Commentary: Broken hearts and minds--depression and incident heart disease and stroke

Abstract

Of several possible ‘psychosocial’ causes of coronary heart disease (CHD), depression probably has the strongest candidature. Since both CHD and cerebrovascular disease (CBVD) share similar pathophysiological roots, it might be expected that if depression caused increased risk of heart attack it would also cause increased risk of stroke. Evidence supporting this latter proposition is both less extensive and less convincing. The public health importance of these issues is considerable, given the contribution of CHD and CBVD to health in general and health inequality in particular and considering the disproportionate experience of depression amongst the disadvantaged. Much of the current cardiovascular disease primary prevention effort is based on population screening with risk stratification and risk reduction treatment for those exceeding a risk threshold. Risk stratification plays less of a role in secondary prevention where risk reduction treatment is generally assumed to be beneficial. Non-causal risk markers may have a role in stratification but are unlikely to lead to effective risk reduction treatment. These points are worth remembering when considering the relevance of epidemiological evidence to clinical practice. New evidence of possible practical relevance is provided by the paper by Hermann Nabi and colleagues in the current issue of the IJE. In the Health and Social Support (HeSSup) study, around 24 000 men and women aged 20–54 years were recruited from the Finnish general population in 1998. They completed a questionnaire that included the Beck Depression Inventory along with health and demographic information. Over the next 7 years they were followed up through linkage to hospital and mortality registers. Baseline questionnaire data were also augmented through linkage to prescription records. HeSSup exemplifies the way that routine administrative data can be used to add value to epidemiological studies. Markers of depression (scores of 10 or above on the Beck inventory or prescriptions for antidepressant medication) at recruitment showed typical associations with increased risk of incident fatal and non-fatal CHD events including an apparent dose–response relation with increasing symptom severity. In general, these associations were not apparent in relation to incident CBVD. HeSSup has several strengths; it was large enough to consider both CHD and CBVD outcomes in the same population; it used a standard instrument to measure exposure to depression and backed this up with antidepressant prescription data; outcomes were ascertained from clinical registries and were thus much less likely to be biased in comparison to endpoints based on symptom self-reports. These strengths should be considered alongside weaknesses common to many observational studies. Depression was only measured at baseline and use of anti-depressant prescriptions as an index of depression, though methodologically ingenious, is problematic. Taking treatment as a measure of disease severity in this context is generally questionable as it ignores the possibility that effective treatment may ameliorate the influence of the risk factor; further, irrespective of their effect on depression, anti-depressant drugs may directly influence cardiovascular disease risk. Alongside these issues of exposure measurement, available covariate information limited the ability to consider questions of confounding and mediation. In addition, the original response rate of HeSSup was relatively low, and older, more socially disadvantaged respondents were under-represented. These factors may have implications for external validity. As the HeSSup investigators argue, prior evidence for a causal influence of depression on CBVD is inconclusive. If anything, HeSSup makes this evidence more conclusively negative. Longer follow-up in older populations alongside better measurement of exposure to depression may provide a different picture but, until it does, there appears no basis to add depression Published by Oxford University Press on behalf of the International Epidemiological Association