Background
For many decades, researchers and clinicians have observed an association between depression and coronary heart disease (CHD), and have tried to discover the mechanisms involved. A fair number of theories have been formulated and investigated, but no conclusive explanation exists as to why depression might lead to CHD and its progression. One potential explanation is that depression is causally related to the development and clinical course of CHD, either directly or through mediating pathways. Suggestive of causality, the association between depression and CHD has been consistently found across different settings and patient groups, as both etiological and prognostic associations between depression and CHD have been found in various studies of research groups worldwide [
1‐
4]. However, in the present review we will argue that, instead of being causally involved, depression can best be defined as a variable risk marker of CHD and its progression. We will discuss this on the basis of Hill’s criteria for causality [
5]. In addition, we will discuss current evidence in the light of the idea that symptoms and subtypes of depression that appear to be a physical consequence of the cardiac disease are most strongly predictive of adverse cardiovascular outcomes.
Principles for establishing causality have been developed by Bradford Hill [
5] (Table
1), which can aid in determining the plausibility of a causal association between two variables. Below, we will discuss how the principles most relevant for the association between depression and CHD (indicated with an *), substantiate the hypothesis that depression is a variable risk marker for CHD.
Table 1
Principles of causality[
5]
Strength | the stronger the association, the more likely is causality* |
Consistency | the association exists in different contexts and times |
Specificity | the association is specific for the variable and one particular outcome* |
Temporality | the variable precedes the outcome* |
Dose–response relationship | an increase in the variable results in an increase in the outcome* |
Plausibility | plausible theories/mechanisms for explaining the association exist* |
Coherence | the causal theory is coherent with existing knowledge |
Experimental manipulation | manipulation of the variable results in changes in the outcome* |
Analogy | similar associations exist between different variables |
The
principle of strength states that causality is more likely in stronger associations. The association between depression and CHD, however, is generally moderate. Otherwise healthy people with depression have an 80% higher risk of dying of CHD or having a myocardial infarction (MI) than those without a depression [
2]. Similarly, depressed patients with established CHD have a two times higher risk of dying of CHD or other causes [
1,
2]. Although the association of depression with development and progression of CHD is generally moderate, its magnitude is similar to that of the associations of other well-known risk factors for CHD, including smoking, hypertension, diabetes, obesity, dietary pattern, alcohol consumption and physical activity [
6]. The causality of each of these factors (including depression), as well as their place in the causal chain towards CHD, remains to be resolved. However, the absence of a strong association suggests the association is not definitely causal. Although causality is not excluded as a possibility, it is also possible that [
7,
8] other factors, such as cardiac disease severity and other health-related risk factors explain at least part of the association between depression and CHD.
The
principle of specificity states that associations are more likely to be causal when they are specific for one variable and one outcome. According to Hill the argument of specificity is difficult to attain, and can only argue strongly in favor of causality, but can never argue against it [
5]. If an association is very specific (that is, one risk factor associated with one disease only), then this is strongly supportive for causality. However, when an association is not specific, this does not necessarily argue against causality. For instance, smoking is associated with many disease outcomes, but may still be causally related to the development of some of the diseases. Like smoking, depression is associated with the outcomes of multiple diseases [
9‐
14], and etiologically, depression is a risk factor for the development and progression of several types of disease [
10], not just CHD. Moreover, other psychological problems have also been found to be associated with the development and progression of CHD, such as anxiety [
15,
16], vital exhaustion [
17,
18], anger and hostility [
19]. Therefore, Hill’s principle of specificity has not been met, meaning that there is no support that depression is definitely a causal risk factor.
The
principle of temporality states that, in a causal association, the determinant variable must precede the outcome. This principle means to determine the absence of causality rather than its presence. That is, an association in which the determinant does not precede the outcome is by definition not causal, but associations in which the determinant does precede the outcome may still be confounded. In the association between depression and CHD, most of the evidence points towards a bidirectional role of depression in CHD. In one direction, depression indeed precedes CHD, as depression in otherwise healthy people is associated with increased risk for the onset of CHD [
2,
20‐
23], with a pooled risk ratio (RR) of between 1.5 and 2.0 [
2,
21,
22]. In addition, in patients with existing CHD, depression is associated with an increased risk of cardiac morbidity and (cardiac) mortality [
1,
2,
24].
In the other direction, CHD is a potential risk factor for depression, as CHD is associated with increased prevalence of depression. An estimated 17% to 27% of CHD patients have major depressive disorder (MDD), compared with around 5% in the general population [
25,
26]. In addition, CHD in older persons without depression at baseline was found to be associated with an increased risk of developing depressive symptoms two years later [
27]. Furthermore, a recent study, using the largest sample (n = 1,117,292) to date, found cardiovascular fitness in healthy young men to predict depression up to 40 years later, even without cardiac disease being present [
28]. This latter study suggests that poor cardiovascular fitness causes depression later in life, but may also mean that both depression and CHD are caused by other factors, such as family environment and inactivity. Thus, the principle of temporality has been met as depression precedes CHD. However, this does not mean that the relation is definitely causal. Instead, the presence of the relationship in the opposite direction (that is, CHD precedes depression) rather suggests against causality.
The
principle of a dose–response relationship states that an association is more likely to be causal when there is a dose–response relationship between the variable and the outcome. There is evidence for such a relationship between self-reported depressive symptoms and cardiovascular outcomes: that is, the more severe the (symptoms of) depression, the stronger the association with CHD [
29,
30]. However, studies comparing the prognostic value of self-reported symptoms of depression with the prognostic value of a diagnosis of MDD report mixed results. Two studies [
31,
32] found a diagnosis of MDD to be more strongly predictive of cardiovascular outcomes in CHD patients. In contrast, two other studies [
33,
34] found self-reported symptoms of depression to be a stronger predictor of cardiovascular outcomes in CHD patients. In addition, we recently found that self-reported symptoms of depression were a stronger predictor of poor cardiovascular outcomes than a diagnosis of MDD in a sample of 2,493 MI patients [
30]. Finally, in our recent study of 767 MI patients, it was found that an increase of symptoms of depression immediately after an MI predicted cardiac events, whether or not these symptoms reached the level of severity of a clinical depression [
35].
Thus, there is evidence for a dose–response relationship between self-reported depressive symptoms and cardiovascular outcomes, which is supportive of the idea of causality. However, a diagnosis of MDD is not necessarily a stronger predictor of cardiovascular outcomes than self-reported depressive symptoms, which seems contradictive. In research, self-report questionnaires are used more often than clinical diagnoses of depression. A characteristic of a Diagnostic Statistical Manual of Mental Disorders (DSM)-IV diagnosis of depression is that by definition, the symptoms of depression cannot be the result of a physical disease or medication use [
36]. In contrast, self-report questionnaires do not distinguish between different causes of symptoms of depression. Therefore, it cannot be excluded that depressive symptoms reported on self-reported questionnaires may be an expression of CHD symptoms, which could explain the association with worse cardiovascular prognosis.
The
principle of plausibility states that causality is more likely when there are plausible theories or mechanisms to explain an association. Indeed, several plausible (mediating) mechanisms have been proposed through which depression can cause poor CHD outcomes. These mechanisms include, among others, elevated inflammation or platelet activation, changes in autonomic nervous system functioning and in hypothalamic-pituitary-adrenal axis functioning [
20,
37]. They may be causally involved in the association between depression and CHD. On the other hand, instead of forming pathways between depression and CHD, they may also confound the association by underlying the development of both depression and CHD. Biological causes for depression are thought to involve vascular disease, atherosclerosis and systemic inflammation [
38], which are processes also strongly involved in cardiac disease. For example, there is evidence that systemic inflammation is involved in the development of both depression and CHD [
39‐
43].
The
principle of experimental manipulation states that an association is more likely to be causal when manipulation of the determinant variable results in changes in the outcome. This may be the most important principle arguing against causality in the association between depression and CHD. Studies using depression treatment methods recommended by clinical guidelines, in which attempts have been made to improve depression (CREATE [
44], SADHART [
45,
47], ENRICHD [
46]) have been moderately effective in doing so, but did not result in subsequent improvement of CHD outcomes. Potentially, the improvements in depression were too small to affect CHD outcomes. On the other hand, it may indicate that depression is not causal of CHD.
In summary, despite decades of research, based on the evidence for criteria of causality discussed above, evidence appears to be against depression as a causal risk factor for CHD. Depression can, therefore, best be conceptualized as a variable risk marker for CHD and its progression [
48].
The consequences of this conceptualization of the association between depression and cardiovascular disease are, most importantly, that variations in depression are associated with variations in CHD and cardiovascular outcomes, but that experimental manipulation of depression does not change the CHD outcomes, as shown. Although depression is considered a mental disorder, for some CHD patients, some depressive symptoms may occur as a physical response to the cardiac disease. For example, fatigue is a symptom of depression, but may also be a consequence of CHD. Therefore, we propose an alternative theory, based on the fact that depression is highly heterogeneous, and typically, those symptoms and subtypes of depression that are most strongly associated with cardiac prognosis are those that are most likely a physical response to cardiac disease. The following areas of research will be discussed in light of the theory of cardiac disease severity as a confounder in the association between depression and CHD: 1) the cardiotoxicity of somatic/affective depressive symptoms; 2) the cardiotoxicity of treatment-resistant depression; and 3) residual confounding.
Competing interests
The authors declare they have no competing interests.
Authors’ contributions
AM drafted the manuscript. PdJ conceived of the paper, and critically reviewed and revised the manuscript. MZ assisted in drafting the manuscript and critically reviewed and revised the paper. All authors read and approved the final manuscript.