Investigating the bidirectional association between cardiovascular diseases and depression

Background: Cardiovascular diseases (CVDs) are the leading cause of disability and mortality globally. Although there has been substantial medical advancement in treating and managing CVDs, surviving CVD patients are at a greater risk of mortality and morbidity. Thus, preventative approaches aiming to identify, manage and control CVD risk factors remain the highest priority. Depression is a leading cause of disability worldwide, and it has been considered a relevant emergent, non-classical risk factor for the onset and poor prognosis of CVDs. Several systematic reviews have been published on this subject, providing evidence that depression is associated with an increased risk of CVD incidence. However, these reviews were limited by incorporating poor study designs and by focusing predominantly on a single CVD outcome. This previously fragmented investigation masked the overall picture of how strongly depression impacts each CVD subtype. At the same time, hypertension is one of the biggest risk factors for CVD; hence, the management and control of hypertension is of the utmost importance. Hypertensive patients mainly rely on antihypertensive treatment with a high dosage regimen and/or a combination of several antihypertensive drugs for the long term to control blood pressure and to consequently prevent the development or complication of CVD. Emerging evidence has investigated the effect of antihypertensive drugs in relation to depression onset, though the exact relationship remains unclear. Given that both hypertension and depression are risk factors for CVD, it becomes important that therapeutic agents to control blood pressure not have deleterious effects toward triggering depressive disorders, as both conditions will have a relevant big impact on patient’s health particularly those at high CVD risk. Objectives: This thesis has two main objectives: (1) updating the evidence of the association between depression and the risk of major subtypes of CVDs and (2) to investigate the association between exposure to antihypertensive drugs and risk of depression incident. Method: For the first objective, I conducted a systematic review and meta-analyses. Depression in the review referred to depressive symptoms or clinical depression and main outcomes of interest were incidence of fatal/non-fatal coronary heart diseases (CHD), heart failure (HF) and stroke, each measured as a single endpoint and reported as hazard ratio (HR) and 95% confidence interval (CI). The results for the systematic review were divided into three main results chapters based on the main outcomes (4-6). For the second objective, a secondary analysis of existing data held in the Glasgow Blood Pressure Clinic (GBPC) was conducted. Exposure was antihypertensive drugs which involves the five major classes including calcium channel blocker (CCB), beta-blocker (BB), angiotensin converting enzyme inhibitor (ACEI), angiotensin receptor blocker (ARB) and thiazide diuretic (TZD). The primary outcome was depression as indicated by the first prescription of antidepressants drug. Main findings of this analysis are presented in chapter 7. Results: Chapter 4 evaluated the relation between depression and risk of stroke. The meta-analysis included 19 studies enrolling 3,154,290 participants, with an average follow-up of 11.2 years. The pooled estimated risk revealed that baseline depression is associated with a 22% (HR = 1.22, 95% CI, 1.11-1.33) increased risk of developing first-ever stroke, with evidence of substantial statistical heterogeneity between studies (I2 = 67%). The magnitude of risk presented in this study is more modest than that previously reported in past systematic reviews for stroke outcomes. Sensitivity analyses were carried out to assess for a possible reverse causality (i.e. depression manifested as an acute sickness response to a subclinical stroke). This was achieved by restricting the analysis to four studies that considered a lag period, excluding stroke events occurring during the first years of follow-up. The results showed that depression remains a statistically strong predictor of stroke incidents with a more pronounced effect, and a wider 95% CI was obtained, which might indicate uncertainty (HR = 1.39, 95% CI, 1.11, 1.74). The statistical positive association remained significant after further restricting the analysis to five studies that measured depression over multiple instants over the follow-up period and modelled depression as a time-varying exposure (HR = 1.33, 95% CI, 1.10, 1.59). This finding suggests that elevated lifetime depressive symptoms among adults can be used as a reliable measure to predict future risk of stroke; however, due to the limited number of studies included to derive these findings, the result should be considered with caution and more work is required to confirm this finding. Subgroup analysis was also performed, and the findings showed that depressed elderly participants aged 65 years or above were at a lower risk of developing stroke than depressed participants at a younger age (< 65 years). However, the group difference showed only a borderline significance (p = .5). The results of this analysis may indicate that depression occurring at an early age might have a more devastating effect than late-life depression, though this finding should be considered with caution given the good heart health condition of elderly patients at baseline. Future epidemiological studies should be carried out on a large-scale to identify the clinical characteristics of participants that make them more prone to developing depression at an early age. Chapter 5 examined the association between depression and incident CHD. The meta-analysis incorporated 23 studies with 33,786299 participants and an average follow-up of 12.4 years. The pooled summary effect showed that the risk of CHD incident increased with depression by 22% (HR= 1.22, 95% CI, 1.13-1.32, p < .000) with evidence of substantial statistical heterogeneity between studies (I2 = 77%). The estimated risk presented in this study is almost identical to the latest review. This study also found that depression is associated with a 24% higher risk of developing myocardial infarction (HR = 1.24, 95% CI, 1.19, 1.29) with no evidence of statistical heterogeneity between studies (I2 = 0%). Sensitivity analyses comprising five cohort studies that considered a lag period provided similar risk estimates (HR = 1.22, 95% CI, 1.01, 1.48). Five studies modelled depression as a time varying exposure; a meta-analysis of these studies revealed an increased risk of incident CHD for depression, though a slightly lower magnitude was observed (HR = 1.17, 95% CI, 1.07, 1.28). Subgroup analysis by type of depression measures showed that the effect of clinical depression is more pronounced (HR = 1.26, 95% CI, 1.20, 1.32; I2 = 0%) than depressive symptoms (HR = 1.17, 95% CI, 1.10, 1.25; I2 = 0%) on risk of CHD incidence. In Chapter 6, I investigated the association between depression and incident HF in a CVD-free population. The meta-analysis was based on only four cohort studies with 2,200,308 participants and an average follow-up of 10.13 years. The main finding revealed that depression was associated with a 17% (HR = 1.17, 95% CI, 1.08, 1.38) increased risk of HF in the absence of CVD events at baseline, with no statistically significant amount of heterogeneity (I2 = 0%). The hypothesis of a dose-response relation was also assessed. Overall, this review identified 12 cohort studies that assessed a dose-response relation between depression and CVD outcomes. For stroke outcomes, four studies suggested a dose-response relation, and two did not confirm this finding (chapter 4). For CHD events, four studies showed no evidence of a dose-response relation and four found that depression increased the risk of CHD incident in a dose-response manner (chapter 5). Importantly, there was substantial heterogeneity in terms of how the studies defined ‘a dose of depression’, which seriously hampered the meta-analysis and drawing of conclusions. Future studies should establish guidance for researchers on the optimal measures of ‘a dose of depression’ to investigate such a relation. Chapter 7 covered the investigation of the association between antihypertensive drugs and the risk of incident depression. This was a retrospective cohort study in which I analysed data of hypertensive patients attending the GBPC, providing secondary and tertiary care service, between January 2005 and March 2013. All patients aged between 18 and 80 years who were newly commenced on antihypertensive drugs were included in this cohort. Exposure to ACEI, ARB, BB, CCB, and diuretics was assessed. Patients were prospectively followed up to the outcome, death, or end of the study. Depression as an outcome in this cohort was defined as patients who filled at least two prescriptions of antidepressants during the study period. Two analyses were performed. The first analysis was on patients who were on antihypertensive monotherapy. Eligible patients had no known history of depression and were on an antihypertensive monotherapy of the same drug class within a 12-month window defined as the exposure period. Patients who died or developed the outcome during the exposure period were excluded. The association between antihypertensive drug classes and depression incidence was investigated using Cox proportional hazards models to estimate HR, and patients who received ACEI therapy were set as the reference group. In this analysis, a dose-response relationship was also investigated, whereby the cumulative defined daily dose (cDDD) of antihypertensives during the exposure period was stratified into tertiles and the lowest tertile was set as the reference group. The second analysis was on patients who were either on antihypertensive monotherapy or polytherapy. In this analysis, eligible patients had an exposure period of 6 months preceded by 6 months of no antihypertensive or antidepressant prescription records. Patients who developed the outcome or died within the six months of the exposure period were excluded. Studied antihypertensive drug classes were additionally included alpha-blocker and centrally acting antihypertensive drugs. CCB and diuretic classes were divided into dihydropyridine CCB and non-dihydropyridine CCB, diuretics, and mineralocorticoids diuretic, correspondingly. Both Cox proportional hazards models and the generalised estimating equation (GEE) were used to investigate the association between antihypertensive drugs and incident depression. The reference group in this analysis was also patients on ACEI therapy. Findings of the monotherapy analysis showed that, among the five major classes of antihypertensive drugs, CCB had the highest risk of developing depression after adjusting for covariates, compared to the ACEI group (HR = 1.39; 95% CI: 1.07, 1.82). Consistence results derived from the polytherapy analysis showed that dihydropyridine CCB was associated with a significantly increased risk of incident depression in comparison to ACEI (HR = 1.38; 95% CI: 1.03, 1.86). The GEE analysis further confirmed this finding (OR = 1.32 95% CI: 1.06, 1.64). The dose-response analysis demonstrated that higher cDDD of ARB was associated with a greater risk of depression, although the association was marginally significant (p = 0.055). Conclusion: This thesis provided evidence that depression imposes a similar level of risk across different CVD subtypes. Future epidemiological studies should examine the dynamic aspects of depressive symptoms in relation to CVD and subclinical CVD, whether the risk of CVD is related to a specific subtype of depression, and the role of antidepressant drugs in this association. The present thesis showed that among population with complicated hypertension, CCB is associated with an increased risk of depression incidence compared to ACEI, supporting findings of previous studies. The risk of developing depression is also linked to ARB, although it might be dose dependent. A well-designed randomised control trial is the optimal study design to validate these findings, and up to that time when a clear association is established, these medications should continue to be used as recommended by the current guidelines for hypertension treatment and CVD prevention