Depression
Depression
Depression
Authors: Yvonne W Leung, PhD1; David B Flora, PhD2; Shannon Gravely, PhD1; Jane Irvine,
Corresponding Authors:
Sherry L. Grace
York University, Faculty of Health
368 Bethune
4700 Keele St
Toronto, ON M3J 1P3
Phone: (416) 736-2100 x.22364
Fax: (416) 736-5774
sgrace@yorku.ca
1
Abstract
Background: Depression is associated with increased cardiac morbidity and mortality in the
general population and in coronary heart disease (CHD) patients. Recent evidence suggests that
patients with new-onset depression post-CHD diagnosis have worse outcomes than those who
had previous or recurrent depression. This meta-analysis investigated timing of depression onset
in established CHD and CHD-free cohorts to determine what timeframe is associated with
greater mortality and cardiac morbidity.
Keywords: coronary heart disease; depression; timing of onset; mortality; morbidity; outcome;
meta-analysis
CHD, coronary heart disease; MI, myocardial infarction; MACEs, major adverse cardiac events;
ES, effect size; RR, risk ratio; HR, Hazard ratio; CI, confidence intervals; DSM, Diagnostic and
2
Introduction:
Depression is common and persistent, with a burden of about 5% in the general population and
20% among cardiac patients [1]. Patients with a depressive history are twice as likely to have
recurrent depression after a myocardial infarction (MI) both in hospital and after hospitalization
[2]. Likewise, among patients who become depressed after an MI, half of them have had a
previous history of depression [3]. In addition to this mental health burden, depression is related
to both greater risk of developing coronary heart disease (CAD [2, 4]) and poorer prognosis
among patients with established CAD [5]. Indeed, several meta-analyses [6-8] have
Recently, there has been particular interest in the timeframe of depression in relation to
patient outcomes, which would have important mechanistic and screening implications. Some
studies have shown that CHD patients with a history of pre-morbid depression have greater
mortality and cardiac morbidity risk compared to patients who have never been depressed [1-3].
Conversely, other research has shown that new-onset depression at the time of a cardiac
hospitalization is particularly prognostic. Our group and others [9-13] have shown that new-
greater risk of mortality and cardiac morbidity when compared to historical or recurrent
depression. However, recent studies [14] report inconsistent findings. To shed light on this
important question, this quantitative review and synthesis tested whether the timeframe of
For the purposes of this meta-analysis, ‘depression’ refers to unipolar clinical depression
3
depression can be classified as pre-morbid or post-morbid onsets. Pre-morbid onset could consist
of 1 of the following: (1) Pre-CHD depression: (1a) CHD-free individuals who had depression
before CHD diagnosis but their post-CHD depression status at the time of cardiac event is
unknown, as their study endpoints were cardiac diagnosis and cardiac death; (1b) History of
Depression Only: CHD patients who had one or more episodes of depression before their current
CHD diagnosis without being currently depressed; (1c) Recurrent depression: CHD patients who
had depression before and after their initial CHD diagnosis; (2) Post-morbid onset refers to New-
onset or Incident depression among CHD patients whose initial depression occurred after their
CHD diagnosis. Non-incident depression refers to recurrent depression, which involves direct
depression status among CHD patients. Finally, new-onset depression is divided into 2a) in-
hospital or within 30 days after hospitalization, and 2b) post-hospitalization onset within 1-year.
The objective of this meta-analysis was to compare the effects on cardiac morbidity and
mortality of the following depression timeframes: I) Each timeframes of depression; II) Pre-
morbid Onset (1a, 1b, and 1c) vs. Post-morbid Onset (2a & 2b); III) Incident/ New-onset (2a &
2b) vs. Non-incident (Recurrent; 1c); and IV) New-onset in-hospital depression vs. New-onset
Methods
Inclusion Criteria
The meta-analysis that the meta-analysis was reported in a manner consistent with MOOSE
4
For a study to be included in this meta-analysis, depression or depressive symptoms had
Historical depression status could also be assessed based on patients’ self-report or indication in
medical records. Results had to be reported for depressed versus non-depressed patients.
Subjects included either CHD-free cohorts or CHD patients of any age. CHD patients
included those who had a confirmed diagnosis of acute coronary syndrome; (unstable angina to
the cohort must be free of cardiac diseases and was followed from prior to CHD onset until
cardiac mortality was ascertained. In CHD cohorts, the studies included in this meta-analysis
were those which ascertained depression status at two time-points at least, with post-diagnosis.
Studies were included if they had a clear indication of historical depression ascertainment and /
or post-morbid depression assessed at time of hospitalization (within one month) and / or post-
Studies were included if outcomes reported included at least one of the following: In
CHD-free cohorts, only studies reported cardiac mortality was included as an outcome. In CHD
cohorts, we extracted outcome in the following order: cardiac mortality, all-cause mortality, a
composite outcome combining cardiac morbidity and mortality, and cardiac morbidity, as
defined as readmission or re-hospitalization due to a major adverse cardiac event (MACEs; i.e.,
MI, need for revascularization, stroke, heart failure, and arrhythmic events).
In addition to the criteria outlined above, the following were also applied: (1) prospective
cohort design with clinical depression or depressive symptoms as the exposure, (2) the timeframe
5
of the depression before, during, or after a cardiac-related hospitalization must be clearly defined
and included in analyses, and (3) paper published in a peer-reviewed English–language journal.
In the case of multiple publications of the same cohort, the most recent article that best addressed
MEDLINE, EMBASE, and PsycINFO were searched for articles published from 1990 to
May 2009. The search strategy included 4 components: CHD, depression, timeframe of
depression, and mortality and cardiac morbidity (see Appendix A for search terms). All the
database searches were conducted by information specialists at the University Health Network.
Next, reference lists of selected papers were hand-searched. Finally, authors of key studies were
Study selection
Using a conservative threshold for exclusion, one investigator (YL) examined all abstracts and
selected articles for full-text examination. Two investigators (YL & SGW) independently used
standardized forms to assess the eligibility of all full-text articles. Inter-observer agreement was
assessed using Cohen’s kappa (κ), and differences were resolved by discussion with a third rater
(SLG). RefWorks [18] software was used to create a database of reference material identified
from included studies as shown in Tables 1 and 2. In some cases, primary authors were contacted
Quality Assessment
6
A study quality assessment manual was prepared based on the Newcastle-Ottawa Scale
for observational studies [19] (Appendix B). This instrument consists of 8 questions assessing
quality in 3 broad categories: patient selection, comparability of study groups, and outcome
assessment. Extra points were assigned to studies that adjusted for cardiac risk factors and anti-
depressant use. Assessment results in a maximum possible score of 9 for each study, with studies
rated 6 or more deemed of highest quality. Each study was rated independently by 2 investigators
Statistical Analyses
version 2 [20]. For each study, effect size (ES) was calculated as the risk ratio (RR) comparing
morbidity. Hazard ratios (HRs) were treated as RRs because HR can be referred to the RR-
averaged-over-time [21]. To pool data across the studies, adjusted RRs with 95% confidence
intervals (CI) or the dichotomous frequency data (depressed vs. not depressed) were retrieved
from each study to compute a pooled ES in RR. A mixed-effects model [22] was adopted which
incorporated a random effects model to combine studies within each timeframe (i.e., new-onset,
recurrent, and history-only in CHD cohort studies and pre-morbid depression in CHD-free cohort
studies) and a fixed-effects model to combine these timeframes to yield an overall ES across
studies as these timeframes were from the same study. To estimate the overall ES, each RR was
weighted by the inverse of its variance, the weighted RR were summed across studies, and then
Publication bias was assessed by means of the Begg and Mazumdar rank correlation test
7
Heterogeneity Analysis:
depression. The statistical heterogeneity between the studies was evaluated using the Cochran's
Sensitivity Analysis:
analyses. The influence of low-quality studies on the pooled estimates was tested in a sensitivity
analysis by including and excluding them, using a test of interaction with a predetermined 2-
tailed α of 0.05 to compare differences between original and corrected ESs [26]. The current
meta-analysis also explored the influence of the method of depression ascertainment, length of
follow-up, adjustment of mortality confounders and study outcome of cardiac and all-cause
Results
A flow diagram of the literature search is shown in Figure 2. Inter-observer agreement was κ
=0.749, indicating “good” [27] concordance between raters for article inclusion decisions.
Ultimately, 22 studies were included in the meta-analysis, yielding 39 ESs. The quality of the
studies was generally good, with 13 out of 22 studies (59.1%) rated as 6 stars or higher on the
Tables 1 and 2 summarize the characteristics of the included CHD cohort and CHD-free
cohort studies, respectively. Studies are presented in alphabetical order based on first author, and
8
in addition Table 1 was organized by depression ascertainment either in-hospital or post-
hospitalization.
Included cohorts reported inception ranging between 1971 and 2001, and had follow-up
from 6 months to 37 years (CHD cohort studies ranged from 6 months to 12 years; CHD-free
cohort studies ranged from 4 to 37 years). Sample sizes ranged from 145 to 62,839, with a total
of 127,590 participants (CHD cohort studies ranged from 145 to 13,708 participants; CHD-free
cohort studies ranged from 660 to 62,839 participants). In CHD-free cohorts, the prevalence of
pre-morbid depression ranged from 2-15%. In CHD cohorts, the prevalence of new-onset
depression ranged from 5.5-27.9%, recurrent depression from 5.1-41.4%, and history of
Depression Measurement
Seven of the 22 studies assessed major depression by clinical interview using DSM-III or
patients with depression through the International Classification of Diseases, 9th Revision (ICD-
depressive symptoms, in which 9 different measures are represented (See Tables 1 and 2).
Study Outcomes
Twelve studies reported cardiac mortality, 6 studies reported all-cause mortality, and 3
studies reported both outcomes separately. Four studies reported a composite endpoint such as
fatal and non-fatal cardiac events, all-cause mortality, and cardiac-related readmission.
Fifteen studies reported adjusted HRs, and four of these adjusted for disease severity.
Eight studies reported crude rates of mortality/ cardiac morbidity. The HRs, crude rates, and
9
The assumption of no publication bias was reasonable, considering that the p values for
the Begg's and Egger's tests based on the 14 studies for which complete data were available in
the original publication or through contact with authors were 0.771 and 0.774, respectively.
Heterogeneity
appeared reasonable. The I2 statistics within depression timeframes ranged from low to high,
although all Q-statistics were significant except that for history of depression only. The results
are as follows: Overall depression (Q= 82.63, df=39, p<0.0001, I2 =53.8%); overall depression in
(Q=30.17, df=12, p=.003; I2 =60.2%); recurrent depression (Q=16.86, df=9, p=0.051; I2 =46.6%);
The overall ES (RR) of depression at any time-period on mortality and cardiac morbidity
in all studies was 1.70 (95% CI 1.48-1.96), indicating that depression is associated with a risk for
mortality and cardiac morbidity that is almost twice the risk for non-depressed patients. Overall
results are displayed in Figure 3, where each white diamond represents the pooled ES and 95%
CI for timeframe of depression and the black diamond represents the overall ES for all studies
combined. The overall ES of depression at any time-period on mortality and morbidity outcomes
among CHD-cohort and CHD-free cohort were 1.76 (95% CI 1.40-2.21) and 1.70 (95% CI 1.41-
2.05) respectively.
10
When comparing across timeframes of depression onset, the magnitude of risk
significantly varied (Q=15.77, df=3, p=0.001). For pre-morbid depression in CHD-free cohorts
(1a) the ES was 1.79 (95% CI 1.45-2.21). In CHD cohorts, the ES for new-onset depression (2a
& 2b) was 2.11 (95% CI 1.66-2.68), while the ESs for recurrent depression (1c) and history only
(1b) were 1.59 (95% CI 1.08-2.34) and 0.76 (95% CI 0.48-1.19) respectively. Q-statistics for
Pre-morbid vs. Post-morbid Depression Onset Timeframe among CHD and CHD-
All CHD and CHD-free cohort studies were grouped by timeframe of depression
according to whether it was a pre- or post-cardiac event or procedure. The pooled ES for post-
morbid onset (new-onset) was 2.11 (95% CI 1.66-2.68), whereas pre-morbid onset (all other
timeframes) was 1.52 (95% CI 1.25-1.84; Fig. 4). The difference in ESs was significant (Q=4.42,
df=1, p=0.036).
All CHD cohort studies were grouped by whether the depression occurred pre- or post-
diagnosis / procedure. The pooled ES of incident depression (2.11 95% CI 1.66-2.68) was greater
than that of non-incident depression (1.59 95% CI 1.08-2.34). This difference in ESs was not
depression during cardiac hospitalization (within 1 month after a CHD event) and 6 defined new-
11
onset post-hospitalization (2-12 months after a CHD event). The ES of new-onset depression in-
hospital was 2.33 (95% CI 1.63-3.32), and post-hospitalization was 1.77 (95% CI 1.37-2.89).
These ESs were not significantly different (Q=1.50, df=1, p=0.22; Fig. 6).
Sensitivity Analysis
To explore the reliability of results, sensitivity analyses were performed with regard to
study quality, method of depression assessment, length of patient follow-up and variable
adjustment. First, 13 high quality studies (6 studies from CHD cohorts [9-12, 28, 29], 7 from
CAD-free cohorts [30-36]) were selected to test the sensitivity of results. The corrected overall
ESs for all timeframes of depression was 1.60 (95% CI 1.40-1.84). The corrected ESs for new-
onset depression was 1.66 (95% CI 1.41-1.95), for recurrent was 1.20 (95% CI 0.68-2.09), for
history of depression only was 0.400 (95% CI 0.180-0.880), and for pre-morbid depression was
1.88 (95% CI 1.40-1.84). These were not significantly different from the ESs estimated using all
22 studies (p=0.632-0.889), except in the case new-onset depression (p=0.001), which was
lower.
onset overall (6 95% CI 1.41-1.95), the relationship remained but the differences become non-
significant (Q=0.501, df=1, p=0.479). When comparing incident (1.66 95% CI 1.41-1.95) to non-
incident depression (1.20 95% CI 0.683-2.09), the finding remained non-significant (Q=1.205,
df=1, p=0.272). In summary, only the new-onset depression estimates were affected by study
The 7 studies of CHD patients that used a “gold standard” depression measure (i.e.,
structured clinical interview or physician diagnosis) were selected [2, 9-11, 13, 29, 37]. The ESs
for these 7 studies were not significantly different from those when all studies were included
12
(ps=0.490-0.919). Therefore, estimates did not appear to be affected by method of depression
ascertainment.
The influence of the length of follow-up among the CHD cohort studies was also
explored. Seven studies with a follow-up length of 1 year or greater were selected [2, 9, 11, 12,
28, 29, 38, 39]. The ESs for these 7 studies were not significantly different than those for all
included studies (p=0.66-0.79). Therefore, estimates did not appear to be affected by length of
follow-up.
The influence of the statistical adjustment for confounding variables such as disease
severity among the CHD cohort studies was explored. Four studies adjusted for CHD disease
severity in their analyses [9, 11, 28, 29]. The ESs for these 4 studies were not significantly
different than the ESs for all included studies (ps=0.33-0.584). Therefore, it can be concluded
Finally, the influence of studies using only cardiac and all-cause mortality outcomes was
tested. Four studies were excluded[10, 13, 29, 37]. The corrected overall ESs for all timeframes
of depression was 1.75 (95% CI 1.47-1.90). The corrected ESs for new-onset depression was
2.33 (95% CI 1.66-3.27), for recurrent was 1.47 (95% CI 0.761-2.84), for history only was 0.710
(95% CI 0.374-1.35), and for pre-morbid depression was 1.76 (95% CI 1.40-2.21). These were
not significantly different from the ESs estimated using all 22 studies (ps= 0.120-0.583).
Discussion
Over and above the psychosocial burden of depression in heart disease, depression is
linked to worse outcomes including mortality and cardiac morbidity. The aim of the current
meta-analysis was to determine the impact of the timing of depression onset relative to the
timing of CHD diagnosis or cardiac event on health outcomes. This meta-analysis confirmed the
13
adverse effect of depression on outcome, with 1.75-times greater cardiac morbidity and mortality
seen in the presence of depression at any time-point. Sensitivity analyses suggest that this
disease severity, or type of outcome. Overall, the magnitude of risk is greatest in patients with
new-onset depression. However, after controlling for study quality, there was no significant
difference in the effect of depression that occurs pre- or post-CHD diagnosis or cardiac event.
Ultimately, the results of this meta-analysis confirm the hazardous effects of comorbid
These results are consistent with a previous meta-analysis demonstrating the negative
effect of depression in CHD-free cohorts [7]. However, the effect of pre-morbid depression on
outcomes among CHD-free and established CHD cohorts appears somewhat contradictory. In the
CHD cohorts, reporting only history without a present episode of depression was unrelated to
cardiac morbidity and mortality. This inconsistency can likely be explained by bias in the self-
report of depression history and/or retention bias. With regard to the latter, it is possible that pre-
morbidly depressed patients who survived to participate in a CHD cohort study had some sort of
survival advantage, while most other depressed individuals died around the time of their CHD
events and thus were not included in the CHD cohorts [9]. Indeed in our previous study [12] and
a study by Lesperance and colleagues [2], historical depression tended to have a protective
effect, and a trend toward a protective effect is seen in this meta-analysis as well (ES=0.76).
CHD patients who experienced one or more depressive episodes prior to their cardiac event and
survived might have been treated with anti-depressant therapy for example, and may have
successfully responded [40]. Moreover, these survivors may have more adaptive coping
strategies and lead a healthier lifestyle, or a stronger social support system, or they may have
14
been more likely to participate in effective psychotherapy. Alternatively, these patients may have
play, and if it is, to determine the explanation for the survival advantage of historical depression
Regarding the issue of depression post-CAD, there has been much debate in the literature
regarding whether incident depression has some sort of acute mechanistic impact on outcome, or
whether the presence of chronic processes are particularly hazardous (see [4, 42-44]). When the
outcomes from CHD-cohort studies were examined, both new-onset and recurrent depression
were significantly related to approximately twice the risk of adverse outcomes. However, the
magnitude of effect for new-onset depression was lessened when only high-quality studies were
tested, and there was no significant difference in the magnitude of effect for new-onset versus
recurrent depression. Overall these results suggest that both new-onset and recurrent depression
in CHD patients are predictive of adverse outcomes. Some of the inconsistencies in the previous
literature regarding timing [13] may be due to smaller sample sizes which we have been able to
overcome in this synthesis, or be explained by different definitions adopted for pre- and post-
CHD depression. Moreover, the effect of retention bias should again be considered, because only
patients who survived to follow-up assessment with recurrent depression would be included in
the analyses. The problem of early mortality cannot be overcome, and therefore comparisons of
new- to late-onset or recurrent depression should be interpreted with particular caution. Clearly,
explanatory mechanisms and theories [4, 45] will be needed to better understand the impact of
post-CHD depression timing. On the whole, depression onset around the time of hospitalization
15
is a risk factor for cardiac events, and this hazard continues when depression is chronic, and
Limitations
Caution is warranted when interpreting these results. First, the current study synthesized
only studies published in English, which may limit the generalizability of these findings. Second,
the severity of depression was not taken into consideration. This may have increased the
heterogeneity of our findings. Third, the definition of depression applied herein may have biased
results. Only 6/22 (27.3%) studies applied DSM-III or -IV criteria for major depression and
1/22(4.5%) applied ICD-9 codes. While the sensitivity analysis showed that results were not
remaining studies could be related (or not) to a major depressive episode. Moreover, dysthymia
and minor depression were not considered, which may have a confounding effect on results.
Fourth, receipt and type of depression treatment and was not taken into consideration, and only a
few of the included studies controlled for these factors in their multivariate analyses [9, 11, 33,
36, 46]. Fifth, history of depression ascertainment was based mainly on self-report as present or
absent which is subject to bias. Sixth, caution is warranted in drawing conclusions regarding
statistical significance between effect sizes after sensitivity analysis due to low power. Finally,
due to the limited amount of research in this area, we adopted a broad definition of CHD and a
composite endpoint, and this may have increased the heterogeneity of findings.
Conclusions
This meta-analysis showed that those with both pre-morbid and post-CHD depression have 1.5-2
times higher risk of mortality and cardiac morbidity. There may be a subgroup of individuals
whose depression is successfully remitted before CHD onset who have a survival advantage.
16
They should be investigated to better understand what interventions can improve prognosis.
Ultimately, new onset depression post-diagnosis may be particularly hazardous, however this
cannot be conclusively determined due to retention or survival biases which cannot be overcome.
specialists Ms. Amy Faulkner and Rouhi Fazelzad for conducting the electronic database
searches. Particular thanks are also due to Ms. Alina Marquez for editing the manuscript. Y
Leung and S Gravely were supported by Canadian Institutes of Health Research (CIHR)
17
References
1. Thombs BD, Bass EB, Ford DE, Stewart KJ, Tsilidis KK, Patel U, Fauerbach JA, Bush DE, &
3. Spijkerman T, de Jonge P, van den Brink RH, Jansen JH, May JF, Crijns HJ, Ormel J.
Depression following myocardial infarction: first-ever versus ongoing and recurrent episodes
4. Lett HS, Blumenthal JA, Babyak MA, Sherwood A, Strauman T, Robins C, Newman NF.
Depression as a risk factor for coronary artery disease: evidence, mechanisms, and treatment.
7. Rugulies R. Depression as a predictor for coronary heart disease: a review and meta-analysis.
18
8. van Melle JP, de Jonge P, Spijkerman TA, Tijssen JG, Ormel J, van Veldhuisen DJ, van den
Brink RH, van den Berg MP. Prognostic association of depression following myocardial
2004;66:814.
9. Carney RM, Freedland KE, Steinmeyer B, Blumenthal JA, de Jonge P, Davidson KW,
Czajkowski SM, Jaffe AS. History of Depression and Survival After Acute Myocardial
10. de Jonge P, Rob H. S., van den Brink RHS, Spijkerman TA, Ormel J. Only Incident
Depressive Episodes After Myocardial Infarction Are Associated With New Cardiovascular
onset depression following myocardial infarction predicts cardiac mortality. Psychosom Med
2008;70:450.
12. Grace SL, Abbey SE, Kapral MK, Fang J, Nolan RP, Stewart DE. Effect of depression on
13. Parker GB, Hilton TM, Walsh WF, Owen CA, Heruc GA, Olley A, Brotchie H, Hadzi-
Pavlovic D. Timing is everything: the onset of depression and acute coronary syndrome
19
14. Glassman AH, Bigger JT,Jr, Gaffney M. Psychiatric characteristics associated with long-term
mortality among 361 patients having an acute coronary syndrome and major depression: seven-
15. Goldston K, Baillie AJ. Depression and coronary heart disease: a review of the
Rev 2008;28:288.
16. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ,
Sipe TA, Thacker SB. Meta-analysis of observational studies in epidemiology: a proposal for
2000;283:2008.
17. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, Clarke M,
Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement for reporting systematic reviews
and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration
BMJ 2009;339:b2700.
19. Wells GA, Shea B, O’Connell D, Peterson J, Welch V, Losos M, Tugwell P. The Newcastle-
Ottawa Scale (NOS) for assessing the quality of non-randomised studies in meta-analyses.
20. Comprehensive Meta-Analysis [computer program]. Version 2. New Jersey, USA; 2006.
20
22. Hedges L. A random effects model for effect sizes. Psychol Bull 1983;93:388.
23. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication
26. Altman DG, Bland JM. Interaction revisited: the difference between two estimates BMJ
2003;326:219.
27. Altman DG. Practical statistics for medical research. London, UK: Chapman and Hall, 1991.
28. Blumenthal JA, Lett HS, Babyak MA,White W, Smith PK, Mark DB, Jones R, Mathew JP,
Newman MF; NORG Investigators. Depression as a risk factor for mortality after coronary artery
29. May HT, Horne BD, Carlquist JF, Sheng X, Joy E, Catinella AP. Depression after coronary
30. Nolen-Hoeksema S, Morrow J, Fredrickson BL. Response styles and the duration of episodes
21
31. Ahto M, Isoaho R, Puolijoki H, Vahlberg T, Kivela SL. Stronger symptoms of depression
predict high coronary heart disease mortality in older men and women. Int J Geriatr Psychiatry
2007;22:757.
32. Ferketich AK, Schwartzbaum JA, Frid DJ, Moeschberger ML. Depression as an antecedent
to heart disease among women and men in the NHANES I study. National Health and Nutrition
33. Ford DE, Mead LA, Chang PP, Cooper-Patrick L, Wang NY, Klag MJ. Depression is a risk
factor for coronary artery disease in men: the precursors study. Arch Intern Med 1998;158:1422.
34. Penninx BW, Guralnik JM, Mendes de Leon CF, Pahor M, Visser M, Corti MC, Wallace RB.
Cardiovascular events and mortality in newly and chronically depressed persons > 70 years of
35. Penninx BW, Beekman AT, Honig A, Deeg DJ, Schoevers RA, van Eijk JT, van Tilburg W.
Depression and cardiac mortality: results from a community-based longitudinal study. Arch Gen
Psychiatry 2001;58:221.
36. Surtees PG, Wainwright NW, Luben RN, Wareham NJ, Bingham SA, Khaw KT. Depression
and ischemic heart disease mortality: evidence from the EPIC-Norfolk United Kingdom
37. Connerney I, Shapiro PA, McLaughlin JS, Bagiella E, Sloan RP. Relation between
depression after coronary artery bypass surgery and 12-month outcome: a prospective study.
Lancet 2001;358:1766.
22
38. Borowicz L,Jr, Royall R, Grega M, Selnes O, Lyketsos C, McKhann G. Depression and
cardiac morbidity 5 years after coronary artery bypass surgery Psychosomatics 2002;43:464.
39. Hata M, Yagi Y, Sezai A, Niino T, Yoda M, Wakui S, Soeda M, Nohata I, Shiono M,
Minami K. Risk analysis for depression and patient prognosis after open heart surgery. Circ J
2006;70:389.
40. Flint AJ, Rifat SL. The effect of treatment on the two-year course of late-life depression Br J
Psychiatry 1997;170:268.
41. Pina IL, Apstein CS, Balady GJ, Belardinelli R, Chaitman BR, Duscha BD, Fletcher BJ, Fleg
JL, Myers JN, Sullivan MJ; American Heart Association Committee on exercise, rehabilitation,
and prevention. Exercise and heart failure: A statement from the American Heart Association
1225.
Carney RM, Freedland KE, Miller GE, Jaffe AS. Depression as a risk factor for cardiac
43. Grippo AJ, Johnson AK. Biological mechanisms in the relationship between depression and
Interactions among higher trait anxiety, sympathetic activity, and endothelial function in the
23
45. Davidson KW, Rieckmann N, Lesperance F. Psychological theories of depression: potential
application for the prevention of acute coronary syndrome recurrence. Psychosom Med
2004;66:165.
46. Whang W, Kubzansky LD, Kawachi I, Rexrode KM, Kroenke CH, Glynn RJ, Garan H,
Albert CM. Depression and risk of sudden cardiac death and coronary heart disease in women:
47. Hui PN, Wan M, Chan WK, Yung PM. An evaluation of two behavioral rehabilitation
programs, qigong versus progressive relaxation, in improving the quality of life in cardiac
48. Branger PJ, van der Wouden JC, Schudel BR, et al. Electronic communication between
treatment of major depression in patients with congestive heart failure. Can J Psychiatry
2003;48:695.
50. Fonarow GC, Albert NM, Curtis AB, et al. Improving Evidence-Based Care for Heart Failure
in Outpatient Cardiology Practices: Primary Results of the Registry to Improve the Use of
Evidence-Based Heart Failure Therapies in the Outpatient Setting (IMPROVE HF). Circulation
2010;122:585.
24
Figure Legends
Figure 4. Forrest Plot of the Effect of Pre-morbid vs. Post-morbid Depression Onset on Mortality
or Cardiac Morbidity among both CHD and CHD-free cohorts
Figure 5. Forrest Plot of the Effect of Incident vs. Non-incident Depression on Mortality or
Cardiac Morbidity among CHD Cohorts Only
Figure 6. Forrest Plot of the Effect of New-onset Depression In-hospital vs. New-onset
Depression Post-hospitalization on Mortality or Cardiac Morbidity in CHD Cohorts
25
Figure 1. Depression Timeframes
1. Pre-morbid
Depression:
1a. Post-CAD Depression Status
Unknown
2. Post-morbid Depression:
2a. New Onset/Incident (In-hospital)
In-hospital/ Post-hospitalization
within 1 mo post- (>1 mo post-CHD
CHD event event)
26
Figure 2. PRISMA Flowchart of Article Inclusion
Studies included in
qualitative synthesis
(n =24)
Studies included in
quantitative synthesis
(meta-analysis)
(n = 22)
27
Figure 3. Forrest Plot of the Effect of Timeframe of Depression on Mortality or Cardiac Morbidity among CHD and CHD-free Cohort
Group by Study & Year Subgroup within study Risk ratio and 95% CI
Subgroup within study
28
Figure 4. Forrest Plot of the Effect of Pre-morbid vs. Post-morbid Depression Onset on Mortality or Cardiac Morbidity among both CHD and
CHD-free Cohort
Group by Study & Year Subgroup within study Risk ratio and 95% CI
Onset
29
Figure 5. Forrest Plot of the Effect of Incident vs. Non-incident Depression on Mortality or Cardiac Morbidity among CHD Cohorts Only
Group by Study & Year Subgroup within study Risk ratio and 95% CI
Incident
30
Figure 6. Forrest Plot of the Effect of New-onset Depression In-hospital vs. New-onset Depression Post-hospitalization on Mortality or Cardiac
Morbidity in CHD Cohorts
Group by Study & Year Subgroup within study Risk ratio and 95% CI
In-hosp
31
Table 1. Characteristics of Selected CHD Cohort Studies by Depression Ascertainment Timeframe
Study Design Sample (size, Depression Depression Depression Ascertained Categories N % Endpoint Results Adjustment Quality
sex, age, CHD) Assessment Ascertainment Pre- In- Post- Score
tool Timeframe Event Hospital Discharge
Blumenthal, Prospective CAG patients CES-D CES-D before No Yes Yes New-onset (Pre- 78 14.1 0-12 yr all- New-onset at 6 A, B, D, F, H 6
2003, U.K. Cohort, w/o history of CABG & 6 mos CABG) cause mos HR=2.17 &N
[28] single site dep N=555, after CABG Persistent mortality (.81-5.81); New-
mean (SD) age, onset Persistent
61 (10.2), 27% HR=2.2 (1.17-
females 4.15); New-onset
Transient HR = .99
(.48-1.99)
No No Yes New-onset 36 6.5
(6 mo Post-
CABG)
No Yes No New-onset (Pre- 108 19.5
CABG) Transient
Carney, Prospective MI patients, DISH; BDI within 28 days No Yes N/A New-onset 370 27.9 29-mo all- All Cause: New- A, B, D, E, 8
2009, U.S cohort N=1328, mean post-hospital cause onset HR=3.1(1.6- H, J, K, L, M
[9] (ENRICHD), (SD) age, 59.9 admission, pre-MI mortality; 6.1); Recurrent &N
multisite (11.8), 49.1% dep was assessed CHD death HR= 2.2 (1.1-4.4).
females by DISH; 6 mo Cardiac Death:
post-enrollment New-onset
HR=2.7 (.9-4.9);
Recurrent HR=1.3
(.8-2)
Yes Yes N/A Recurrent 550 41.4
Connerney, Prospective CABG patients, Clinical Clinical Interview No N/A Yes New-onset 39 12.6 12-mo CHD *New-onset= N/A 5
2001, Cohort, N=309, mean Interview few days after (In-hospital) mortality/ 11/39 (28.2%);
U.S.[37] single site age=65, 33% (USNIMH); CABG re-admission Recurrent=6/24
female BDI (25%); History
Only=6/48
(12.5%); No
Dep=19/198
(9.6%)
Yes N/A Yes Recurrent 24 7.8
32
Grace, Prospective ACS patients, BDI In-hosp, hist of dep No Yes N/A New-onset 130 17.3 5-yr all- New-onset A, E, G & O 7
2005, Cohort, N=750, mean was assessed by (In-hospital) cause HR=1.75 (1.14-
Canada multisite (SD) age, 61.6 self-report mortality 2.69), Recurrent
[12] (11.9), 35% questionnaire Yes Yes N/A Recurrent 105 14.0 dep HR=.98 (.56-
females 1.72), History of
dep only HR=.38
Yes No N/A History only 69 9.2 (.14-1.05)
Hata, 2006, Prospective CABG patients CES-D In-hosp after No Yes N/A New-onset (In- 97 21.5 2-yr all- Hospital Death: N/A 4
Japan [47] Cohort, free of dep CABG hospital) cause New-onset= 11/97
single site N=452, mean mortality (11.3%); No Dep=
(SD) age, 65 0 (0%)Late Death:
(11), 35.2% New-onset=10/97
females (10.3%); No Dep=
10 (2.8%)
All Death:
New-onset= 21/97
(21.6%); No Dep=
10/355 (2.8%)
No No N/A No Dep 355 78.5
Lesperance, Prospective MI patients, DIS; BDI In-hosp, 6 mos & No Yes N/A New-onset 20 9.0 18-mo all- *New-onset in- N/A 4
1996, Cohort, N=222, mean 12 mos (In-hospital) cause hosp= 2/20 (10%);
Canada [2] single site (SD) age, 59.6 mortality Recurrent = 5/15
(24-88), 22% Yes Yes N/A Recurrent 15 6.8 (40%); History
females Only=1/46 (2.2%);
No Dep=9/141
Yes No N/A History Only 46 20.7 (6.4%)
33
Parker, Prospective ACS patients, CIDI; a CIDI (in-hosp; No Yes N/A New-onset (Post- 25 5.5 2-12 mo Death & N/A 5
2008, U.S. Cohort, N=451, mean checklist for retrospective to ACS; Incident) composite Readmission
[13] single site (SD) age, 65.7 DSM-IV dep pre-ACS)& 1 mo outcome: Rates: Pre-ACS=
(12.2), 29.7% symptoms post-discharge readmission/ 7 (12.7%); Post-
females (DSM-IV checklist) CHD ACS= 15 (32.6%);
mortality Non-incident
(Recurrent)=12/50
(24%); Incident
(New-onset)= 8/25
(32%); *History
Only= 2/26 (7.7%);
*No Dep= 55/346
(15.9%)
Yes Yes N/A Chronic (Pre- 30 6.7
ACS; Non-
incident)
Yes No Yes Recurrent (Post- 23 5.1
ACS; Non-
incident)
Yes No N/A History Only 27 6.0
Sorensen, Prospective MI patients, MDI at discharge; No Yes N/A New-onset 46 6.6 1-yr all *New-onset= 2/46 N/A 4
2006, Cohort, N=694, mean history of cause (4.3%);
Denmark multisite (SD) age, 59.2 depression was mortality Recurrent=0 (0%);
[48] (9.6); 24.2% assessed at Yes Yes N/A Recurrent 12 1.7 History Only=1/37
females discharge (2.7%); No Dep=
9/599 (1.5%)
Yes No N/A History Only 37 5.3
34
Post-hospitalization Depression Ascertainment (>1 month to 12 months post-CHD event)
Borowicz, Prospective CABG patients, CES-D 1 month after No N/A Yes New-onset 35 24.1 5-yr CHD *New-onset=6/35 N/A 5
2002, U.S. Cohort, N= 145, mean CABG morbidity (17.1%);
[38] single site (SD) age, 62.7 Recurrent=0 (0%);
(9.5), 22.1% History Only=1/8
females (12.5%); No Dep=
9/92 (9.8%)
Yes N/A Yes Recurrent 10 6.9
Dickens, Prospective MI patients, N= HADS; 313 HADS pre-MI dep No N/A Yes New-onset 71 16.1 12-mo CHD New-onset, D, K, N & P 6
2008, U.S. Cohort, 440, mean (SD) MI patients (retrospective); 12 mortality HR=2.33 (1.05-
[11] single site age, 60 (11.1), were mo post-MI 5.16), Recurrent,
29.6% females interviewed HR=.31(.07-1.36);
by SCAN History Only=no
data.
Yes N/A Yes Recurrent 96 21.8
de Jonge, Prospective MI patients, N= CIDI 3 & 12 mos post- Yes N/A Yes Recurrent 53 11.3 12-mo CHD New-onset A, B, C & N 6
2006, Cohort, 468, mean (SD) (assessed MI mortality & HR=1.76 (1.06-
Netherland multisite age, 61 (11.4), incident/non- CHD-related 2.93), Recurrent
[10] 19.7% females incident dep) re- HR=1.39 (0.74-
admissions 2.61); *History
Only= 5/22
(22.7%)
No N/A Yes New-onset 66 14.1
35
May, 2009 , Prospective CHD pts, IDC-9 Post CHD No N/A Yes New-onset Post- 1377 10.0 5.6-yr CHD New-onset A, B, D, E, 9
U.S. [29] Cohort, N=13,708, mean diagnosis morbid morbidity HR=1.5 (1.38- F, G, I, L &
multisite age=52, 55% (Heart 1.63) N
(Inter- females Failure)
mountain
Heart
Collaborative
Study)
No N/A No No Dep 12331 90.0
Parashar, Prospective MI patients, PHQ In-hosp & 1 mo No Yes Yes New-onset 371 19.8 6-mo all- *New-onset= N/A 4
2006, U.S. Cohort N=1875, mean Persistent cause 15/371 (4%);
[49] (PREMIER), (SD) age= 60.5 (at 1 month) mortality Recurrent dep=
multisite (12.5), 31.6% Yes No Yes Recurrent 125 6.7 4/125 (3.2%);
females History Only=
3/109 (2.8%), No
Yes No No History Only 109 5.8 Dep= 28/1270
(2.2%)
ACS, acute coronary syndrome; BDI, Beck Depression Inventory, CABG, coronary artery bypass surgery; CES-D, Center for Epidemiological Studies Depression scale; CHD, coronary artery disease; CIDI, Composite International
Diagnostic Interview; DASS, Depression Anxiety Stress Scale; DIS, Diagnostic interview schedule; DISH, depression interview and structured Hamilton; Dep, depression; DSM, Diagnostic Statistics Manual; HADS; Hospital Anxiety &
Depression Scale; HR, hazard ratio; ICD; International Classification of Disease; IHD, Ischemic Heart Disease; In-Hosp, in-hospital; MDD, major depression disorder; MDI, major depression inventory; MI, Myocardial Infarction; Mos,
months; SCAN, Schedule for Clinical Assessment in Neuropsychiatry; USNIMH, US National Institute of Mental Health.
A, age, B sex, C, education, D, disease severity, E, comorbidities, F, smoking status, G, hypertension, H, diabetes, I, family history of CHD, J, history of CHD, K, antidepressant use, L, medication use, M, initial depression score, N, cardiac
function, O, cardiac diagnosis, P, recurrent cardiac events
*data obtained from authors.
36
Table 2 Characteristics of Selected CHD-free Cohort Studies
Depression
Ascertained
Depression
Depression Assessment Quality
Study Sample (size, sex, age, CHD) Assessment Timing Pre-Event Categories N % Endpoint Results Adjustment Score
Anda, 1993, CHD free cohort, N=2801, GWBS baseline Yes Pre-morbid (at baseline) 148 5 12.4 yrs CHD Pre-morbid dep A, B, C, D, F, H, 7
U.S. (NHEFS) mean age 57.5, 52.4% mortality RR =1.5 (1.0-2.3) I, J, K, M & R
[30] females
No No Dep 2653 95
Cole, 1998 CHD free cohort, N=5053, Self-reported baseline Yes Pre-morbid (at baseline) 154 3 12 yrs CHD Pre-morbid dep N/A 5
[50] mean age=61.8 (8.3), 100% physician mortality CHD mortality rate
males diagnosed RR= 1.20 95% CI
depression 0.53 to 2.71.
No No Dep 4899 97
Ferketich, CHD free cohort N=7903 , CES-D baseline Yes Pre-morbid (at baseline) 1154 15 8.3 yrs CHD pre-morbid dep E, D, H, I, L, M 7
2000, U.S. mean age=55.1 (14.2), 63.2% mortality RR= 2.34 (1.54- &O
(NHANES I) female 3.56) for men; 0.74
[32] (0.40-1.48) for
No No Dep 6749 85 women
Ford, 1998, CHD free cohort (medical self-report baseline Yes Pre-morbid (at baseline) 132 11 37 yrs CHD pre-morbid dep A, H, I, J, K, L, 6
U.S. [33] students) N=1190, mean symptoms & mortality RR= 1.8 (.56-5.75) N & Q.
age=26, 0% females treatment;
response
validated by No No Dep 1058 89
physicians
Penninx, CHD free cohort, N=3701, CES-D baseline, 3 and 6 Yes Pre-morbid (at baseline) 256 7 4-yrs CHD Pre-morbid newly A, B, G, H, I, M 7
1998, U.S. mean age, 78.3 (range 70- yrs follow-ups; 1, mortality dep mortality &R
(EPESE) [34] 103), 66.2% females 2, 4, 5 yrs HR=1.45 (1.02-
telephone follow- 2.08); pre-morbid
Yes Pre-morbid Chronic (3 & 221 6
up chronic dep
6 yrs before baseline)
mortality HR=.91
(.59-1.41)
No No Dep 3224 87
Penninx 2001, CHD free cohort N=2397 , CES-D; DIS baseline Yes Pre-morbid major dep 43 2 4.2 yrs Minor Depression: A, B, C, G, H, I, 7
Netherland mean age=69.8 (8.7), 54.5% Cardiac pre-morbid RR= L, M & R
(LASA) [35] female mortality 1.6 (1.0-2.8)
Major Depression:
Yes Pre-morbid minor dep 282 12 pre-morbid RR=
3.8 (1.4-10.6)
No No Dep 2072 86
Surtees, CHD free cohort, N=19,649, HLEQ baseline, 12 mos Yes Pre-morbid (at baseline) 586 3 8.5-yrs CHD Pre-morbid major A, B, E, H, I, J, 7
2008, U.K. age (range 41-80), 58% mortality dep CHD mortality K, L, M, R & R
(EPIC- females HR=2.67 (1.54-
NORFOLK) 4.64)
[36] No Pre-morbid (within past 1030 5
12 mos)
No No Dep 16592 84
Whang, 2009, CHD free cohort, N=62839, MHI-5; anti- baseline (1992), Yes Pre-morbid (at baseline) 4994 8 4 yrs CHD Pre-morbid dep A, B, H, I, J, K, 4
U.S. (Nurse age (range 30-55), 100% depressant use 1996 & 2000 mortality CHD mortality rate L, M, N, P & S
Health Study) females = 64 (1.28%); No
[46] Dep CHD mortality
No No Dep 57845 92 rate=477 (.82%)
106193
CES-D, Center for for Epidemiological Studies Depression scale; CIDI, Composite International Diagnostic Interview; CHD, coronary heart disease; DIS, Diagnostic interview schedule; Dep, depression; GWBS, General Well-Being
Schedule; HLEQ, Health & Life Experience Questionnaire; HR, hazard ratio; IHD, Ischemic Heart Disease; In-Hosp, in-hospital; MDD, major depression disorder; MHI, Mental Health Index; MI, Myocardial Infarction; MMPI, Minnesota
Multiphasic Personality Inventory; Mos, months; ZSDS, Zung Self-report Depressive Symptoms.
A, age, B sex, C, education, D, race, E, socioeconomic status, F, marital status, G, comorbidities, H, smoking status, I, hypertension/blood pressure, J, hyperlipidemia, K, physical activity, L, diabetes, M, body mass index, N, family history of
CHD, O, history of CVD, P, supplement use, Q, graduation year, R, alcohol use, S, medication use.
*data obtained from authors.
Table 3. Summary of Primary Results
New Onset
History Only Pre-morbid Recurrent New Onset (In-hospital) (Post-hospitalization)
History Only
RR=0.76 (95% CI 0.48-1.19) - Q=10.72, p=0.001 Q=6.04, p=0.014 Q=14.26, p<0.0001 Q=5.67, p=0.017
Recurrent
RR=1.59 (95% CI 1.08-2.34) - Q=2.00, p=0.157 Q=0.192, p=0.662
PREVIOUS READMIT$
RECUR$ RE-ADMIT$
CHRONIC$$ RE-HOSPITAL$
PRESIST$ REHOSPITAL$
AFTER
POST-
CAD DEPRESSION TIMING OUTCOMES
FIRST
FIRST-EVER
NEW$
INCIDENT
EPISOD$
TRANSIENT
IN-HOSPITAL
CURRENT$
Appendix B Quality Assessment
Study ID: Author(s) Year:
Reviewer: Y Leung S Gravely S Grace Other:
Note: A study can be awarded a maximum of one star for each numbered item within the Selection and
Outcome categories. A maximum of two stars can be given for Comparability
Selection
1) Representativeness of the exposed cohort
a) truly representative of the average depressed patients in the community
b) somewhat representative of the average depressed patients in the community
c) selected group of users eg nurses, volunteers
d) no description of the derivation of the cohort
2) Selection of the non exposed cohort
a) drawn from the same community as the exposed cohort
b) drawn from a different source
c) no description of the derivation of the non exposed cohort
3) Ascertainment of exposure
a) secure record (eg surgical records)
b) structured interview
c) written self report
d) no description
4) Demonstration that outcome of interest was not present at start of study
a) yes
b) no
Comparability
1) Comparability of cohorts on the basis of the design or analysis
a) study controls for disease severity/ modifiable CHD risk factors (the most important factor)
b) study controls for history of cardiac events/ non-modifiable CHD (family history) risk factors
Outcome
1) Assessment of outcome
a) independent blind assessment
b) record linkage
c) self report
d) no description
2) Was follow-up long enough for outcomes to occur
a) yes (select an adequate follow up period for outcome of interest)
b) no
3) Adequacy of follow up of cohorts
a) complete follow up - all subjects accounted for
4
2
b) subjects lost to follow up unlikely to introduce bias - small number lost - > 80 % follow up, or
description provided of those lost)
c) follow up rate < 80 % and no description of those lost
d) no statement
4
3