but fell short of statistical significance. Collectively, these results suggest that our adherence measures are capturing meaningful differences in health behaviors across respondents. Self-report measures are among the most frequently used methods to assess medical adherence (DiMatteo, 2004). Some evidence indicates that self-report measures provide higher adherence estimates than more objective measures (Shi et al., 2010), whereas other evidence suggests the opposite (DiMatteo, 2004).

77

Nonetheless, a recent meta-analysis (Shi et al., 2010) revealed a moderate correlation (pooled r ¼ 0.45; range: 0.24e0.87) between medication adherence measured by self-report and the Medication Events Monitoring System (MEMS). Furthermore, self-reports of adherence to medication and lifestyle recommendations have been associated with a wide range of health outcomes (DiMatteo et al., 2002; Gehi et al., 2007; Wrotniak et al., 2005). Thus, individual differences (i.e., relative standing) on self-report measures of medical adherence reflect clinically relevant differences across people. 1.2.3. Covariates Covariates in the statistical models were: age (years), sex (0 ¼ male, 1 ¼ female), three race/ethnicity dummy variables (RE1: 0 ¼ non-Hispanic White, 1 ¼ non-Hispanic Black, RE2: 0 ¼ nonHispanic White, 1 ¼ Mexican American, RE3: 0 ¼ non-Hispanic White, 1 ¼ Other), education (1 ¼ less than 9th grade, 2 ¼ 9the12th grade (no diploma), 3 ¼ high school grad/GED or equivalent, 4 ¼ some college or associate degree, 5 ¼ college graduate or above), self-reported diabetes (0 ¼ no, 1 ¼ yes), body mass index (BMI; kg/m2), current smoking status (0 ¼ no, 1 ¼ yes), and alcohol consumption (drinks per day). During the household interview, demographic and health-related questions were administered. The primary NHANES race/ethnicity variable has five levels: nonHispanic White, non-Hispanic Black, Mexican American, Other Hispanic, and Other Race including Multi-Racial. Because the Other Hispanic (n ¼ 192e253) and Other Race (n ¼ 80e113) groups were much smaller than the other racial/ethnic groups and because NHANES recommends not combining the Mexican American and Other Hispanic groups (www.cdc.gov/nchs/data/nhanes/ analyticnote_2007-2010.pdf), the Other Hispanic and Other Race groups were collapsed to create an Other group. Then, three race/ ethnicity dummy variables were created comparing the nonHispanic White group to the non-Hispanic Black group (RE1), Mexican American group (RE2), and Other group (RE3). Respondents were asked to report their highest level of education completed and whether they had ever been diagnosed with diabetes by a health professional. Those who reported a past diabetes diagnosis were coded as having self-reported diabetes. We classified those who reported smoking at least 100 cigarettes during their lifetime and indicated that they now smoke cigarettes every day or some days as current smokers. BMI (kg/m2) was computed from height and weight measurements obtained during the MEC physical examination. During the MEC interview, respondents reported the frequency and quantity of consumption of various alcoholic beverages, from which we computed a drinks per day variable. Specifically, we first calculated the number of days drinking alcohol during the past year, which was then multiplied by the average number of drinks consumed per drinking day during the past year to give the total number of drinks consumed. Next, we divided this value by 365 to compute drinks per day during the past year. 1.3. Data analysis 1.3.1. Data cleaning Prior to establishing the subsamples, the following data transformation and imputations were performed. Because the drinks per day variable was positively skewed, we log transformed it to normalize the distribution. For respondents missing one PHQ-9 item (n ¼ 71), we imputed the missing value using the mean of the other eight items for that individual. For diabetes, answers of don't know (n ¼ 9) and “borderline” or “pre-diabetes” responses (n ¼ 221) were coded as “no.” Missing values were imputed with between-subject means for education, rounded closest to the

78

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

nearest whole number (n ¼ 7), and with between-subject, gender stratified means for BMI (n ¼ 26 male; n ¼ 26 female) and alcohol consumption (n ¼ 6 male, n ¼ 8 female). Due to the large number with missing smoking data (see Fig. 1), we decided to exclude these respondents rather than impute their missing values. 1.3.2. Primary analyses To examine the relationships among depressive symptom severity and self-reported adherence to multiple medication and lifestyle recommendations, we ran two logistic regression models e demographic-adjusted and fully-adjusted e in each of the five subsamples. The predictor variable was the continuous PHQ-9 score, and the criterion variables were the five dichotomous adherence variables. Covariates in the demographics-adjusted models were age, sex, the three race/ethnicity dummy variables (RE1, RE2, and RE3), and education. In addition to these covariates, diabetes, BMI, current smoking status, and alcohol consumption were entered in the fully-adjusted models. These potential confounders could produce a spurious relationship between depressive symptoms and medical adherence, given that they all have been found to predict future depression (Klungsøyr et al., 2006; Luppino et al., 2015; Mezuk et al., 2008; Paschall et al., 2005) and could theoretically influence medical adherence. To illustrate the potential clinical relevance of significant effects, we reran the logistic regression models with our depressive symptoms variable dichotomized at an established clinical cut point (nondepressed: PHQ-9 < 10; depressed: PHQ-9  10) (Kroenke and Spitzet, 2002). All estimates from the logistic regressions were weighted using NHANES examination sample weights, which account for the complex survey design, survey nonresponse, and post-stratification (Centers for Disease Control and Prevention (CDC), 2004). By incorporating sample weights, analyses yield estimates representative of the U.S. civilian, non-institutionalized population. We conducted our analyses with SAS statistical software (version 9.3) using the survey logistic regression procedure. 1.3.3. Exploratory analyses We also explored whether depressive symptom-medical adherence relationships were moderated by demographic factors. In these exploratory fully-adjusted models, we tested 25 PHQ9  demographic factor interactions terms [PHQ-9  five demographic factors (age, gender, three race/ethnicity dummy variables) for the five adherence criterion variables]. Only three interactions were significant (p < .05): the RE2 by PHQ-9 interaction (OR ¼ 1.32, 95% CI ¼ 1.01e1.73, p ¼ .043) and RE3 by PHQ-9 interaction (OR ¼ 1.64, 95% CI ¼ 1.20e2.26, p ¼ .002) for the physical activity recommendation and the Age by PHQ-9 interaction (OR ¼ 0.85, 95% CI ¼ 0.75e0.96, p ¼ .009) for the antihypertensive medication recommendation. However, when we applied a Bonferroni correction (Bland and Altman, 1995) to account for the 25 exploratory tests, all three interactions were not significant (Bonferroni corrected critical p ¼ .002). Because it is likely that these interactions are type I errors, we decided not to interpret these interaction results. 2. Results As is shown in Table 1, the mean age across the subsamples ranged from 56 to 61 years, and just over half of the respondents were women. Approximately half of the respondents were racial/ ethnic minorities, and 12e14% had less than 9th grade education. About a quarter of the respondents had diabetes, and about a sixth were current smokers. The mean BMI fell in the range indicative of obesity (BMI  30), and respondents had 0.1 drinks per day on average.

The mean PHQ-9 score fell in the minimal depression range across all five subsamples (see Table 1); however, 9e10% had a score  10, which is indicative of clinically significant depressive symptoms (Kroenke and Spitzet, 2002). As can also be seen in Table 1, self-reported adherence to the five recommendations intended to prevent CVD ranged from 77% (increase physical activity) to 87% (take antihypertensive medication). These adherence rates are higher than those reported in a meta-analysis by DiMatteo (79% for medication, 72% for exercise, and 59% for diet) (DiMatteo, 2004). As is shown in Table 2, results of the demographic-adjusted logistic regression models revealed that depressive symptom severity was negatively related to adherence to the cholesterol management recommendations to control/lose weight (p ¼ .002) and to increase physical activity (p < .001). More specifically, a 1-SD increase in PHQ-9 score (z4.5 points) was associated with a 19% lower odds of adherence to the control/lose weight recommendation and a 28% lower odds of adherence to the increase physical activity recommendation. In contrast, PHQ-9 score was not related to the odds of adherence to BP management recommendation to take antihypertensive medication (p ¼ .19) or the cholesterol management recommendations to eat fewer high fat/high cholesterol foods (p ¼ .13) or to take cholesterol medication (p ¼ .84). The pattern of results was the same in the in fully-adjusted models, which further adjusted for diabetes, BMI, current smoking status, and daily alcohol intake (see Table 2). In these models, depressive symptom severity was negatively related to adherence to the control/lose weight (p ¼ .02) and increase physical activity (p < .001) recommendations but was not related to the take antihypertensive medication (p ¼ .21), eat fewer high fat/high cholesterol foods (p ¼ .40), or take cholesterol medication (p ¼ .90) recommendations. Of the covariates, current smoking status (5/5 models), diabetes (4/5 models), age (3/5 models), race/ethnicity (3/ 5 models), and BMI (3/5 models) were consistently associated with adherence to the medication and lifestyle recommendations (see Table 2). In general, smokers and individuals with higher BMIs were less likely to report being adherent, whereas individuals with diabetes, older adults, and African Americans were more likely to report being adherent. To illustrate the potential clinical relevance of our findings, we reran the logistic regression models for the control/lose weight and increase physical activity recommendations with the dichotomized depressive symptoms variable (nondepressed: PHQ-9 < 10; depressed: PHQ-9  10). For both depressed and nondepressed adults, Fig. 2 displays adherence rates to the control/lose weight recommendation (10% absolute difference) and the increase physical activity recommendation (16% absolute difference). In the Weight subsample, depressed adults were 36% less likely to adhere to the control/lose weight recommendation than nondepressed adults in the demographic-adjusted model (OR ¼ 0.64, 95% CI: 0.44e0.92, p ¼ 0.02) and 28% less likely in the fully-adjusted model (OR ¼ 0.72, 95% CI: 0.49e1.07, p ¼ 0.10), which fell short of significance. In the Activity subsample, depressed adults were 62% less likely to adhere to the increase physical activity recommendation than nondepressed adults in the demographic-adjusted model (OR ¼ 0.38, 95% CI: 0.26e0.56, p < .001) and 58% less likely in the fully-adjusted model (OR ¼ 0.42, 95% CI: 0.29e0.61, p < .001). 3. Discussion Our objective was to examine the relationships between depressive symptoms and self-reported adherence to medication and lifestyle recommendations intended to prevent CVD. In a large, diverse sample of Americans with self-reported hypertension and/ or hypercholesterolemia, we found that depressive symptom

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

79

Table 2 Logistic regression models examining associations of depressive symptoms with self-reported adherence to five medication and lifestyle recommendations intended to prevent cardiovascular disease. Adherence variables

Demographic-adjusted Modelsa

Fully-adjusted Modelsb

Significant covariates (p < .05)

OR

95% CI

OR

95% CI

(OR, 95% CI)

Take antihypertensive medication (n ¼ 3313)

0.92

0.81e1.04

0.93

0.82e1.05

Eat fewer high fat/cholesterol foods (n ¼ 2924)

0.91

0.80e1.03

0.94

0.82e1.08

Control/lose weight (n ¼ 2177)

0.81*

0.71e0.92

0.86*

0.75e0.98

Increase physical activity (n ¼ 2540)

0.72*

0.63e0.83

0.75*

0.65e0.86

Take cholesterol medication (n ¼ 2266)

0.99

0.86e1.13

0.99

0.86e1.14

Age (1.06, 1.05e1.07) RE2 (0.50, 0.33e0.75) Diabetes (2.43, 1.55e3.81) Smoking (0.58, 0.40e0.84) Age (1.03, 1.02e1.04) Sex (1.37, 1.00e1.87) Diabetes (1.75, 1.19e2.58) Smoking (0.54, 0.42e0.69) RE1 (1.56, 1.12e2.16) Diabetes (1.59, 1.03e2.44) BMI (0.95, 0.93e0.97) Smoking (0.50, 0.34e0.75) RE1 (1.49, 1.07e2.07) BMI (0.97, 0.96e0.99) Smoking (0.69, 0.50e0.95) Age (1.05, 1.04e1.06) Diabetes (2.32, 1.48e3.63) BMI (1.03, 1.00e1.06) Smoking (0.64, 0.41e0.99)

Note. Odd ratios (ORs) indicate the change in odds of adherence for every 1-SD increase in PHQ-9 score (z4.5 points). Race/ethnicity was modeled using three dummy variables: RE1: 0 ¼ non-Hispanic White, 1 ¼ non-Hispanic Black, RE2: 0 ¼ non-Hispanic White, 1 ¼ Mexican American, RE3: 0 ¼ non-Hispanic White, 1 ¼ Other. * p < .05. a Adjusted for age, sex, race/ethnicity (3 dummy variables), education, and NHANES sample weights. b Adjusted for age, sex, race/ethnicity (3 dummy variables), education, diabetes, body mass index (BMI), current smoking status, drinks per day, and NHANES sample weights.

severity was inversely related to adherence to the cholesterol management recommendations to control/lose weight and to increase physical activity. Compared to nondepressed adults, depressed adults were 28e36% less likely to adhere to the control/ lose weight recommendation and 58e62% less likely to adhere to the increase physical activity recommendation, depending on whether the models were demographic adjusted (age, gender, and race/ethnicity) or fully adjusted (demographic factors, diabetes, BMI, smoking, and alcohol use). Depressive symptoms, however, were not associated with adherence to the BP management

Fig. 2. Self-reported adherence rates (%) to the control/lose weight and increase physical activity recommendations for depressed verses nondepressed adults. Respondents with a Patient Health Questionnaire-9 (PHQ-9) score  10 were coded as depressed, and those with a PHQ-9 < 10 were coded as nondepressed (Kroenke and Spitzet, 2002).

recommendation to take antihypertensive medication or the cholesterol management recommendations to eat fewer high fat/ high cholesterol foods or to take cholesterol medication. None of the examined relationships varied with age, gender, or race/ ethnicity in exploratory interaction analyses. To our knowledge, this study is the first to examine the depressionemedical adherence relationship in a sample representative of the U.S. population. Moreover, our finding that depressive symptom severity was most strongly associated with poorer adherence to physical activity recommendations may be of particular importance. Existing studies suggest that physical inactivity is among the strongest mediators of the relationship between depressive symptoms and future CVD events (Hamer et al., 2008; Whooley et al., 2008). Therefore, our findings raise the possibility that poor adherence to lifestyle recommendations to prevent CVD, especially physical activity recommendations, may partially explain the more than 50% greater risk of developing CVD of depressed adults (Van der Kooy et al., 2007). The present results extend to the U.S. population prior results which indicate that lower emotional well-being is associated with poorer adherence to exercise recommendations among adults at elevated CVD risk (Trivedi et al., 2008). Although we are not aware of studies examining the link between depressive symptoms and adherence to weight management recommendations in adults at elevated risk but free of clinical CVD, a recent meta-analysis demonstrated that depression is inversely related to weight control over time (Blaine, 2008). Elevated depressive symptoms could have a negative influence on several of the stages that comprise adherence to medical advice (Dunbar-Jacob et al., 2010). For instance, depression-related low self-efficacy and pessimism about the future (Alloy and Ahrens, 1987), possibly including the potential benefits of lifestyle changes, could dissuade a patient from attempting to initiate the recommended behavioral changes. In addition, depression-related anhedonia, fatigue, or psychomotor retardation (American Psychological Association, 2000) could

80

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81

decrease the likelihood of a patient initiating effortful behavioral changes, such as increasing physical activity, even after deciding to make an attempt. It is worth noting that the observed association between depressive symptoms and poorer adherence to the weight management recommendation may, in part, reflect poorer adherence to the physical activity recommendation, as it has been long recommended that individuals alter their diet and increase their activity level to control or lose weight (National Institutes of Health (1998)). The absence of associations between depressive symptom severity and adherence to medication and dietary recommendations conflict with findings from most other studies of adults at elevated CVD risk (Grenard et al., 2011; Trivedi et al., 2008). While our findings raise the possibility that depressive symptoms are more strongly associated with adherence to physical activity and weight management recommendations than medication and dietary recommendations, our null results should not be viewed as definitive evidence of absence due to our adherence assessment method. Although many previous studies used a self-report measure of adherence, the yeseno questions we utilized could not capture the degree of adherence. This, in turn, may have reduced our statistical power, potentially leading to type II errors. Furthermore, our cardiovascular medication adherence rates are considerably higher than those reported in recent meta-analyses, which could have restricted variability in these outcomes and reduced our statistical power (Chowdhury et al., 2013; Naderi et al., 2012). Of relevance, depressive symptoms were related to a numerically, but not statistically significant, lower odds of adherence to antihypertensive and dietary recommendations (see Table 2). Limitations. Although the large and diverse sample and examination of adherence to multiple medical recommendations are study strengths, there are also important limitations. One limitation is that the NHANES data are cross-sectional, which prevents directionality of the associations from being determined. Specifically, adherence to control/lose weight recommendations could result in weight loss, which may then improve depressive symptoms. However, weight loss has not been convincingly related to improvements in depressive symptoms (Fabricatore et al., 2011). Moreover, adherence to the increase physical activity recommendation could result in increased exercise, which has been shown to improve depressive symptoms in clinical trials (Mead et al., 2008). Another limitation is that we could not distinguish between respondents who did not receive medical recommendations from those who did not recall receiving such recommendations. Both of these groups were not administered the adherence items and, thus, had to be excluded from our study. A third limitation is that the NHANES adherence items were self-report, yeseno questions, which did not capture the degree of adherence (e.g., the number of minutes and intensity of physical activity). Consequently, we were unable to determine whether respondents reporting adherence met current guideline recommendations, such as those for physical activity (US Department of Health and Human Services (2008)). In addition, the nature of these questions likely (a) reduced our statistical power (discussed above) and (b) resulted in overreporting of adherence across all recommendations due to socially desirable responding (Sankar et al., 2007). In accordance with the latter notion, the present rates of adherence to cardiovascular medications are higher than those reported in recent meta-analyses (Chowdhury et al., 2013; Naderi et al., 2012). Depressive symptoms have been weakly to moderately correlated with reduced socially desirable responding (Beck et al., 1988; Radloff, 1977). However, it is unlikely that reduced socially desirable responding accounts for our results, as we observed that depressive symptom severity was inversely related to some, but not all, of the recommendations. A final

limitation is some respondents may have been misclassified as being non-adherent to the antihypertensive medication recommendation, given that this adherence item asked respondents who had been previously told to take an antihypertensive medication if they were currently taking it. Some respondents' prescriptions might have been discontinued because they experienced significant side effects or their BP was successfully managed with lifestyle changes alone. This possible misclassification could explain why we did not observe a link between depressive symptoms and antihypertensive medication adherence. Although not as feasible in population-based samples, future studies utilizing quantitative and objective measures of adherence to medication (e.g., MEMS) and lifestyle (e.g., actigraphy) recommendations intended to prevent CVD would likely provide more accurate estimates of adherence rates and may detect relationships that we did not. 4. Conclusions Our findings indicate that American men and women with elevated depressive symptoms report being less adherent to cholesterol management recommendations to control/lose weight and to increase physical activity. Thus, depressed adults appear to be a subpopulation at increased risk for nonadherence to provider lifestyle recommendations, and this poor adherence may be one of the mechanisms by which depression increases CVD risk. These findings highlight the need for (1) longitudinal investigations of depressioneadherence relationships and (2) clinical trials examining whether screening for and addressing patients' depression when providers deliver lifestyle recommendations results in improved adherence rates, cholesterol levels, and ultimately CVD event rates. Addressing depression may be particularly important in the context of primary prevention, as adherence to these recommendations tends to be poorer than to secondary prevention recommendations (Naderi et al., 2012). Recent findings, however, suggest that depression treatment alone is not sufficient to improve medical adherence in patients with depression (Kronish et al., 2012). Moreover, a recent Cochrane review concluded that current interventions to improve medication adherence are of limited effectiveness (Nieuwlaat et al., 2014). Thus, an approach blending evidence-based depression treatments (e.g., antidepressant medication and/or cognitive-behavioral therapy) with novel interventions designed to improve adherence should be evaluated as a potential strategy for addressing the lower medical adherence of depressed patients. Given the size of the depressed population (z9.5% in this study), such a strategy has the potential to reduce CVD morbidity and mortality at the U.S. population level. 5. Funding The authors have no support or funding to report. References Alloy, L.B., Ahrens, A.H., 1987. Depression and pessimism for the future: biased use of statistically relevant information in predictions for self versus others. J. Personal. Soc. Psychol. 52 (2), 366e378. Retrieved from. http://www.ncbi.nlm. nih.gov/pubmed/3559896. American Psychological Association, 2000. Diagnostic and Statistical Manual of Mental Disorders, fourth ed. American Psychiatric Association, Washington, D.C. Bane, C., Hughes, C.M., McElnay, J.C., 2006. The impact of depressive symptoms and psychosocial factors on medication adherence in cardiovascular disease. Patient Educ. Couns. 60 (2), 187e193. http://dx.doi.org/10.1016/j.pec.2005.01.003. Beck, A.T., Steer, R.A., Carbin, M.G., 1988. Psychometric properties of the Beck depression inventory: twenty-five years of evaluation. Clin. Psychol. Rev. 8 (1), 77e100. http://dx.doi.org/10.1016/0272-7358(88)90050-5. Blaine, B., 2008. Does depression cause obesity?: a meta-analysis of longitudinal

J. Berntson et al. / Social Science & Medicine 138 (2015) 74e81 studies of depression and weight control. J. Health Psychol. 13 (8), 1190e1197. http://dx.doi.org/10.1177/1359105308095977. Bland, J.M., Altman, D.G., 1995. Multiple significance tests: the Bonferroni method. Br. Med. J. 310 (6973), 170. http://dx.doi.org/10.1136/bmj.310.6973.170. Carney, R.M., Freedland, K.E., Veith, R.C., 2005. Depression, the autonomic nervous system, and coronary heart disease. Psychosom. Med. 67 (Suppl. 1), S29eS33. http://dx.doi.org/10.1097/01.psy.0000162254.61556.d5. Centers for Disease Control and Prevention (CDC), 2004. NHANES Analytic Guidelines. Chowdhury, R., Khan, H., Heydon, E., Shroufi, A., Fahimi, S., Moore, C., Stricker, B., et al., 2013. Adherence to cardiovascular therapy: a meta-analysis of prevalence and clinical consequences. Eur. Heart J. 34 (38), 2940e2948. http://dx.doi.org/ 10.1093/eurheartj/eht295. €ltner, A., Ladwig, K.-H., Dempe, C., Jünger, J., Hoppe, S., Katzenberger, M.-L., Mo Herzog, W., et al., 2013. Association of anxious and depressive symptoms with medication nonadherence in patients with stable coronary artery disease. J. Psychosom. Res. 74 (2), 122e127. http://dx.doi.org/10.1016/ j.jpsychores.2012.12.003. DiMatteo, M.R., 2004. Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med. Care 42 (3), 200e209. Retrieved from. http://www.ncbi.nlm.nih.gov/pubmed/15076819. Dimatteo, M.R., Lepper, H.S., Croghan, T.W., 2000. Depression is a risk factor for noncompliance with medical treatment: meta-analysis of the effects of anxiety and depression on patient adherence. Arch. Intern Med. 160, 2101e2107. DiMatteo, M.R., Giordani, P.J., Lepper, H.S., Croghan, T.W., 2002. Patient adherence and medical treatment outcomes: a meta-analysis. Med. care 40 (9), 794e811. http://dx.doi.org/10.1097/01.MLR.0000024612.61915.2D. Dunbar-Jacob, J., Houze, M., Kramer, C., Luyster, F., McCall, M., 2010. Adherence to medical advice: process and measurement. In: Steptoe, A. (Ed.), Handbook of Behavioral Medicine: Methods and Applications. Springer, New York, pp. 83e95. Fabricatore, A., Wadden, T., Higginbotham, A., Faulconbridge, L., Nguyen, A., Heymsfield, S., Faith, M., 2011. Intentional weight loss and changes in symptoms of depression: a systematic review and meta-analysis. Int. J. Obes. (2005) 35 (11), 1363e1376. http://dx.doi.org/10.1038/ijo.2011.2. Gehi, A.K., Ali, S., Na, B., Whooley, M.A., 2007. Self-reported medication adherence and cardiovascular events in patients with stable coronary heart disease: the heart and soul study. Arch. Intern. Med. 167 (16), 1798e1803. http://dx.doi.org/ 10.1001/archinte.167.16.1798. Go, A.S., Mozaffarian, D., Roger, V.L., Benjamin, E.J., Berry, J.D., Borden, W.B., Bravata, D.M., et al., 2013. Heart disease and stroke statisticse2013 update: a report from the American heart association. Circulation 127 (1), e6ee245. http://dx.doi.org/10.1161/CIR.0b013e31828124ad. Gonzalez, J.S., Peyrot, M., McCarl, L. a, Collins, E.M., Serpa, L., Mimiaga, M.J., Safren, S. a, 2008. Depression and diabetes treatment nonadherence: a meta-analysis. Diabetes Care 31 (12), 2398e2403. http://dx.doi.org/10.2337/dc08-1341. Grenard, J.L., Munjas, B. a, Adams, J.L., Suttorp, M., Maglione, M., McGlynn, E. a, Gellad, W.F., 2011. Depression and medication adherence in the treatment of chronic diseases in the United States: a meta-analysis. J. General Intern. Med. 26 (10), 1175e1182. http://dx.doi.org/10.1007/s11606-011-1704-y. Hamer, M., Molloy, G.J., Stamatakis, E., 2008. Psychological distress as a risk factor for cardiovascular events: pathophysiological and behavioral mechanisms. J. Am. Coll. Cardiol. 52 (25), 2156e2162. http://dx.doi.org/10.1016/ j.jacc.2008.08.057. Joynt, K.E., Whellan, D.J., O'Connor, C.M., 2003. Depression and cardiovascular disease: mechanisms of interaction. Biol. Psychiatry 54 (3), 248e261. http:// dx.doi.org/10.1016/S0006-3223(03)00568-7. Klungsøyr, O., Nygård, J.F., Sørensen, T., Sandanger, I., 2006. Cigarette smoking and incidence of first depressive episode: an 11-year, population-based follow-up study. Am. J. Epidemiol. 163 (5), 421e432. http://dx.doi.org/10.1093/aje/kwj058. Kop, W.J., Gottdiener, J.S., 2005. The role of immune system parameters in the relationship between depression and coronary artery disease. Psychosom. Med. 67 (Suppl. 1), S37eS41. http://dx.doi.org/10.1097/01.psy.0000162256.18710.4a. Kroenke, K., Spitzet, R.L., 2002. The PHQ-9: a new depression measure. Psychiatr. Ann. 32 (9), 509e521. Kroenke, K., Spitzer, R.L., Williams, J.B.W., 2001. The PHQ-9: validity of a brief depression severity measure. J. Gen. Intern Med. 16, 606e613. Kronish, I.M., Rieckmann, N., Burg, M.M., Edmondson, D., Schwartz, J.E., Davidson, K.W., 2012. The effect of enhanced depression care on adherence to risk-reducing behaviors after acute coronary syndromes: findings from the COPES trial. Am. Heart J. 164 (4), 524e529. http://dx.doi.org/10.1016/ j.ahj.2012.07.024. ntara, C., Davidson, K.W., 2013. The Kronish, I.M., Rieckmann, N., Burg, M.M., Alca psychosocial context impacts medication adherence after acute coronary syndrome. Ann. Behav. Med. a Publ. Soc. Behav. Med. 47 (2), 158e164. http:// dx.doi.org/10.1007/s12160-013-9544-0. Luppino, F.S., de Wit, L.M., Bouvy, P.F., Stijnen, T., Cuijpers, P., Penninx, B.W.J.H., Zitman, F.G., 2015. Overweight, obesity, and depression. Arch. General Psychiatry 67 (3), 220e229. Manea, L., Gilbody, S., McMillan, D., 2012. Optimal cut-off score for diagnosing depression with the patient health questionnaire (PHQ-9): a meta-analysis. CMAJ: Can. Med. Assoc. J. ¼ J. de l'Assoc. Med. Can. 184 (3), E191eE196. http:// dx.doi.org/10.1503/cmaj.112004. McConnell, S., Jacka, F.N., Williams, L.J., Dodd, S., Berk, M., 2005. The relationship between depression and cardiovascular disease. Int. J. Psychiatry Clin. Pract. 9

81

(3), 157e167. http://dx.doi.org/10.1080/13651500510029138. Mead, G.E., Morley, W., Campbell, P., Greig, C.A., McMurdo, M., Lawlor, D.A., 2008. Exercise for depression. Cochrane Database Syst. Rev. 3, CD004366. Mezuk, B., Eaton, W.W., Albrecht, S., Golden, S.H., 2008. Depression and type 2 diabetes over the lifespan: a meta-analysis. Diabetes Care 31 (12), 2383e2390. http://dx.doi.org/10.2337/dc08-0985. Moroshko, I., Brennan, L., O'Brien, P., 2011. Predictors of dropout in weight loss interventions: a systematic review of the literature. Obes. Rev. Off. J. Int. Assoc. Study Obes. 12 (11), 912e934. http://dx.doi.org/10.1111/j.1467789X.2011.00915.x. Naderi, S.H., Bestwick, J.P., Wald, D.S., 2012. Adherence to drugs that prevent cardiovascular disease: meta-analysis on 376,162 patients. Am. J. Med. 125 (9), 882e887. http://dx.doi.org/10.1016/j.amjmed.2011.12.013. National Center for Health Statistics, 2009. Bloop Pressure e BPQ. Retrieved June 26, 2014, from. http://www.cdc.gov/nchs/data/nhanes/nhanes_09_10/bpq_f.pdf. National Institutes of Health, 1998. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults-the evidence report. Obes. Res. 6 (Suppl. 2), 51Se209S. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and T. of H. B. C. in A. (Adult T. P. I, 2002. Third report of the national cholesterol education program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (Adult treatment panel III) final report. Circulation 106, 3143e3421. http://dx.doi.org/10.1001/ archinte.1991.00400060019005. Nemeroff, C.B., Musselman, D.L., 2000. Are platelets the link between depression and ischemic heart disease? Am. Heart J. 140 (4), S57eS62. http://dx.doi.org/ 10.1067/mhj.2000.109978. Nieuwlaat, R., Wilczynski, N., Navarro, T., Hobson, N., Jeffery, R., Keepanasseril, A., Agoritsas, T., et al., 2014. Interventions for enhancing medication adherence. Cochrane Database Syst. Rev. http://dx.doi.org/10.1002/ 14651858.CD000011.pub4. Issue 11, Art. No.: CD000011. Paschall, M.J., Freisthler, B., Lipton, R.I., 2005. Moderate alcohol use and depression in young adults: findings from a national longitudinal study. Am. J. Public Health 95 (3), 453e457. http://dx.doi.org/10.2105/AJPH.2003.030700. Patten, S.B., Schopflocher, D., 2009. Longitudinal epidemiology of major depression as assessed by the brief patient health questionnaire (PHQ-9). Compr. Psychiatry 50 (1), 26e33. http://dx.doi.org/10.1016/j.comppsych.2008.05.012. Patten, S.B., Williams, J.V.A., Lavorato, D.H., Eliasziw, M., 2009. A longitudinal community study of major depression and physical activity. General Hosp. Psychiatry 31 (6), 571e575. http://dx.doi.org/10.1016/j.genhosppsych.2009.08.001. Radloff, L.S., 1977. The CES-d scale: a self-report depression scale for research in the general population. Appl. Psychol. Meas. 1 (3), 383e401. Sankar, A.P., Nevedal, D.C., Neufeld, S., Luborsky, M.R., 2007. What is a missed dose? implications for construct validity and patient adherence. AIDS Care 19 (6), 775e780. http://dx.doi.org/10.1080/09540120600708501. Shi, L., Liu, J., Fonseca, V., Walker, P., Kalsekar, A., Pawaskar, M., 2010. Correlation between adherence rates measured by MEMS and self-reported questionnaires: a meta-analysis. Health Qual. Life Outcomes 8, 99. http://dx.doi.org/10.1186/ 1477-7525-8-99. Sumlin, L.L., Garcia, T.J., Brown, S. a, Winter, M. a, García, A. a, Brown, A., Cuevas, H.E., 2015. Depression and adherence to lifestyle changes in type 2 diabetes: a systematic review. Diabetes Educ. 40 (6), 731e744. http://dx.doi.org/ 10.1177/0145721714538925. Teixeira, P.J., Going, S.B., Sardinha, L.B., Lohman, T.G., 2005. A review of psychosocial pre-treatment predictors of weight control. Obes. Rev. 6, 43e65. Trivedi, R.B., Ayotte, B., Edelman, D., Bosworth, H.B., 2008. The association of emotional well-being and marital status with treatment adherence among patients with hypertension. J. Behav. Med. 31 (6), 489e497. http://dx.doi.org/ 10.1007/s10865-008-9173-4. US Department of Health and Human Services, 2008. 2008 Physical Activity Guidelines for Americans. US Department of Health and Human Services, Washington, DC. Van der Kooy, K., van Hout, H., Marwijk, H., Marten, H., Stehouwer, C., Beekman, A., 2007. Depression and the risk for cardiovascular diseases: systematic review and meta analysis. Int. J. Geriatr. Psychiatry 22 (7), 613e626. http://dx.doi.org/ 10.1002/gps.1723. van der Wal, M.H.L., Jaarsma, T., Moser, D.K., Veeger, N.J.G.M., van Gilst, W.H., van Veldhuisen, D.J., 2006. Compliance in heart failure patients: the importance of knowledge and beliefs. Eur. Heart J. 27 (4), 434e440. http://dx.doi.org/10.1093/ eurheartj/ehi603. Whooley, M.A., de Jonge, P., Vittinghoff, E., Otte, C., Moos, R., Carney, R.M., Ali, S., et al., 2008. Depressive symptoms, health behaviors, and risk of cardiovascular events in patients with coronary heart disease. JAMA 300 (20), 2379e2388. http://dx.doi.org/10.1001/jama.2008.711. Wittkampf, K.A., Naeije, L., Schene, A.H., Huyser, J., van Weert, H.C., 2007. Diagnostic accuracy of the mood module of the patient health questionnaire: a systematic review. Gen. Hosp. Psychiatry 29, 388e395. Wrotniak, B.H., Epstein, L.H., Paluch, R.A., Roemmich, J.N., 2005. The relationship between parent and child self-reported adherence and weight loss. Obes. Res. 13 (6), 1089e1096. http://dx.doi.org/10.1038/oby.2005.127. Ziegelstein, R.C., Fauerbach, J.A., Stevens, S.S., Romanelli, J., Richter, D.P., Bush, D.E., 2000. Patients with depression are less likely to follow recommendations to reduce cardiac risk during recovery from a myocardial infarction. Arch. Intern. Med. 160 (12), 1818e1823. Retrieved from. http://www.ncbi.nlm.nih.gov/ pubmed/10871976.