In this study, we showed that long-term stress disorders (post-traumatic stress and adjustment disorders) or anxiety were associated with increased OHCA rate in the general population. The association persisted when the outcome was corrected for common OHCA risk factors. This association applied to men and women and was independent from the presence of cardiovascular disease.
Comparison with previous studies
Stress-related disorders and OHCA
Our finding that stress-related disorders or anxiety were associated with increased rate of OHCA are supported by previous studies.4–12 A case–control study showed that cardiac arrest patients had experienced a higher number of major life events in the year prior to the cardiac arrest than had matched controls, and that experiencing major life events were associated with a higher risk of cardiac arrest.6 These findings were supported by a case-crossover study, where higher occurrence of major life events in the month compared with the previous 5 months prior to cardiac arrest was reported.7 In contrast, a case–control study in 81 American women described no difference in experiencing stressful life events within 6 months prior to cardiac death compared with matched controls, although cases experienced more often the death of a significant other than matched controls.8 Previous studies, however, had limited sample size.6–8 Also, the presence of stressful life event was based on interview with spouses or family, thereby increasing the risk of information bias. Finally, Song et al investigated the risk of cardiovascular disease, including cardiac arrest, in a large cohort of 136 637 patients with stress-related disorders compared with matched controls, and found a significant higher risk of cardiac arrest in those with stress-related disorders.4 The study, however, relied solely on inpatient and outpatient diagnosis to define cardiac arrest, and omitted patients who died before reaching the hospital. Such inclusion criteria may result in important inclusion bias, particularly since psychiatric disorders have been associated with lower survival rates after OHCA.13 In our study, the capture of OHCA is nearly complete due to our collaboration with all EMS departments. This allowed us to enrol both patients who survived to hospital admission and those who died before reaching the hospital, thereby minimising inclusion bias by the use of an unselected cohort of OHCA patients. Further, misclassification of the outcome may have occurred in the study by Song et al4 since the cohort was not specifically designed to study OHCA and properly adjudicating OHCA victims is challenging.
Anxiety and OHCA
Our finding that anxiety was associated with higher rate of OHCA is in line with a previous study reporting a trend towards an increased risk of sudden cardiac death associated with high anxiety scores in a sample of women with no history of cardiovascular disease or cancer.9 Other studies of men with anxiety have described that high anxiety scores were associated with increased risk of sudden cardiac death.10 11 These previous studies, however, included small number of cardiac arrest patients,9–11 and no adjustments for important risk factors of OHCA, such as heart failure,10 11 diabetes mellitus11 and/or QT-prolonging drugs,9–11 were performed. Finally, studies in populations at risk for OHCA are equivocal. One study showed that anxiety was associated with an increased risk of ventricular arrhythmias 1 year after ICD implantation,12 while another study reported no association between anxiety and ventricular arrhythmias.17
Are long-term stress conditions and anxiety independent risk factors for OHCA?
Numerous studies demonstrated associations of stress-related disorders or anxiety with cardiovascular outcomes such as myocardial infarction, heart failure and cerebrovascular disease.4 5 These disease states might be biological mediators in the causal pathway of OHCA. Consequently, these disorders could contribute, to some extent, to increased OHCA rate associated with stress-related disorders and anxiety observed in our study. Nevertheless, both stress-related disorders and anxiety remained significantly associated with OHCA after controlling for these variables in the multivariable analyses, suggesting that it is unlikely that traditional risk factors of OHCA alone explain this relationship. Moreover, the elevated OHCA rate associated with stress-related disorders and anxiety in subjects without cardiovascular disease provides additional support for the notion that both stress-related disorders and anxiety may be directly related to OHCA.
The relation between stress and anxiety with sudden death has been long established10 11 18 and is likely mediated by the activity of the sympathetic autonomic nervous system.19 20 Indeed, a decrease in heart rate variability is a measure for an increase of the activity of the sympathetic nervous system and has been used as a biomarker for the increased risk of cardiac mortality.21 On the other hand, treatment with β-adrenoceptor blocking drugs is recommended to reduce cardiac mortality and prevent ventricular tachyarrhythmias.22 The exact mechanistic explanation of the proarrhythmic effects of increased sympathetic activity on the heart is still incompletely known. Increased sympathetic neural activity leads to an increase in heart rate, release of neurotransmitters into the circulation and local release of neurotransmitters in the heart.20 23 Each of these factors may potentially influence cardiac electrophysiology and facilitate ventricular arrhythmias and OHCA.20 23 Local release of catecholamines is related to changes in local action potential duration. The local effects have a sidedness, with stimulation of the left stellate ganglion leading to first a short increase and then a prolonged decrease of action potential duration in the left lateral and posterior wall, while right stellate stimulation shows these changes in the right ventricular wall and the left ventricular anterior wall of pigs.24 A shortening of action potential duration (and of the effective refractory period) is proarrhythmic in the setting of reentrant arrhythmias in two ways. First, it leads to shortening of the wavelength and thereby to facilitation of reentrant arrhythmias.25 Second, if the shortening of the action potential duration occurs adjacent to a region of myocardial ischaemia, it may cause an increase in the end-systolic ‘current of injury’ and thereby promote the genesis of closely coupled premature beats that may initiate reentry.26 Generalised release of noradrenalin from the adrenal glands has been studied by superfusing myocardial fibres or cells with fluids containing increased catecholamines. Superfusion human myocytes with norepinephrine leads to action potential prolongation and triggered activity based on early after depolarisations.27
In addition to a biological mechanism, behavioural and psychosocial factors may also contribute to OHCA risk. Stress-related disorders and anxiety often lead to unhealthy lifestyle, such as smoking and lower physical activity, which in turn may increase the risk of OHCA. In our study, we could not adjust for these factors since we had no data on these features to include in our statistical analyses. On the other hand, if such behaviours lie in the causal pathway between our exposure and OHCA, then treating them as confounders in the multivariable analyses, might underestimate the overall impact of stress-related disorders and anxiety on OHCA risk. Moreover, it is unlikely that knowledge of these factors would have altered our conclusions considering that we have adjusted for all the relevant cardiovascular comorbidities. Finally, other psychiatric disorders, such as the presence of depression, may contribute to OHCA risk. However, it is unlikely that depression accounts for our observed association in this study, since we adjusted for depression in the multivariable analyses.
Strengths and limitations
A main strength of this study is its population-based design in which large and unselected number of OHCA patients were obtained from nationwide databases, thereby minimising the risk of inclusion bias and selection bias and increasing the likelihood that our results are applicable to the community at large. Also, DANCAR was specifically designed to study OHCA, which allows accurate data collection for each OHCA patient. Nevertheless, the lack of information from primary care might have led to underestimation of the number of patients with milder forms of stress-related disorders or anxiety. Consequently, our results may not be directly applicable to subjects with less severe stress/anxiety or daily stress. Another limitation is that data on disease severity were not available in our analyses since information regarding diagnosis was defined as a binary variable. Third, we cannot exclude that residual confounders may have affected our results as we had no information on several risk factors such as smoking and alcohol consumption that may have contributed to the observed associations. Also, we had no data on left ventricular ejection fraction to include in our multivariable analyses. However, given the highly unpredictable way in which OHCA occurs, it is very difficult, if not impossible, to obtain such data shortly before the occurrence of OHCA in a uniform manner across the studied population. Fourth, higher prevalence of comorbidities among OHCA cases may result in higher diagnosis of stress-related conditions compared with non-OHCA controls in which having stress-related conditions is probably less diagnosed, as non-OHCA controls may seek less medical care. This, in turn, could result in overestimation of the OHCA risk associated with long-term stress conditions and anxiety. To deal with this, we performed an additional subgroup analyses by selecting only individuals without the comorbidities from table 1 (ie, cardiovascular comorbidities, depression) and without the use of cardiovascular drugs, QT-prolonging drugs and antidepressants. By doing this, we aimed to make cases and controls more comparable with respect to seeking medical care. Here, we confirmed our main findings that long-term stress conditions and anxiety (online supplemental table 6) were associated with higher risk of OHCA. Finally, the diagnosis codes used to identify stress-related disorders or anxiety have not been validated previously which may lead to misclassification bias. However, the majority of the information recorded in the Danish National Patient Registry has undergone scrutiny for data quality and shows high validity.28