Depression, anxiety and major adverse cardiovascular and cerebrovascular events in patients following coronary artery bypass graft surgery: a five year longitudinal cohort study

Background Although depression and anxiety have been implicated in risk for major adverse cardiovascular and cerebrovascular events (MACCE), a theoretical approach to identifying such putative links is lacking. The objective of this study was to examine the association between theoretical conceptualisations of depression and anxiety with MACCE at the diagnostic and symptom dimension level. Methods Before coronary artery bypass graft (CABG) surgery, patients (N = 158; 20.9 % female) underwent a structured clinical interview to determine caseness for depression and anxiety disorders. Depression and anxiety disorders were arranged into the distress cluster (major depression, dysthymia, generalized anxiety disorder, post-traumatic stress disorder) and fear cluster (panic disorder, agoraphobia, social phobia). Patients also completed the self-report Mood and Anxiety Symptom Questionnaire, measuring anhedonia, anxious arousal and general distress/negative affect symptom dimensions. Incident MACCE was defined as fatal or non-fatal; myocardial infarction, unstable angina pectoris, repeat revascularization, heart failure, sustained arrhythmia, stroke or cerebrovascular accident, left ventricular failure and mortality due to cardiac causes. Time-to-MACCE was determined by hazard modelling after adjustment for EuroSCORE, smoking, body mass index, hypertension, heart failure and peripheral vascular disease. Results In the total sample, there were 698 cumulative person years of survival for analysis with a median follow-up of 4.6 years (interquartile range 4.2 to 5.2 years) and 37 MACCE (23.4 % of total). After covariate adjustment, generalized anxiety disorder was associated with MACCE (hazard ratio [HR] = 2.79, 95 % confidence interval [CI] 1.00-7.80, p = 0.049). The distress disorders were not significantly associated with MACCE risk (HR = 2.14; 95 % CI .92-4.95, p = 0.077) and neither were the fear-disorders (HR = 0.24, 95 % CI .05-1.20, p = 0.083). None of the symptom dimensions were significantly associated with MACCE. Conclusions Generalized anxiety disorder was significantly associated with MACCE at follow-up after CABG surgery. The findings encourage further research pertaining to generalized anxiety disorder, and theoretical conceptualizations of depression, general distress and anxiety in persons undergoing CABG surgery.


Introduction
Depression dominates recent understandings of the putative links between negative emotions and major adverse cardiovascular and cerebrovascular events (MACCE, e.g. myocardial infarction, stroke) in patients with coronary heart disease (CHD) [1]. However, despite depression treatment with psychotherapy and antidepressant interventions [2,3], a consistent reduction in MACCE remains elusive in the population with comorbid depression and CHD [3], raising questions about the focus of interventions [4]. Indeed, depression is but one of a spectrum of disorders (e.g. anxiety, post-traumatic stress disorder and panic disorder) which are purported to deleteriously affect CHD outcomes [5][6][7][8][9]. In fact, most negative emotional risk factors for MACCE share a common predisposition to negative affectivity (NA), also known as neuroticism, evident at both the theoretical [10,11] and measurement level [12][13][14]. Past attempts to clarify the risk of MACCE attributable to depression independent of NA and anxiety have not been consistently supported [12,[15][16][17][18][19], and a theoretical approach to this field is lacking [13]. Therefore the aim of this study is to utilize theoretical conceptualizations of depression, NA and anxiety at the diagnostic disorder, cluster and symptom level. Such theoretical conceptualizations will, in turn, be employed to predict the prognostic MACCE outcome in a sample of patients undergoing coronary artery bypass graft (CABG) surgery.
Contemporary understandings of psychiatric nomenclature indicate that NA is the most ubiquitous feature of depression and anxiety disorders [10,[20][21][22][23]. Beyond the NA commonality, recent research also shows that certain depression and anxiety disorders relate more strongly to each other, and do not necessarily fall within prescribed diagnostic categories of anxiety and depression. Specifically, previous empirical work largely supports a theoretical model with at least two groups of affective disorders. The first is collectively labelled the distress disorders and is comprised by major depression, dysthymia, post-traumatic stress disorder and generalized anxiety disorder (GAD) [10,[20][21][22][23][24]. Research also supports that symptoms of anhedonia are a dimensional marker for the distress disorders, and therefore, not only for major depression [10]. The second group of disorders are collectively labeled the fear disorders which is comprised of panic disorder, agoraphobia and social phobia [10,[20][21][22][23][24]. Symptoms of anxious arousal are a dimensional marker for the fear disorders [10], and therefore, not for other traditional anxiety disorders such as obsessive-compulsive disorder and post-traumatic stress disorder.
Along these lines, prior research in CHD populations has most commonly examined uni-polar and dysthymic depression subtypes [25][26][27] and anhedonia in relation to MACCE [28][29][30][31]. Also, some evidence in CHD populations suggests that post-traumatic stress disorder is associated with MACCE recurrence [32]. By contrast, GAD findings are more mixed [33][34][35] and there are too few studies on other anxiety disorder subtypes [36]. We are only aware of one study that assess anhedonia contemporaneously with anxious arousal and NA dimensions of the distress and fear clusters [37]. With these limitations in mind, we sought to extend the analysis of MACCE to 4.6 year follow-up. Based on the predominant research to date, we hypothesized that the distress disorder cluster and major depression especially would be significantly associated with MACCE at follow-up. Secondly, we hypothesized that the anhedonia dimension would be significantly associated with MACCE at follow-up.

Patients
Informed consent and ethics approval was obtained for this study (The University of Adelaide Human Research Ethics Committee approval # H-010-2007, The Flinders Medical Centre Human Research Ethics Committee approval 112/067) and the methods have been reported previously [37,38]. Briefly, recruitment took place at the Flinders Medical Centre, South Australia, between February 2007 and March 2009. Patients were considered eligible if aged ≥ 18 years and undergoing CABG with cardiopulmonary bypass, and with or without concomitant valve procedures. From 252 eligible patients 84 patients were ineligible: declined (n = 23), communication difficulty (n = 4), participating in another research trial (n = 10), health reasons (n = 2), developmental disorder (n = 2), dementia (n = 1), living in a remote community with no contact details (n = 11), late addition to surgical list (n = 2), on the hospital ward <24 h (n = 16), time constraints/admitted on weekend (n = 13). From 252 eligible patients we recruited 168 and 10 patients were excluded further: surgery postponed indefinitely (n = 1), withdrawal of consent (n = 1), current psychosis and/or taking anti-psychotic medications (n = 3), current or past alcohol and/ or substance abuse (n = 5). This left a total sample of 158 patients (63 % participation rate). A flow chart of participants through the study is shown in Fig. 1. Non-respondents and excluded patients were more likely to identify as Aboriginal or Torres Strait Islander χ 2 (1) = 5.85, p = .02 but were otherwise not discrepant from participants on baseline demographic and comorbid conditions. Medical data were prospectively collected by medical officers at pre-surgical consultation and entered directly onto an electronic database with quality assurance maintained at weekly database meetings by the third author. Data definitions utilize those of the Australian Society of Cardiac and Thoracic Surgeons [39] including permanent stroke, cerebrovascular accident or central neurological deficit persisting for longer than 72 h, myocardial infarction (2 or more of: enzyme level elevation; new wall motion abnormalities; serial EGG showing new Q waves).

Structured interview
Patients were assessed a median of 3 days preoperatively (interquartile range 1-3 days). The MINI International Neuropsychiatric Interview (MINI) 5.0.0 [40,41] was employed to determine psychiatric disorders by an intern psychologist (first author, with 1,000 h clinical psychology experience, employed 0.4 full-time equivalent in the study hospital). The MINI has high sensitivity and specificity to detect current mood and anxiety disorders, with Kappa coefficients (κ = .86 -.96) suggesting favorable agreement with Diagnostic and Statistical manual of Mental Disorders-IV (DSM-IV) [40,42]. The MINI hierarchical diagnostic criteria stipulate that a GAD diagnosis cannot be made with a concurrent major depression diagnosis thus precluding comorbidity of these disorders and providing the advantage of classifying only primary affective disorders. Disorders were arranged into the distress cluster (major depression, dysthymia, post-traumatic stress disorder and GAD) and the fear disorder cluster (panic disorder, agoraphobia, and social phobia). We also considered affective disorders with more than 10 % prevalence at baseline as candidates for analysis in relation to MACCE.

Self-report sistress assessments Mood and anxiety symptom questionnaire
The self-report mood and anxiety symptom questionnaire (MASQ) was used to measure anhedonia, anxious arousal and general NA [43]. Based on the work of Wardenaar et al. [44,45], we constructed a 30-item short form where 10-items each are allocated to an anxious arousal, anhedonia/low positive affect and general NA scale. Example items of anxious arousal include "Was trembling or shaking"; "Had hot or cold spells". The anhedonic depression scale utilizes reverse-keyed items assessing positive emotional experiences including "Felt like I had a lot to look forward to"; "Felt like I was having a lot of fun". Example items of the MASQgeneral NA scale include "Felt irritable"; "Worried a lot about things". The MASQ has been found to fit the three-dimensional model (general NA, anhedonia, anxious arousal) indicating good construct validity with high discriminant validity [44], and was recently validated in the English language [46]. Previous psychometric research with the MASQ has established good psychometric properties of this questionnaire [44,[46][47][48][49][50][51][52]. In the present sample satisfactory internal consistency was observed (Cronbach's alpha coefficients; general NA = 0.88; anhedonia = 0.84; anxious arousal = 0.77).

Major adverse cardiovascular and cerebrovascular events
The Australian Institute of Health and Welfare's National Death Index was utilized to determine mortality data until the study census date 31st December 2013, according to the International Classification of Diseases (ICD) 10th Revision codes [53]. Hospital admission after discharge from the index CABG procedure was ascertained from patient medical records and electronic admission data linkage between hospitals in ICD code [53]. The MACCE endpoint was defined as fatal or non-fatal hospitalisation for myocardial infarction, unstable angina pectoris, repeat revascularization, sustained arrhythmia, stroke or cerebrovascular accident, heart failure, left ventricular failure and mortality due to cardiac causes, as consistent with previous research [54]. In this manner, any non-cardiac death is censored from the analyses at the date of death. All MACCE were inspected blinded to psychological distress scores.

Statistical analysis
Data analysis was performed with SPSS® 22.0 statistical software package (SPSS Inc., Chicago, IL). Descriptive comparisons were made with the t-test, the chi-square test or the Fisher's exact test as appropriate. All statistical tests were two-tailed, an alpha value p ≤ .05 was considered statistically significant, and no adjustment was made for multiple comparisons based on the recommendations of Rothman [55]. Adjusted Cox proportional model hazard ratios (HR) and 95 % confidence intervals (CIs) were used to determine the risk of MACCE associated with negative emotions. Candidate covariates for hazard models were selected a priori based on the literature to cover the covariates that are associated with depression and anxiety (independent variables) [56]. We also selected covariates associated with MACCE risk and cardiac surgery morbidity outcomes (dependent variables) [57,58]. The candidate covariates included smoking, body mass index, hypertension, heart failure, peripheral vascular disease, and the European System for Cardiac Operative Risk Evaluation [59] (EuroSCORE). The EuroSCORE is calculated from 17 risk factors including age, sex, left ventricular dysfunction, previous cardiac surgery, elevated creatinine and concomitant procedures among others, and is associated with MACCE and survival in the long-term [57,58]. The proportionality of hazards assumption was checked initially by entering covariates as interactions with time and also ascertained graphically in final models via examination of the log-minus-log plot of survival function, and the Schoenfield residuals.
We examined the associations between clusters, disorders and symptom dimensions with future MACCE in three respective models. Model 1 was comprised of covariates and the diagnostic categories of prevalent affective disorders (GAD, panic disorder and major depression). Model 2 was comprised of covariates and the affective disorders were arranged into distress and fear clusters. Model 3 concerned the symptom dimensions of anhedonia, anxious arousal and general NA as measured with the MASQ.

Descriptive characteristics
The final sample consisted of 158 CABG patients between 36 and 87 years (mean age = 64.7 years ± 10.6, 20.9 % women, 11.4 % concomitant valve surgery). Baseline characteristics stratified by MACCE are shown in Table 1. Patients experiencing a MACCE were significantly older and were characterized by a higher proportion of hypertension, heart failure and peripheral vascular disease. In the total sample, there were 698 cumulative person years of survival for analysis with a median follow-up of 4.6 years (interquartile range 4.2 to 5.2). There were 37 MACCE events (23.4 % of total), most commonly deaths due to CHD (n = 15), and nonfatal myocardial infarction (n = 13), incident heart failure (n = 5) and stroke (n = 4).

Prevalence of affective disorders
Diagnostic interview indicated that major depression was most common (n = 27, 17.1 %), followed by panic disorder (n = 12, 10.8 %) and GAD (n = 16, 10.2 %). In total, there were 39 (24.7 %) patients meeting at least one diagnosis of the distress cluster and 21 (13.3 % of total) participants meeting criteria from the fear cluster.

Risk factors for major adverse cardiovascular events
Examination of the affective disorders, clusters and symptom dimensions are shown in Table 2. Model 1 suggested that only GAD was significantly associated with an increased risk of MACCE (adjusted HR = 2.79, 95 % CI 1.00 to 7.80, p = .05). Fig. 2 depicts the divergence in cumulative survival curves for MACCE in the period after CABG according to GAD status before surgery and is evident within the first year. Neither depression nor panic disorder was associated with MACCE (both p >. 20).
[Model 2 shows the association between the distress disorders and MACCE (adjusted HR = 2.14, 95 % CI .92 to 4.95, p = .08) and the fear disorders and MACCE (adjusted HR = .24, 95 % CI .05 to 1.20, p = .08). Although both associations were above conventional significance, there was a trend for an increased risk for MACE associated with the distress disorders and a reduced risk associated with the fear disorders.
Model 3 shows that anhedonia, anxious arousal and general NA were not associated with MACCE (all p >. 30). Covariates significantly associated with MACCE in Models 1 to 3 included heart failure (HR 2.31 to 2.40) and EuroSCORE (HR 1.05 to 1.06).

Sensitivity analyses
Considering that there was a significant association between GAD and MACCE, but a relatively limited number of MACCE, we performed sensitivity analyses showing the change in effect size attributable to GAD with adjustment for each individual covariate. This analytical strategy involved entering only a single covariate into the GAD model. This approach has been adopted previously [35] and is less prone to overfitting the Cox regression model. The results indicated that a 5 % change in the GAD-MACCE hazard ratio was evident for panic disorder, hypertension and tobacco smoking, though GAD remained significantly associated with MACCE (Table 3). In our second sensitivity analysis, we assessed whether theoretical conceptualizations of depression and anxiety were associated with the acute cardiovascular events of myocardial infarction and cardiac death (combined n = 28). In unadjusted analyses, it was evident that only GAD was related to acute cardiovascular events (unadjusted HR = 2.92, 95 % CI 0.98 to 8.64, p = .054).

Discussion
This study was the first to comprehensively examine the MACCE risk attributable to affective disorders at the  [33,35]. With respect to the arrangement of disorders into clusters, it was evident that the distress disorders were not significantly associated with increased MACCE risk, and therefore not supporting the hypothesis. The symptom dimensions anhedonia and anxious arousal were not significantly associated with MACCE, thus not supporting the hypothesis nor previous work [30,31]. The study showed no significant association between MACCE and the distress or fear disorders, their symptom dimensions, panic disorder or major depression. Nonetheless, the GAD findings align with some larger studies [33,35], albeit contrasting to Parker and colleagues' cohort [34]. Previously in a large epidemiological survey, Goodwin and colleagues [60] showed that GAD was most strongly associated with lower CHD risk in cross-sectional analyses whereas the mood disorders were not. Potential mechanisms underlying the association with CHD include GAD patients' propensity toward diminished heart rate variability [61], elevated heart rate, smoking, and hypertension [62], parallel to what has been reported amongst depressed CHD and CABG surgery patients [63][64][65][66]. Our arrangement of disorders into the distress cluster, inclusive of GAD and depression, did not show a significant association with MACCE (p = .08). However, this finding was possibly limited by the infrequent occurrence of other disorders especially dysthymia and post-traumatic stress disorder, considering that an emerging literature documents an association between these disorders and MACCE in CHD populations [25,27,67].
When symptom dimensions were examined, there was no significant association between MASQ anhedonia, anxious arousal or general NA with subsequent MACCE. The findings with respect to anhedonia contrast to Denollet and colleagues study [28] which reported that a four-item anhedonia measure was associated with MACCE in percutaneous coronary intervention patients. Also, Leroy and colleagues' study [31] showed that the Chapman Physical Anhedonia Scale was associated with twofold increase in MACCE at 3-year follow-up after acute coronary syndrome. Further assessment of the anhedonia symptom dimension may provide insight as to specific underlying mechanisms linking depression and MACCE. For example, the anhedonia symptom dimension is highly associated with biological mechanisms of cardiopathogenesis such as the hypothalamus-pituitary-adrenal axis [68] and the metabolic syndrome [69]. Moreover, symptom dimensions are also associated with other risk factors for cardiopathogenesis such as exposure to childhood trauma, social disadvantage, and adversity in adult life [70].
The current findings may have clinical relevance for the population with GAD who are facing CABG surgery, or comorbid CHD and anxiety disorders generally [71]. Specifically, the findings raise the possibility that psychological interventions targeting GAD are warranted in CABG surgery patients. Recently it was shown that collaborative care programs were effective for a reduction in generalized anxiety and depression symptoms [72,73]. Moreover, exercise, anxiolytic use and cognitive behavioral therapy for GAD were associated with a reduction in somatic depressive symptoms among persons with comorbid depression disorder-GAD and heart failure [74]. In the population undergoing CABG surgery, prior interventions have indicated small to medium treatment effect sizes regarding reduction in depression and anxiety symptoms [75][76][77][78]. Moreover, non-pharmacological intervention is especially advantageous in the CABG population given that selective serotonin re-uptake inhibitors may pose a morbidity risk particularly relating to postoperative hemorrhage [79][80][81]. This study's main methodological strength is the delineation of disorders at the diagnostic and cluster level and the measurement of symptom dimensions. There are several limitations to the generalizability of these results including that the psychiatric assessment occurred before CABG surgery. It cannot be ruled out that some persons experienced a worsening of their mental health postoperatively and possibly developed psychiatric comorbidities not quantified in this study. Furthermore, we were not able to identify persons receiving treatment for mental disorders after CABG surgery which may affect the occurrence of MACCE [82]. It is also possible too that anxiety levels were especially higher than normal in the CABG patients during the pre-operative period. With respect to psychiatric comorbidity that is common in persons with depression, we cannot extend our findings to the externalizing cluster of disorders [24], as patients with alcohol and substance abuse were excluded. Also, low base rates were observed for some disorders. Moreover, hierarchical exclusion rules stipulated by the MINI preclude comorbidity between these disorders and would inevitably result in lower depression and GAD prevalence. Consequently, analysis of cardiac morbidity was constrained to a combined MACCE endpoint with adjustment for a limited number of covariates. Indeed, the width of the confidence intervals suggests that future studies may benefit from larger samples. Finally, the participation rate was under-represented by Indigenous Australian peoples partly because persons living in rural and remote areas without a fixed residential address were excluded. As lower access to medical services may disadvantage Indigenous Australian peoples with CHD [83], the findings may not generalize to these cultural groups.
In conclusion, analysis with various theoretical conceptualizations of negative emotions suggested that only GAD was significantly associated with MACCE after CABG surgery. The non-significant association between MACCE and the distress-cluster may warrant further investigation in larger samples. Further research concerning combinations of disorders and negative emotions may contribute to clinical intervention in the population with CHD.

Informed consent
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. Informed consent was obtained from all patients for being included in the study.

Competing interests
The article processing charge was funded by the German Research Foundation (DFG) and the Albert Ludwigs University Freiburg in the funding programme Open Access Publishing. This research was supported by a postgraduate scholarship generously provided by the Sir Robert Menzies Foundation to PJT. PJT is supported by the National Health and Medical Research Council of Australia (Neil Hamilton Fairley -Clinical Overseas Fellowship #1053578). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors declare that they have no competing interests.