Sudden cardiac death (SCD) constitutes a significant global public health burden, with some estimates of its mortality burden as high as 20% of all deaths in industrialized countries (1–3). SCD refers to any unexpected death within 1 h of the onset of cardiac arrest symptoms.

When the death is not witnessed, the timeline expands to 24 h (4). SCD can be the first presentation of cardiovascular disease, and almost half of all SCD-victims have no previously diagnosed heart condition (1, 5).

In the past 20 years, cardiovascular mortality has decreased significantly in high-income countries (6), especially in groups with higher risk for SCD such as patients with coronary artery disease (CAD) and heart failure (HF) (7). However, recent studies from the U.S. still report a staggering incidence of cardiac arrest with over 350,000 cases out-of-hospital (3) and 290,000 in hospital (8), annually.

Identifying and targeting modifiable risk factors for SCD can improve survival for at-risk patients by preventing the onset of SCD. Yet, risk prediction for SCD is complex. The propensity for sudden death is due to1a combination of intrinsic factors, such as genetic or acquired heart diseases, and transient factors that can trigger an SCD event (7). These factors can be unmodifiable, such as age and gender or modifiable, such as ischemic heart disease (IHD), smoking, low-level physical activity, atrial fibrillation (AF), and type 2 diabetes mellitus (T2DM). Although numerous meta-analyses on risk factors of SCD have been published, there is not yet a complete and succinct summary of the research, that can be applied clinically.

Here we perform an umbrella review to summarize the existing evidence concerning risk and protective factors associated with SCD among published meta-analyses. In accordance with best research practices, we rank the evidence of existing meta-analyses in this topic according to sample size, strength of the association, and existence of diverse biases (9, 10).

This umbrella meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (11) reporting guidelines and the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) guidelines (12) (Appendix 1). The study protocol was registered in the prospective registry of systematic reviews, PROSPERO (CRD42020216363).

In this study, we reviewed 55 articles concerning the risk and protective factors of SCD. Despite most of the associations being statistically significant, only a minority of them provided convincing evidence. Our meta-analyses of observational studies showed that the presence of ERP on ECG, current smoking, and T2DM were important risk factors, while physical activity was an important protective factor. In patients with HF, the use of CRT-D compared to CRT-P was the most important protective factor. Sensitivity analyses limited to prospective cohort studies did not alter marginally the main results. Our meta-analyses of RCTs showed that in patients with HF taking MRAs or SGLT-2 inhibitors, and the use of ICDs and CRT-Ds were important protective factors. Furthermore, the association of AF, HTN, and T2DM in patients with cardiovascular comorbidities with the risk of SCD was supported by highly suggestive evidence. Beta-blockers and ICDs were protective factors from SCD with moderate evidence, in certain subpopulations (Table 3).

ERP is defined as an elevation of the QRS-ST junction, J-point, and QRS notching in multiple ECG leads, and is high prevalent in middle-aged individuals (51). Although ERP in most cases can be considered benign, it is a marker of increased heterogeneity of ventricular repolarization, which might increase the risk of ventricular fibrillation (24). It is also possible that an ERP pattern can serve as a surrogate ECG marker of certain conditions known to predispose to repolarization heterogeneity, such as myocardial infarction (MI), hypokalemia, and HF. Accordingly, we found that patients with ERP are at increased risk for SCD. However, in one prospective analysis (24), ERP association with SCD was supported only by weak evidence, so future large prospective cohort studies would be of value to clarify the credibility of this association. Other electrocardiographic features, namely, the existence of premature ventricular contractions (PVCs) and microvolt T-wave alternans provided low credibility.

One of the modifiable risk factors of SCD identified in our analysis was smoking. Smoking can lead to increased blood pressure, resting heart rate, and risk of T2DM, AF, and MI, which are all risk factors of SCD (29, 52). The association between smoking and SCD can also be explained by biological mechanisms, as smoking increases the risk of ventricular arrhythmias possibly due to altered ventricular recovery time (53). Furthermore, nicotine has been shown to induce different cardiac arrhythmias in animal models, such as bradycardia, atrioventricular block, and ventricular tachyarrhythmia (54). However, other comparisons between ever, former, or never smokers and a dose-response association between smoking and SCD showed weak evidence, which may be due to the small number of patients (n < 1,000) in the included primary studies. Smoking as a risk factor is modifiable, and the risk of cardiovascular disease is reduced by 39% as soon as 5 years after cessation (55). Therefore, interventions targeting this risk factor may be able to have a significant impact on SCD incidence.

T2DM increased the risk for SCD by two-fold in our analysis. Several mechanisms have been postulated to explain the association between T2DM and SCD, such as myocardial disease due to atherosclerosis, inflammation-mediated associated with uremia and HTN, potassium imbalances, and arrhythmogenic effects secondary to autonomic neuropathy (56). Interventions to reduce the prevalence of T2DM, such as diet and physical activity modifications, may therefore reduce the risk of SCD indirectly. In fact, physical activity was found in our analysis to be a significant protective factor for SCD. Physical activity is important for controlling metabolic risk factors including obesity, HTN, T2DM (57), CAD (58), and HF (59), all of which are risk factors for SCD.

In patients with HF, the implantation of ICD is the most important protective factor against SCD. We found that more than two-thirds as many patients with a CRT indication are protected from SCD when they receive ICD compared to only CRT-P, with convincing evidence level. This observation is supported by other meta-analyses of RCTs, which found that ICD reduces the risk of SCD by more than half compared to standard medical treatment, in patients with reduced ejection fraction (EF). These results are consistent both in IHD patients and non-IHD patients (34, 38), with a high GLE and without significant heterogeneity.

More than ten medications have been studied for the risk of SCD, in different patient populations. Androgen deprivation therapy, macrolides, antipsychotics, and Parkinson's drugs, were evaluated in observational studies as risk factors for SCD. All were significant risk factors for SCD but, none showed high epidemiological credibility. However, in the performed sensitivity analysis, when the criterium of more than 1,000 cases per association was omitted, the association of antipsychotics with SCD was upgraded to highly convincing for risperidone and convincing for the antipsychotics' haloperidol, quetiapine, and thioridazine, a finding in line with the literature (60). The risk of SCD is high in psychiatric patients, owing to a large extent to psychotropic drugs. Different mechanisms have been introduced to explain this association (such as the increased torsadogenic effect of a psychotropic drug and the synergic effect of different proarrhythmic drugs) in the coexistence/or not of pre-existing congenital cardiopathies (such as long-QT and Brugada syndrome) (60). Predicting the safety of potential proarrhythmic medicines is a top priority (61). Thus, measures such as the use of pharmacogenetics (i.e., how genes affect the way a person responds to medications) might have relevant clinical implications, particularly for idiosyncratic adverse drug reactions, such as in the case of the use of antipsychotics and other drugs and the risk of SCD (62).

More than six medicines, including amiodarone, beta-blockers, statins, ACE-inhibitors, MRA, SGLT-2, omega-3 fatty acids, and other antihypertensive drugs were tested in meta-analyses including only RCTs. All were found to be statistically protective against SCD, but only SGLT-2 and MRA associations were supported by a high GLE. It is also important to note that these medicines are used for the treatment of heart failure (63) and arrhythmias (7) in a population already at high risk and the generalizability of these findings can be limited.

Concerning imaging-related risk factors for SCD, the presence of LGE in MRI examination was associated with SCD but was supported by weak evidence (64). The lack of strong evidence can be attributed to the small number of patients included in the original studies. When this criterium was omitted from our grading, the level of evidence was raised to convincing in patients with non-IHD. Thus, larger prospective cohort studies can be of value. Similarly, the association of the reduced LVEF with SCD was statistically significant but only supported by weak evidence in the primary analysis and highly supportive only when the criterium of n > 1,000 patients was omitted. This finding is surprising as low LVEF is the criterion used most commonly during the last decades to find patients eligible for ICD therapy for primary prevention of SCD. The small number of primary studies and the issue of low reproducibility of the measurement of LVEF in clinical settings can possibly explain this finding (7).

In pediatric patients with HCM traditional risk factors, such as extreme LV hypertrophy and non-sustain VT, didn't show significant associations, while others as previous history of syncope or adverse cardiac event (aborted SCD or sustain ventricular tachycardia), were significant, albeit with weak evidence. This finding could be explained by the observational study design, the small sample size, and the critically low quality of the included studies, a fact which was also annotated in the latest published guidelines (65). Hence, larger cohort studies are of great importance for optimizing risk stratification for HCM in children.

Other interesting factors associated with the risk of SCD but not included in previous meta-analyses that fulfill the inclusion criteria of our umbrella review involve risk factors such as gender in young, episodes of supraventricular tachycardias, and COVID-19 infection. Data from observational studies show that the incidence of SCD in young men is lower compared to young women, indicating that SCD due to potentially inherited cardiac diseases is less often in young women (66). Even if this factor is not modifiable, it can lead to further research about young women's protection mechanisms against SCD. SVTs have been reported to be the etiology of sudden cardiac arrest in 5% of all patients with aborted sudden death, including 7 of 13 patients without preexcitation on their baseline ECG (67). There is a subgroup of patients with SVTs with a rapid ventricular rate in which cardiac arrest may be a manifestation even in pediatric patients. Thus, electrophysiology testing must be considered, especially in the pediatric population (68). There is evidence of an increased incidence of ventricular arrhythmias and SCD in COVID-19 patients (69), while a recent meta-analysis found a higher prevalence of SCD during the COVID-19 pandemic compared to the pre-pandemic period (70). Several mechanisms have been proposed to explain the possible association of SARS-CoV-2 infection with increased SCD risk and arrhythmogenesis, including direct myocardial injury, oxygen demand-supply mismatch due to hypoxia, hypercoagulability, and adverse effects of medications for COVID-19 (71). However, reliable data assessing SARS-CoV-2 infection as a potential risk factor for SCD is still missing.

In the current guidelines (7), there is an emphasis on establishing screening and prevention programs for SCD. However, no clear recommendations for population screening have been provided due to a paucity of evidence (7). To the best of our knowledge, this is the first umbrella review providing evidence concerning the associations of modifiable risk and protective factors with the risk of SCD. Our results indicate that people who smoke tobacco and have a sedentary lifestyle, who are diagnosed with DM, AF, or HTN, as well as those who have ERP on their ECG, are at increased risk for SCD. Therefore, these factors should be considered in the design of future studies on SCD prevention. Another implication from the present study is that it identifies several protective factors such as MRA or SGLT-2 inhibitors, and the use of ICD in patients with HF. The use of them should be emphasized whenever possible in patients at high risk of SCD.

There were several modifiable risk factors with only a weak level of evidence (e.g., pre-diabetes, BMI, PVCs, etc.), in the general population and subpopulations tested. This is likely due to the limited number of available cohorts and the small number of participants available for the subpopulation analysis. Larger cohorts may be helpful for further elucidating the role of these modifiable risk factors, by providing more evidence about these associations.

Our umbrella review provides a broad picture of the non-genetic factors that have been studied for SCD. However, this study has also several limitations. First, in meta-analyses that included observational data, the associations which were supported by high epidemiological credibility can be considered strong evidence, but they cannot imply causality. On the contrary, meta-analyses which include RCTs provide data mostly in patients already at high risk, and therefore is less generalizable to the general population. Thus, our study yields risk factors with proven significant associations to SCD but does not allow conclusions as to their clinical value in primary prevention. Second, grading of meta-analyses which include observational data can provide only warnings concerning the presence of systematic biases and not proof about the nature of these biases (72, 73). Thus, only a description of the results and sources of bias has been made. Third, although a large number of risk and protective factors for SCD were included in this analysis, there may be other important factors not included, as they have not been evaluated in previously published meta-analyses, like the New York Heart Association score. In addition, potential associations of genetic factors with SCD were not assessed, as genetic causality is tested with other analytic approaches -i.e., Mendelian randomization studies- rather than pairwise meta-analysis, which was defined as the unit of analysis in the present review.

In this umbrella review, we mapped the epidemiological evidence on non-genetic factors associated with SCD as identified in previously published meta-analyses. Even though SCD is a prevalent medical issue, we were only able to identify a small number of risk factors associated with SCD and even fewer with high epidemiological credibility. The association between SCD and the following risk factors were supported by convincing and highly supported evidence: lifestyle risk factors, like the lack of physical activity and smoking; comorbidities, like AF and DM; the use of medications, like MRA or SGLT-2 inhibitors; ECG features, like ERP; and the use of ICD. Further investigation with targeted interventions in these populations is the first step toward a better strategy for SCD prevention.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

EC, DT, and ED designed the study. DK and EC performed a comprehensive screening of the literature, selected the studies included in the meta-analysis, and abstracted the data items. DT and ED performed the statistical analysis. DT and EC drafted the manuscript. EC, DT, DK, LOK, AS, DV, EF, AA, CT, CA, and FB interpreted the results and edited the manuscript critically. All the co-authors have read and accepted this version of the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.