Introduction

Myocardial infarction (MI) is a stressful life-threatening event that is often sudden, uncontrollable, and unanticipated. As such, it has been recognized as a potentially traumatic event which may evoke posttraumatic stress reactions [1]. Indeed, studies indicate that 4–18 % of MI patients are classified as having acute stress disorder (ASD) [2–5], a diagnosis that typically involves anxiety responses [6] and is manifested in elevated levels of dissociative, intrusive, avoidant, and hyperarousal symptoms during hospitalization. Although a diagnosis of ASD applies only to the first month after exposure to the traumatic event [6], studies have also shown that it is a marker for long-term distress. More specifically, severity of ASD symptoms during hospitalization has been shown to predict subsequent posttraumatic stress disorder (PTSD) [7–10] and impairment in health-related quality of life [7]. Studies have recognized the adverse effect of anxiety and stress symptoms on individuals’ health. Prospective studies of Vietnam veterans found that PTSD was a predictor of mortality from heart disease [11], as well as of all-cause mortality [12, 13]. Similarly, anxiety was found to predict premature all-cause and cardiovascular death in middle-aged women [14].

PTSD and anxiety were also shown to seriously hinder the recovery of coronary heart disease patients. PTSD was associated with hospital readmission [15] and was evaluated as doubling the risk for subsequent adverse cardiac events among patients with acute coronary syndrome [16]. Similarly, studies have demonstrated that anxiety during the first 3 months after hospitalization predicted major cardiac events among MI patients [17–20], patients with stable coronary artery disease [21], and patients who underwent artery bypass graft surgery [22].

Moreover, there are indications that PTSD and anxiety predict mortality among coronary heart disease patients. Dao and colleagues [23] found that PTSD was associated with increased in-hospital mortality among coronary artery bypass graft surgery patients, and Ladwig and colleagues [24] found that PTSD symptoms, assessed among patients with implantable cardioverter-defibrillators at various times after the implantation (3–142 months), predicted 5-year mortality. Similar effects were documented with regard to anxiety [19, 25, 26].

Yet, some studies suggest that the picture with regard to the role that stress and anxiety play in coronary heart disease patients’ recovery is less clear. There is evidence showing that anxiety during the first 3 months after hospitalization is not associated with either 2- [27] or 5- year [28] mortality. Others report that only persistent anxiety is a risk factor for major adverse cardiovascular events and mortality among coronary heart disease patients [29], while still, others found that anxiety increased cardiac mortality only among female patients [30]. One study even showed that anxiety might be considered a protective factor, as high levels of anxiety among patients treated with percutaneous coronary interventions were positively associated with survival and reduced risk for major cardiac events during the first 5 years after the intervention [31].

Studies also suggest that not all stress symptom clusters have the same adverse effect on individuals’ health. Some studies conducted among survivors of various traumatic events have pointed to the role of emotional numbness [32] in health impairment, while others have emphasized avoidance [33] or arousal [32–34].

A similar inconclusive picture has been observed in the few studies that have evaluated the independent role played by stress symptom clusters in health among coronary heart disease patients. Shemesh et al. [35] reported a nonsignificant trend (p = 0.06) of association between avoidance and serious adverse events, defined as death (from cardiac or other causes) or cardiac-related hospitalizations. Edmondson and colleagues [36], however, found that only intrusive symptoms—assessed 1 month after hospitalization for acute coronary syndrome—predicted the combination of major adverse cardiac events and all-cause mortality, three and a half years after hospitalization. Both studies assessed stress symptoms after the first month post-MI; they did not evaluate in-hospital stress symptoms, in general, or acute dissociation, in particular.

The Current Study

ASD symptoms are considered an early marker for subsequent stress symptoms in trauma survivors, in general [37], and in coronary heart disease patients, in particular [38]. While there is some indication that stress symptoms constitute a risk for mortality among coronary heart disease patients, the unique role played by each of the symptom clusters in the acute phase of hospitalization is unclear. Furthermore, most of the studies assessed stress and anxiety symptoms after patients’ discharge from the hospital and were restricted to the first few years after the cardiac event; there has therefore been no empirical evidence with regard to the role played by acute symptoms over a longer period of time.

The current study aims to examine the role of ASD symptom clusters—expressed during hospitalization—in predicting all-cause mortality over a 15-year period after the MI. Although previous studies have suggested that not all stress symptom clusters exert the same adverse effect on coronary heart disease patients’ health, the findings with regard to each clusters’ effect are inconclusive. Thus, a set of exploratory examinations will assess the independent prediction of each cluster to all-cause mortality. These predictions will be examined while adjusting for factors that have been shown to be related to mortality: age, sex [39], education [40], and left ventricular ejection fraction [41]. Previous studies have indicated that ASD symptoms are often comorbid with in-hospital depression [42] which in and of itself is a recognized risk factor for mortality among coronary heart disease patients [43]. Furthermore, it has been shown that when anxiety is comorbid with depression, the risk for all-cause mortality among coronary heart disease patients increases [44]. Thus, severity of depressive symptoms was also covariated.

Methods

Sample and Study Design

The participants were 196 MI patients, drawn from all patients admitted to the cardiac intensive care units of three medical centers between November 1998 and October 1999, who met the following inclusion criteria: (a) they fulfilled the MI diagnostic criteria of that period: typical clinical symptomatology, electrocardiographic evidence of MI, and typically elevated serum levels of myocardial enzymes; (b) they were 70 years of age or younger; and (c) they were not suffering from any other life-threatening illness. In all, 245 patients met the inclusion criteria. Of these, 196 were assessed during hospitalization (M = 3.69, SD = 2.72 days after admission; response rate = 80 %). Data regarding mortality was missing for three individuals; thus, the sample for the current analyses included 193 MI patients.

Most of the study participants (n = 163; 85 %) were male. At the time of hospitalization, 35 % of the participants (n = 67) were 50 years old or younger, 41 % (n = 79) were between 51and 60, and the remainder (24 %; n = 47) was between 61 and 70 (M = 54; SD = 8.27 years). Twenty-seven percent (n = 49) had fewer than 12 years of formal education, 32 % (n = 60) had 12 years, and the remainder (41 %, n = 50) had more than 12 years.

The study was undertaken after institutional Helsinki committees approved the research design. Informed consent was obtained from all participants.

Measures

Background Variables

Participants were asked to provide information about their sex, age, and level of education. In addition, they were asked to indicate whether they experienced each of 12 stressful life events (e.g., death of significant other, accident, war) during the previous 24 months. Each respondent received a pre-MI life event score, which was the number of events that occurred in the previous 24 months.

ASD Symptoms

ASD symptoms were measured during hospitalization by the Stanford Acute Stress Reaction Questionnaire (SASRQ) [45, 46]. This self-report questionnaire consists of 30 items describing dissociative symptoms (10 items, e.g., “I felt a sense of timelessness,” “I experienced myself as though I were a stranger”), intrusive symptoms (six items, e.g., “I had repeated distressing dreams of the event,” “I had repeated and unwanted memories of the event”), avoidant symptoms (six items, e.g., “I tried to avoid thoughts about the event,” “I avoided contact with people who reminded me of the event”), and hyperarousal symptoms (six items, e.g., “I would jump in surprise at the least thing,” “I felt irritable or had outbursts of anger”). Two additional items reflected impairment in functioning but were not relevant to the respondents’ condition during hospitalization. Respondents were asked to consider the items specifically in relation to their heart attack and to rate on a six-point Likert scale, ranging from 0 to 5, the extent to which they suffered from each of the symptoms. Severity of each ASD symptom cluster was calculated as the mean of the relevant items. In addition, a total ASD symptom score was calculated, as the mean of all items.

Previous studies have indicated the scale’s psychometric properties among survivors of various events, including floods, firestorm, and child abuse [45, 46]. The scale has been shown to have high convergent validity as demonstrated by high correlations with other measures of intrusion and avoidance (the Impact of Event Scale (IES)), as well as high predictive validity, indicated by a high correlation with subsequent PTSD [46]. Cronbach’s alpha for the current sample was high for the ASD symptom score (0.94) as well as for each of the symptom cluster subscales (0.80–0.85), indicating high reliability.

Depression

Depression was assessed by the Beck Depression Inventory (BDI-I) [47], the most commonly used measure to assess depression. It consists of 21 items describing various manifestations of depression. Total scores range between 0 and 63, with higher scores indicating more severe depression. Internal consistency among the 21 items was high (0.88) in the current sample, indicating high reliability.

Medical Measures

Health-related risk factors and clinical data were gathered from the patients’ hospital records. This data included the following risk factors: hyperlipidemia, diabetes, hypertension, cigarette smoking, and body mass index. Cardiac prognostic measures included family history of cardiovascular disease, prior MI, location of MI (anterior /nonanterior MI), left ventricular ejection fraction, left ventricular end-systolic diameter, left ventricular end-diastolic diameter, peak creatine phosphokinase level, lactate dehydrogenase level, primary ventricular fibrillation, percutaneous coronary intervention, coronary artery bypass graft surgery, prescription of beta blockers and angiotensin-converting-enzyme inhibitors, and length of hospitalization.

Vital Status Assessment

Vital status and date of death were longitudinally retrieved from the Administration of Population of the Israeli Ministry of Interior. The endpoint for this study was 15-year all-cause mortality.

Data Analysis

Analyses were performed with SPSS v.21. Overall, 5 % of the data were missing. To account for the missing data, multiple imputation was employed, using ten datasets [48]. The pattern of correlations between ASD symptoms and quantitative measures was calculated using Spearman correlations, and a series of t tests was conducted to examine the relationship between ASD symptoms and dichotic risk factors and clinical measures.

T tests were conducted to examine differences between the decedents and the survivors in background variables, health-related risk factors, clinical variables, stress-related variables, depression, and ASD symptoms. A series of X 2 analyses was conducted to examine the differences between the decedents and the survivors in dichotic risk factors and clinical measures.

To examine the predictive role of the ASD symptom score for all-cause mortality among MI patients, we conducted a series of hierarchical Cox regression analyses (also called Cox proportional hazards models), adjusting for age, sex, education, left ventricular ejection fraction, and depression. As was shown by Vittinghoff and McCulloch [49], Cox models can be used with a ratio of five outcome events per predictor variable. After we conducted the initial series of regression analyses, another series of regression analyses was conducted, for each of the ASD symptom clusters separately. Due to the exploratory nature of the analyses for the four ASD symptom clusters, a familywise Bonferroni adjustment for four comparisons was applied; accordingly, the p value accepted as significant was 0.0125.

Results

Associations Between Background and Clinical Variables and ASD Symptoms

A series of Spearman correlations indicated that the correlations of ASD symptoms with age, number of years of education, left ventricular end-systolic diameter, left ventricular end-diastolic diameter, peak creatine phosphokinase levels, lactate dehydrogenase levels, and number of days of hospitalization were not significant. Avoidance was negatively associated with level of education (r = −0.15, df = 193, p = 0.05). All four ASD symptom clusters, as well as the ASD symptom score, were positively associated with exposure to pre-MI stressful life events (rs 0.16–0.22, df = 193, p < 0.05). Finally, the ASD symptom score and all four ASD symptom clusters were positively associated with depression (rs 0.40–0.53, df = 193, p < 0.001).

A series of t tests further indicated that ASD symptoms were not related to sex, family history of cardiovascular disease, previous MI, risk factors (hyperlipidemia, diabetes, hypertension, smoking, and body mass index), or the other cardiac clinical measures (MI location, ventricular fibrillation, percutaneous coronary intervention, coronary artery bypass graft surgery, and prescription of angiotensin-converting-enzyme inhibitors). Prescription of beta blockers was associated with hyperarousal; those for whom they were prescribed had lower levels of hyperarousal than those for whom they were not (p < 0.05).

Finally, the four ASD symptom clusters were positively associated with each other (rs 56–0.74, df = 193, p < 0.001).

Predictors for All-Cause Mortality After a MI

Forty-two (21.8 %) of the MI patients died at some point during the 15-year follow-up period. Fourteen percent of the 42 died during the first year post-MI, 19 % died 1 to 5 years later, 33 % died 5 to 10 years later, and the remainder (33 %) died 10–15 years post-MI.

Table 1 presents means, standard errors, or distribution of ASD symptoms, background variables, risk factors, and clinical measures, among the two groups. As can be seen, the decedents had been somewhat older during hospitalization and had had higher levels of left ventricular end-systolic diameter and left ventricular end-diastolic diameter, than had the survivors. In addition, the two groups differed in severity of depression, dissociation, hyperarousal, and the ASD symptom score, as the decedents had reported higher levels of depression, dissociation, hyperarousal, and the ASD symptom score during hospitalization than had the survivors.

Table 1 Means, standard errors, and frequencies of demographic variables, risk factors, clinical measures, and ASD symptoms, according to all-cause mortality
Full size table

A series of hierarchical Cox regression analyses was conducted to examine the predictive role of both the ASD symptom score and depression, adjusting for age, sex, education, and left ventricular ejection fraction. Unstandardized coefficients, standard errors, hazard ratios, and 95 % confidence intervals for hazard ratios for these models are presented in Table 2.

Table 2 Unstandardized coefficients, standard errors, hazard ratios, and 95 % confidence intervals for predicting the liability for all-cause mortality by ASD symptom score and depression, adjusting for age, sex, education, and left ventricular ejection fraction
Full size table

As can be seen, age was a significant predictor for all-cause mortality, indicating that the higher the age during hospitalization, the shorter the MI patients’ survival time. For every 1-year increase in the patient’s age, the likelihood for mortality increased by 4–5 %. The ASD symptom score did not predict all-cause mortality. When its predictive role was examined separately, depression was a significant predictor for all-cause mortality (Table 2, model 2). Yet, when the regression model included both the ASD symptom score and depression, the contribution of depression to all-cause mortality was nonsignificant.

Table 3 presents the results of a series of hierarchical Cox regression analyses, conducted to examine the predictive role of each of the ASD symptom clusters in all-cause mortality, adjusting for age, sex, education, left ventricular ejection fraction, and depression. The first model, examining the predictive role of dissociative symptoms, was significant (i.e., p < 0.05, corrected for multiple comparisons using the familywise Bonferroni adjustment; Table 3, model 1). Acute dissociative symptoms were significant predictors for all-cause mortality, indicating that the higher the level of in-hospital dissociation, the shorter the MI patient’s survival time. For every one-point increase in dissociative symptoms, the likelihood for mortality increased by 49 %. To demonstrate the cumulative effect of dissociative symptoms on the likelihood for all-cause mortality, respondents were categorized as either high or low dissociation, according to the median score of dissociation (median = 0.615; M = 1.65, SD = 0.77 for the high-dissociation group; M = 0.18, SD = 0.20 for the low-dissociation group). Figure 1 presents the cumulative survival percentage of MI patients with high and low dissociation. As can be seen, the risk for mortality was higher for the high-dissociation group across all 15 years of assessment. Adjusting for age, sex, education, left ventricular ejection fraction, and depression, none of the other ASD symptom clusters (intrusion, avoidance, and hyperarousal symptoms) predicted all-cause mortality (Table 3, models 2–4).

Table 3 Unstandardized coefficients, standard errors, hazard ratios, and 95 % confidence intervals for predicting the liability for all-cause mortality by ASD symptoms, adjusting for age, sex, education, left ventricular ejection fraction, and depression
Full size table
Fig. 1
figure 1

Cumulative survival percentage of MI patients according to in-hospital dissociation

Full size image

Finally, two hierarchical Cox regression models indicated that when acute intrusion (model 2) and avoidance (model 3) were controlled for, in-hospital depression was a significant predictor for all-cause mortality, indicating that the higher the level of depression during hospitalization, the shorter the MI patient’s survival time. For every one-point increase in the patient’s level of depression, the likelihood for mortality increased by 5 %. Yet, applying the familywise Bonferroni adjustment for multiple comparisons indicated that only model 2 was significant.

Discussion

To the best of our knowledge, this is the first study to examine the prediction of ASD symptoms on all-cause mortality among coronary heart disease patients. The findings of this 15-year longitudinal study show that the severity of in-hospital ASD symptoms, in general, and dissociative and hyperarousal symptoms, in particular, is associated with all-cause mortality. Adjusting for age, sex, education, left ventricular ejection fraction, and depression, only dissociative symptoms remain a significant predictor of all-cause mortality. That is, the higher the level of in-hospital dissociative symptoms, the shorter the MI patients’ survival time.

The findings of this study emphasize the pathogenic effect of dissociative symptoms among hospitalized MI patients. High levels of dissociation during and immediately after traumatic events can be viewed as a compensatory mechanism [50] or a defense [51] against intolerable distress. Indeed, the high correlations between dissociation and the other ASD symptom clusters found in this study are consistent with previous reports that dissociation is positively associated with emotional and physical hyperarousal [50, 52–54]. Thus, high levels of dissociation do not prevent anxiety and distress; instead, they may trigger or be triggered by them [55]. As such, high levels of dissociation may be viewed as a marker of extreme distress in the immediate aftermath of trauma and as a marker for subsequent vulnerability [56, 57].

One should take into account the possibility that high levels of dissociation might predict all-cause mortality, with no relation to the occurrence of an MI. Yet, studies that have analyzed cause of death among coronary heart disease patients in large samples that were longitudinally assessed 13–20 years after diagnosis showed that 50–81 % of deaths were attributable to coronary heart disease [58–60]. Further, although dissociative symptoms are part of the ASD diagnosis [6] and despite longitudinal evidence showing that these symptoms increase following exposure to stress [61], in the absence of a control group, one cannot rule out the possibility that the reported high levels of dissociation among the decedents were unrelated to the MI. Actually, pre-MI vulnerability does not preclude the possibility that post-MI dissociative symptoms reflect a reaction to the traumatic event. As suggested by Butler and colleagues [62] in their diathesis-stress model of dissociative symptoms, these symptoms may appear as a reaction to exposure to a traumatic event among those who have a prior capacity to dissociate: symptoms which, in the absence of the trigger, were not expressed.

The findings of the current study should be considered in view of the revision that was made to the diagnosis of ASD in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [6]. The decision to revise the construct of the diagnosis criteria—from a required presentation of each of the four symptom clusters to a total symptom count—was based on two arguments. According to the first argument, ASD is a highly heterogeneous disorder; acknowledging the diversity of acute stress reactions would therefore increase identification of acutely distressed individuals who might benefit from early identification and intervention [63]. Indeed, the differential prediction of all-cause mortality according to these symptom clusters suggests that attention should be given to each cluster separately.

The second argument for the DSM-5’s revision was that the DSM-IV’s emphasis on dissociative symptoms was overly restrictive [63]. The findings of the current study point to the significance of acute dissociation over the long term and may serve as a reminder for mental health professionals that ASD symptoms, in general, and dissociation, in particular, should be considered not only as precursors for subsequent stress symptoms but also as having other, wider detrimental effects.

A few explanations may be offered for the effect of immediate and enduring distress on an individual’s health. Acute and long-term posttraumatic distress may cause neurochemical and physiological changes that can promote premature mortality. Such changes may be manifested as dysregulations of the activity of the hypothalamic- pituitary-adrenocortical (HPA) axis, often observed among PTSD patients [64], or altered inflammation, demonstrated in MI patients with PTSD [65]. Telomere shortening—a cellular marker of biological aging [66] which has been shown to be associated with all-cause mortality in patients with stable coronary heart disease [67]—is another possible factor, as was seen among individuals with PTSD [68, 69]. Further, studies show that PTSD in MI patients is associated with nonadherence, which in turn relates to poor medical outcomes [70]. Risk behaviors such as smoking, alcohol consumption, reckless driving, or engagement in dangerous activities—which were shown to be associated with PTSD, in general [71, 72], and dissociation, in particular (e.g., [72–74])—may also explain the increased risk for all-cause mortality found in our study. Finally, a recent study demonstrated that PTSD in coronary heart disease patients is associated with sleep disturbances [75] which are, in and of themselves, associated with all-cause mortality [76].

The findings of this study should be considered in light of its limitations. First, the small sample size restricted the number of covariates that could be examined in one model and may have led to unstable estimates. Although the effect of dissociation on all-cause mortality was replicated when alternative models were tested, further research with a larger sample is needed in order to validate these findings. Second, since the inclusion criteria specified the age of 70 or younger and a greater number of male patients than female patients inhabit this age group [77], caution should be employed when generalizing from the findings. A third limitation was the reliance on self-report questionnaires for the assessment of ASD symptoms. Although the self-report is the most common and widely used method to assess acute stress symptoms and the SASRQ has been shown to be a highly valid and reliable measure [45], the well-known limitations of self-reports must be taken into consideration. Another limitation of the study is the lack of information regarding participants’ history of anxiety and depression. Although ASD symptoms, which were measured as a reaction to the MI, were only mildly associated with participants’ exposure to stressful life events before the MI—which is a well-known risk factor for distress—one should take into account the possibility that the ASD symptoms reflected prior vulnerability. Finally, due to the exploratory nature of the current study, further research is required. Such research should be conducted among larger samples and among survivors of other life-threatening events in order to determine the generalizability of our findings to other events, in general, and other medical conditions, in particular.

Studies among MI patients and trauma survivors, in general, have documented the long-term implications of acute dissociation on survivors’ emotional and functional adjustment [7]. The findings of the current study show that the ramifications of dissociative symptoms in the immediate aftermath of an MI may be much deeper and longer-lasting than originally thought; these ramifications were manifested in an increased risk for all-cause mortality. Identifying individuals with high levels of dissociative symptoms while they are still hospitalized and under professional observation may therefore be useful not only in terms of helping them lessen their immediate distress, but also in terms of decreasing their risk for all-cause mortality.