|Home | About | Journals | Submit | Contact Us | Français|
Major depression (MD) is a risk factor for mortality after acute myocardial infarction (MI). Recent reports suggest that the level of risk may depend upon whether the comorbid MD is a first or a recurrent episode. The purpose of this study was to compare survival in post-MI participants from the ENRICHD clinical trial with a first episode of MD to those with recurrent MD.
Survival was compared over a median of 29 months in 370 patients with an initial episode of MD, 550 with recurrent MD, and 408 who were free of depression.
After adjusting for an all-cause mortality risk score, initial Beck Depression Inventory score, and the use of SSRI antidepressants, patients with a first episode of major depression had poorer survival (18.4% all-cause mortality) than those with recurrent major depression (11.8%) (HR=1.4; 95% CI: 1.0-2.0), p=.05). Both first depression (HR= 3.1; 95% CI: 1.6-6.1; p= 0.001) and recurrent major depression (HR=2.2; 95% CI: 1.1-4.4; p= 0.03) had significantly poorer survival than did the nondepressed patients (3.4%). A secondary analysis of deaths classified as probably due to a cardiovascular cause resulted in similar hazard ratios, but the difference between depression groups was not significant.
Both initial and recurrent episodes of MD predict shorter survival after acute MI, but initial MD episodes are more strongly predictive than recurrent episodes. Exploratory analyses suggest that this cannot be explained by more severe heart disease at index, poorer response to depression treatment, or a higher risk of cerebrovascular disease in patients with initial MD episodes.
Numerous studies have established depression as a risk factor for mortality following acute myocardial infarction(1-3), although a few studies have failed to replicate these findings.(3) One possible explanation for the negative findings is that not all patients with depression are at high risk for mortality, and the proportion of truly high-risk depressed patients may vary across samples. One of the factors that may affect the level of risk is whether the major depressive episode at the time of the MI is the patient's first. That is, whether a patient has a history of major depressive episodes prior to their index MI, or whether the episode occurring in conjunction with the MI is the first episode .(4, 5)
In the first study to address this question, Lespérance and his colleagues found that patients with recurrent major depression tended to be at higher risk for mortality than those with an initial episode.(6) This finding was not unexpected: the more exposure to a risk factor, the higher the risk. Having hypertension for ten years, for example, is associated with a higher risk than having it for one year. Thus, depressed patients with a history of earlier depressive episodes, and therefore more exposure to its cardiotoxic effects, should be at greater risk for cardiac events than those with a first episode. However, three recent studies have found that initial episodes of depression may carry more risk than recurrent depression.(7-9)
Recent clinical trials have provided evidence that patients with a recurrent depressive episode following an MI may respond better to antidepressants than those with an initial episode.(10, 11) It is possible, therefore, that patients with a first episode of depression have a form of depression that is less responsive to standard treatments. These patients may be at higher risk for mortality following an MI because their depressions are more likely to persist even if treated.
The primary purpose of this study was to examine the relationship between initial vs. recurrent major depression and survival in post-MI patients who were enrolled in the Enhancing Recovery In Coronary Heart Disease (ENRICHD) clinical trial. The secondary aims were to explore medical, demographic, and psychiatric differences between these subgroups in order to generate hypotheses about why medical prognosis may differ because of depression history, and to determine whether depression history affects response to treatment.
Patients admitted between October 1996 and October 1999 to coronary care units at eight ENRICHD clinical trial sites (Washington University, St. Louis, MO; Duke University, Durham, NC; Harvard University, Boston, MA/Yale University, New Haven, CT [combined]; Stanford University, Stanford CA; University of Miami, Miami FL; University of Alabama, Birmingham AL; University of Washington, Seattle WA; and Rush Presbyterian Hospital, Chicago IL) for an acute MI were screened for eligibility within 28 days following hospital admission. The study protocol was approved by institutional review boards from each participating site. Myocardial infarction was documented by cardiac enzymes and by chest pain compatible with acute MI, characteristic evolutionary ST-T changes, or new Q waves. Details of the methods and design of the ENRICHD clinical trial are available elsewhere.(12, 13) Patients were excluded from ENRICHD if they:
Patients who fulfilled the eligibility criteria and provided informed consent were administered the ENRICHD social support instrument(12) and the Depression Interview and Structured Hamilton.(14) The DISH is a semistructured interview that was developed for ENRICHD to diagnose current depressive episodes in cardiac patients according to the DSM-IV criteria, to determine the severity of depression using the 17-item Hamilton Rating Scale for Depression, and to screen for other psychiatric disorders. The interview also is designed to differentiate between first and recurrent major depressive episodes. There is a high level of diagnostic agreement between DISH interviews administered to cardiac patients by trained research nurses and structured interviews administered by trained clinicians (weighted kappa =0.86).(14) Patients also completed the Beck Depression Inventory(15), a 21-item measure of the self-reported severity of depression.
Medically eligible patients who met the symptom criteria for either major or minor depressive episode for at least 7 rather than the usual 14 days could be enrolled in ENRICHD if they had a prior episode of major depression. However, only patients who met the full DSM-IV criteria for major depression, including a 14-day or longer duration, were included in the analysis.
Patients who met all medical inclusion criteria, but who did not meet the depression criteria of the ENRICHD trial, were recruited for a nondepressed comparison group for an ancillary study from three of the ENRICHD sites (Washington University, St. Louis, MO; Duke University, Durham, NC; Harvard University, Boston, MA/Yale University, New Haven, CT). Additional criteria for enrollment in the comparison group included a score of less than 10 on the BDI and no previous episodes of major depression.
Patients enrolled in the clinical trial were randomly assigned to either the study intervention or to usual care. Intervention patients received individual, and in some cases, group cognitive behavior therapy weekly for up to six months. Patients with severe depression (HAM-D>24), and those who did not show at least a 50% decrease in BDI scores after 5 weeks, were referred to study psychiatrists for consideration of pharmacotherapy. Most of these patients were started on sertraline 50 mg/day, and the dosage was subsequently adjusted up to a maximum of 200mg/day, if needed. Nearly all of these patients received the antidepressant for 12 months.(13) The nondepressed comparison patients did not receive a study-related intervention, but they were followed for the duration of the study.
Follow-up assessments were performed annually beginning 6 months after enrollment. The primary outcome of the ENRICHD trial was the combined endpoint of all-cause mortality or recurrent nonfatal acute MI. Standardized, group-masked classification of the major end points, including probable cardiovascular and noncardiovascular death, was performed by the ENRICHD Medical Endpoints Committee. Death certificates were obtained to document all deaths. The primary endpoint for the present study was all-cause mortality. Suspected cardiovascular mortality was also evaluated as a secondary endpoint.
Chi-square tests and analyses of variance were used to compare demographic and medical variables across the three groups (i.e., initial episode of depression, recurrent depression, nondepressed controls). Kaplan-Meier estimates and covariateadjusted Cox proportional hazards regression models were used to describe the relationship between depression subgroup and survival. Schoenfeld(16) and Martingale residuals(17) were used to test the Cox model assumptions of proportional hazards and linearity of continuous covariates, respectively. Variable-by-time interaction terms were also calculated to test the proportional hazards assumption. Kaplan-Meier survival curves for the three groups were compared with the log rank statistic.(18)
A previously published, weighted index of independent risk factors for all-cause mortality in the ENRICHD trial was used to adjust for possible confounders.(19) All major risk factors and cardiac treatments were initially considered in this model, including smoking, hypertension, gender, current heart failure, and being discharged on beta blockers. The final risk score, representing the best set of independent predictors for all-cause mortality for the participants in the ENRICHD clinical trial, included age, diabetes, LVEF, creatinine level, prior MI, history of pulmonary disease, prior transient ischemic attack or stroke, history of heart failure, Killip class at time of the index MI, and treatment with vasodilators. Baseline BDI scores, and SSRI antidepressant use during the study, which was previously shown to be associated with improved survival in the
ENRICHD trial (20), were also included in the adjusted models. Because SSRI use varied throughout the follow up period, the effect of SSRI use was modeled as a timedependent covariate, as described by Taylor and colleagues (20).
Multiple imputation (SAS Proc MI) was used to impute missing data which occurred in 3%-10% of cases for the covariables included in the statistical models.(21) Survival outcomes were not included in the imputation model. All analyses were performed on 50 completed data sets in which missing values were replaced with values estimated from observed variables. The resulting model estimates were then combined for statistical inference. SAS 9.1.3 software (SAS Institute, Inc., Raleigh, NC) was used to perform all statistical analyses.
Nine hundred twenty patients enrolled in the ENRICHD study who met the full DSM-IV criteria for major depression at the time of enrollment, including the >14 day episode duration criterion, and 408 patients who were free of depression but otherwise eligible for the ENRICHD study, were included in the present analyses. Three hundred seventy (40%) of the patients who met the criteria for major depression at enrollment reported this to be their first depressive episode, and 550 (60%) reported having one or more previous episodes. Comparisons of the medical and demographic characteristics of the three groups are presented in Table 1. There are many demographic and medical differences between both of the depressed groups and the nondepressed patients, including a higher proportion of women and a higher prevalence of diabetes. These differences have been reported in earlier studies.(22, 23)
During a median follow-up of 29 months, 3.4% of the nondepressed patients died, compared to 18.4% of the patients with first-time major depression, and 11.8% of those with recurrent major depression. There was no difference in survival between intervention vs. usual care arms for either first depressive episode (20% vs. 17%, p = 0.23) or recurrent depression cases (11% vs. 12%, p = .61). Consistent with the results of the ENRICHD trial (13), there were no differences in survival overall between patients who received the intervention and those in the usual care arm (log rank: χ2= .17; df =1; p = 0.68). Thus, treatment arm assignment was not included in the primary analyses.
Kaplan-Meier survival curves for the three groups are presented in Figure 1. The omnibus test of differences in survival among the groups was significant (log-rank test: χ 2= 38.4; df = 2; p=0.001). Post-hoc comparisons of survival among the three groups with a Tukey adjustment for multiple comparisons showed that both recurrent (χ2= 10.4; df =1; p=0.004) and first-time (χ2= 38.2; df=1; p < 0.001) depressed groups differed from the nondepressed group. First-time and recurrent depression groups also differed from each other (χ2= 5.8; df =1; p=0.04).
Unadjusted hazard ratios (HRs) were estimated from a Cox proportional odds model for patients with first-time major depression and those with recurrent major depression, each compared to the reference group of nondepressed patients. Both first-time major depression (HR, 5.2; 95% CI: 2.9-9.3; p< 0.001) and recurrent major depression (HR, 3.3; 95% CI: 1.8-5.9; p<0.001) predicted shorter survival. Patients with a first time depression were at higher risk than those with recurrent depression (HR= 1.6; 95% 1.2 - 2.2; p=0.008). Adjusting for the ENRICHD all-cause mortality risk score, initial BDI score, and SSRI antidepressant use, the difference in adjusted survival between those with recurrent major depression and those with a first episode of major depression remained significant (HR=1.4; 95% CI: 1.0-2.0; p=.05). Similarly, following adjustment for covariates, survival time for those with first-time (HR= 3.1; 95% CI: 1.6-6.1; p=0.001) and recurrent major depression HR=2.2; 95% CI: 1.1-4.4; p=0.03), remained significantly less than for nondepressed patients.
In a secondary analysis, unadjusted hazard ratios for cardiovascular-specific mortality were again estimated using Cox regression. Death likely due to cardiovascular causes occurred in 2.5 % of the nondepressed patients, 11.1 % of the patients with first-time major depression, and 7.6 % of those with recurrent major depression. First-time major depression (HR = 4.5; 95% CI: 2.3 -9.0; p< 0.001) and recurrent major depression (HR = 3.0; 95% CI: 1.5-6.1; p=0.002) predicted shorter survival from cardiovascular-related mortality, although the difference in survival between the two groups only approached significance (HR=1.5; 95% CI: .97-2.3; p = 0.07). After adjusting for the ENRICHD all-cause mortality risk score, initial BDI score, and SSRI antidepressants, the effects for first time major depression remained significant (first-time major depression HR = 2.7; 95% CI: 1.2-6.0; p= 0.02), while the effect for recurrent major depression was only marginal (HR=2.1; 95% CI: .9-4.9; p<0.09). Cardiovascular-related mortality did not differ between the recurrent major depression (reference group) and first episode of major depression subgroups (HR=1.3; 95% CI: 0.80-2.0; p=0.31).
To determine whether 6-month depression remission rates differed between the two depressed groups, chi-squared analyses were performed separately for patients in the intervention and usual care groups. Remission of depression was defined in the ENRICHD trial as a score of BDI < 7 at the 6-month assessment. The 6-month BDI was completed by 80% of the ENRICHD participants. In the intervention group, 57% of those with first-time depression remitted by six months, compared to 41% of those who enrolled with recurrent depression (χ 2 = 8.4, df = 1; p = 0.004). In the usual care group, 40% of the patients with first-time depression, and 20% of those who enrolled with recurrent depression, remitted at six months (χ 2 = 16.0, df = 1; p=0.001). However, the difference between the intervention and usual care groups in 6-month BDI change scores did not differ by initial (mean = 3.3) versus recurrent major depressive episode (mean = 3.9; F = 0.18; df = 1; p = 0.67), although the baseline levels did (first = 19.4 ± 8.2 vs. recurrent = 22.2 ± 8.4), suggesting that the difference in remission rates may be due to lower baseline depression scores in the group with an initial depressive episode.
Participants in the ENRICHD clinical trial with a first episode of major depression had poorer survival (18.4% all-cause mortality) than those with recurrent major depression (11.8%), and both groups had significantly poorer survival than did the nondepressed patients (3.4% all-cause mortality). These differences persisted after adjusting for baseline BDI score, SSRI antidepressant use, and other medical and demographic predictors of mortality. The results of a secondary analysis of cardiovascular deaths found similar results, but the two depression groups did not differ even though the hazard was similar to that for all-cause mortality (1.3 vs. 1.4). Only 83 of the 133 total deaths were classified as being clearly cardiovascular-related, suggesting that insufficient statistical power may explain the lack of significance.
The findings of the primary analysis differ somewhat from those of five other studies of first-time vs. recurrent depression and survival following MI. In a secondary analysis, Lespérance and colleagues found an 18-month post-MI mortality rate of 40% in 15 patients with recurrent major depression, compared to 10% in 20 patients with first-time depression.(6) In a more recent study of a cohort of 750 post-ACS patients, Grace et al. found that depression (BDI>10) at the time of the MI was a significant predictor of all-cause mortality, but only in patients who had never before been depressed.(9) Thus, only new onset depression without prior depression predicted survival. It is not clear, however, whether these patients were depressed before or only after the MI.
In a study of 468 patients, de Jonge and his colleagues administered a depression interview at 3 and 12 months following an acute MI.(7) They found that of the patients who became depressed in the year following the MI, only those who were depressed for the first time after the MI were at increased risk of cardiovascular mortality and cardiovascular readmissions. Like Grace et al., those who became depressed but who had a history of earlier episodes of depression were not at greater risk for cardiac events.
Dickens and his colleagues studied 588 patients with a recent MI.(8) Patients who were depressed at 12 months after the MI, but not those who reported having been depressed in the week before the MI (HADS > 17), were at greater risk for cardiovascular-related death during the follow-up period. It is not clear from this report whether patients were depressed at or some time during the previous 12 months.
In the most recent report, Parker and his colleagues administered a depression interview to 489 patients following ACS, and interviewed them again by phone at one and 12 months.(24) Neither a history of depression prior to the MI nor depression at the time of the MI was found to be associated with a combined endpoint of cardiovascular mortality or cardiac hospitalization. However, similar to Grace et al. and deJonge et al., depression that began after an ACS admission was a risk factor for future cardiac events. However, unlike Grace et al. and de Jonge et al., the risk was present in patients with a depression following the event regardless of whether there were prior episodes of depression. That is, the timing of the onset of depression in relation to the MI, and not the whether it was an initial or recurrent depressive episode, determined its impact. Unfortunately, the ENRICHD interviewers did not determine the precise onset of the current depressive episode so this could not be evaluated in the present study.
Although there are methodological differences among these five studies of post-MI or ACS patients, including the use of interview-based depression diagnostic criteria vs. self report inventories, different endpoints (all-cause mortality, cardiovascular-related mortality, and cardiac events), and different time points for assessment of depression, none of these differences seem to explain the contradictory findings. Determining depression history is difficult, at best, and often unreliable.(25) It seems likely that this difficulty is magnified when interviewing someone who just experienced a life-threatening medical event. Thus, it is possible that differences in depression history ascertainment may at least partially explain some of the variation in the findings.
Nevertheless, the preponderance of evidence now seems to suggest that an initial episode of depression following an acute MI carries a higher risk of death than does a recurrent episode, especially if its onset occurs after the acute cardiac event. Why patients with an initial depressive episode would tend to be at higher risk than those depressed patients with prior depressive episodes is not clear.
Two earlier studies found that patients who were having their first major depressive episode at the time of diagnostic coronary angiography had more severe coronary artery disease than did those patients with a recurrent episode of depression.(26, 27) Furthermore, first episodes of depression relatively late in life could be manifestations of cerebrovascular disease in some cases.(28, 29) Late life depression has been associated with white matter hyperintensities on MRI scans, suggesting cerebrovascular abnormalities, and with mild to moderate cognitive impairment.(30) Thus, one explanation for the present findings is that patients with an initial episode of depression may have more significant coronary artery disease or cerebrovascular disease than those with recurrent major depression, placing them at higher risk for mortality.
Coronary angiography, MRI scans, and tests of cognitive functioning were not routinely performed on ENRICHD participants before or after the index MI. However, risk factors for coronary artery disease and cerebrovascular disease were determined from systematic chart reviews. ENRICHD patients with initial depressive episodes were younger than those with recurrent depression by an average of two years. Mean systolic blood pressure was slightly higher (4mm/Hg), but well within normal limits (mean = 126.7± 20.2 mm/Hg). Patients with initial episodes were more likely to have coronary bypass surgery following their MI, but no more likely to have coronary angioplasty. On the other hand, they were less likely to smoke, and slightly less likely to have had a history of peripheral vascular disease than were those with recurrent depression. There were no differences between the groups in history of stroke, prior MI, revascularization, hypercholesterolemia, total serum cholesterol levels, or family history of stroke or heart disease. Thus, there was little evidence that these patients were at higher risk for cerebrovascular disease, or that they had more significant coronary artery disease prior to the index MI. Nevertheless, neither of these possibilities can be entirely ruled out.
It has also been suggested that acute MI patients with initial depressive episodes may have more severe coronary heart disease, compared to those with recurrent depression.(7, 31) However, little evidence was found for this in the present study. There were no differences between these groups in Killip class, LVEF, heart failure, prior MI, the ENRICHD all-cause mortality risk score, or the presence of other medical comorbidities. One interesting and unexpected finding was that a higher proportion of patients with initial major depressive episodes had received thrombolytic therapy, compared to patients with recurrent major depression. There were no differences between these subgroups in the proportion of patients with documented Q-wave MIs. In any case, receiving thrombolytic therapy was not associated with survival in this study.
Patients with recurrent depression had slightly more severe depression, were more likely to have a family history of depression, to be female, to have less perceived social support, and to be on antidepressants at the baseline assessment. Although the data were only available for about 80% of cases, patients with initial or recurrent major depression experienced similar changes in depression severity over the first six months after the acute MI. However, because their depressions were slightly more severe at baseline, those with recurrent depression also remained slightly more depressed than those with first-time depression.
Initial major depressive episodes were more likely to remit than were recurrent episodes, both in the intervention and in the usual care groups. However, this may have been due to lower baseline depression scores in the patients who were depressed for the first time. In any case, this finding is not consistent with the results of two recent antidepressant clinical trials.
Unlike our study, Lespérance et al.(11) found no difference in HAM-D scores at baseline between initial and recurrent depressive episodes (29.6±6.8; vs. 29.8±6.7), but patients with recurrent depression showed a larger response to citalopram than did patients in their first major depressive episode. Glassman et al.(10) reported a better response to sertraline in patients with recurrent major depression compared to patients experiencing an initial major depressive episode. Although about 20% of patients in the ENRICHD intervention arm received sertraline, most were treated solely with cognitive behavior therapy. It is possible that patients with initial major depressive episodes following acute MI respond better to psychotherapy, and those with recurrent depression respond better to antidepressant therapy. However, the Lespérance et al. study found that interpersonal psychotherapy had little effect on either type of depressive episode.
One possible explanation for the finding that a first episode of depression is a greater risk for mortality than a recurrent depression is that those patients with prior episodes of depression at greatest risk died as a result of the index MI. Those who survived the index MI may have been at lower risk for reasons that are not presently known. These patients may remain at lower risk in the months following the index MI.
One of the limitations of this study is that the sample was composed of participants enrolled in a clinical trial or in an ancillary study of that trial. As a result, the findings may not generalize to all post-MI patients, as patients who were too sick or debilitated to participate in the trial were excluded. Furthermore, we know little about the duration of the depression episodes that were identified at baseline, the rates of relapse and recurrence, peak severity, or the cumulative exposure to depression over the total follow-up period for either group of depressed patients. This is a shortcoming of this and of most previous studies. A better understanding of the course of depression after acute MI and its relationship to mortality risk remains one of the most important goals in this area of study.(32) Finally, it is possible that high risk patients with prior episodes of depression were less likely to volunteer for the study than those with a first episode of depression, although we have no evidence for this. More than half of the patients in our sample had recurrent depression, similar to that reported in a previous study of depression in post MI patients (31).
In summary, in a large sample of well-characterized post-MI patients with interview-based depression diagnoses, those with initial episodes of major depression tended to have poorer survival than those with recurrent major depression, but both had poorer survival than did patients without depression. Exploratory analyses suggest that the greater risk of mortality seen among patients with an initial episode of major depression cannot be explained by more severe cardiac illness at index or by being at higher risk for more severe coronary artery or cerebrovascular disease. Future research should attempt to confirm this finding, and seek to identify the factors contributing to a greater risk of mortality following acute MI in this common subtype of major depression.
This research was supported in part by Grant No. 2 RO-1HL58946 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland, and from the Lewis and Jean Sachs Charitable Lead Trust
Disclosures:Dr. Carney receives Zoloft from Pfizer, Inc., for an NIH-funded clinical trial. Dr. Jaffe receives research support from Dade-Behring, Beckman-Coulter, and Ortho Diagnostics, and he is a consultant for Dade-Behring, Beckman-Coulter, Ortho Diagnostics, Critical Diagnostics, Liposcience, Bayer, and Pfizer.