Depression and 1-Year Prognosis in Unstable Angina FREE

AFTER MYOCARDIAL infarction, 35% to 45% of patients show some degree of depression.^1- 2 Similar rates have been reported with other manifestations of coronary artery disease.^3- 4 Although depression has a clear impact on quality of life,⁵ is a major cause of disability,⁶ and may influence compliance with medications and lifestyle changes,⁷ its impact can be even more profound. Studies have demonstrated that depression is an independent predictor of cardiac mortality over 6 to 18 months following myocardial infarction and that the risk is not restricted to major depression.^2,8- 10 Unstable angina and myocardial infarction share many clinical and pathophysiological characteristics and are often described together as acute coronary syndromes.^11- 13 Moreover, unstable angina accounts for the majority of coronary care hospitalizations.¹⁴ Despite recent progress in the management of an acute episode of unstable angina, the incidence of recurrent coronary events remains high in the year following admission,¹¹ justifying the search for additional prognostic factors and interventions. This study was designed to evaluate the impact of depression on 1-year cardiac prognosis following hospitalization for an episode of unstable angina.

The Research Ethics Board of the Montreal Heart Institute approved the study on May 12, 1994. All consecutive patients admitted between May 17, 1994, and July 24, 1996, for suspected unstable angina (defined as new-onset angina within 2 months, exacerbation of previous angina with pain at rest or with minimal exercise, prolonged chest pain lasting ≥30 minutes, or angina within 2 weeks following discharge for myocardial infarction) were considered eligible if there was documentation of coronary artery disease prior to hospital discharge. This documentation included ischemic electrocardiographic ST-T changes, previously documented myocardial infarction, a positive nuclear treadmill test result, or coronary angiographic evidence of blockage of 50% or greater in at least 1 major coronary artery. Patients were excluded if the unstable angina was secondary to another disease or if an acute myocardial infarction was documented during the admission. Also, patients who had undergone a coronary angioplasty or bypass surgery within the previous 6 months were excluded because of the different underlying pathophysiologic conditions involved in early episodes of unstable angina following these procedures.^15- 16 Of the 1388 surviving subjects who met the study definition of unstable angina, 67 were excluded because of another disease likely to limit survival to less than 1 year, 108 because of cognitive impairment or insufficient knowledge of French or English, and 138 because of early discharge. Finally, another 283 subjects for whom bypass surgery was planned to take place during the index admission were excluded because of the potential impact of surgery on 1-year cardiac prognosis,¹⁷ and the fact that the evaluation of depression could have been affected by the surgical procedure. Of the remaining 792 eligible patients, 448 provided informed consent and 432 participated in baseline interviews, with 430 completing the primary measure of depression (the Beck Depression Inventory [BDI]¹⁸). Men were more likely to accept study participation and complete interviews than women (57% vs 48%; P=.01). Participants were also on average 4 years younger than nonparticipants (P<.001). No other data were available concerning patients who did not provide informed consent.

A research assistant carried out structured baseline interviews a mean of 5 days (SD, 2.7) after admission. These included questions about social and demographic characteristics and our primary measure of depression, the 21-item self-report BDI.¹⁸ A cutoff score of 10 or greater out of 63 was applied to the BDI to identify individuals with at least mild to moderate symptoms of depression.¹⁹ We also included a modified version of the Diagnostic Interview Schedule (DIS), an interview administered by a research assistant that assesses the psychiatric criteria for major depression.²⁰ Interviewers were blind to BDI scores at the time of administration of the DIS. Our previous work with the 2 indices after myocardial infarction indicated that while the DIS was a good predictor of cardiac mortality during the first 6 months, over 12- and 18-month periods it was not as strong a predictor as the BDI.¹⁰ Therefore, the BDI was selected as the primary measure of depression for the current study. We planned to include 600 patients to assure a power of at least 80% to be able to detect a doubling in risk of cardiac events associated with BDI scores of 10 or greater. This was based on an estimated combined cardiac event rate of 10% over 1 year, a prevalence of elevated BDI scores of 33%, and 2-tailed P≤.05.

Electrocardiograms on arrival at the emergency department (n=427) and those recorded during chest pain later during the admission (n=236) were read by an experienced research nurse blind to other study data. Evidence of myocardial ischemia included any of the following in 2 contiguous leads: new persistent or transient ST-segment depression of 0.1 mV or greater or T-wave inversion, or transient (<20 minutes) ST-segment elevation of 0.1 mV or greater.

Left ventricular ejection fractions were available for 404 patients (316 during the index hospitalization). This included 324 based on contrast angiography, 56 based on radionuclide ventriculography, and 24 based on 2-dimensional echocardiography. Results of coronary angiograms were also accessible for 411 patients, including 344 obtained during the index admission. Data from a previous admission were used only when there was no evidence of a subsequent myocardial infarction or revascularization procedure prior to the index admission. Angiograms were blindly coded by an experienced cardiac radiologist using the Bypass Angioplasty Revascularization Investigation rules to assess the severity of coronary atherosclerosis.²¹ A reduction in lumen diameter of 50% or more was considered significant unless it was corrected by a patent graft with less than 50% stenosis. The other baseline medical characteristics were extracted from hospital charts.

The primary study outcome was the combination of cardiac mortality and nonfatal myocardial infarction (major cardiac events) during the 12 months following discharge. This outcome was obtained for all patients on the basis of a 12-month telephone interview with patients or family members, Quebec Medicare data, and hospital chart searches. Quebec Medicare data provide information on virtually all physician services carried out in the province. Hospital records for all cardiac readmissions were abstracted, including data on procedures, enzyme levels, chest pain, and electrocardiographic changes. The diagnosis of nonfatal myocardial infarction was based on a rise in creatine kinase levels with a diagnostic elevation in creatine kinase MB fractions in conjunction either with new Q-wave or ST-T changes suggestive of acute ischemia or with typical chest pain lasting 30 minutes or longer. Readmissions for unstable angina were identified using index criteria. Study cardiologists (P.T. and M.J.) blind to depression data independently classified all deaths as cardiac or noncardiac based on all available data from hospital records, ambulance records, and death and autopsy reports (when available). Disagreements were discussed to obtain a consensus. All dates of death were confirmed by death certificates or hospital records.

Statistical analyses were carried out using SPSS for Windows (version 8.5).²² All tests were 2-tailed. Logistic regression analysis and the χ² statistic²³ were used to assess the odds ratio for cardiac events in depressed vs nondepressed patients and to compare the baseline characteristics of depressed and nondepressed patients and of those who did and did not experience a cardiac event. We assessed the impact of depression after control for each baseline variable by carrying out separate hierarchical logistic regression analyses, forcing in each predictor in the first step followed by the addition of depression in the second step. In addition, stepwise multiple logistic regression analysis was used to select the baseline variables independently (P≤.05) related to depression, and hierarchical logistic regression analysis was used to calculate the adjusted impact of depression after control for these variables. Finally, in order to assess the importance of depression in relation to multiple baseline predictors, depression and all other baseline variables significantly related to major cardiac events were entered into a backward stepwise logistic regression analysis confirmed with forward stepping. Criteria for entry and removal were based on the likelihood ratio test with enter and remove limits set at P≤.05 and P≥.10, respectively.

Patients' ages ranged from 31 to 87 years (mean, 62.1; SD, 10.6). There were 123 women (28.6%). Fifty-four percent had a previous myocardial infarction, 16.5% had 3-vessel disease (n=411), 21.3% had left ventricular ejection fractions of 0.45 or lower (n=404), and 48.4% underwent angioplasty during the index admission. Overall, 41.4% were depressed according to the BDI (score ≥10), including 15.1% who met the psychiatric criteria for major depression according to the DIS. By 1 year, 16 patients had died, 13 of cardiac causes (including 6 fatal myocardial infarctions). There were 2 deaths caused by cancer and 1 caused by meningitis. In addition, 15 patients survived a myocardial infarction. The overall 1-year rate of major cardiac events, including cardiac deaths and nonfatal myocardial infarctions, was 6.5% (n=28).

As Table 1 shows, patients who were depressed (BDI ≥10) were significantly more likely to die of cardiac causes or to experience nonfatal myocardial infarctions than other patients (odds ratio, 4.68; 95% confidence interval [CI], 1.94-11.27; P<.001). Indeed, 21 of the 28 major cardiac events occurred in the 41.4% of patients with depression. Kaplan-Meier survival curves for major cardiac events (with noncardiac deaths censored on the day of the death) in relation to elevated BDI scores appear in Figure 1. The curves separated early and kept diverging during follow-up. In addition, the differences were significant when cardiac deaths and nonfatal myocardial infarctions were considered separately. Rates of readmission for unstable angina were only marginally related to depression.

Within the group of depressed patients (BDI ≥10), those who met the psychiatric criteria for major depression on the DIS did not differ significantly on any outcomes from those who did not. However, the rates of events tended to be lower in those who met the DIS criteria. For example, 7.7% of the patients with major depression experienced a major cardiac event in comparison with 14.2% of those with elevated BDI scores who did not meet the criteria for major depression (P=.20). Furthermore, only 1 of the 9 cardiac deaths among patients with elevated BDI scores occurred among those with major depression (P=.14). Nonetheless, the number of events in the depression subgroups is small, and the study was not planned to have sufficient power to draw conclusions about the presence or absence of prognostic differences between them.

Depressed patients (BDI ≥10) were undistinguishable from nondepressed patients in terms of age, education, hypertension, diabetes, previous angioplasty, baseline electrocardiographic evidence of ischemia, angioplasty during admission, number of diseased coronary arteries, and prescription at discharge of aspirin, hypolipidemics, β-blockers, and calcium channel blockers (Table 2). Not surprisingly, depressed patients were more likely to be female, to be unmarried, to report that they had no close friends, and to be current smokers. However, depressed patients were also more likely to have had previous myocardial infarctions, admissions for unstable angina, and bypass surgery, and to have left ventricular ejection fractions of 0.45 or lower. Nitrate therapy and triple anti-ischemic therapy (the combination of β-blockers, calcium channel blockers, and nitrates) were also more common among depressed patients.

In summary, there were differences between the depressed (BDI ≥10) and nondepressed patients on 10 of the 24 background variables. Control for each of these individual variables resulted in odds ratios for depression ranging from 4 to 5 (see Table 3). However, as in most risk factor studies, many of these variables were interrelated. For example, women were more likely to be unmarried (P<.001), patients with a previous myocardial infarction were more likely to have low left ventricular ejection fractions (P<.001), and patients with previous admissions for unstable angina were more likely to have had previous bypass surgery (P<.001). Therefore, multivariate stepwise procedures were used to identify the group of baseline factors independently (P<.05) related to depression assessed by the BDI. Four variables emerged: sex, smoking, previous bypass surgery, and prescription of nitrates (goodness of fit χ²₇=2.89; P=.90). The impact of depression on cardiac events was only slightly attenuated and remained significant after simultaneous adjustment for these 4 independent imbalances (adjusted odds ratio, 3.82; 95% CI, 1.54-9.52; P=.002). Thus, there was little evidence that depression's impact on cardiac events was attributable to observed background differences between the depressed and nondepressed patients.

In addition to depression (BDI ≥10), the significant bivariate predictors of major cardiac events included previous coronary bypass surgery, baseline electrocardiographic evidence of ischemia, absence of angioplasty during admission, left ventricular ejection fraction of 0.45 or lower, 3-vessel disease, prescription of nitrates, and prescription of triple anti-ischemic therapy. Older age, previous myocardial infarction, and previous unstable angina were also marginally related to the occurrence of major cardiac events. However, as shown by the adjusted odds ratios in Table 3, depression continued to have a significant impact after control for each of the other separate baseline predictors. Furthermore, there was no evidence of significant interactions between depression and any of the other predictors; that is, depression had an impact in both the presence and absence of other prognostic variables.

In order to compare the predictive importance of depression with that of the other baseline factors, stepwise multiple logistic regression analysis was used to identify the independent predictors of major cardiac events. The model building was based on the 392 patients (91.2%) with complete data on predictive variables. The crude odds ratio for cardiac events associated with depression in these patients was 5.51 (95% CI, 2.16-14.04; P<.001). While more depressed than nondepressed patients had missing data (11.2% vs 7.1%; P=.14), only 2 major cardiac events (1 depressed patient) occurred in the patients not included in the multivariate analysis.

The final multivariate model included electrocardiographic evidence of ischemia, left ventricular ejection fraction of 0.45 or lower, the number of diseased vessels, and depression (Table 4). The adjusted odds ratio for depression was 6.73 (95% CI, 2.43-18.64; P<.001). Thus, depression had a significant 1-year prognostic impact independent of and comparable with that of other known predictors of cardiac events.

To our knowledge, this is the first prospective study documenting that depression is associated with worse prognosis in patients with unstable angina. We found that depression increased the risk of cardiac death or nonfatal myocardial infarction more than 4-fold. The prognostic impact of depression remained after adjustment for the major background variables associated with depression. Furthermore, our data strongly suggest that depression is an independent risk factor for cardiac events. The impact of depression was statistically independent of and at least as important as left ventricular function, extent of coronary atherosclerosis, and baseline electrocardiographic evidence of ischemia, all objectively measured baseline predictors of prognosis. Furthermore, no significant interactions were found between depression and the other baseline risks, so that the risk associated with depression was observed in both the presence and absence of other prognostic variables.

This report strengthens the evidence that depression is a risk factor for subsequent cardiac events in patients with established coronary artery disease^24- 25 and demonstrates that the risk is not restricted to the post–myocardial infarction period.^2,8- 10 Given the increasing number of hospitalizations for unstable angina in comparison with myocardial infarction,¹⁴ these results are of particular clinical importance.

Of all the psychological conditions that have been reported to predict outcomes in patients with coronary artery disease, the epidemiological findings are most consistent for depression.²⁶ Secondary analyses of several large longitudinal data sets also suggest that depression may influence the incidence of coronary artery disease.^27- 30 Although a number of potential behavioral and physiological mechanisms have been proposed to explain the link between depression and cardiac events,³¹ including decreased compliance with medications or recommended lifestyle changes,⁷ hyperactivity of the sympathoadrenal system,³² diminished heart rate variability,³³ and enhanced platelet responsiveness,³⁴ it is important to underscore that no direct causal relationship has yet been established.

Other research has indicated that depression is often overlooked or untreated by clinicians in medical settings.^35- 36 To explore this issue, we examined Quebec Medicare data and found that only 66 study patients were seen at least once by a psychiatrist during the index admission or during the year after discharge. This involved 9.5% of the patients with baseline BDI scores lower than 10 and 23.6% of those with BDI scores of 10 or greater (including 29.2% of those who met the DIS criteria for major depression and 20.4% of those who did not). Furthermore, based on hospital charts, only 17 of the 178 depressed patients (including 9 subjects with major depression) were prescribed antidepressants at discharge or later during the year, precluding any attempt at assessing the impact of treatment.

As in our post–myocardial infarction research,¹⁰ patients did not need to experience full-blown major depressive episodes to be at increased risk. In fact, in the current study there was a tendency toward higher cardiac event rates among patients with mild to moderate levels of depressive symptoms on the BDI who did not meet the diagnostic criteria for major depression than among those who did meet the criteria. However, the difference was not statistically significant. In order to have had sufficient power to reliably compare the 2 depression subgroups, we would have needed a sample at least twice as large as that in the current study. Furthermore, the DIS interviews for assessing major depression took place on medical wards with little privacy during relatively short hospitalizations for unstable angina. In this context, diagnostic evaluation using the DIS may not provide as accurate a measure of a major depressive episode as it does in other settings. In contrast, the self-report format of the BDI may facilitate acknowledgment of depressive symptoms and is probably a preferable approach to in-hospital screening for patients at increased cardiac risk associated with depression. It is also possible that more severely depressed patients were less likely to take part in this study because of other exclusion criteria and refusal. However, the distinction between major and subthreshold or minor depression may have little meaning in terms of prognosis in cardiac disease. Both may be presentations of the same chronic affective illness that fluctuates over time, independently of medical comorbidity.^37- 39 Thus, the increased cardiac risk may be associated with the underlying depressive illness rather than the severity of its current presentation.

Study limitations include the fact that the 20% of patients who had bypass surgery during the index admission were excluded. Their in-hospital prognosis is related to their surgery, and their overall 1-year cardiac prognosis is good.¹⁷ Because it is unlikely that any 1-year prognostic factors, psychological or physical, could be detected in this group without a sizable increase in sample size, they were not included in the study. In addition, many patients who were well enough to be discharged early were not available for interviews, and this may also have biased the sample toward a more severely ill group. Also, 45% of eligible patients refused participation. Refusal was particularly common in women and elderly patients, a pattern similar to that in other recent psychosocial studies of patients with coronary artery disease.⁴⁰ Thus, our results apply to patients with unstable angina who do not have surgical treatment during the index admission and the types of patients who are willing and able to complete 1-hour baseline interviews before discharge. Our research was conducted only in one university hospital specialized in cardiac care, and this also limits the generalizability of the findings. Furthermore, the relatively small number of major cardiac events and the wide CIs of our statistical models are also limitations. However, the clinical variables in our data set included all the currently used predictors of prognosis, and the end points were objective and clinically important, reinforcing the prognostic significance of depression.

While this report documents the prognostic importance of depression following an admission for unstable angina, it also raises many questions. Longitudinal studies are needed to establish whether the cardiovascular risk is linked to a severe episode of depression or to a broad spectrum of presentations of depression. Research exploring the biological mechanisms underlying the depression-related increase in cardiovascular risk could help identify the subset of depressed patients at higher risk and provide more specific pathophysiological factors on which to base treatment to modify that risk. Depression, as measured by the BDI, has a similar prognostic impact in patients with unstable angina and those with myocardial infarction. This suggests that these 2 groups of patients could be combined in trials targeting depression. More importantly, it extends the clinical significance of depression from patients with myocardial infarction to the large group of patients with an acute coronary syndrome.

Accepted for publication October 14, 1999.

This study was sponsored by grants from the Medical Research Council of Canada and the Pharmaceutical Manufacturers' Association of Canada Health Program, Ottawa, Ontario, in collaboration with Pfizer Canada Inc, Montreal, Quebec; the Fonds de la Recherche en Santé du Québec, Montreal; the Pierre David Fund, Montreal; and the Montreal Heart Institute Research Fund.

We thank the patients who generously took part in study interviews and telephone calls; Marie Drolet, Lise Grenier, and Mélanie Richard for their work in data collection and coding; Ginette Gravel, MSc, and Aline Masson, MSc, for data analysis and outcome verification; Doris Morisette, RN, for carrying out electrocardiographic readings; Jacques Lespérance, MD, Marie-Josée Dussault, and Colette Desjardins for coding results of angiograms; and the Régie de l'Assurance Maladie du Québec and the Commission d'Accès à l'Information du Québec for providing Medicare data.

Corresponding author: François Lespérance, MD, Research Center, Montreal Heart Institute, 5000 Bélanger St E, Montreal, Quebec, Canada H1T 1C8 (e-mail: lesperf@icm.umontreal.ca).