University of Groningen
Shades of a blue heart
Moreira da Rocha de Miranda Az, Ricardo
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Moreira da Rocha de Miranda Az, R. (2018). Shades of a blue heart: An epidemiological investigation of depressive symptom dimensions and the association with cardiovascular disease. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Chapter 5
Investigating The Longitudinal
Association of Arterial Stiffness
and Carotid Intima-Media
Thickness With Depressive
Symptom Dimensions in
Middle-Aged Individuals
de Miranda Azevedo R, Roest AM, Groenewolde NA, Seldenrijk A, Penninx BWJH, de Jonge P. Investigating the Longitudinal Association of Arterial Stiffness and Carotid Intima-Media Thickness With Depressive Symptom Dimension in Middle-Aged Individuals. Psychol Med. (To be submitted). 130 23. de Miranda Azevedo R, Roest AM, Carney RM, et al. A bifactor model of the beck depression inventory and its association with medical prognosis after myocardial infarction. Health Psychology. 2016. 24. Wilhelmson, K., Eklund,K. Positive effects on life satisfaction following health-promoting interventions for frail older adults: A randomized controlled study. Health Psychology Research. 2013;1(1). 25. Guiry E, Conroy RM, Hickey N, Mulcahy R. Psychological response to an acute coronary event and its effect on subsequent rehabilitation and lifestyle change. Clin Cardiol. 1987;10(4):256-260. 26. Suarez L, Beach SR, Moore SV, et al. Use of the patient health questionnaire-9 and a detailed suicide evaluation in determining imminent suicidality in distressed patients with cardiac disease. Psychosomatics. 2015;56(2):181-189. 27. Alsen P, Brink E, Brandstrom Y, Karlson BW, Persson L. Fatigue after myocardial infarction: Relationships with indices of emotional distress, and sociodemographic and clinical variables. Int J Nurs Pract. 2010;16(4):326-334. 28. Sandor B, Nagy A, Toth A, et al. Effects of moderate aerobic exercise training on hemorheological and laboratory parameters in ischemic heart disease patients. PLoS One. 2014;9(10):e110751. 29. Dunn SL. Hopelessness and depression in myocardial infarction. ProQuest Information & Learning; 2006. 30. Valtonen M, Laaksonen DE, Laukkanen J, et al. Leisure-time physical activity, cardiorespiratory fitness and feelings of hopelessness in men. BMC Public Health. 2009;9:204-2458-9-204. 31. Fried EI, Nesse RM. Depression sum-scores don't add up: Why analyzing specific depression symptoms is essential. BMC Med. 2015;13:72-015-0325-4. 32. Ormel J, De Jonge P. Unipolar depression and the progression of coronary artery disease: Toward an integrative model. Psychother Psychosom. 2011;80(5):264-274.
133
INTRODUCTION
The relation between depression and heart disease is complex and appears to be bidirectional. Depression in otherwise healthy individuals has been shown to predict the onset of heart disease and cardiovascular
mortality 1. Individuals with both depression and coronary heart disease
have a risk of mortality two times higher as compared to individuals with
CHD that do not have depression 2. Moreover, evidence suggests that
cardiovascular fitness early in life predicts depression later in life. In a longitudinal study of more than a million Swedish men, lower levels of cardiovascular fitness were associated with the onset of depression later
in life 3.
Atherosclerosis is a critical contributing factor to the onset of heart disease. Atherosclerosis is a vascular condition in which the arterial walls thicken due to the accumulation of fat, calcium and other substances. Despite being asymptomatic most of its course, atherosclerosis of the carotid artery can be assessed through several non-invasive ways, such as by measuring the thickness of the two innermost layers of the carotid artery wall (“carotid intima-media thickness”; CIMT) or assessing the
presence of carotid plaques 4. Another useful non-invasive way to assess
subclinical heart disease is by measuring arterial stiffness 5. Alterations in
CIMT 6-8 and arterial stiffness 9-11 are established precursors to the onset of
heart disease.
Due to the growing interest in the association between depression and cardiovascular health, some studies investigated if markers of CIMT are associated with depressive symptoms. This association has been investigated in several cross-sectional studies and findings were mixed.
Some studies reported an association between CIMT and depression 4,12,13,
while others did not 14,15. Several studies also demonstrated an association
between depression and markers of arterial stiffness such as
augmentation index 5, distensibility coefficient 5,16, pulse wave velocity 16,17
and forearm hyperemic reactivity 18.
A limitation of these studies is that they assessed depression only through a clinical diagnosis or through a sum score of depressive symptoms, rather than looking at depressive symptom dimensions. The
132
ABSTRACT
BACKGROUND: Arterial stiffness and carotid intima-media thickness (CIMT) are potential mechanisms explaining the association between depression and heart disease. The objectives of this study were to assess whether markers of subclinical vascular pathology are associated with depressive symptom dimensions in middle-aged individuals, and if this association is stable over time.
METHODS: The sample included 600 participants enrolled in the Netherlands Study of Depression and Anxiety (NESDA). CIMT, presence of plaque and arterial stiffness (represented by augmentation index) were the main predictors. Depressive symptoms were measured at the same time-point as the vascular pathology indicators, as well as one and two years later with the Inventory of Depressive Symptoms Self-Report. Mood/cognition and anxiety/somatic symptom dimensions were used as outcome variables in linear mixed models.
RESULTS: Augmentation index was associated with increased overall mood/cognition (β=.007; 95%CI=0.00-0.01; p=.031) and anxiety/somatic symptoms (β=.006; 95%CI=0.00-0.01; p=.025). CIMT and presence of plaque were not associated with depressive symptom dimensions. The association did not change across time.
CONCLUSIONS: Arterial stiffness, but not CIMT or presence of plaque, was significantly associated with both depressive symptom dimensions. The association was moderate and stable over time. Future studies should assess these associations in older individuals with more advanced vascular pathology.
133
INTRODUCTION
The relation between depression and heart disease is complex and appears to be bidirectional. Depression in otherwise healthy individuals has been shown to predict the onset of heart disease and cardiovascular
mortality 1. Individuals with both depression and coronary heart disease
have a risk of mortality two times higher as compared to individuals with
CHD that do not have depression 2. Moreover, evidence suggests that
cardiovascular fitness early in life predicts depression later in life. In a longitudinal study of more than a million Swedish men, lower levels of cardiovascular fitness were associated with the onset of depression later
in life 3.
Atherosclerosis is a critical contributing factor to the onset of heart disease. Atherosclerosis is a vascular condition in which the arterial walls thicken due to the accumulation of fat, calcium and other substances. Despite being asymptomatic most of its course, atherosclerosis of the carotid artery can be assessed through several non-invasive ways, such as by measuring the thickness of the two innermost layers of the carotid artery wall (“carotid intima-media thickness”; CIMT) or assessing the
presence of carotid plaques 4. Another useful non-invasive way to assess
subclinical heart disease is by measuring arterial stiffness 5. Alterations in
CIMT 6-8 and arterial stiffness 9-11 are established precursors to the onset of
heart disease.
Due to the growing interest in the association between depression and cardiovascular health, some studies investigated if markers of CIMT are associated with depressive symptoms. This association has been investigated in several cross-sectional studies and findings were mixed.
Some studies reported an association between CIMT and depression 4,12,13,
while others did not 14,15. Several studies also demonstrated an association
between depression and markers of arterial stiffness such as
augmentation index 5, distensibility coefficient 5,16, pulse wave velocity 16,17
and forearm hyperemic reactivity 18.
A limitation of these studies is that they assessed depression only through a clinical diagnosis or through a sum score of depressive symptoms, rather than looking at depressive symptom dimensions. The
132
ABSTRACT
BACKGROUND: Arterial stiffness and carotid intima-media thickness (CIMT) are potential mechanisms explaining the association between depression and heart disease. The objectives of this study were to assess whether markers of subclinical vascular pathology are associated with depressive symptom dimensions in middle-aged individuals, and if this association is stable over time.
METHODS: The sample included 600 participants enrolled in the Netherlands Study of Depression and Anxiety (NESDA). CIMT, presence of plaque and arterial stiffness (represented by augmentation index) were the main predictors. Depressive symptoms were measured at the same time-point as the vascular pathology indicators, as well as one and two years later with the Inventory of Depressive Symptoms Self-Report. Mood/cognition and anxiety/somatic symptom dimensions were used as outcome variables in linear mixed models.
RESULTS: Augmentation index was associated with increased overall mood/cognition (β=.007; 95%CI=0.00-0.01; p=.031) and anxiety/somatic symptoms (β=.006; 95%CI=0.00-0.01; p=.025). CIMT and presence of plaque were not associated with depressive symptom dimensions. The association did not change across time.
CONCLUSIONS: Arterial stiffness, but not CIMT or presence of plaque, was significantly associated with both depressive symptom dimensions. The association was moderate and stable over time. Future studies should assess these associations in older individuals with more advanced vascular pathology.
135 METHODS Sample The study sample included participants enrolled in “The Netherlands Study of Depression and Anxiety – NESDA.” NESDA is a Dutch longitudinal cohort study that aims to investigate affective disorders in subjects recruited from several different settings of health care, such as outpatient psychiatric clinics, primary care and the community. Participants were excluded if they had a diagnosis of another psychiatric disorder not investigated in NESDA (i.e. psychotic disorders, obsessive compulsive disorder, bipolar disorder) or if they were not fluent in the Dutch language. The total NESDA sample was composed of 2329 participants with a current or remitted depressive and/or anxiety disorder and 652 controls. All participants provided written informed consent, and the Medical Ethical Committees of the universities involved in the study approved the study protocol. More details of the NESDA cohort (e.g. methods, recruitment, rationale
and strategy) are available elsewhere 23.
In the present study, we used data from the first (“W1”), second (“W2”) and third (“W3”) follow-up measurement waves of the NESDA study. For the sake of simplicity, we will name the first follow-up measurement wave of NESDA (“W1”) as our “baseline” measurement, the second follow-up measurement wave (“W2”) as the first follow-up measurement wave and the third follow-up measurement wave (“W3”) as the second follow-up measurement wave.
Carotid Intima Media Thickness (CIMT)
Cardiovascular measurements were assessed at the baseline of the present study (W1), from June 2007 until July 2009 at the Amstelland Hospital (Amstelveen, the Netherlands). From the 2,596 subjects participating in this measurement wave, 1,624 were excluded due to: consumption of insulin for diabetes treatment (N = 7) or living far from the hospital (N = 1,617). A total of 972 participants were eligible for the sub-study. Ultrasound data of the carotid artery was recorded in 649 participants, of who 179 had a lifetime diagnosis of depressive or anxiety
134
presentation of depression in cardiac patients differs from somatically
healthy people 19. Ormel and de Jonge hypothesized that depression in
patients with cardiovascular disease is a combination of two prototypical subtypes: a cognitive/affective, which is marked by psychosocial vulnerability, and a somatic/affective, marked by inflammation, sickness
behavior and atherosclerosis 20. Relatively few studies have examined the
association between depressive symptom dimensions and subclinical vascular pathology. One previous cross-sectional study reported that the association between depressive symptoms with CIMT was largely explained by somatic/affective symptoms of depression in an older
community sample 21. Another cross-sectional study reported that neither
the cognitive/affective nor the somatic/affective dimensions were
associated with increased CIMT or plaque presence in young adults 22. To
date, no studies investigated the longitudinal association of CIMT and arterial stiffness with depressive symptom dimensions.
The objective of the present study is to investigate whether markers of CIMT and arterial stiffness are associated with depressive symptom dimensions, and whether this association is stable over time in a sample of depressed and non-depressed middle-aged individuals.
135 METHODS Sample The study sample included participants enrolled in “The Netherlands Study of Depression and Anxiety – NESDA.” NESDA is a Dutch longitudinal cohort study that aims to investigate affective disorders in subjects recruited from several different settings of health care, such as outpatient psychiatric clinics, primary care and the community. Participants were excluded if they had a diagnosis of another psychiatric disorder not investigated in NESDA (i.e. psychotic disorders, obsessive compulsive disorder, bipolar disorder) or if they were not fluent in the Dutch language. The total NESDA sample was composed of 2329 participants with a current or remitted depressive and/or anxiety disorder and 652 controls. All participants provided written informed consent, and the Medical Ethical Committees of the universities involved in the study approved the study protocol. More details of the NESDA cohort (e.g. methods, recruitment, rationale
and strategy) are available elsewhere 23.
In the present study, we used data from the first (“W1”), second (“W2”) and third (“W3”) follow-up measurement waves of the NESDA study. For the sake of simplicity, we will name the first follow-up measurement wave of NESDA (“W1”) as our “baseline” measurement, the second follow-up measurement wave (“W2”) as the first follow-up measurement wave and the third follow-up measurement wave (“W3”) as the second follow-up measurement wave.
Carotid Intima Media Thickness (CIMT)
Cardiovascular measurements were assessed at the baseline of the present study (W1), from June 2007 until July 2009 at the Amstelland Hospital (Amstelveen, the Netherlands). From the 2,596 subjects participating in this measurement wave, 1,624 were excluded due to: consumption of insulin for diabetes treatment (N = 7) or living far from the hospital (N = 1,617). A total of 972 participants were eligible for the sub-study. Ultrasound data of the carotid artery was recorded in 649 participants, of who 179 had a lifetime diagnosis of depressive or anxiety
134
presentation of depression in cardiac patients differs from somatically
healthy people 19. Ormel and de Jonge hypothesized that depression in
patients with cardiovascular disease is a combination of two prototypical subtypes: a cognitive/affective, which is marked by psychosocial vulnerability, and a somatic/affective, marked by inflammation, sickness
behavior and atherosclerosis 20. Relatively few studies have examined the
association between depressive symptom dimensions and subclinical vascular pathology. One previous cross-sectional study reported that the association between depressive symptoms with CIMT was largely explained by somatic/affective symptoms of depression in an older
community sample 21. Another cross-sectional study reported that neither
the cognitive/affective nor the somatic/affective dimensions were
associated with increased CIMT or plaque presence in young adults 22. To
date, no studies investigated the longitudinal association of CIMT and arterial stiffness with depressive symptom dimensions.
The objective of the present study is to investigate whether markers of CIMT and arterial stiffness are associated with depressive symptom dimensions, and whether this association is stable over time in a sample of depressed and non-depressed middle-aged individuals.
137
Sydney, Australia). More details of the measurement procedure are
reported elsewhere 5. Measurements of arterial stiffness were also
assessed at the baseline wave of the present study (“W1”). To represent central arterial stiffness in the analyses, we used the augmentation index as a continuous predictor variable. Augmentation index consists of the ratio between the late systolic pressure augmentation minus pressure at early systolic peak (PA) and the central pulse pressure (PP). In accordance with previous work, augmentation index was standardized to a heart rate
of 75 BPM 25. Augmentation index can be calculated with the following
formula: !"#$%&'(')*& !"#$% = !" !! × 100 !" !"#$"%&'(" Depressive symptom dimensions The Dutch translation of the Inventory of Depressive Symptoms - Self Report (IDS-SR) was used. The IDS-SR consists of 30 equally weighted items, rated on a four-point scale (range 0-3). The items “weight” and “appetite” are scored separately for increase and decrease. Three symptom dimensions were identified in previous work, namely
anxiety/somatic, mood/cognition and sleep symptom dimensions 26,27. In
the present study, only the anxiety/somatic and mood/cognition dimensions were used because a large body of work has been conducted assessing similar symptom dimensions, namely somatic/affective and
cognitive/affective 28,29. In the present study we used factor scores of
depressive symptom dimensions extracted from the baseline (“W1”), the first (“W2”) and the second (“W3”) follow-up measurement waves.
Statistical analysis
Confirmatory Factor Analysis (CFA) of the IDS-SR
A confirmatory factor analysis (CFA) was performed to generate the factor scores of the symptom dimensions at each of the measurement waves. The factor structure followed previous work conducted in the
136
disorders and 470 did not. Comparisons with eligible non-participants indicated that participants were older, and were less likely to have
depression or anxiety disorders 4.
B-mode images were taken from participants at the following spots: bilateral common carotid artery at 10mm proximal to carotid dilatation, carotid bifurcation between dilatation and flow divider and internal carotid artery at 10mm distal to tip of flow divider. Additionally, the distance between the media-adventitia interfaces and the far wall lumen-intima was measured at all available segments. The measurements
followed a valid and standardized protocol reported elsewhere 24.
All the measurements were conducted with an Acuson Aspen ultrasound instrument equipped with a near-field L7 array of 5-10 MHz broadband transducer (Siemens, Erlangen, Germany). The within-sonographer reproducibility was assessed through Pearson’s correlation (r = .98). The reproducibility of CIMT analysis was also assessed, and Pearson’s correlation coefficients of r = 0.99 and r = 0.93 were found for
within and between-reader correlations, respectively 4.
We used total CIMT as a main predictor variable in the analyses (i.e. the average of the bilateral common carotid artery, carotid bifurcation and internal carotid artery). We also used presence of plaque in the near and far walls common carotid artery, carotid bifurcation and internal carotid artery as a dichotomous main predictor variable. An increased thickness ≥1.10mm of the intimal and medial layers in relation to their
adjacent segments indicated presence of plaque 4.
Arterial stiffness (Central): Radial Artery Pulse Wave Analysis - Augmentation Index.
Participants had their brachial diastolic and systolic blood pressures measured at the right upper arm, using oscillometric equipment (Dinamap PRO100; GE Medical Systems, Tampa, Florida) while lying in a supine position, during three series of two consecutive readings in W1. Radial artery waveforms were recorded with an applanation tonometer interfaced with SphygmoCor software (2000 version 7; AtCor Medical,
137
Sydney, Australia). More details of the measurement procedure are
reported elsewhere 5. Measurements of arterial stiffness were also
assessed at the baseline wave of the present study (“W1”). To represent central arterial stiffness in the analyses, we used the augmentation index as a continuous predictor variable. Augmentation index consists of the ratio between the late systolic pressure augmentation minus pressure at early systolic peak (PA) and the central pulse pressure (PP). In accordance with previous work, augmentation index was standardized to a heart rate
of 75 BPM 25. Augmentation index can be calculated with the following
formula: !"#$%&'(')*& !"#$% = !" !! × 100 !" !"#$"%&'(" Depressive symptom dimensions The Dutch translation of the Inventory of Depressive Symptoms - Self Report (IDS-SR) was used. The IDS-SR consists of 30 equally weighted items, rated on a four-point scale (range 0-3). The items “weight” and “appetite” are scored separately for increase and decrease. Three symptom dimensions were identified in previous work, namely
anxiety/somatic, mood/cognition and sleep symptom dimensions 26,27. In
the present study, only the anxiety/somatic and mood/cognition dimensions were used because a large body of work has been conducted assessing similar symptom dimensions, namely somatic/affective and
cognitive/affective 28,29. In the present study we used factor scores of
depressive symptom dimensions extracted from the baseline (“W1”), the first (“W2”) and the second (“W3”) follow-up measurement waves.
Statistical analysis
Confirmatory Factor Analysis (CFA) of the IDS-SR
A confirmatory factor analysis (CFA) was performed to generate the factor scores of the symptom dimensions at each of the measurement waves. The factor structure followed previous work conducted in the
136
disorders and 470 did not. Comparisons with eligible non-participants indicated that participants were older, and were less likely to have
depression or anxiety disorders 4.
B-mode images were taken from participants at the following spots: bilateral common carotid artery at 10mm proximal to carotid dilatation, carotid bifurcation between dilatation and flow divider and internal carotid artery at 10mm distal to tip of flow divider. Additionally, the distance between the media-adventitia interfaces and the far wall lumen-intima was measured at all available segments. The measurements
followed a valid and standardized protocol reported elsewhere 24.
All the measurements were conducted with an Acuson Aspen ultrasound instrument equipped with a near-field L7 array of 5-10 MHz broadband transducer (Siemens, Erlangen, Germany). The within-sonographer reproducibility was assessed through Pearson’s correlation (r = .98). The reproducibility of CIMT analysis was also assessed, and Pearson’s correlation coefficients of r = 0.99 and r = 0.93 were found for
within and between-reader correlations, respectively 4.
We used total CIMT as a main predictor variable in the analyses (i.e. the average of the bilateral common carotid artery, carotid bifurcation and internal carotid artery). We also used presence of plaque in the near and far walls common carotid artery, carotid bifurcation and internal carotid artery as a dichotomous main predictor variable. An increased thickness ≥1.10mm of the intimal and medial layers in relation to their
adjacent segments indicated presence of plaque 4.
Arterial stiffness (Central): Radial Artery Pulse Wave Analysis - Augmentation Index.
Participants had their brachial diastolic and systolic blood pressures measured at the right upper arm, using oscillometric equipment (Dinamap PRO100; GE Medical Systems, Tampa, Florida) while lying in a supine position, during three series of two consecutive readings in W1. Radial artery waveforms were recorded with an applanation tonometer interfaced with SphygmoCor software (2000 version 7; AtCor Medical,
139
activity related energy expenditure in comparison with rest time multiplied by the total of minutes spent while performing the physical
activity 32.
Depressive (major depression, dysthymia) and anxiety (generalized anxiety (social phobia, panic disorder and/or agoraphobia) disorders were assessed between baseline (W1) and the second follow-up measurement wave (W3). This assessment followed the diagnostic criteria of the Composite International Diagnostic Interview (CIDI; WHO version 2.1). Linear Mixed Models
We conducted linear mixed models to assess the association between subclinical heart disease and depressive symptoms. A separate model was built for each of the predictors (CIMT total, presence of plaque and augmentation index). Coefficients (β), 95% confidence intervals and p-values of the linear mixed model were reported. All models included the following covariates: age, sex, years of education, smoking, BMI and physical activity. To evaluate if the association is stable over time, an additional model including an interaction term between each predictor and time was conducted. In these models, a p-value of 0.05 was used to indicate statistical significance. An “identity” covariance structure was used for every model. The “identity” covariance structure sets equal variances for random effects and all covariances to zero. A total of 12 linear mixed models were built, for each of the three predictors and two outcomes, one including main effects of predictors, and one including an interaction term between predictors and time. The analyses were conducted using Stata 13.0 for Windows.
Interaction models and subgroup analyses
We assessed for potential interactions between predictors and the following variables that could confound the association between subclinical heart disease and depressive symptom dimensions: age, sex and diagnosis of depressive/anxiety disorders (between baseline and
138
NESDA sample 26,27. The factor structure consisted of a 2-factor model;
including a “mood/cognition” dimension (items 5 “feeling sad”, 8 “reactivity of mood”, 10 “quality of mood”, 11/12 “appetite disturbance”, 15 “concentration and decision making”, 16 “self-criticism and blame”, 17 “future pessimism”, 18 “suicidal thoughts”, 19 “interest in people/activities”, 20 “energy/fatigability”, 22 “interest in sex” and 29 “interpersonal sensitivity”) and an “anxiety/somatic” dimension (items 6 “feeling irritable”, 23 “psychomotor retardation”, 24 “psychomotor agitation”, 25 “aches and pain”, 26 “sympathetic arousal”, 27 “panic/phobic symptoms”, 28 “constipation/diarrhea” and 30 “leaden paralysis”). Items on appetite were rescored into a single ‘change of
appetite’ (item 11/12) variable. Factor covariance was freely estimated.
MPLUS 7.0 was used for all the CFA models. We assessed the goodness of fit of the CFA models based on three fit indices: The Root Mean Square Error of Approximation (RMSEA), the Tucker-Lewis Index (TLI) and the Comparative Fit Index (CFI). We used standard cutoffs to assess the
goodness of fit of the CFA models 30,31. To increase statistical power, all
CFA models were conducted on the whole NESDA sample.
Factor scores of the mood/cognitive and the anxiety/somatic depressive symptom dimensions were available from 647, 596 and 558 participants on baseline (W1), first (W2) and second (W3) follow-up measurement waves, respectively.
Covariates
Several covariates were added to the present analysis as they could confound the association between markers of subclinical cardiovascular disease and depressive symptom dimensions. The following sociodemographic characteristics were included: age, sex and years of education. Health-related variables were also added: smoking status (categorized as current non-smoker or smoker); body mass index (i.e. weight in kilograms divided by the height in meters squared) and level of physical activity (assessed through the International Physical Activity Questionnaire). The International Physical Activity Questionnaire measures the weekly MET-minutes, which is the ratio between physical
139
activity related energy expenditure in comparison with rest time multiplied by the total of minutes spent while performing the physical
activity 32.
Depressive (major depression, dysthymia) and anxiety (generalized anxiety (social phobia, panic disorder and/or agoraphobia) disorders were assessed between baseline (W1) and the second follow-up measurement wave (W3). This assessment followed the diagnostic criteria of the Composite International Diagnostic Interview (CIDI; WHO version 2.1). Linear Mixed Models
We conducted linear mixed models to assess the association between subclinical heart disease and depressive symptoms. A separate model was built for each of the predictors (CIMT total, presence of plaque and augmentation index). Coefficients (β), 95% confidence intervals and p-values of the linear mixed model were reported. All models included the following covariates: age, sex, years of education, smoking, BMI and physical activity. To evaluate if the association is stable over time, an additional model including an interaction term between each predictor and time was conducted. In these models, a p-value of 0.05 was used to indicate statistical significance. An “identity” covariance structure was used for every model. The “identity” covariance structure sets equal variances for random effects and all covariances to zero. A total of 12 linear mixed models were built, for each of the three predictors and two outcomes, one including main effects of predictors, and one including an interaction term between predictors and time. The analyses were conducted using Stata 13.0 for Windows.
Interaction models and subgroup analyses
We assessed for potential interactions between predictors and the following variables that could confound the association between subclinical heart disease and depressive symptom dimensions: age, sex and diagnosis of depressive/anxiety disorders (between baseline and
138
NESDA sample 26,27. The factor structure consisted of a 2-factor model;
including a “mood/cognition” dimension (items 5 “feeling sad”, 8 “reactivity of mood”, 10 “quality of mood”, 11/12 “appetite disturbance”, 15 “concentration and decision making”, 16 “self-criticism and blame”, 17 “future pessimism”, 18 “suicidal thoughts”, 19 “interest in people/activities”, 20 “energy/fatigability”, 22 “interest in sex” and 29 “interpersonal sensitivity”) and an “anxiety/somatic” dimension (items 6 “feeling irritable”, 23 “psychomotor retardation”, 24 “psychomotor agitation”, 25 “aches and pain”, 26 “sympathetic arousal”, 27 “panic/phobic symptoms”, 28 “constipation/diarrhea” and 30 “leaden paralysis”). Items on appetite were rescored into a single ‘change of
appetite’ (item 11/12) variable. Factor covariance was freely estimated.
MPLUS 7.0 was used for all the CFA models. We assessed the goodness of fit of the CFA models based on three fit indices: The Root Mean Square Error of Approximation (RMSEA), the Tucker-Lewis Index (TLI) and the Comparative Fit Index (CFI). We used standard cutoffs to assess the
goodness of fit of the CFA models 30,31. To increase statistical power, all
CFA models were conducted on the whole NESDA sample.
Factor scores of the mood/cognitive and the anxiety/somatic depressive symptom dimensions were available from 647, 596 and 558 participants on baseline (W1), first (W2) and second (W3) follow-up measurement waves, respectively.
Covariates
Several covariates were added to the present analysis as they could confound the association between markers of subclinical cardiovascular disease and depressive symptom dimensions. The following sociodemographic characteristics were included: age, sex and years of education. Health-related variables were also added: smoking status (categorized as current non-smoker or smoker); body mass index (i.e. weight in kilograms divided by the height in meters squared) and level of physical activity (assessed through the International Physical Activity Questionnaire). The International Physical Activity Questionnaire measures the weekly MET-minutes, which is the ratio between physical
141 Table 1. Descriptive statistics at baseline (N = 647) Predictor variable N (%); Mean (± SD) Age 46.5 (± 12.1) % of men 225 (35%) % of smokers 174 (27%) BMI 25.4 (±4.5) Years of education 13.2 (±4.5) % of participants with low physical activity 118 (20%) % of participants with moderate physical activity 256 (43%) % of participants with high physical activity 226 (37%) CIMTtotal 0.66 (± 0.14) % with presence of plaque 95 (15%) Augmentation index 14 (± 14.88) % Diagnosed with depressive or anxiety disorders 188 (29%)
BMI: Body mass index; CIMTtotal: Total carotid intima-media thickness.
140
second follow-up measurement wave of the present study). Following previous recommendations, a p-value of .20 was used for flagging
potential interactions 33. When a potential interaction effect was flagged,
subgroup analyses were performed. In subgroup analyses, models were fitted separately for each of the predictor variables. The same covariate adjustment performed in main analyses was used in subgroup analyses.
RESULTS
Table 1 displays the descriptive statistics of the sample on the baseline measurement wave (W1) of the present study. Depressive symptoms Table 2 shows the factor loadings of the items of the IDS-SR and fit indices of the models across measurement waves. Model fit was very good and of comparable magnitude across all three measurement waves. Linear mixed models – Main analyses Table 3 shows the results of the linear mixed models according to the outcome used. Augmentation index was associated with both depressive symptom dimensions (mood/cognition, β = 0.01 [0.00 – 0.01], p = .03; anxiety/somatic, β = 0.01 [0.00 – 0.01], p = .03). Neither total CIMT (mood/cognition, β = -.50 [-1.07 – 0.07], p = .09; anxiety/somatic, β = -.40 [-0.89 – 0.10]. p = .12) nor presence of plaque (mood/cognition, β = .09 [-0.09 – 0.26], p = .35; anxiety/somatic, β = .09 [-0.06 – 0.25]. p = .25) was associated with depressive symptom dimensions. None of the interaction terms between predictors and time were significantly associated with depressive symptom dimensions. This implicates that the association between subclinical heart disease and depressive symptom dimensions was stable over time.
141 Table 1. Descriptive statistics at baseline (N = 647) Predictor variable N (%); Mean (± SD) Age 46.5 (± 12.1) % of men 225 (35%) % of smokers 174 (27%) BMI 25.4 (±4.5) Years of education 13.2 (±4.5) % of participants with low physical activity 118 (20%) % of participants with moderate physical activity 256 (43%) % of participants with high physical activity 226 (37%) CIMTtotal 0.66 (± 0.14) % with presence of plaque 95 (15%) Augmentation index 14 (± 14.88) % Diagnosed with depressive or anxiety disorders 188 (29%)
BMI: Body mass index; CIMTtotal: Total carotid intima-media thickness.
140
second follow-up measurement wave of the present study). Following previous recommendations, a p-value of .20 was used for flagging
potential interactions 33. When a potential interaction effect was flagged,
subgroup analyses were performed. In subgroup analyses, models were fitted separately for each of the predictor variables. The same covariate adjustment performed in main analyses was used in subgroup analyses.
RESULTS
Table 1 displays the descriptive statistics of the sample on the baseline measurement wave (W1) of the present study. Depressive symptoms Table 2 shows the factor loadings of the items of the IDS-SR and fit indices of the models across measurement waves. Model fit was very good and of comparable magnitude across all three measurement waves. Linear mixed models – Main analyses Table 3 shows the results of the linear mixed models according to the outcome used. Augmentation index was associated with both depressive symptom dimensions (mood/cognition, β = 0.01 [0.00 – 0.01], p = .03; anxiety/somatic, β = 0.01 [0.00 – 0.01], p = .03). Neither total CIMT (mood/cognition, β = -.50 [-1.07 – 0.07], p = .09; anxiety/somatic, β = -.40 [-0.89 – 0.10]. p = .12) nor presence of plaque (mood/cognition, β = .09 [-0.09 – 0.26], p = .35; anxiety/somatic, β = .09 [-0.06 – 0.25]. p = .25) was associated with depressive symptom dimensions. None of the interaction terms between predictors and time were significantly associated with depressive symptom dimensions. This implicates that the association between subclinical heart disease and depressive symptom dimensions was stable over time.
143 Table 2. Factor loadings and fit indices of confirmatory factor analyses across time points. Item Baseline (W1); N = 2,505) First follow-up assessment wave (W2); N = 2,316 Second follow-up assessment wave (W3); N = 2,149 Mood/Cognition Factor 5. Feeling sad .878 .864 .893 8. Reactivity of mood .696 .664 .685 10. Quality of Mood .822 .840 .838 11/12. Appetite disturbance .569 .595 .611 15. Concentration and decision Making .774 .775 .775 16. View of Myself .774 .777 .795 17. View of My Future .809 .793 .800 18. Thoughts of Death or Suicide .762 .727 .715 20. Energy Level .780 .772 .785 22. Interest in Sex .593 .592 .599 29. Interpersonal Sensitivity .650 .668 .708 Anxiety/Somatic Factor 6. Feeling Irritable .757 .770 .772 23. Feeling slowed down .802 .827 .800 24. Feeling restless .663 .688 .706 25. Aches and pains .539 .569 .566 26. Other bodily symptoms .601 .638 .660 27. Panic/Phobic symptoms .552 .598 .559 28. Constipation/diarrhea .438 .469 .490 30. Leaden Paralysis/Physical Energy .823 .814 .804 Fit indices CFI .959 .954 .959 TLI .954 .948 .954 RMSEA .068 (.065 - .071) .070 (.068 - .073) .068 (.065 - .071) CFI: Comparative fit index; TLI: Tucker-Lewis index; RMSEA: Root mean square error of approximation 142 Linear mixed models – Main analyses
Table 3 shows the results of the linear mixed models according to the
outcome used. Augmentation index was associated with both depressive symptom dimensions (mood/cognition, β = 0.01 [0.00 – 0.01], p = .03; anxiety/somatic, β = 0.01 [0.00 – 0.01], p = .03). Neither total CIMT (mood/cognition, β = -.50 [-1.07 – 0.07], p = .09; anxiety/somatic, β = -.40 [-0.89 – 0.10]. p = .12) nor presence of plaque (mood/cognition, β = .09 [-0.09 – 0.26], p = .35; anxiety/somatic, β = .09 [-0.06 – 0.25]. p = .25) was associated with depressive symptom dimensions. None of the interaction terms between predictors and time were significantly associated with depressive symptom dimensions. This implicates that the association between subclinical heart disease and depressive symptom dimensions was stable over time.
Interaction models and sensitivity analyses
We tested for interaction effects between predictors and sex, age and diagnosis of depressive and anxiety disorders. Potential interactions were found only between two predictor variables (CIMT and augmentation index) and diagnosis of depressive and anxiety disorders (Table 4). Subgroup analyses indicated that augmentation index was associated with cognitive/affective and anxiety/arousal depressive symptom dimensions only in participants diagnosed with depressive and/or anxiety disorder. CIMT was not associated with cognitive/affective or anxiety/arousal depressive symptoms in none of the subgroups. The association of augmentation index and depressive symptom dimensions in individuals diagnosed with depression and/or anxiety was stable over time (see Table 5).
143 Table 2. Factor loadings and fit indices of confirmatory factor analyses across time points. Item Baseline (W1); N = 2,505) First follow-up assessment wave (W2); N = 2,316 Second follow-up assessment wave (W3); N = 2,149 Mood/Cognition Factor 5. Feeling sad .878 .864 .893 8. Reactivity of mood .696 .664 .685 10. Quality of Mood .822 .840 .838 11/12. Appetite disturbance .569 .595 .611 15. Concentration and decision Making .774 .775 .775 16. View of Myself .774 .777 .795 17. View of My Future .809 .793 .800 18. Thoughts of Death or Suicide .762 .727 .715 20. Energy Level .780 .772 .785 22. Interest in Sex .593 .592 .599 29. Interpersonal Sensitivity .650 .668 .708 Anxiety/Somatic Factor 6. Feeling Irritable .757 .770 .772 23. Feeling slowed down .802 .827 .800 24. Feeling restless .663 .688 .706 25. Aches and pains .539 .569 .566 26. Other bodily symptoms .601 .638 .660 27. Panic/Phobic symptoms .552 .598 .559 28. Constipation/diarrhea .438 .469 .490 30. Leaden Paralysis/Physical Energy .823 .814 .804 Fit indices CFI .959 .954 .959 TLI .954 .948 .954 RMSEA .068 (.065 - .071) .070 (.068 - .073) .068 (.065 - .071) CFI: Comparative fit index; TLI: Tucker-Lewis index; RMSEA: Root mean square error of approximation 142 Linear mixed models – Main analyses
Table 3 shows the results of the linear mixed models according to the
outcome used. Augmentation index was associated with both depressive symptom dimensions (mood/cognition, β = 0.01 [0.00 – 0.01], p = .03; anxiety/somatic, β = 0.01 [0.00 – 0.01], p = .03). Neither total CIMT (mood/cognition, β = -.50 [-1.07 – 0.07], p = .09; anxiety/somatic, β = -.40 [-0.89 – 0.10]. p = .12) nor presence of plaque (mood/cognition, β = .09 [-0.09 – 0.26], p = .35; anxiety/somatic, β = .09 [-0.06 – 0.25]. p = .25) was associated with depressive symptom dimensions. None of the interaction terms between predictors and time were significantly associated with depressive symptom dimensions. This implicates that the association between subclinical heart disease and depressive symptom dimensions was stable over time.
Interaction models and sensitivity analyses
We tested for interaction effects between predictors and sex, age and diagnosis of depressive and anxiety disorders. Potential interactions were found only between two predictor variables (CIMT and augmentation index) and diagnosis of depressive and anxiety disorders (Table 4). Subgroup analyses indicated that augmentation index was associated with cognitive/affective and anxiety/arousal depressive symptom dimensions only in participants diagnosed with depressive and/or anxiety disorder. CIMT was not associated with cognitive/affective or anxiety/arousal depressive symptoms in none of the subgroups. The association of augmentation index and depressive symptom dimensions in individuals diagnosed with depression and/or anxiety was stable over time (see Table 5).
145
Table 4. Interaction models
Mood/Cognition factor Anxiety/Somatic factor
Fixed effect β (95% CI) p β (95% CI) p
CIMT × Sex .075 (-0.74 - 0.89) .856 -.009 (-0.72 - 0.70) .980 CIMT × Age .013 (-0.03 - 0.05) .506 .008 (-0.03 - 0.04) .649 CIMT × Diagnosis of depression and/or anxiety .668 (-0.10 - 1.43) .087 .601 (-0.07 - 1.28) .078 Presence of plaque × Sex -.141 (-0.46 - 0.18) .393 -.114 (-0.40 - 0.17) .426 Presence of plaque × Age -.009 (-0.03 - 0.01) .439 -.008 (-0.03 - 0.01) .400 Presence of plaque × Diagnosis of depression and/or anxiety .024 (-0.31 - 0.36) .888 . 030 (-0.27 - 0.33) .842 Augmentation Index × Sex .002 (-0.01 - 0.01) .639 -.001 (-0.01 - 0.01) .714 Augmentation Index × Age -.001 (-0.01 - 0.01) .394 -.001 (-0.00 - 0.01) .281 Augmentation Index × Diagnosis of depression and/or anxiety .008 (0.00 - 0.01) .041 . 006 (-0.00 - 0.01) .078 All predictors were included in separate models. Interaction terms and main effects were calculated in different models. All models included age, sex, BMI, level of physical activity, smoking and level of education. Statistically significant associations are marked in bold 144 Table 3. Linear mixed models – Main analyses (N = 600) Mood/Cognition factor Anxiety/Somatic factor
Fixed effect β (95% CI) p β (95% CI) p
CIMT - .500 (-1.07 - 0.07) .087 -.399 (-0.90 - 0.10) .115
CIMT × Time .010 (-0.15 - 0.17) .903 -.003 (-0.14 - 0.13) .971
Presence of plaque .086 (-0.09 - 0.26) .345 .091 (-0.06 - 0.25) .246
Presence of plaque ×
Time -.015 (-0.08 - 0.05) .657 -.016 (-0.07 - 0.04) .581 Augmentation Index .007 (0.00 - 0.01) .031 .006 (0.00 - 0.01) .025 Augmentation Index × Time -.001 (-0.00 - 0.00) .506 -.001 (-0.00 - 0.00) .352 All predictors were included in separate models. Interaction terms and main effects were calculated in different models. All models included age, sex, BMI, level of physical activity, smoking and level of education. Statistically significant associations are marked in bold.
145
Table 4. Interaction models
Mood/Cognition factor Anxiety/Somatic factor
Fixed effect β (95% CI) p β (95% CI) p
CIMT × Sex .075 (-0.74 - 0.89) .856 -.009 (-0.72 - 0.70) .980 CIMT × Age .013 (-0.03 - 0.05) .506 .008 (-0.03 - 0.04) .649 CIMT × Diagnosis of depression and/or anxiety .668 (-0.10 - 1.43) .087 .601 (-0.07 - 1.28) .078 Presence of plaque × Sex -.141 (-0.46 - 0.18) .393 -.114 (-0.40 - 0.17) .426 Presence of plaque × Age -.009 (-0.03 - 0.01) .439 -.008 (-0.03 - 0.01) .400 Presence of plaque × Diagnosis of depression and/or anxiety .024 (-0.31 - 0.36) .888 . 030 (-0.27 - 0.33) .842 Augmentation Index × Sex .002 (-0.01 - 0.01) .639 -.001 (-0.01 - 0.01) .714 Augmentation Index × Age -.001 (-0.01 - 0.01) .394 -.001 (-0.00 - 0.01) .281 Augmentation Index × Diagnosis of depression and/or anxiety .008 (0.00 - 0.01) .041 . 006 (-0.00 - 0.01) .078 All predictors were included in separate models. Interaction terms and main effects were calculated in different models. All models included age, sex, BMI, level of physical activity, smoking and level of education. Statistically significant associations are marked in bold 144 Table 3. Linear mixed models – Main analyses (N = 600) Mood/Cognition factor Anxiety/Somatic factor
Fixed effect β (95% CI) p β (95% CI) p
CIMT - .500 (-1.07 - 0.07) .087 -.399 (-0.90 - 0.10) .115
CIMT × Time .010 (-0.15 - 0.17) .903 -.003 (-0.14 - 0.13) .971
Presence of plaque .086 (-0.09 - 0.26) .345 .091 (-0.06 - 0.25) .246
Presence of plaque ×
Time -.015 (-0.08 - 0.05) .657 -.016 (-0.07 - 0.04) .581 Augmentation Index .007 (0.00 - 0.01) .031 .006 (0.00 - 0.01) .025 Augmentation Index × Time -.001 (-0.00 - 0.00) .506 -.001 (-0.00 - 0.00) .352 All predictors were included in separate models. Interaction terms and main effects were calculated in different models. All models included age, sex, BMI, level of physical activity, smoking and level of education. Statistically significant associations are marked in bold.
147
Discussion Main findings
The present study assessed whether CIMT, presence of plaque and arterial stiffness are differentially associated with depressive symptom dimensions in middle-aged individuals. Augmentation index was modestly but significantly associated with both mood/cognition and anxiety/somatic depressive symptom dimensions. Moreover, the course of the association of augmentation index and depressive symptom dimensions was stable over time. Neither total CIMT nor presence of plaque were associated with any of the depressive symptom dimensions.
We assessed for potential interaction effects between markers of subclinical heart disease with sex, age and diagnosis of depressive and/or anxiety disorders in the association with depressive symptom dimensions. There was a potential interaction between augmentation index and diagnosis of depressive and/or anxiety disorders in the association with both depressive symptom dimensions. In subgroup analyses, augmentation index was modestly but significantly associated with increased levels of both depressive symptom dimensions only in participants that have been previously diagnosed with depressive or anxiety disorders. The course of the association of augmentation index with depressive symptom dimensions was stable over time. In subgroup analyses, CIMT was not associated with any of the depressive symptom dimensions. Subclinical heart disease and depressive symptoms The longitudinal association between subclinical heart disease and depressive symptoms has been previously investigated, and mixed findings have been reported 34,35. Prugger and colleagues reported that both total CIMT and presence of plaque are associated with depressive symptoms in a large sample of French participants 34. Nonetheless, another longitudinal
study conducted in elderly participants, reported that subclinical heart disease (as measured by presence of plaque and aortic calcification) did
not increase the risk of late life incident depression 35. 146 Ta bl e 5. L in ea r m ix ed m od els – Subg roup ana ly se s Par tic ipant s w ith a di agnos is of de pr es si ve or anxi et y di so rd er s (N = 1 75 ) Par tic ipant s w ithout a di agnos is of de pr es si ve or anxi et y di so rd er s ( N = 4 25 ) M ood/ Cog ni tion fa ct or An xi ety /S om ati c fa cto r M ood/ Cog ni tion fa ct or An xi ety /S om ati c fa cto r Fi xed ef fec t β ( 95 % C I) p β ( 95 % C I) p β ( 95 % C I) p β ( 95 % C I) p CM ITTot al -.0 97 (-0. 93 - 0. 74) .8 21 -.0 91 (-0. 81 - 0. 62) .8 03 -.2 08 (-0. 80 - 0. 39) .4 92 -.1 41 (-0. 67 - 0. 38) .5 98 CM ITTo ta l X T im e .0 19 (-0. 32 - 0. 35) .9 13 -.0 03 (-0. 29 -0. 28) .9 86 -.0 74 (0 .2 5 - 0. 10) . 4 05 -.0 68 (-0. 22 - 0. 08) .3 71 Augm ent at ion Inde x .0 10 (0 .0 0 - 0. 02) .0 20 .0 08 (0 .0 0 - 0. 02) .0 27 .0 04 (-0. 00 - 0. 01) .1 83 .0 04 (-0. 00 - 0. 10 ) .1 38 Augm ent at ion Inde x X Ti m e .0 02 (-0. 00 - 0. 01) .2 95 .0 02 (-0. 00 - 0. 00) .2 81 -.0 01 (-0. 00 0. 00 ) .1 37 -.0 01 (-0. 00 - 0. 00) .0 62 CM IT : Ca rot id int im a m edi a thi ck ne ss Al l pr edi ct or s w er e inc lude d in se pa ra te m ode ls. Int er ac tion te rm s a nd m ai n ef fe ct s w er e ca lc ul at ed in di ffe re nt m ode ls. Al l m od el s i nc lud ed ag e, sex , B M I, lev el o f ph ys ic al ac tiv ity , s m ok in g an d lev el o f ed uc at io n. St at ist ic al ly si gn ifi can t as so ci at io ns a re m ar ke d in bol d.
147
Discussion Main findings
The present study assessed whether CIMT, presence of plaque and arterial stiffness are differentially associated with depressive symptom dimensions in middle-aged individuals. Augmentation index was modestly but significantly associated with both mood/cognition and anxiety/somatic depressive symptom dimensions. Moreover, the course of the association of augmentation index and depressive symptom dimensions was stable over time. Neither total CIMT nor presence of plaque were associated with any of the depressive symptom dimensions.
We assessed for potential interaction effects between markers of subclinical heart disease with sex, age and diagnosis of depressive and/or anxiety disorders in the association with depressive symptom dimensions. There was a potential interaction between augmentation index and diagnosis of depressive and/or anxiety disorders in the association with both depressive symptom dimensions. In subgroup analyses, augmentation index was modestly but significantly associated with increased levels of both depressive symptom dimensions only in participants that have been previously diagnosed with depressive or anxiety disorders. The course of the association of augmentation index with depressive symptom dimensions was stable over time. In subgroup analyses, CIMT was not associated with any of the depressive symptom dimensions. Subclinical heart disease and depressive symptoms The longitudinal association between subclinical heart disease and depressive symptoms has been previously investigated, and mixed findings have been reported 34,35. Prugger and colleagues reported that both total CIMT and presence of plaque are associated with depressive symptoms in a large sample of French participants 34. Nonetheless, another longitudinal
study conducted in elderly participants, reported that subclinical heart disease (as measured by presence of plaque and aortic calcification) did
not increase the risk of late life incident depression 35. 146 Ta bl e 5. L in ea r m ix ed m od els – Subg roup ana ly se s Par tic ipant s w ith a di agnos is of de pr es si ve or anxi et y di so rd er s (N = 1 75 ) Par tic ipant s w ithout a di agnos is of de pr es si ve or anxi et y di so rd er s ( N = 4 25 ) M ood/ Cog ni tion fa ct or An xi ety /S om ati c fa cto r M ood/ Cog ni tion fa ct or An xi ety /S om ati c fa cto r Fi xed ef fec t β ( 95 % C I) p β ( 95 % C I) p β ( 95 % C I) p β ( 95 % C I) p CM ITTot al -.0 97 (-0. 93 - 0. 74) .8 21 -.0 91 (-0. 81 - 0. 62) .8 03 -.2 08 (-0. 80 - 0. 39) .4 92 -.1 41 (-0. 67 - 0. 38) .5 98 CM ITTo ta l X T im e .0 19 (-0. 32 - 0. 35) .9 13 -.0 03 (-0. 29 -0. 28) .9 86 -.0 74 (0 .2 5 - 0. 10) . 4 05 -.0 68 (-0. 22 - 0. 08) .3 71 Augm ent at ion Inde x .0 10 (0 .0 0 - 0. 02) .0 20 .0 08 (0 .0 0 - 0. 02) .0 27 .0 04 (-0. 00 - 0. 01) .1 83 .0 04 (-0. 00 - 0. 10 ) .1 38 Augm ent at ion Inde x X Ti m e .0 02 (-0. 00 - 0. 01) .2 95 .0 02 (-0. 00 - 0. 00) .2 81 -.0 01 (-0. 00 0. 00 ) .1 37 -.0 01 (-0. 00 - 0. 00) .0 62 CM IT : Ca rot id int im a m edi a thi ck ne ss Al l pr edi ct or s w er e inc lude d in se pa ra te m ode ls. Int er ac tion te rm s a nd m ai n ef fe ct s w er e ca lc ul at ed in di ffe re nt m ode ls. Al l m od el s i nc lud ed ag e, sex , B M I, lev el o f ph ys ic al ac tiv ity , s m ok in g an d lev el o f ed uc at io n. St at ist ic al ly si gn ifi can t as so ci at io ns a re m ar ke d in bol d.
149
Many studies demonstrated that symptom dimensions of depression are differentially associated with several cardiovascular-related variables. Somatic/affective but not cognitive/affective symptoms of depression have been shown to be associated with adverse medical
prognosis 28, and elevated inflammatory markers, such as C-reactive
protein (CRP), white blood cell count (WBC) and N-terminal pro-brain
natriuretic peptide (NTproBNP) 37 in patients with heart disease.
Moreover, lower heart rate variability has been shown to be associated with somatic/affective but not with cognitive/affective symptoms of
depression in patients with heart disease 38 and in adolescents 39. Markers
of visceral obesity such as waist circumference and waist-to-hip ratio have also been shown to be associated with somatic/affective but not with
cognitive/affective symptoms of depression in the elderly 40. Previous
evidence also suggests that questionnaires that include somatic/affective
symptoms can yield inflated total scores in patients with heart disease 41.
Findings of the present study suggest that there is not a differential association between subclinical heart disease and symptom dimensions of depression in middle-aged adults. Nonetheless, a large body of evidence suggests a differential association between depressive symptom dimensions and cardiovascular-related variables. Therefore, we suggest that authors of future studies assessing this association should consider including additional post-hoc analyses in which depressive symptom dimensions are taken into account in order to further clarify the conditions under which differential associations arise.
Limitations and strengths
Findings of the present study should be interpreted in the light of the study limitations. Our study sample was rather young as compared with previous studies. Despite not being a limitation per se, that could have limited our power to detect statistically significant associations. Moreover, our follow-up time was only two years, which is substantially
shorter than other longitudinal studies on the same topic 34,35.
The present study also has some points of strength. We used a
148
Moreover, the previous studies also had important methodological differences if compared to the present study. For instance, the Three-City Study was composed of a substantially larger sample (N = 4,125) of older
participants (mean age = 73 years) 34. In the present study, 649
participants were included, and these participants had a mean age of 46 years. The young age of the sample of the present study poses as a possible explanation for the absence of strong associations between subclinical heart disease and depressive symptom dimensions. Previous evidence demonstrates that subclinical heart disease is more prevalent in
older participants 14. Nonetheless, it is interesting that associations were
found for arterial stiffness in the present study, suggesting that this global measure of vascular pathology might be more sensitive than carotid measures in our younger sample. Future studies may consider replicating these analyses in older participants.
Depressive symptom dimensions and markers of heart disease
To our knowledge, only one study assessed the longitudinal association between subclinical heart disease and depressive symptom
dimensions 36. However, in this study depressive symptom dimensions
were modelled as predictors of subclinical heart disease. Authors reported that elevated baseline overall depressive symptoms were associated with CIMT. Nonetheless, post-hoc analyses showed that somatic/vegetative but not cognitive/affective symptoms of depression were associated with 3-year changes in CIMT.
Somatic/affective symptoms of depression also explained the
association of depression and subclinical heart disease, in another study 21.
Authors suggested that total depressive symptom scores might be inflated as a result of somatic/affective symptoms of depression that are secondary to subclinical heart disease. Based on this belief, authors postulated that depression might be an epiphenomenon of subclinical heart disease, rather than causally related to heart disease. Nonetheless, this study has been conducted in cross-sectional data, and that makes it difficult to infer on causality.
149
Many studies demonstrated that symptom dimensions of depression are differentially associated with several cardiovascular-related variables. Somatic/affective but not cognitive/affective symptoms of depression have been shown to be associated with adverse medical
prognosis 28, and elevated inflammatory markers, such as C-reactive
protein (CRP), white blood cell count (WBC) and N-terminal pro-brain
natriuretic peptide (NTproBNP) 37 in patients with heart disease.
Moreover, lower heart rate variability has been shown to be associated with somatic/affective but not with cognitive/affective symptoms of
depression in patients with heart disease 38 and in adolescents 39. Markers
of visceral obesity such as waist circumference and waist-to-hip ratio have also been shown to be associated with somatic/affective but not with
cognitive/affective symptoms of depression in the elderly 40. Previous
evidence also suggests that questionnaires that include somatic/affective
symptoms can yield inflated total scores in patients with heart disease 41.
Findings of the present study suggest that there is not a differential association between subclinical heart disease and symptom dimensions of depression in middle-aged adults. Nonetheless, a large body of evidence suggests a differential association between depressive symptom dimensions and cardiovascular-related variables. Therefore, we suggest that authors of future studies assessing this association should consider including additional post-hoc analyses in which depressive symptom dimensions are taken into account in order to further clarify the conditions under which differential associations arise.
Limitations and strengths
Findings of the present study should be interpreted in the light of the study limitations. Our study sample was rather young as compared with previous studies. Despite not being a limitation per se, that could have limited our power to detect statistically significant associations. Moreover, our follow-up time was only two years, which is substantially
shorter than other longitudinal studies on the same topic 34,35.
The present study also has some points of strength. We used a
148
Moreover, the previous studies also had important methodological differences if compared to the present study. For instance, the Three-City Study was composed of a substantially larger sample (N = 4,125) of older
participants (mean age = 73 years) 34. In the present study, 649
participants were included, and these participants had a mean age of 46 years. The young age of the sample of the present study poses as a possible explanation for the absence of strong associations between subclinical heart disease and depressive symptom dimensions. Previous evidence demonstrates that subclinical heart disease is more prevalent in
older participants 14. Nonetheless, it is interesting that associations were
found for arterial stiffness in the present study, suggesting that this global measure of vascular pathology might be more sensitive than carotid measures in our younger sample. Future studies may consider replicating these analyses in older participants.
Depressive symptom dimensions and markers of heart disease
To our knowledge, only one study assessed the longitudinal association between subclinical heart disease and depressive symptom
dimensions 36. However, in this study depressive symptom dimensions
were modelled as predictors of subclinical heart disease. Authors reported that elevated baseline overall depressive symptoms were associated with CIMT. Nonetheless, post-hoc analyses showed that somatic/vegetative but not cognitive/affective symptoms of depression were associated with 3-year changes in CIMT.
Somatic/affective symptoms of depression also explained the
association of depression and subclinical heart disease, in another study 21.
Authors suggested that total depressive symptom scores might be inflated as a result of somatic/affective symptoms of depression that are secondary to subclinical heart disease. Based on this belief, authors postulated that depression might be an epiphenomenon of subclinical heart disease, rather than causally related to heart disease. Nonetheless, this study has been conducted in cross-sectional data, and that makes it difficult to infer on causality.
