Abstract
Background
Life‐threatening ventricular arrhythmias (LTVAs) can occur in individuals without apparent structural heart disease, yet the cardiometabolic and inflammatory determinants remain unclear. We aimed to identify clinical, sociodemographic, and lifestyle determinants of LTVAs in structurally normal hearts.
Methods
We included 423 186 UKB (UK Biobank) participants after excluding coronary artery disease, heart failure, cardiomyopathy, and valvular and congenital heart disease at baseline and during follow‐up. The primary end point was incident LTVAs, defined as ventricular tachyarrhythmias, cardiac arrest, sudden cardiac death, and related resuscitation or ablation procedures. Cox models estimated adjusted hazard ratios (HRs) with 95% CIs. Model discrimination was assessed using Harrell's concordance index and explained variance (R 2).
Results
Over a median 13.3 (interquartile range, 12.6–14.0) years, 1116 LTVAs occurred. Cases were older (61 versus 57 years) and more often men (56.1% versus 42.6%). Hypertension (HR, 2.21 [95% CI, 1.90–2.57]), current smoking (HR, 1.78 [95% CI, 1.48–2.13), and male sex (HR, 1.67 [95% CI, 1.45–1.92]) showed the strongest associations with LTVAs. Additional factors associated with increased risk included nonideal sleep, previous smoking, physical inactivity, older age, and higher high‐sensitivity C‐reactive protein. Low‐density lipoprotein cholesterol showed a modest inverse association (HR, 0.89 [95% CI, 0.82–0.97]). Age and hypertension showed the highest discrimination (concordance index, 0.629 and 0.615; R 2 0.0082 and 0.0101), and overall model discrimination was modest. Findings were consistent across sexes and in sensitivity analyses.
Conclusions
In adults without structural heart disease, hypertension, smoking, and male sex show the strongest associations with LTVAs, with additional contributions from age, inflammation, and other lifestyle factors.
Keywords: inflammation, lifestyle, life‐threatening ventricular arrhythmias, sociodemographic, structural heart disease
Subject Categories: Cardiovascular Disease, Sudden Cardiac Death, Arrhythmias

Nonstandard Abbreviations and Acronyms
- ARIC
Atherosclerosis Risk in Communities
- LTVAs
life‐threatening ventricular arrhythmias
- SCD
sudden cardiac death
- TDI
Townsend Deprivation Index
- UKB
UK Biobank
Clinical Perspective.
What Is New?
This large prospective study of 423 186 adults without structural heart disease identified hypertension, current smoking, and male sex as the strongest factors associated with incident life‐threatening ventricular arrhythmias, with additional contributions from older age, systemic inflammation, and adverse lifestyle factors.
Most traditional cardiometabolic risk factors, including body mass index, diabetes, and most lipid parameters, showed limited associations with life‐threatening ventricular arrhythmias after multivariable adjustment, suggesting that arrhythmic risk in structurally normal hearts differs from pathways leading to structural cardiovascular disease.
What Are the Clinical Implications?
Common clinical characteristics and inflammatory biomarkers may help identify individuals at increased risk of life‐threatening ventricular arrhythmias before the development of overt structural heart disease and support earlier preventive risk assessment, while future risk evaluation should incorporate electrophysiological, imaging, and genetic markers to improve discrimination.
Life‐threatening ventricular arrhythmias (LTVAs), including sustained ventricular tachycardia, ventricular fibrillation, and cardiac arrest, are the most critical cardiac rhythm disturbances and a leading cause of sudden cardiac death (SCD) worldwide. 1 , 2 SCD constitutes 15% 20% of all deaths in industrialized nations, highlighting its significant public health impact. 3 In the United States, the annual incidence of SCD is estimated at 230 000 to 350 000 cases, with reported ranges extending from <170 000 to >450 000 depending on data sources and definitions. 1 Europe also faces a high burden, with the European Sudden Cardiac Arrest Network consortium reporting ≈249 500 SCDs per year in the European Union, at incidence rates of 36.8 to 39.7 per 100 000 person‐years, predominantly affecting men. 4 The risk of SCD significantly increases with age, from about 1 per 100 000 person‐years in children to >200 per 100 000 in older adults. 2 Traditional culprits such as coronary artery disease (CAD) account for ≈75% to 80% of SCD cases in Western countries, yet its dominance has waned, with an increasing burden from hypertensive cardiomyopathy and nonischemic conditions. 3 , 4 , 5 Notably, many SCD cases occur without prior symptoms or a diagnosed cardiovascular condition, emphasizing the need to identify determinants of malignant ventricular arrhythmias in individuals without structural heart disease, a population in whom arrhythmic risk remains poorly characterized despite the absence of overt cardiac pathology. 1 , 2 , 3 , 5
The risk of LTVAs is influenced by a combination of modifiable and nonmodifiable cardiometabolic and inflammatory factors. While hypertension, diabetes, smoking, and advanced age are well‐established determinants of cardiovascular disease and SCD, their relevance to malignant ventricular arrhythmias in individuals without structural heart disease remains unclear. 1 , 2 , 6 Most prior studies have focused on populations with established structural substrates, such as CAD or heart failure, where arrhythmogenesis is driven by scar formation and ventricular remodeling. 6 As a result, it is uncertain whether these risk factors retain predictive value in structurally normal hearts. Mechanistically, hypertension, smoking, inflammation, and metabolic dysfunction may promote arrhythmogenesis through ion channel remodeling, impaired calcium handling, autonomic imbalance, and myocardial fibrosis. 6 Despite the substantial proportion of SCD occurring without overt structural disease, this population remains underrepresented in prior research. Studying this group presents a crucial opportunity to identify risk factors that contribute to LTVA risk and to develop early prevention strategies in the general population.
Therefore, leveraging the UKB (UK Biobank), a large prospective cohort with comprehensive phenotyping and long‐term follow‐up, we aimed to evaluate the cardiometabolic and inflammatory determinants of LTVAs in adults without structural heart disease. Specifically, we sought to assess whether established cardiovascular risk factors remain associated with malignant ventricular arrhythmias in this low‐risk population and to quantify their relative contributions to risk. By focusing on structurally normal hearts, this study provides insight into early arrhythmic risk evaluation before the development of overt cardiac disease.
METHODS
Data Statement
All data are publicly available in the UKB repository.
Study Population
The UKB is a large‐scale prospective cohort that recruited >500 000 participants aged 37 to 73 years between 2006 and 2010 from 22 centers across the United Kingdom. Study design and protocols have been described previously. 7 The UKB study was approved by the UK National Health Service National Research Ethics Service (Reference: 11/NW/0382), and all participants gave written informed consent in accordance with the Declaration of Helsinki. The present analysis was conducted under UKB application number 217390. Of 502 128 participants, we excluded 78 942 with a history of LTVAs or structural heart disease at baseline or during follow‐up, yielding a final analytic cohort of 423 186. Structural heart disease was defined as CAD, heart failure, cardiomyopathy, valvular heart disease, or congenital heart disease. Structural heart disease was defined on the basis of clinically recorded diagnoses in linked hospital data (International Classification of Diseases, Tenth Revision [ICD‐10] codes), and systematic imaging (eg, echocardiography or cardiac magnetic resonance imaging) was not available for cohort‐wide exclusion.
Assessment of Outcome
The primary end point was incident LTVAs, ascertained through linked hospital admission and mortality records. LTVAs were defined using ICD‐10 and Office of Population Censuses and Surveys Classification of Interventions and Procedures, Version 4 codes for ventricular tachyarrhythmias (I47.0, I47.2, I49.0), cardiac arrest and SCD (I46.0, I46.1, I46.9), and relevant resuscitation or ablation procedures (K57.6, K64.1, X50.3, X50.4, X50.8, X50.9). To evaluate the robustness of findings and address potential heterogeneity, the outcomes were categorized into 3 definitions as secondary outcomes reflecting increasing specificity: specific LTVAs (I47.0, I47.2, I49.0), LTVAs without SCD (I47.0, I47.2, I49.0, K57.6, K64.1), and SCD only (I46.0, I46.1, I46.9, X50.3, X50.4, X50.8, X50.9). Follow‐up extended from baseline assessment until the first occurrence of LTVAs or the censoring date, whichever came first. Data were censored on May 31, 2022.
Statistical Analysis
Baseline characteristics were summarized as medians with interquartile ranges for continuous variables and percentages for categorical variables. Group comparisons used Mann–Whitney U tests for continuous variables and χ2 tests for categorical variables. Associations of demographic, clinical, and lifestyle variables with incident LTVAs were examined using Cox proportional hazards models, with results expressed as hazard ratios (HRs) and 95% CIs. Candidate covariates included age, sex, body mass index (BMI), systolic blood pressure (SBP), ethnicity, Townsend Deprivation Index (TDI), smoking status (never, previous, current), alcohol intake (noncurrent, ideal, nonideal), physical activity, sleep (6–8 hours a day treated as ideal, otherwise nonideal), diet, high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol (LDL‐C), total cholesterol, triglycerides, hs‐CRP (high‐sensitivity C‐reactive protein), glycated hemoglobin (HbA1c), estimated glomerular filtration rate, hypertension, and diabetes. To account for potential confounding by treatment, antihypertensive medications, glucose‐lowering therapies, and lipid‐lowering medications were included as covariates in the multivariable models. Multicollinearity among predictors was assessed using variance inflation factors. All variables showed acceptable collinearity (variance inflation factor <5), except for total cholesterol and LDL‐C, for which LDL‐C was retained in the final models due to its greater clinical relevance. Predictor evaluation was performed using complementary approaches. Multivariable Cox regression models were used to identify variables associated with LTVAs after adjustment. In addition, each predictor was evaluated separately in a single‐predictor Cox proportional hazards model, and Harrell's concordance index and explained variance (R 2) were calculated to assess its stand‐alone discriminatory ability and contribution to explained variation. The reported concordance index values represent apparent (in‐sample) discrimination and were not corrected for optimism using bootstrap resampling or cross‐validation. Sequential multivariable models were further compared with evaluate the incremental improvement in model discrimination associated with the addition of cardiometabolic, sociodemographic, and lifestyle variables. Accordingly, predictor ranking reflects stand‐alone predictive performance rather than adjusted effect size. Cumulative incidence curves were derived and stratified by key predictors. A 2‐sided P<0.05 was considered statistically significant. Analyses were performed in R version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria).
Handling of Missing Data
Missing values were handled using multiple imputation with the “missRanger” package 8 in R, which uses random forests with predictive mean matching. A total of 16 variables with missingness up to 14.45% were imputed. These included demographic markers (ethnicity, TDI), clinical measures (estimated glomerular filtration rate, HbA1c, SBP, hs‐CRP), lipid biomarkers (total cholesterol, LDL‐C, high‐density lipoprotein cholesterol, triglycerides), and lifestyle factors (BMI, smoking status, alcohol intake, physical activity, sleep, diet). Outliers in estimated glomerular filtration rate (values >200 mL/min per 1.73 m2; n=84) and HbA1c (values >100 mmol/mol; n=372) were treated as missing before imputation. All available variables, including age and sex, were used as predictors in the imputation model. Imputation was performed before applying exclusion criteria for structural heart disease to preserve the underlying data structure. Five imputed data sets were generated using 50 trees, predictive mean matching (pmm.k) =3, max.depth=8, sample.fraction=0.8, and min.node.size=10. Following imputation, no missing values remained across variables. Imputation quality was evaluated using standard diagnostics (Tables S1–S5). Distributions of continuous variables (mean±SD, median) were highly consistent between observed and imputed values. Similarly, proportions of categorical variables were comparable across observed and imputed data, indicating no evidence of systematic bias and good performance of the imputation model.
All inferential analyses were conducted across the 5 imputed data sets, and results were pooled using Rubin's rules to appropriately account for both within‐ and between‐imputation variability. This included Cox proportional hazards models, effect estimates (HRs with 95% CIs), and measures of model performance, including Harrell's concordance index and explained variance (R 2). Two exceptions were made for presentation purposes. Baseline characteristics (Table 1) were summarized using a single imputed data set to provide a clear and interpretable description of the study population. Kaplan–Meier curves were also generated from a single imputed data set for visualization.
Table 1.
Baseline characteristics of participants by LTVAs incidence status
| Overall (n = 423,186) | No LTVAs (n = 422,070) | LTVAs (n = 1,116) | P value* | |
|---|---|---|---|---|
| Age, years | 57.0 (49.0–62.0) | 57.0 (49.0–62.0) | 61.0 (55.0–65.0) | < 0.001 |
| Women, n (%) | 242554 (57.3) | 242064 (57.4) | 490 (43.9) | < 0.001 |
| BMI, kg/m2 | 26.5 (24.0–29.6) | 26.5 (24.0–29.6) | 27.2 (24.3–30.6) | < 0.001 |
| sBP, mmHg | 135.5 (124.0–148.5) | 135.5 (124.0–148.5) | 138.0 (127.0–151.5) | < 0.001 |
| Ethnicity | 0.422 | |||
| White, n (%) | 399940 (94.5) | 398879 (94.5%) | 1061 (95.1%) | |
| Other ethnic backgrounds‡, n (%) | 23246 (5.5) | 23191 (5.5%) | 55 (4.9%) | |
| Townsend deprivation index | −2.2 (−3.7–0.4) | −2.2 (−3.7–0.4) | −1.9 (−3.5–0.9) | < 0.001 |
| Smoking status | < 0.001 | |||
| Current, n (%) | 42702 (10.1) | 42522 (10.1%) | 180 (16.1%) | |
| Never, n (%) | 239440 (56.6) | 238945 (56.6%) | 495 (44.4%) | |
| Previous, n (%) | 141044 (33.3) | 140603 (33.3%) | 441 (39.5%) | |
| Alcohol intake | 0.019 | |||
| Noncurrent, n (%) | 31766 (7.5) | 31658 (7.5%) | 108 (9.7%) | |
| Ideal alcohol intake, n (%) | 263318 (62.2) | 262659 (62.2%) | 659 (59.1%) | |
| Nonideal alcohol intake, n (%) | 128102 (30.3) | 127753 (30.3%) | 349 (31.3%) | |
| Regular Physical activity | < 0.001 | |||
| Yes, n (%) | 311988 (73.7) | 311222 (73.7%) | 766 (68.6%) | |
| No, n (%) | 111198 (26.3) | 110848 (26.3%) | 350 (31.4%) | |
| Sleep | < 0.001 | |||
| Ideal, n (%) | 368360 (87.0) | 367448 (87.1%) | 912 (81.7%) | |
| Nonideal, n (%) | 54826 (13.0) | 54622 (12.9%) | 204 (18.3%) | |
| Diet | 0.099 | |||
| Healthy, n (%) | 65816 (15.6) | 65636 (15.6%) | 180 (16.1%) | |
| Intermediate healthy, n (%) | 302016 (71.4) | 301248 (71.4%) | 768 (68.8%) | |
| Poor, n (%) | 55354 (13.1) | 55186 (13.1%) | 168 (15.1%) | |
| Laboratory data | ||||
| LDL‐C, mmol/L | 3.6 (3.0–4.1) | 3.6 (3.0–4.1) | 3.5 (2.9–4.0) | < 0.001 |
| TC, mmol/L | 5.7 (5.0–6.5) | 5.7 (5.0–6.5) | 5.6 (4.8–6.4) | < 0.001 |
| TG, mmol/L | 1.4 (1.0–2.1) | 1.4 (1.0–2.1) | 1.6 (1.1–2.3) | < 0.001 |
| HDL‐C, mmol/L | 1.4 (1.2–1.7) | 1.4 (1.2–1.7) | 1.4 (1.1–1.6) | < 0.001 |
| Hs‐CRP, mg/L | 1.3 (0.6–2.6) | 1.3 (0.6–2.6) | 1.8 (0.9–3.7) | < 0.001 |
| HbA1c, mmol/mol | 35.0 (32.6–37.5) | 35.0 (32.6–37.5) | 35.9 (33.5–39.0) | < 0.001 |
| eGFR, mL/min/1.73 m² | 86.2 (76.6–96.9) | 86.3 (76.6–96.9) | 84.5 (74.7–97.5) | 0.019 |
| Comorbidities | ||||
| Hypertension†, n (%) | 105591 (25.0) | 105054 (24.9%) | 537 (48.1%) | < 0.001 |
| Diabetes mellitus †, n (%) | 28041 (6.6) | 27886 (6.6%) | 155 (13.9%) | < 0.001 |
Baseline characteristics of the cohort are represented as median (interquartile range) or n (%).
BMI, body mass index; eGFR, estimated glomerular filtration rate (MDRD formula);. HbA1c, glycated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; LDL‐C, low‐density lipoprotein cholesterol; LTVAs, life‐threatening ventricular arrhythmias; sBP, systolic blood pressure; TC, total cholesterol; TDI, Townsend Deprivation Index; TG, triglycerides.
Comparisons between participants with and without LTVAs.
Comorbidities include diagnoses present at baseline or recorded during follow‐up.
Other ethnic backgrounds include Mixed, Asian or Asian British, Black or Black British, Chinese, and Other ethnic groups.
For sensitivity analyses, we excluded participants with missing baseline covariates, resulting in a complete‐case cohort of 367 180 individuals. After further excluding those with prior LTVAs or structural heart disease, 310 417 participants remained for analysis. The results from this complete‐case cohort were compared with those obtained from the multiply imputed data sets to evaluate the robustness of our findings.
RESULTS
Baseline Characteristics
The baseline characteristics of participants are presented in Table 1. Among 423 186 participants (median age, 57 years [interquartile range, 49–62]; 57.3% women) followed for a median of 13.3 years (interquartile range, 12.6–14.0), 1116 developed LTVAs. Compared with those without events, affected individuals were older (median, 61 versus 57 years; P<0.001), more often men (56.1% versus 42.6%; P<0.001), and had greater socioeconomic deprivation (TDI, –1.9 versus −2.2; P<0.001). They also had higher BMI (27.2 versus 26.5 kg/m2), higher prevalence of current (16.2% versus 10.1%) and previous (39.5% versus 33.3%) smoking, less regular physical activity (68.6% versus 73.7%), and more nonideal sleep (18.3% versus 12.9%) (all P<0.001). Furthermore, the LTVA group had less favorable cardiometabolic profiles, including elevated triglycerides (1.6 versus 1.4 mmol/L), hs‐CRP (1.8 versus 1.3 mg/L), and HbA1c (35.9 versus 35.0 mmol/mol) (all P<0.001). Finally, hypertension (48.1% versus 24.9%) and diabetes (13.9% versus 6.6%) were significantly more common in the LTVAs group (both P<0.001).
Predictors of LTVAs
A wide range of demographic, clinical, and lifestyle characteristics were significantly associated with the risk of LTVAs (Figure 1). As anticipated, age and sex were associated with risk of LTVAs: each SD increase in age was associated with a higher risk (HR, 1.06 [95% CI, 1.05–1.07]; P<0.001), while men had a 67% higher risk compared with women (HR, 1.67 [95% CI, 1.45–1.92]; P<0.001). Among traditional cardiometabolic risk factors, hypertension was the strongest predictor, conferring a >2‐fold increase in risk (HR, 2.21 [95% CI, 1.90–2.57]; P<0.001). In contrast, SBP showed a modest inverse association after adjustment (HR, 0.993 [95% CI, 0.989–0.996]; P<0.001), while use of antihypertensive medications was not significant. Diabetes and HbA1c were not associated with LTVAs. With respect to lipid‐related variables, LDL‐C was associated with a lower risk of LTVAs (HR, 0.89 [95% CI, 0.82–0.97]; P=0.01), whereas high‐density lipoprotein cholesterol and triglycerides were not significant. Use of lipid‐lowering medications was also associated with reduced risk (HR, 0.81 [95% CI, 0.68–0.98]; P=0.029). Systemic inflammation remained an important determinant, with higher hs‐CRP associated with increased risk (HR, 1.03 [95% CI, 1.02–1.04]; P<0.001). Socioeconomic deprivation was also associated with elevated risk (TDI: HR, 1.02 [95% CI, 1.00–1.04]; P=0.019), while ethnicity did not significantly influence outcomes. Lifestyle factors played an important role, with current smoking strongly associated with risk (HR, 1.78 [95% CI, 1.48–2.13]; P<0.001) and former smoking showing a more modest but significant association (HR, 1.21 [95% CI, 1.05–1.38]; P=0.006). Nonideal sleep duration (HR, 1.22 [95% CI, 1.04–1.43]; P=0.017) and lack of regular physical activity (1.19, 1.04–1.36; P=0.011) were also associated with increased risk after multivariable adjustment. BMI and diet quality were not associated with LTVAs. Ideal alcohol intake, compared with being a noncurrent drinker, was associated with a slightly lower risk (0.81, 0.66–0.995; P=0.045).
Figure 1. Multivariable Cox regression analysis of predictors of LTVAs in the overall population.

BMI indicates body mass index; eGFR, estimated glomerular filtration rate (Modification of Diet in Renal Disease formula); HbA1c, glycated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; LDL‐C, low‐density lipoprotein cholesterol; LTVA, life‐threatening ventricular arrhythmia; and TDI, Townsend Deprivation Index.
Formal interaction analyses were performed to assess potential effect modification by sex, age, and systemic inflammation. Overall, no statistically significant interactions were observed between key predictors and sex or age (all P for interaction>0.05), indicating that associations were generally consistent across subgroups (Table S3). There was borderline evidence of interaction for hs‐CRP with sex (P=0.064) and age (P=0.087), suggesting possible modest heterogeneity in the association between inflammation and LTVA risk; however, these did not reach statistical significance. In addition, no significant interactions were observed between hs‐CRP and cardiometabolic factors, including hypertension, smoking, sleep, physical inactivity, or LDL‐C (all P for interaction>0.05).
Overall, hypertension, current smoking, and male sex showed the strongest associations with LTVAs in multivariable models, while older age, nonideal sleep, previous smoking, lower physical activity, and higher hs‐CRP were also associated with risk, albeit with more modest effect sizes.
When stratified by sex (Figure 2), the associations of risk factors were generally consistent, although some differences in magnitude were observed. In men, hypertension (HR, 2.23 [95% CI, 1.83–2.71]), and current smoking (HR, 1.76 [95% CI, 1.39–2.23]) were the primary predictors, with age (HR, 1.06 [95% CI, 1.05–1.07]), hs‐CRP (HR, 1.02 [95% CI, 1.01–1.03]) and previous smoking (HR, 1.35 [95% CI, 1.13–1.61]) also showing significant associations. In women, hypertension (HR, 2.19 [95% CI, 1.74–2.75]) and current smoking (HR, 1.81 [95% CI, 1.37–2.40]) remained the strongest predictors, along with age (HR, 1.05 [95% CI, 1.03–1.06]) and hs‐CRP (HR, 1.03 [95% CI, 1.02–1.05]). Although LDL‐C reached statistical significance only in women, the LDL‐C×sex interaction was not significant (P for interaction=0.32), suggesting that the apparent difference between sexes should be interpreted cautiously. Other cardiometabolic and lifestyle factors showed generally similar directions of association across sexes, although effect sizes were modest and some associations did not reach statistical significance in stratified analyses, likely reflecting reduced statistical power.
Figure 2. Multivariable Cox regression analysis of predictors of LTVAs stratified by sex.

A, Women; (B) men. BMI indicates body mass index; eGFR, estimated glomerular filtration rate (Modification of Diet in Renal Disease formula); HbA1c, glycated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; LDL‐C, low‐density lipoprotein cholesterol; LTVA, life‐threatening ventricular arrhythmia; and TDI, Townsend Deprivation Index.
Predictive Performance of Predictors
To determine the relative predictive contributions of individual risk factors, we ranked them by concordance index and explained variance (R 2) (Figure 3). Age and hypertension were the most discriminative variables—concordance indices, 0.629 (95% CI, 0.629–0.629) and 0.615 (95% CI, 0.615–0.615), respectively—followed by hs‐CRP (HR, 0.582 [95% CI, 0.575–0.588]), HbA1c (HR, 0.574 [95% CI, 0.569–0.579]), smoking status (HR, 0.568 [95% CI, 0.566–0.570]), and sex (HR, 0.568 [95% CI, 0.568–0.568]). Other cardiometabolic and lifestyle factors showed limited individual discrimination, clustering closer to a concordance index of 0.50. In terms of explained risk, hypertension (R 2=0.0101) and age (0.0082) accounted for the largest proportions, with smaller contributions from sex, smoking, HbA1c, and hs‐CRP.
Figure 3. Discrimination ability and relative importance of predictors for predicting LTVAs.

A, Discrimination ability (Harrell's concordance index). B, Relative importance (explained variance, R 2). Each predictor was evaluated separately in a single‐predictor Cox proportional hazards model. Higher values indicate greater stand‐alone discrimination or explained variation, respectively. These estimates are unadjusted for the other candidate predictors. BMI indicates body mass index; eGFR, estimated glomerular filtration rate (Modification of Diet in Renal Disease formula); HbA1c, glycated hemoglobin; HDL‐C, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; LDL‐C, low‐density lipoprotein cholesterol; LTVA, life‐threatening ventricular arrhythmia; SBP, systolic blood pressure; TDI, Townsend Deprivation Index; and TG, triglycerides.
Model‐based discrimination analyses further confirmed the incremental value of adding different domains of risk factors (Table 2). In the overall population, a base model including only age and sex achieved a concordance index of 0.647 (95% CI, 0.647–0.647). Adding traditional cardiometabolic risk factors (blood pressure, HbA1c, renal function, lipids, hypertension, medications, and diabetes) improved the model performance to 0.692 (95% CI, 0.691–0.694). Incorporation of sociodemographic characteristics (ethnicity and TDI) yielded a modest additional gain (HR, 0.696 [95% CI, 0.694–0.697]). The greatest incremental improvement came from including lifestyle behaviors (BMI, smoking, alcohol intake, sleep, diet, and physical activity), resulting in a final concordance index of 0.704 (95% CI, 0.702–0.705). Results were consistent by sex. In men, model performance increased from 0.634 (95% CI, 0.634–0.634) with age and sex alone to 0.698 (95% CI, 0.697–0.699) in the fully adjusted model. In women, the corresponding concordance index rose from 0.622 (95% CI, 0.622–0.622) to 0.693 (95% CI, 0.690–0.696).
Table 2.
Discrimination of Models for Incident LTVAs
| Models | Concordance index (95% CI) | ||
|---|---|---|---|
| Overall population | Men | Women | |
| Model 1 | 0.647 (0.647– 0.647) | 0.634 (0.634– 0.634) | 0.622 (0.622–0.622) |
| Model 2 | 0.692 (0.691–0.694) | 0.682 (0.681–0.684) | 0.682 (0.680–0.684) |
| Model 3 | 0.696 (0.694–0.697) | 0.687 (0.685–0.688) | 0.686 (0.684–0.688) |
| Model 4 | 0.704 (0.702–0.705) | 0.698 (0.697–0.699) | 0.693 (0.690–0.696) |
Model 1 (basic): age, sex. Model 2 (model 1 plus traditional cardiometabolic factors): model 1 plus systolic blood pressure, hypertension, glycated hemoglobin, diabetes, estimated glomerular filtration rate, low‐density lipoprotein cholesterol, high‐density lipoprotein cholesterol, triglycerides, antihypertensive medications, glucose‐lowering therapies, and lipid‐lowering medications, and high‐sensitivity C‐reactive protein. Model 3 (model 2 plus sociodemographic factors): model 2 plus ethnicity and Townsend Deprivation Index. Model 4 (model 3 plus lifestyle factors): model 3 plus body mass index, smoking status, alcohol intake, sleep, diet quality, and regular physical activity. LTVA indicates life‐threatening ventricular arrhythmia.
Cumulative Incidence of Important Predictors
Cumulative incidence curves (Figure 4) demonstrated clear gradients in LTVA risk across major predictors. The overall incidence rate was 0.198 per 1000 person‐years. For smoking status, separation of the curves was evident early in follow‐up and widened progressively over time. Current smokers experienced the steepest rise in LTVA incidence (incidence rate, 0.315 per 1000 person‐years), with previous smokers showing intermediate risk (0.235 per 1000 person‐years), and never‐smokers consistently maintaining the lowest event rates (0.155 per 1000 person‐years, log‐rank P<0.001). Hypertension was also associated with marked differences in cumulative incidence. Participants with hypertension had a substantially higher incidence of LTVAs compared with those without hypertension (0.382 versus 0.137 per 1000 person‐years), and this difference became more pronounced with longer follow‐up (log‐rank P<0.001). Participants in the oldest age tertile had the highest incidence rate (0.308 per 1000 person‐years), compared with 0.174 and 0.106 per 1000 person‐years in the middle and youngest tertiles, with curves diverging progressively during follow‐up (log‐rank P<0.001). Systemic inflammation, assessed by hs‐CRP, further stratified risk. Individuals in the highest hs‐CRP tertile had the greatest incidence rate (0.264 per 1000 person‐years), compared with 0.197 and 0.133 per 1000 person‐years in the middle and lowest tertiles, respectively (log‐rank P<0.001).
Figure 4. Cumulative incidence of LTVAs according to major predictors.

A, Hypertension status; (B) smoking status; (C) age group; (D) hs‐CRP. hs‐CRP indicates high‐sensitivity C‐reactive protein; and LTVA, life‐threatening ventricular arrhythmia.
Sensitivity Analyses in the Complete‐Case Cohort
In the complete‐case cohort of 310 417 participants, 817 developed LTVAs during follow‐up (Table S4). Baseline differences between those with and without LTVAs were consistent with the main analysis: Affected individuals were older (61.0 versus 57.0 years, P<0.001), more frequently men (57.3% versus 43.4%, P<0.001), and more likely to have hypertension (47.7% versus 24.9%) or diabetes (13.1% versus 6.4%, both P<0.001). Higher prevalence of current and previous smoking, nonideal sleep, lower rates of regular physical activity, and elevated BMI, SBP, triglycerides, and HbA1c were also observed in the LTVA group (all P<0.05).
Multivariable Cox regression analysis in the complete‐case cohort (Figures S1–S5) confirmed the principal predictors of LTVAs. Hypertension remained the strongest determinant (HR, 2.12 [95% CI, 1.78–2.54]), followed by current smoking (HR, 1.82 [95% CI, 1.47–2.27]), and male sex (HR, 1.73 [95% CI, 1.47–2.04]; all P<0.001). Previous smoking (HR, 1.34 [95% CI, 1.15–1.57]) and lack of regular physical activity (HR, 1.21 [95% CI, 1.04–1.41]) were also associated with higher risk, while older age (HR, 1.05 [95% CI, 1.04–1.06]) and hs‐CRP (HR, 1.03 [95% CI, 1.02–1.04]) remained significant biomarkers. Other factors, including lipid measures, BMI, diet, and alcohol intake, were not associated.
Model discrimination was largely unchanged compared with the main cohort (Table S5). A model with age and sex alone yielded a concordance index of 0.644 (95% CI, 0.625–0.664), which improved to 0.690 (95% CI, 0.671–0.709) with addition of traditional risk factors. Incorporation of sociodemographic variables modestly increased the concordance index to 0.694 (95% CI, 0.675–0.712), and inclusion of lifestyle factors yielded a final concordance index of 0.703 (95% CI, 0.684–0.721). Ranking of predictors by discrimination and explained variance (Figure S2) mirrored the primary analysis. Age and hypertension remained the most discriminative predictors (concordance index, 0.624 [95% CI, 0.605–0.644] and 0.613 [95% CI, 0.596–0.631], respectively), followed by hs‐CRP, HbA1c, smoking status, and sex. In terms of explained variance, hypertension (R 2=0.0098) and age (R 2=0.0078) accounted for the largest proportion, with smaller contributions from smoking, sex, hs‐CRP, and HbA1c.
Overall, the complete‐case analyses were highly consistent with the primary analyses based on multiple imputation. The magnitude and direction of associations for key predictors, including hypertension, current smoking, male sex, older age, and hs‐CRP, were largely unchanged. Model discrimination was also similar between approaches (concordance index, 0.703 in the complete‐case cohort versus ≈0.704 in the imputed cohort), supporting the robustness of the findings to different missing data assumptions.
Sensitivity Analyses for Secondary Outcomes
In sensitivity analyses using alternative end point definitions (Figures S3–S5), a clear gradient was observed across levels of specificity. A total of 279 events were identified for specific LTVAs, 312 events for LTVAs without SCD, and 837 events for SCD only. Hypertension and male sex remained consistently associated with increased risk across all end point definitions. In contrast, smoking showed a borderline association in the most specific LTVAs definition but became statistically significant in the broader end points (LTVAs without SCD and SCD only). The number of identified predictors increased with broader end point definitions. For the SCD‐only end point, additional associations were observed for diabetes and lifestyle factors, whereas the more specific LTVA definitions identified a more limited set of predictors. Overall, these findings demonstrate consistent associations for key predictors across end points, with variation in the strength and number of associations according to end point specificity.
DISCUSSION
LTVAs are typically associated with structural heart disease, such as CAD, heart failure, or cardiomyopathies, which provide substrates for arrhythmogenesis. To isolate predictors in individuals without these conditions, we excluded participants with structural heart disease at baseline and during follow‐up. This design enabled evaluation of arrhythmic risk determinants in the absence of structural substrates, addressing a key gap in current risk evaluation frameworks that are largely derived from diseased populations. Finally, in this large prospective cohort of >420 000 adults without structural heart disease, hypertension, current smoking, and male sex showed the strongest associations with LTVAs. Older age, nonideal sleep, previous smoking, lower physical activity, and elevated hs‐CRP were also associated with increased risk, although effect sizes were modest. Most lipid parameters, glycemic markers, and BMI were not associated after multivariable adjustment; however, LDL‐C demonstrated a modest inverse association that should be interpreted cautiously, as it may reflect residual confounding or treatment effects rather than a direct biological mechanism. Associations were generally consistent across sexes, with some variation in magnitude. Formal interaction analyses did not identify statistically significant effect modification by sex, age, or inflammatory status, suggesting that the associations of major cardiometabolic and lifestyle factors with LTVAs are broadly consistent across subgroups. However, the overall modest discrimination of the models reflects the multifactorial nature of LTVAs and the low baseline risk in individuals without structural heart disease. These findings should therefore be interpreted as providing pathogenetic and risk‐stratification insights rather than as a stand‐alone predictive tool.
Importantly, our findings demonstrate that traditional cardiometabolic risk factors retain strong associations with malignant ventricular arrhythmias even before the development of overt structural heart disease, suggesting that arrhythmic vulnerability may arise earlier in the disease continuum than previously appreciated. In structurally normal hearts, arrhythmogenesis is likely driven by an interplay of subclinical electrical remodeling, autonomic imbalance, and inflammation‐mediated electrophysiological alterations.
Nonmodifiable Demographic Factors
Our findings align with previous studies recognizing age as a primary determinant of SCD and ventricular arrhythmias, with risk increasing significantly in middle and older adulthood. 1 , 2 Large community‐based studies have consistently shown age as a strong predictor of SCD, even in populations without overt cardiovascular disease. 9 , 10 Mechanistically, aging may promote electrophysiological instability through ion channel remodeling, increased myocardial fibrosis, and impaired autonomic regulation, creating a substrate vulnerable to malignant ventricular arrhythmias. 6 Our results confirm that older individuals face a substantially increased risk of LTVAs, even without structural heart disease, highlighting that age alone can stratify vulnerability. Sex differences in arrhythmic risk are also well established, with men exhibiting a substantially higher risk of SCD across most age groups. 1 , 2 In large community‐based cohorts, male sex consistently emerged as an important predictor and contributed significantly to validated SCD risk models. 9 Interestingly, men not only experience higher SCD incidence but also more frequently present with shockable rhythms like ventricular tachycardia or fibrillation, contributing to their greater event burden. 3 Conversely, women more often present with pulseless electrical activity, a nonshockable rhythm associated with poorer resuscitation outcomes. 3 While men consistently outnumber women in SCD incidence, women have poorer survival rates following cardiac arrest despite their lower overall risk. 5 These sex differences in arrhythmic outcomes likely stem from a complex interplay of electrophysiological substrates, hormonal influences, and arrest phenotypes. 6 Our findings confirm that men face a markedly increased risk of LTVAs, even without structural heart disease, indicating that sex alone can stratify vulnerability.
Major Clinical Predictor: Hypertension
The role of hypertension in arrhythmic risk has been inconsistently described. While some studies identify it as a strong risk factor for malignant ventricular arrhythmias, others suggest its influence is primarily mediated through structural heart disease. 1 , 2 The ARIC (Atherosclerosis Risk in Communities) study found elevated blood pressure to be the most potent determinant of SCD among metabolic syndrome components, with an HR of 1.81 (95% CI, 1.45–2.27), surpassing impaired fasting glucose or dyslipidemia. 11 Recent decades have also seen a rise in hypertension prevalence among patients with sudden cardiac arrest, indicating its increasing relevance. 5 Hypertension has been incorporated into risk scores to improve SCD prediction and is highlighted as a major cardiovascular risk factor associated with increased SCD susceptibility. 10 In our cohort of individuals without structural heart disease, hypertension emerged as the strongest predictor, which supports a direct arrhythmogenic role of hypertension beyond its contribution to structural remodeling, potentially mediated through electrophysiological and autonomic alterations that precede overt cardiac disease. Interestingly, while diagnosed hypertension was a strong predictor of LTVAs, baseline SBP showed a weak, nonsignificant inverse association (Figures 1 and 2). This discrepancy may be due to the difference between chronic disease status and a single measurement, as well as the effect of antihypertensive treatment. Individuals with diagnosed hypertension likely have longer‐term vascular and myocardial changes, contributing to higher risk regardless of current blood pressure readings. These findings suggest that arrhythmic vulnerability is better captured by hypertension as a chronic condition rather than by isolated blood pressure measurements.
Lifestyle Predictors
Multiple studies consistently identify smoking as a strong predictor of SCD. Current smoking nearly doubles SCD risk and is a key component in validated risk models. 9 The impact of tobacco exposure is graded and persistent, with former smokers also showing an increased risk compared with never‐smokers. 12 , 13 Smoking cessation significantly reduces this risk, highlighting both immediate and long‐term benefits. 14 Mechanistically, smoking contributes to arrhythmic outcomes through heightened sympathetic tone, oxidative stress, endothelial dysfunction, and promoting a proarrhythmic myocardial substrate. In our cohort of individuals without structural heart disease, the gradient in cumulative incidence across smoking categories mirrored earlier studies, with clear separation of risk emerging early in follow‐up and widening progressively over time. When we differentiated between current and previous smoking, a marked gradient in risk was observed. Current smoking was associated with nearly a 2‐fold higher hazard of LTVAs (HR, 1.78 [95% CI, 1.48–2.13]), underscoring the substantial ongoing impact of active tobacco exposure. Previous smoking also conferred a significant but more modest excess risk (HR, 1.21 [95% CI, 1.05–1.38]), suggesting that adverse cardiovascular effects persist even after cessation. The elevated risk among former smokers may reflect irreversible myocardial or vascular changes accrued during years of smoking, while the much higher risk in current smokers indicates additive effects of continued exposure. Together, these findings highlight both the immediate benefits of smoking cessation and the importance of preventing initiation in the first place, given that some degree of residual risk may remain despite quitting.
Sleep duration is an emerging factor in cardiovascular health. Our analysis defined ideal sleep as 6 to 8 hours per night. Participants with nonideal sleep durations (shorter or longer) showed a significantly higher risk of LTVAs (HR, 1.22) compared with those with ideal sleep. This aligns with prior research linking both insufficient and excessive sleep to adverse cardiovascular outcomes. For instance, in the Kuopio Ischemic Heart Disease cohort, men sleeping >10 hours per night had increased risks of all‐cause death, coronary death, and sudden cardiac death. 15 Extreme sleep patterns may indicate underlying physiological dysregulation, potentially contributing to an arrhythmogenic substrate even in individuals without structural heart disease. This highlights sleep health as a crucial lifestyle factor in arrhythmic risk.
Physical activity also emerged as a significant lifestyle factor influencing arrhythmic risk. Individuals reporting regular physical activity had a significantly lower incidence of LTVAs compared with inactive participants, with lack of regular activity associated with higher risk (HR, 1.19). This aligns with previous UKB research showing that higher levels of moderate to vigorous activity reduce the risk of atrial and ventricular arrhythmias. 16 Additionally, evidence from Tulppo et al demonstrated that in patients with CAD, physical inactivity was associated with a >2‐fold increased risk of SCD. 17 Regular activity may protect against arrhythmias by favorably modulating autonomic tone, reducing inflammation, improving endothelial function, and preventing adverse myocardial remodeling. Our results highlight physical inactivity as a modifiable contributor to malignant ventricular arrhythmias and support guideline‐recommended activity levels as a cornerstone of primary prevention, even among individuals without structural heart disease.
In this analysis of adults without structural heart disease, alcohol intake was not a predictor of LTVAs, consistent with previous UKB studies. 18 , 19 These studies found no consistent association between alcohol consumption and SCD or fatal ventricular arrhythmias, with only a weak U‐shaped pattern for SCD in 1 study. 18 , 19 Together, these findings suggest that total alcohol consumption is largely neutral on arrhythmic outcomes in the general population. Our results extend this by showing alcohol intake has minimal impact on arrhythmic risk in individuals with structurally normal hearts. The observed lower risk in the “ideal intake” group is likely due to confounding, such as the “sick quitter” phenomenon, rather than a protective effect of alcohol. Compared to stronger predictors like hypertension, smoking, physical inactivity, and inflammation, alcohol appears to contribute minimally to risk evaluation for LTVAs in structurally normal hearts.
BMI, a measure of adiposity influenced by lifestyle, is a known risk factor for CAD, heart failure, and atrial fibrillation, but its direct link to malignant ventricular arrhythmias is less clear. In our study, BMI was not associated with LTVAs after adjustment, likely because the risk from obesity is often mediated through structural conditions like ischemic cardiomyopathy or heart failure, which were excluded. Prior reviews suggest that obesity‐related increases in ventricular arrhythmias and SCD are largely explained by comorbidities rather than BMI itself, although mechanisms like corrected QT prolongation and autonomic imbalance may play a role. 20 Junttila et al further observed that while obesity‐related cardiomyopathy was a significant cause of SCD in the past, its proportion has decreased, with hypertensive and obesity/diabetes‐related heart disease becoming more prominent. 5 These findings suggest that obesity primarily increases arrhythmic risk indirectly through structural and metabolic pathways, and BMI alone may not be sufficient to drive malignant arrhythmogenesis in individuals with structurally normal hearts.
Together, these observations are consistent with the integrated framework of autonomic imbalance and inflammation‐mediated electrophysiological alterations described above, even in the absence of structural disease.
Other Biomarkers
Our findings emphasize the role of systemic inflammation, indicated by hs‐CRP, in arrhythmic risk. Inflammation is increasingly linked to adverse electrophysiological remodeling. In the Physicians' Health Study, the highest quartile of baseline CRP was associated with a nearly 3‐fold increased risk of SCD, after multivariate adjustment for traditional risk factors and lipid levels. 21 This suggests that inflammation actively promotes arrhythmogenic vulnerability. In our cohort without structural heart disease, higher hs‐CRP remained associated with increased LTVAs, highlighting its relevance beyond ischemic substrates. As hs‐CRP is inexpensive and easily measured, it could aid in identifying individuals at elevated risk who might otherwise go undetected. This reinforces inflammation as an active driver of arrhythmic risk and suggests potential for novel preventive strategies targeting inflammatory pathways.
Glycemic status is linked to cardiovascular disease, with diabetes being a major risk factor for CAD, heart failure, and SCD. However, in our analysis, HbA1c and diabetes itself were not predictors of LTVAs. This suggests that subclinical glycemic dysregulation, rather than overt diabetes, may contribute to arrhythmogenic vulnerability in the absence of structural heart disease. Elevated HbA1c may indicate cumulative metabolic stress leading to systemic inflammation, autonomic imbalance, and microvascular injury, creating a substrate for arrhythmias. These findings align with studies showing diabetes significantly increases SCD risk, with progressively higher odds ratios for borderline diabetes, diabetes without microvascular disease, and diabetes with microvascular complications. 20 , 22 Electrophysiological disturbances in diabetes, such as prolonged corrected QT and repolarization instability, also contribute to arrhythmogenic vulnerability. 20 Taken together, these findings suggest that glycemic markers may reflect broader metabolic and inflammatory processes relevant to arrhythmic risk, but their role in risk evaluation appears limited in this population without structural heart disease.
Lipid parameters like LDL‐C, high‐density lipoprotein cholesterol, total cholesterol, and triglycerides are key determinants of atherosclerotic cardiovascular disease and contribute significantly to CAD risk and its arrhythmic complications. In our analysis restricted to individuals without structural heart disease, most lipid measures were not associated with LTVAs after multivariable adjustment, supporting the concept that lipid‐related arrhythmic risk is largely mediated through structural pathways, including coronary atherosclerosis, ischemia, and ventricular remodeling.
Although the inverse association appeared more evident in women than in men, formal interaction testing did not demonstrate significant effect modification by sex (P for interaction=0.32). Therefore, this apparent difference may reflect chance variation or reduced statistical power in the stratified analyses rather than a true biological sex difference. This finding should be interpreted with caution, as it may reflect residual confounding, the influence of lipid‐lowering therapy, or selection effects, rather than a direct or causal protective effect of LDL‐C. Consistent with this, 1 study found that lipid alterations in survivors of sudden cardiac arrest disappeared when compared with CAD controls, indicating lipids influence arrhythmic risk mainly through the CAD pathway. 23 Overall, these findings indicate that while lipid management remains essential for cardiovascular disease prevention, lipid levels alone appear to have limited utility for arrhythmic risk evaluation in individuals without overt structural heart disease.
The lack of consistent associations for traditional cardiometabolic factors such as lipid levels, BMI, and diabetes in the fully adjusted models may suggest differences between pathways leading to structural cardiovascular disease and those underlying arrhythmogenesis in structurally normal hearts. These factors may exert their arrhythmic effects primarily through the development of structural substrates, which were excluded in the present study. In contrast, inflammation, reflected by hs‐CRP, and factors influencing autonomic and electrophysiological stability may play a more direct role in ventricular arrhythmogenesis in this population. These findings support a relatively more direct role of inflammatory and electrophysiological mechanisms in arrhythmogenesis compared with traditional cardiometabolic pathways in structurally normal hearts.
Socioeconomic Factors
Sociodemographic factors influenced arrhythmic risk in our study. Greater socioeconomic deprivation, indicated by higher TDI scores, was associated with increased LTVAs. This aligns with research linking deprivation to adverse health behaviors and comorbidities that can heighten arrhythmic susceptibility. 24 , 25 While racial and ethnic differences in SCD risk are documented, with Black adults showing higher SCD incidence, 24 our UK‐based cohort found no significant association between race and ethnicity and LTVAs after adjustment. This may be due to the UKB's limited racial and ethnic diversity and the exclusion of structural heart diseases, which often mediate racial and ethnic disparities. These findings emphasize the importance of socioeconomic factors in arrhythmia prevention, suggesting that in structurally normal hearts, biological racial and ethnic differences may play a lesser role than social and environmental determinants.
Sensitivity Analyses for Alternative End Points
In sensitivity analyses using alternative end point definitions, a clear gradient was observed across levels of specificity. Hypertension and male sex remained robust predictors across all end points, including specific LTVAs, LTVAs without SCD, and SCD only. In contrast, smoking demonstrated a borderline association in the most specific LTVA definition but became significant in broader end points. This pattern likely reflects both limited statistical power in the most specific analysis and the broader systemic effects of smoking, which may contribute to both arrhythmogenesis and pathways leading to cardiac arrest and sudden death. Notably, the SCD‐only end point yielded stronger and more numerous associations, including for diabetes and lifestyle factors, suggesting the contribution of nonarrhythmic mechanisms. In contrast, more specific LTVA definitions identified a narrower set of predictors consistent with direct electrophysiological susceptibility. These findings highlight the importance of end point definition in arrhythmia research and support the robustness and biological plausibility of our primary results.
Clinical and Research Implications
These findings have potential implications for early arrhythmic risk evaluation. Hypertension, smoking, male sex, older age, and systemic inflammation were identified as the strongest predictors of LTVAs, even in individuals without structural heart disease. This emphasizes the importance of managing traditional and modifiable risk factors. Although model discrimination was modest, this is expected in a population without structural heart disease and reflects the multifactorial and low‐incidence nature of LTVAs. Rather than serving as stand‐alone screening tools, these findings support the use of readily available cardiometabolic and inflammatory markers for risk enrichment. Blood pressure control and smoking cessation are crucial for preventing cardiovascular disease and may also reduce the risk of malignant arrhythmias. Lifestyle factors like regular physical activity and adequate sleep are also important determinants, highlighting the need for behavioral counseling in arrhythmia prevention. While lipids and BMI showed limited associations, hs‐CRP and HbA1c may offer incremental value in risk assessment, suggesting the integration of inflammatory and metabolic markers may contribute to future risk evaluation frameworks. Importantly, individuals without structural heart disease but with hypertension or active smoking may represent a previously underrecognized but clinically relevant subgroup for early prevention strategies.
Mechanistically, cardiometabolic risk factors may induce subtle changes in ion channel function and calcium handling, promoting repolarization heterogeneity, while lifestyle‐related exposures contribute to sympathetic predominance and reduced vagal tone. Concurrently, systemic inflammation, reflected by elevated hs‐CRP, may modulate myocardial conduction, gap junction function, and microstructural integrity, collectively increasing susceptibility to ventricular arrhythmias even in the absence of overt structural disease. From a research standpoint, our findings point to several areas for future investigation. The sustained risk in former smokers and those with mild glycemic abnormalities suggests long‐term or irreversible mechanisms that need further study. Inflammation, indicated by hs‐CRP, warrants attention, and future research should explore whether anti‐inflammatory interventions can reduce arrhythmic outcomes. The association between socioeconomic deprivation and LTVA risk highlights the need to incorporate social determinants of health into arrhythmia research and prevention. The modest predictive power of current models indicates significant unexplained residual risk, emphasizing the need for novel biomarkers, imaging, and genetic approaches to improve risk prediction in structurally normal populations. Future efforts should focus on integrating clinical factors with electrophysiological, imaging, and genetic data to improve discrimination and enable clinically actionable prediction models. Prospective intervention studies are crucial to determine if modifying these risk factors can effectively reduce LTVA incidence.
Strengths and Limitations
This study's strengths lie in its large sample size, prospective design, and >13 years of follow‐up, providing significant power to assess associations between demographic, clinical, and lifestyle factors and incident LTVAs. Robust outcome ascertainment was ensured through linked hospital admission and mortality records. The comprehensive assessment of cardiometabolic risk factors, sociodemographics, lifestyle behaviors, and biomarkers allowed for detailed analysis of their contributions to arrhythmic risk. Standardized baseline assessments and national health record linkage enhanced data reliability. Extensive adjustment for confounding and the use of multiple imputation and sensitivity analyses bolstered the robustness of the findings.
Several limitations should be acknowledged. First, the UKB cohort is not fully representative of the general population. Participants tend to be healthier, less socioeconomically deprived, and predominantly of European ancestry, which may introduce selection bias and limit the generalizability of our findings to broader and more diverse populations. Second, only baseline risk factor measurements were used, which may not reflect longitudinal changes over time. Although multiple imputation was applied to address missing data, some degree of bias may persist; however, complete‐case analyses yielded consistent results. Third, structural heart disease was defined on the basis of clinically recorded diagnoses in linked hospital data, and systematic imaging was not available for all participants. As a result, subclinical or undiagnosed structural abnormalities may have been present at baseline. Such misclassification is likely nondifferential with respect to the studied exposures and would tend to bias associations toward the null, potentially underestimating true effect sizes. Additionally, excluding participants who developed clinically diagnosed structural heart disease during follow‐up helps mitigate but does not eliminate the influence of latent disease at baseline. Fourth, although the overall number of events was sufficient for primary analyses, the relatively low event rate in this population may limit statistical power for subgroup and interaction analyses, and such findings should be interpreted with caution. In addition, the use of a composite end point encompassing multiple arrhythmic events and procedures may introduce heterogeneity, as these outcomes differ in underlying mechanisms and clinical context. While sensitivity analyses using more specific end point definitions demonstrated consistent associations for key predictors, variation in the strength and number of associations across end points highlights the importance of cautious interpretation. Fifth, nonarrhythmic death represents a competing risk, particularly in older populations with long follow‐up. As standard Cox models treat such events as censoring, the estimated associations may not fully reflect the cumulative incidence of LTVAs in the presence of competing risks. Future studies using formal competing risk methods are warranted to better quantify absolute risk. Sixth, model discrimination was evaluated using apparent concordance index estimates without optimism correction. Although the large sample size is expected to minimize optimism, some overestimation of model performance remains possible. Finally, although variables were identified as strongly associated with LTVAs after multivariable adjustment, these findings should not be interpreted as indicating causal independence, as residual confounding and correlations between predictors may persist. Further mechanistic and interventional research is needed to determine whether modifying these risk factors can reduce LTVA risk.
Conclusions
In this large prospective study of >420 000 adults rigorously screened for the absence of structural heart disease, we identified hypertension, smoking, male sex, older age, nonideal sleep, physical inactivity, and elevated hs‐CRP as the strongest cardiometabolic and inflammatory predictors of LTVAs. In contrast, most lipid parameters, BMI, diet, and alcohol intake were not associated with risk, although LDL‐C showed a modest inverse association that should be interpreted with caution. By focusing on structurally normal hearts, this study isolated demographic, clinical, and lifestyle determinants of arrhythmic risk in the absence of CAD or cardiomyopathy, offering a novel perspective. The finding that common, modifiable cardiometabolic factors significantly drive arrhythmic vulnerability in this context highlights opportunities for risk evaluation beyond traditional disease substrates. These findings suggest that malignant ventricular arrhythmias may originate from early cardiometabolic and inflammatory disturbances before the development of structural heart disease, supporting a paradigm shift toward earlier prevention.
Sources of Funding
The study was supported by the Career Awards for Medical Scientists Innovation Fund for Medical Sciences (2023‐I2M‐C&T‐B‐067), National High Level Hospital Clinical Research Funding (2025‐GSP‐GG‐22 and 2025‐GSP‐GG‐39). The funding source had no role in study design; in the collection, analysis, and interpretation of data; in writing the report; and in the decision to submit the article for publication.
Disclosures
None.
Supporting information
Tables S1–S5
Figures S1–S5
Acknowledgments
Author contributions: Study conception and design: H.‐D.Z., L.D., M.T.; analysis and interpretation of data: H.‐D.Z., L.D., M.T.; drafting of the article: H.‐D.Z., L.D., M.T.; critical revision of the article for intellectual content: H.‐D.Z., L.D., Y.‐D.L., K.Z., Y.‐J.Q., Z.‐H.J., F.‐H.P., X.‐X.Y., Y.‐J.S., M.T.; final approval of the article: H.‐D.Z., L.D., Y.‐D.L., K.Z., Y.‐J.Q., Z.‐H.J., F.‐H.P., X.‐X.Y., Y.‐J.S., M.T.; statistical expertise: H.‐D.Z., M.T.; obtaining of funding: H.‐D.Z.; administrative, technical, or logistic support: H.‐D.Z., M.T.; collection of data: H.‐D.Z., L.D.
This work was presented orally at the Heart Rhythm conference, Chicago, IL, April 23–26, 2026.
This manuscript was sent to Sula Mazimba, MD, MPH, Associate Editor, for review by expert referees, editorial decision, and final disposition.
Supplemental Material is available at https://www.ahajournals.org/doi/suppl/10.1161/JAHA.126.050176
For Sources of Funding and Disclosures, see page 14.
References
- 1. Al‐Khatib SM, Stevenson WG, Ackerman MJ, Bryant WJ, Callans DJ, Curtis AB, Deal BJ, Dickfeld T, Field ME, Fonarow GC, et al. AHA/ACC/HRS guideline for Management of Patients with Ventricular Arrhythmias and the prevention of sudden cardiac death: a report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines and the Heart Rhythm Society. Circulation. 2017;138:e210–e271. doi: 10.1161/CIR.0000000000000548 [DOI] [PubMed] [Google Scholar]
- 2. Zeppenfeld K, Tfelt‐Hansen J, de Riva M, Winkel BG, Behr ER, Blom NA, Charron P, Corrado D, Dagres N, de Chillou C, et al. 2022 ESC guidelines for the management of patients with ventricular arrhythmias and the prevention of sudden cardiac death. Eur Heart J. 2022;43:3997–4126. doi: 10.1093/eurheartj/ehac262 [DOI] [PubMed] [Google Scholar]
- 3. Hayashi M, Shimizu W, Albert CM. The spectrum of epidemiology underlying sudden cardiac death. Circul Res. 2015;116:1887–1906. doi: 10.1161/CIRCRESAHA.116.304521 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Empana JP, Lerner I, Valentin E, Folke F, Böttiger B, Gislason G, Jonsson M, Ringh M, Beganton F, Bougouin W, et al. Incidence of sudden cardiac death in the European Union. J Am Coll Cardiol. 2022;79:1818–1827. doi: 10.1016/j.jacc.2022.02.041 [DOI] [PubMed] [Google Scholar]
- 5. Junttila MJ, Hookana E, Kaikkonen KS, Kortelainen ML, Myerburg RJ, Huikuri HV. Temporal trends in the clinical and pathological characteristics of victims of sudden cardiac death in the absence of previously identified heart disease. Circ Arrhythm Electrophysiol. 2016;9:9. doi: 10.1161/CIRCEP.115.003723 [DOI] [PubMed] [Google Scholar]
- 6. Delisle BP, George AL Jr, Nerbonne JM, Bass JT, Ripplinger CM, Jain MK, Hermanstyne TO, Young ME, Kannankeril PJ, Duffy JF, et al. Understanding mechanisms of sudden cardiac death: a report from the National Heart, Lung, and Blood Institute workshop, part 2: population and clinical considerations. Circ Arrhythm Electrophysiol. 2021;14:e010190. doi: 10.1161/CIRCEP.121.010190 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, Downey P, Elliott P, Green J, Landray M, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Schwarz DF, Konig IR, Ziegler A. On safari to random jungle: a fast implementation of random forests for high‐dimensional data. Bioinformatics. 2010;26:1752–1758. doi: 10.1093/bioinformatics/btq257 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Deo R, Norby FL, Katz R, Sotoodehnia N, Adabag S, DeFilippi CR, Kestenbaum B, Chen LY, Heckbert SR, Folsom AR, et al. Development and validation of a sudden cardiac death prediction model for the general population. Circulation. 2016;134:806–816. doi: 10.1161/circulationaha.116.023042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Waks JW, Sitlani CM, Soliman EZ, Kabir M, Ghafoori E, Biggs ML, Henrikson CA, Sotoodehnia N, Biering‐Sørensen T, Agarwal SK, et al. Global electric heterogeneity risk score for prediction of sudden cardiac death in the general population: the atherosclerosis risk in communities (ARIC) and cardiovascular health (CHS) studies. Circulation. 2016;133:2222–2234. doi: 10.1161/CIRCULATIONAHA.116.021306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hess PL, Al‐Khalidi HR, Friedman DJ, Mulder H, Kucharska‐Newton A, Rosamond WR, Lopes RD, Gersh BJ, Mark DB, Curtis LH, et al. The metabolic syndrome and risk of sudden cardiac death: the atherosclerosis risk in communities study. J Am Heart Assoc. 2017;6:e006103. doi: 10.1161/JAHA.117.006103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Wennberg P, Eliasson M, Hallmans G, Johansson L, Boman K, Jansson JH. The risk of myocardial infarction and sudden cardiac death amongst snuff users with or without a previous history of smoking. J Internal Med. 2007;262:360–367. doi: 10.1111/j.1365-2796.2007.01813.x [DOI] [PubMed] [Google Scholar]
- 13. Aune D, Schlesinger S, Norat T, Riboli E. Tobacco smoking and the risk of sudden cardiac death: a systematic review and meta‐analysis of prospective studies. Eur J Epidemiol. 2018;33:509–521. doi: 10.1007/s10654-017-0351-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Goldenberg I, Jonas M, Tenenbaum A, Boyko V, Matetzky S, Shotan A, Behar S, Reicher‐Reiss H. Current smoking, smoking cessation, and the risk of sudden cardiac death in patients with coronary artery disease. Arch Internal Med. 2003;163:2301–2305. doi: 10.1001/archinte.163.19.2301 [DOI] [PubMed] [Google Scholar]
- 15. Khan H, Kella D, Kunutsor SK, Savonen K, Laukkanen JA. Sleep duration and risk of fatal coronary heart disease, sudden cardiac death, cancer death, and all‐cause mortality. Am J Med. 2018;131:1499–1505 e1492. doi: 10.1016/j.amjmed.2018.07.010 [DOI] [PubMed] [Google Scholar]
- 16. Elliott AD, Linz D, Mishima R, Kadhim K, Gallagher C, Middeldorp ME, Verdicchio CV, Hendriks JML, Lau DH, La Gerche A, et al. Association between physical activity and risk of incident arrhythmias in 402 406 individuals: evidence from the UK biobank cohort. Eur Heart J. 2020;41:1479–1486. doi: 10.1093/eurheartj/ehz897 [DOI] [PubMed] [Google Scholar]
- 17. Tulppo MP, Kiviniemi AM, Lahtinen M, Ukkola O, Toukola T, Perkiomaki J, Junttila MJ, Huikuri HV. Physical activity and the risk for sudden cardiac death in patients with coronary artery disease. Circ Arrhythm Electrophysiol. 2020;13:e007908. doi: 10.1161/CIRCEP.119.007908 [DOI] [PubMed] [Google Scholar]
- 18. Park CS, Choi J, Choi J, Lee KY, Ahn HJ, Kwon S, Lee SR, Choi EK, Kwak SH, Oh S. Alcohol is neither a risk factor nor a protective factor for sudden cardiac death and/or fatal ventricular arrhythmia: a population‐based study with genetic traits and alcohol consumption in the UK biobank. Heart Rhythm. 2024;21:1820–1826. doi: 10.1016/j.hrthm.2024.04.097 [DOI] [PubMed] [Google Scholar]
- 19. Tu SJ, Gallagher C, Elliott AD, Linz D, Pitman BM, Hendriks JML, Lau DH, Sanders P, Wong CX. Alcohol consumption and risk of ventricular arrhythmias and sudden cardiac death: an observational study of 408,712 individuals. Heart Rhythm. 2022;19:177–184. doi: 10.1016/j.hrthm.2021.09.040 [DOI] [PubMed] [Google Scholar]
- 20. Remme CA. Sudden cardiac death in diabetes and obesity: mechanisms and therapeutic strategies. Ca J Cardiol. 2022;38:418–426. doi: 10.1016/j.cjca.2022.01.001 [DOI] [PubMed] [Google Scholar]
- 21. Albert CM, Ma J, Rifai N, Stampfer MJ, Ridker PM. Prospective study of C‐reactive protein, homocysteine, and plasma lipid levels as predictors of sudden cardiac death. Circulation. 2002;105:2595–2599. doi: 10.1161/01.CIR.0000017493.03108.1C [DOI] [PubMed] [Google Scholar]
- 22. Jouven X, Lemaître RN, Rea TD, Sotoodehnia N, Empana JP, Siscovick DS. Diabetes, glucose level, and risk of sudden cardiac death. Eur Heart J. 2005;26:2142–2147. doi: 10.1093/eurheartj/ehi376 [DOI] [PubMed] [Google Scholar]
- 23. Norby FL, Nakamura K, Fu Q, Venkatraman V, Sundararaman N, Mastali M, Reinier K, Salvucci A, Jui J, Van Eyk JE, et al. A panel of blood biomarkers unique to sudden cardiac arrest. Heart Rhythm. 2022;20:414–422. doi: 10.1016/j.hrthm.2022.12.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Deo R, Safford MM, Khodneva YA, Jannat‐Khah DP, Brown TM, Judd SE, McClellan WM, Rhodes JD, Shlipak MG, Soliman EZ, et al. Differences in risk of sudden cardiac death between blacks and whites. J Am Coll Cardiol. 2018;72:2431–2439. doi: 10.1016/j.jacc.2018.08.2173 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Zhao D, Post WS, Blasco‐Colmenares E, Cheng A, Zhang Y, Deo R, Pastor‐Barriuso R, Michos ED, Sotoodehnia N, Guallar E. Racial differences in sudden cardiac death. Circulation. 2019;139:1688–1697. doi: 10.1161/CIRCULATIONAHA.118.036553 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Tables S1–S5
Figures S1–S5
