Expanding the knowledge on the long-term bidirectional relationship between depressive symptoms and coronary heart disease (CHD) is key for decreasing the excess morbidity and mortality associated with both conditions, and for reducing healthcare costs. Patients with CHD exhibit a higher incidence of depression or subclinical depression.1 In addition, depressive symptoms have been associated with up to a 4.5-fold increase in the incidence of CHD, even in the presence of other relevant risk factors.1 The prospective association between these two conditions is strong and consistent across different age strata, gender and racial groups.2 In addition, many studies have provided evidence for a synergistic effect of CHD and depression on mortality, as well as on a poorer prognosis with regard to clinical outcomes, treatment adherence and functional indicators. For instance, depressed postmyocardial infarct patients are more likely to withdraw from exercise programmes, and depressed patients with CHD are less likely to adhere to low-dose aspirin therapy.1 Based on the available evidence, UK general practitioners have received extra remuneration for identifying depression in patients with CHD since 2006,3 and a pilot trial of a nurse-led personalised care, which combines case management and self-management theory, have suggested promising results for the treatment of patients with both CHD and depression.4

The Whitehall study II (https://www.ucl.ac.uk/whitehallII) has provided an important contribution to uncover the bidirectional relationship between psychological suffering and cardiovascular diseases in a broad-spectrum population. In their Heart manuscript, they expand prior evidence by linking CHD to long-term trajectories of psychological distress.5 Mood disorders are generally episodic, exhibiting fluctuations throughout their course; thus, the severity of symptoms at one-time point may not be the best way to apprehend the longitudinal burden of such disorders.6 In a recently published review, Fiedorowicz6 suggested that the risk of vascular mortality related to mood disorders might depend on the amount of time that a patient is symptomatic. Using Group-based Trajectory Modelling (GBTM), Virtanen et al have identified four developmental psychological trajectories over a 21-year follow-up of the Whitehall II cohort: ‘persistently low’, ‘persistently intermediate’, ‘intermediate to low’ and ‘persistently high’. A developmental trajectory is defined as sequences of transitions from one state to another. Identifying groups of individuals following similar developmental patterns over time can allow a better understanding of the long-lasting relation between CHD and depressive symptoms.

One advantage of the GBTM approach is the capacity to qualitatively identify distinct developmental progressions that are not readily identifiable using prior classification rules. Hence, GBTM allows distinctive latent developmental trajectories to emerge from the data and the identification of subpopulations with similar trajectories. By doing so, GBTM facilitates the examination of factors that may predict the trajectory group membership.7 The results from the Whitehall II study suggest that CHD might have some predictive role in the developmental trajectory of psychological distress from midlife to old age. Even considering that the trajectory groups drawn from the analysis of Virtanen et al are only approximations of a more complex reality,7 the authors showed that prevalent CHD predicts worse progression slopes of psychological distress, when compared with CHD-free participants. A similar result (ie, worse trajectories of psychological distress) was also found for study participants presenting with obesity and smoking at baseline.

The course of chronic illnesses is known to be a dynamic and complex one, and accounting for this in epidemiological and clinical studies is one of the greatest challenges faced by health researchers. Life course epidemiology has emerged to address the long-term and cumulative effects on health of various exposures across the lifespan, and the Whitehall study II has contributed enormously in this field. In a similar way, GBTM addresses the forthcoming health burden predicted by exposures at a given time. The results reported in the paper of Virtanen et al indicate that CHD might not only increase the occurrence of depression, as previously described, but it might also predict a poorer long-term emotional burden of depressive disorders. Because the ascertainment of baseline CHD included the first three cohort waves (1985–1988 to 1991–1994), that is, prevalent and incident CHD cases, a reasonable variability in disease duration was expected. Even though the number of baseline CHD cases was small (n=201), we wonder whether the duration of CHD diagnosis, as well as the disease severity at baseline, would also have influenced the trajectories of psychological distress over time. Besides, as one looks forward in time to draw psychological trajectories from midlife to old age, the impact of disease management (for both CHD and depression) and differential survival is likely to play an important role in shaping group trajectories and their associations with a prevalent condition in a fixed time point. The risks of myocardial infarct and death among depressed patients with CHD were substantially increased by the presence of behavioural risk factors, such as smoking and low levels of physical activity.1 The longer the psychological trajectory, the more likely it is to be influenced by these multiple factors. Similarly, the influence of differential survival on the detection of group-based trajectories is probably increased as the cohort gets older.

The authors provided three possible explanations for their findings, assuming that they have captured real long-term effects of CHD, smoking and obesity on the psychological distress trajectory: (1) vascular depression hypothesis, (2) short-term adjustment disorder in response to a severe chronic disease and its impact on the quality of life, and (3) shared systemic inflammation origin. One would expect some kind of dose–response gradient in the OR linking CHD to the two trajectories of psychological distress, when compared with a ‘persistently low trajectory’. The authors justified the finding that the OR for ‘intermediate to low trajectory’ was higher than the OR for ‘persistently intermediate trajectory’ (which was not statistically significant) as an indication of the short-term adjustment response hypothesis. This might well be the case, but it may also have resulted from successful treatment of prolonged psychological distress within the study period, as screening and treatment for depression is increasing among patients with CHD. The lack of association between prevalent hypertension and diabetes with psychological distress trajectories weakens the ‘vascular depression’ hypothesis as a possible explanation for Virtanen et al findings. On the other hand, the fact that the magnitude of the OR remains about the same after adjustment for history of psychological distress reinforces the possibility of late onset depressive symptoms, which is a characteristic of ‘vascular depression’. Finally, the common systemic inflammation mechanism of CHD and depression remains a plausible possibility, but results are inconsistent as diabetes and hypertension were not associated with worse developmental psychological trajectories. Moreover, recent findings tend to link systemic inflammation only to atypical or somatic depression;8 nevertheless, the testing of such hypothesis is beyond the scope of a GBTM analysis.

The dissemination of novel analytical methods such as GBTM, in combination with high-quality observational data from many cohort studies currently under way, will soon allow researchers to confirm (or not) the findings from the Whitehall II study in other settings or populations, thus helping to unravel the implications of poorer psychological distress trajectories on the quality of life and survival of patients with CHD, as well as on healthcare load and costs. It is reasonable to postulate that other chronic diseases also related to depression, such as diabetes, might similarly predict poorer depressive trajectories, and such question needs a systematic investigation. The Whitehall results reinforce the view that bringing down obesity, smoking and CHD might also reduce the long-term burden of psychological distress. They also endorse the opinion that screening and treating mental health problems of individuals with CHD is not a one-time approach, but a continuous goal. It is reasonably accepted that psychotherapy and antidepressant medications, especially selective serotonin uptake inhibitors, are safe and reduce depressive symptoms in patients with cardiovascular diseases.9 Besides, some studies showed that treating depressive disorders in patients with CHD may improve cardiovascular outcomes, although these effects appear small and the findings are inconsistent.1

In conclusion, we believe that there is sufficient support in the existing literature for a positive answer to both questions raised by Virtanen et al at the end of their paper: whether being on a long-term trajectory of symptoms predict incident CHD, or whether a combination of long-term symptoms and CHD increases the risk of premature death. It would be hard to expect the contrary given the results from previous longitudinal studies on depression or depressive mood and CHD. A further question that GBTM might help to answer relates to whether improving the management of CHD would change the developmental progression of psychological distress from midlife to old age.