Article Text
PDF
Statistics from Altmetric.com
Question
Question: Is depression after myocardial infarction (MI) associated with risk of all-cause mortality and further cardiovascular events after adjusting for severity of cardiac disease and other health-related variables?
Outcomes: All-cause mortality and both fatal and non-fatal cardiovascular events.
Methods
Design: Systematic review and individual patient meta-analysis.
Data sources: MEDLINE (PubMed), EMBASE and PyscINFO were searched between 1975 and 5 January 2011, with search alerts set-up to identify studies published after this date. In addition, relevant reviews and articles were cross-referenced.
Study selection and analysis: Prospective studies that investigated the association between postinfarction depression and all-cause mortality and/or cardiovascular events. Depression had to be assessed using validated methods (self-report questionnaire or standardised diagnostic interview) within 3 months of hospital admission for myocardial infarction (MI). Study authors were contacted to obtain original data on patient demographics, depression, disease severity, comorbidities, medication use and outcomes. Depression scores from individual studies were standardised to z-scores. Studies reporting time-to-event data were combined to calculate HZ using multilevel Cox proportional hazard regression analysis. All studies, including those which reported dichotomous outcome data only (event vs no event), were combined to calculate ORs using logistic regression analysis. The base-case model analysed the association between depression severity and all-cause mortality/cardiovascular events, adjusting for age and gender. The fully adjusted model further adjusted for markers of disease severity (left ventricular ejection fraction (LVEF), Killip class, past-MI) and other health factors (diabetes, smoking, BMI).
Main results
Individual patient data was available for 16 studies including 10 175 participants. Prevalence of depression post-MI was between 11% and 15% based on diagnostic interview, and 17–69% based on self-report questionnaire. Mean follow-up was 3.2 years. After adjusting for age and gender, depression was associated with increased risk of all-cause mortality and cardiovascular events. Each SD increase in depression z-score was associated with HR of 1.32 (95% CI 1.26 to 1.38, 10 studies) for all-cause mortality and HR of 1.19 (95% CI 1.14 to 1.24, 7 studies) for cardiovascular events. In the fully adjusted model, the association between depression and all-cause mortality was attenuated by 28% (HR 1.23, 95% CI 1.15 to 1.31, 3 studies), and the association with cardiovascular events was attenuated by 25% (HR 1.12, 95% CI 1.01 to 1.25, 2 studies). When only cardiovascular disease variables were considered, it was found that they were responsible for most of the attenuation of the full model, both for all-cause mortality and cardiovascular events. Results were similar in the logistic regression analyses.
Conclusions
Depression after myocardial infarction is associated with increased risk of all-cause mortality and cardiovascular events. The association is attenuated after adjustment for cardiac disease severity and other health variables, but remains significant.
Abstracted from
Commentary
Many prior individual studies and multiple traditional meta-analyses have found associations between depression and adverse cardiac outcomes following myocardial infarction (MI), often controlling for at least some relevant medical covariates. The major innovation of this work is that it uses an individual patient data meta-analysis approach, which combines all raw patient data across analysed studies into a single database. This allows for a richer and more sophisticated accounting for covariates by adjusting consistently for the same variables across studies. This approach allowed the authors to carefully account for several key markers of cardiac disease severity, such as left ventricular ejection fraction, across thousands of patients.
The authors’ findings continue to confirm depression as a risk factor for adverse cardiac events post-MI, even when accounting for cardiac disease severity in multiple ways. The analysis was limited by only obtaining patient-level data from approximately half (16/30) of the appropriate studies. Furthermore, even among these 16 studies, only some could be included in the various analyses looking at covariates. Still, all analyses included thousands of patients and the authors thoughtfully explored the characteristics of included versus excluded studies whenever possible.
From a research standpoint, the authors have now identified several specific markers of cardiac disease severity that clearly attenuate the connections between depression and adverse outcomes and must be included in future prospective studies examining depression's link to outcomes in cardiac patients. This work also reinforces the ongoing need to find effective, broadly accessible treatments for depression in cardiac patients that impact cardiovascular outcomes. At this stage, existing depression treatments and care models clearly improve mood symptoms and quality of life in cardiac patients. However, there is not yet definitive data that any specific depression treatment or model reduces the occurrence of adverse cardiac events in patients with post-MI or other persons with heart disease.
Importantly, this work does not impact important questions about routine depression screening post-MI. Routine assessment for depression is still only indicated if there is a clear pathway by which positive-screen patients can obtain formal assessment, initiation of treatment and longitudinal care.
Footnotes
-
Competing interests JCH has received grants to study depression in cardiac patients in the past (American Heart Association; Scientist Development Grant 0735530T and American Heart Association; Grant-in-Aid 10GRNT3450015).