Social jet lag is associated with incident cardiovascular disease independent of sleep duration and cardiac genetic risk

Abstract

Background

Cardiovascular disease (CVD) remains the leading cause of mortality worldwide, yet many events occur in individuals not fully explained by traditional clinical and lifestyle factors. Social jet lag reflects misalignment between biological timing and socially imposed sleep schedules and may represent an overlooked behavioral circadian cardiovascular risk factor.

Objective

To determine whether accelerometer‐derived social jet lag is independently associated with incident CVD beyond established clinical factors, sleep‐related characteristics, and cardiac genetic risk.

Methods

Sleep timing was assessed using wrist accelerometry in 51,562 UK Biobank participants. Social jet lag was defined as the absolute difference between weekday and weekend mid‐sleep timing. Participants were prospectively followed for incident CVD, including myocardial infarction, ischemic heart disease, heart failure, stroke, and atrial fibrillation (3853 events).

Results

Higher social jet lag was associated with increased CVD incidence after multivariable adjustment. Each standard deviation increase in social jet lag increased composite CVD risk (hazard ratio [HR] 1.05, 95% confidence interval [CI] 1.02–1.09, p = 0.0015). Individuals with ≥2 h of social jet lag had greater CVD risk (HR 1.30, 95% CI 1.11–1.54, p = 0.0014). Associations were strongest for myocardial infarction, ischemic heart disease, and heart failure. Severe social jet lag remained associated with CVD even among individuals with normal sleep duration, whereas sleep duration itself was not independently associated with risk. Social jet lag also remained independently associated with CVD after accounting for cardiac genetic risk.

Conclusion

Social jet lag was associated with incident CVD independent of sleep duration and cardiac genetic susceptibility, supporting a role for circadian misalignment in cardiovascular pathophysiology.

Introduction

Cardiovascular disease (CVD) remains the leading cause of death worldwide, accounting for approximately 20.5 million deaths in 2021 and nearly one‐third of all global mortality [1]. Despite advances in risk prediction, many events occur in individuals whose conventional risk profiles do not fully explain their risk [2, 3]. This residual risk highlights the role of nontraditional exposures that operate chronically and remain unrecognized across the life course. Circadian misalignment has emerged as one such exposure, particularly in the form of social jet lag, defined as the mismatch between endogenous circadian timing and externally imposed sleep–wake schedules. Unlike acute sleep deprivation or overt sleep disorders, social jet lag reflects persistent phase shifts between workdays and free days driven by social demands rather than pathology [4, 5]. It is highly prevalent, affecting nearly 70% of adults by at least 1 h and approximately one third by 2 h or more in industrialized populations [6, 7, 8, 9]. However, whether this common, behaviorally driven circadian misalignment increases the risk of incident CVD remains unclear. In this study, we examined the prospective association between social jet lag and incident CVD in a large population‐based cohort.

Social jet lag captures a distinct dimension of circadian disruption, reflecting recurrent shifts in circadian phase between workdays and free days rather than differences in sleep duration or quality [10]. As such, it represents a form of temporal misalignment between biological and social time that conventional sleep metrics do not capture. Evidence indicates that this misalignment is associated with adverse cardiometabolic profiles and early vascular alterations, including obesity, insulin resistance, hypertension, and subclinical atherosclerosis [11, 12, 13]. Notably, these associations are often observed independent of sleep duration, suggesting that circadian timing itself may contribute to cardiometabolic risk. However, existing evidence is largely confined to intermediate risk factors and subclinical disease, and direct associations with incident cardiovascular events remain limited. In addition, many studies rely on self‐reported sleep timing, which may introduce measurement error and obscure true exposure–outcome relationships.

The rapid expansion of wearable technologies has transformed the ability to measure sleep and circadian behaviors in real‐world settings. Wrist‐worn accelerometers enable continuous and objective assessment of sleep timing across multiple days, offering greater precision and ecological validity than self‐reported measures. Large‐scale studies have demonstrated the feasibility of capturing habitual sleep timing and circadian patterns at population scale using wearable data [14, 15]. These approaches are particularly well suited for quantifying social jet lag, which reflects recurring behavioral phase shifts embedded within everyday life. Wearable‐derived sleep metrics therefore represent scalable digital biomarkers that can quantify circadian misalignment while reducing exposure misclassification. Such tools provide an opportunity to more accurately evaluate how habitual circadian disruption contributes to long‐term disease risk and are increasingly recognized as important components of digital medicine and population health research [16, 17]. In this context, objective assessment of social jet lag may offer a practical and scalable approach to identify circadian misalignment as a clinically relevant behavioral exposure.

In parallel, genetic susceptibility plays a critical role in CVD. Polygenic risk scores for coronary artery disease can stratify individuals according to inherited cardiac risk but do not account for behavioral exposures that may modify disease trajectories [18, 19]. Emerging evidence suggests that lifestyle and environmental factors can meaningfully influence genetic risk, underscoring the importance of integrating behavioral and genetic determinants of disease [20]. In this context, circadian misalignment may represent a complementary and potentially modifiable exposure that refines risk stratification. Integrating wearable‐derived circadian metrics with cardiac genetic risk information may therefore provide a more comprehensive framework for understanding cardiovascular vulnerability.

To address these gaps, we investigated the prospective association between objectively measured social jet lag and incident CVD in a large population‐based cohort. Sleep timing was derived from wrist‐worn accelerometer recordings, enabling precise quantification of habitual circadian behavior under free‐living conditions. Participants were followed longitudinally for major cardiovascular outcomes, including myocardial infarction, ischemic heart disease, heart failure, stroke, and atrial fibrillation. We further evaluated exposure–response relationships across multiple definitions of social jet lag, including continuous measures, clinically defined thresholds, and fine‐grained categories, to assess potential nonlinear associations. In addition, analyses incorporated a cardiac genetic risk score for coronary artery disease to determine whether social jet lag contributes to cardiovascular risk beyond inherited cardiac susceptibility. By integrating objective wearable‐derived circadian metrics with prospective clinical outcomes and genetic risk information, this study aimed to establish social jet lag as a measurable behavioral exposure with relevance for cardiovascular risk prediction in the general population.

Methods

Study design and reporting standards

This study was conducted as a prospective cohort analysis within the UK Biobank resource. All analyses adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for cohort studies. Ethical approval for UK Biobank was obtained from the North West Multi‐Centre Research Ethics Committee, and all participants provided written informed consent at recruitment. The present analyses were conducted under an approved UK Biobank research application and complied with UK Biobank governance and data access policies.

Study population and analytic samples

UK Biobank recruited over 500,000 participants across the United Kingdom between 2006 and 2010. Participants were eligible for the present analysis if they had valid wrist‐worn accelerometer data enabling objective assessment of sleep timing and social jet lag, along with longitudinal follow‐up for cardiovascular outcomes.

Participants with prevalent CVD at baseline were excluded, defined using self‐reported medical history, hospital admission records, and death registry data prior to baseline assessment. Individuals with missing data for social jet lag, outcome ascertainment, or key covariates were also excluded. The final analytic cohort comprised 51,562 participants free of CVD at baseline. The number of excluded participants and reasons for exclusion are summarized in Fig. 1.

Fig. 1.

Study flow diagram. Selection of participants from the UK Biobank cohort to derive the analytic sample used to investigate associations between social jet lag and incident CVD. CVD, cardiovascular disease.

For analyses involving genetic risk scores or imaging phenotypes, sample sizes varied according to data availability and are reported in the Results.

Accelerometer‐derived sleep and activity data

Objective sleep and rest–activity patterns were derived from wrist‐worn triaxial accelerometer recordings collected in the UK Biobank accelerometer sub‐study. Participants wore Axivity AX3 devices continuously on the dominant wrist for seven consecutive days. Raw acceleration signals were processed using the UK Biobank pipeline to generate hourly sleep probability estimates across the 24‐h day.

From these processed outputs, 24‐h sleep probability profiles were derived separately for weekdays and weekends, providing a standardized representation of the probability of sleep within each hourly bin. This approach enabled robust estimation of habitual sleep timing while minimizing variability arising from individual nights.

Circular estimation of mid‐sleep timing

For each participant, weekday and weekend sleep timing was summarized using 24‐element vectors representing hourly sleep probabilities (“sleep weekday hour average” and “sleep weekend hour average”). This approach enabled robust estimation of habitual sleep timing while avoiding participant‐level noise associated with individual nights.

To obtain a representative mid‐sleep timing from each 24‐h sleep probability distribution, we applied a circular centroid (weighted circular mean) approach that accounts for the periodic nature of clock time.

Let $t_i$ denote the center of each hourly bin (0.5–23.5 h). Clock time was mapped to an angle on the unit circle:

$${\theta }_i=2\pi \left(\frac{t_i}{24}\right)$$

The circular mean angle ($\stackrel{-}{\theta }$) was calculated as

$$\stackrel{-}{\theta }=\mathrm{atan}2\left(\sum _i{w}_i\sin \left({\theta }_i\right),\sum _i{w}_i\cos \left({\theta }_i\right)\right)$$

where $w_i$ denotes the sleep probability weight for hourly bin $i$, and ${\theta }_i$ denotes the angular representation of the corresponding clock time. The resulting centroid angle was transformed back into clock time on the 24‐h scale as

$$\stackrel{-}{t}=24\left(\frac{\stackrel{-}{\theta }}{2\pi }\right)$$

with values wrapped to the 0–24‐h range where necessary.

This circular approach accounts for the periodic nature of clock time and avoids discontinuities at midnight that can arise with linear averaging. Mid‐sleep timing was estimated separately for weekdays (MSW) and weekends/free days (MSF) [21, 22].

Definition of social jet lag

Social jet lag was defined as the absolute difference between weekend and weekday mid‐sleep timing [23, 24]:

$$\mathrm{Social}\mathrm{jet}\mathrm{lag}(\mathrm{h})=\mathrm{MSF}-\mathrm{MSW}$$

Absolute values were used so that both delayed and advanced free‐day sleep timing relative to weekdays contributed equivalently to the measure of circadian misalignment. Thus, negative differences (e.g., earlier free‐day sleep timing relative to weekdays) were retained through absolute transformation rather than excluded. Social jet lag was analyzed as both a continuous variable (per hour increase) and a categorical variable (<1 h, 1–2 h, and ≥2 h), consistent with prior literature [25, 26, 27]. The <1 h category served as the reference group. The analytic cohort was not restricted exclusively to employed participants because participant‐specific employment schedules during accelerometer assessment were unavailable in the UK Biobank. Therefore, weekday and weekend sleep timing were used as population‐level proxies for socially constrained and freer sleep schedules.

Validation of social jet lag using raw accelerometer data (GGIR)

To assess the robustness of the vector‐based sleep timing estimates, an independent validation was performed using raw accelerometer (.cwa) files in a representative subset of participants. Raw data were processed using the GGIR package (version 3.3.0) [28], which identifies the sleep period time (SPT) window based on sustained periods of low movement and temperature‐informed heuristics.

For each valid night, GGIR provided estimates of sleep onset (SPT start), wake time (SPT end), and sleep duration. Nightly mid‐sleep timing was calculated as the midpoint between SPT start and SPT end. Mid‐sleep estimates were aggregated separately for weekdays and weekends using the same circular centroid framework described above, and social jet lag was recalculated using the identical formula.

Concordance between GGIR‐derived and vector‐derived estimates demonstrated close agreement, supporting the validity of the vector‐based sleep timing approach for large‐scale analyses.

Cardiovascular outcomes and follow‐up

The primary outcome was incident CVD, defined as the first occurrence of myocardial infarction, ischemic heart disease, heart failure, stroke, or atrial fibrillation after baseline assessment. Outcomes were ascertained through linkage to Hospital Episode Statistics and national death registries using ICD‐10 codes.

Follow‐up time was calculated from baseline assessment to the first cardiovascular event, death, loss to follow‐up, or administrative censoring, whichever occurred first. Secondary analyses examined cardiovascular mortality and composite endpoints. Outcome definitions and censoring procedures were applied consistently across analyses.

Covariates

Covariates were selected a priori and included demographic factors (age and sex), socioeconomic status (Townsend deprivation index), cardiometabolic factors (body mass index, blood pressure, diabetes, cholesterol, and HbA1c), lifestyle factors (smoking and alcohol use), and sleep‐related variables (sleep duration, chronotype, insomnia symptoms, snoring, and shift work status). Ethnicity was additionally evaluated in sensitivity analyses, because ethnicity data were available for a subset and the cohort was predominantly White. Sleep variability was evaluated separately in exploratory attenuation analyses because it represents a broader measure of day‐to‐day sleep instability that may overlap conceptually with social jet lag. Sleep‐related variables were included to estimate the association of social jet lag beyond established sleep characteristics and cardiometabolic risk factors, although some variables may also partially reflect intermediate pathways linking circadian misalignment to cardiovascular risk. All covariates were assessed at or prior to accelerometer acquisition and included consistently across models.

Statistical analysis

Associations between social jet lag and incident CVD were examined using Cox proportional hazards regression models. Follow‐up time was used as the time scale. Social jet lag was modeled as both a continuous and categorical exposures.

To examine the combined effects of circadian misalignment and sleep irregularity, participants were additionally classified into four circadian phenotypes based on social jet lag (<2 h vs. ≥2 h) and sleep variability (regular vs. irregular sleep). Sleep irregularity was defined using the upper quartile of accelerometer‐derived sleep duration variability, consistent with the threshold‐based approach used for other circadian variability metrics. The reference phenotype consisted of individuals with low social jet lag and regular sleep.

Models were adjusted for demographic, socioeconomic, lifestyle, cardiometabolic, and sleep‐related covariates. Nonlinear associations were assessed using restricted cubic spline models. Stratified and interaction analyses were conducted to evaluate effect modification by cardiac genetic risk and clinical factors.

Proportional hazards assumptions were assessed using Schoenfeld residuals. All analyses were performed using Python (lifelines package). Raw accelerometer validation analyses were conducted using the GGIR package (version 3.3.0).

Ethics, data access, and reproducibility

This study was conducted under an approved UK Biobank research application. Data were de‐identified prior to access and analyzed within the UK Biobank Research Analysis Platform in accordance with governance and data security policies. Code used for data processing and analysis is available from the authors upon reasonable request, subject to UK Biobank data access restrictions.

Results

Baseline characteristics of the study population

After exclusion of participants with prevalent CVD at baseline and those with missing covariates, a total of 51,562 individuals were included in the primary prospective analysis (Fig. 1). Participants were categorized according to social jet lag into <1 h (n = 41,101), 1–2 h (n = 8436), and ≥2 h (n = 2025). Baseline characteristics across social jet lag categories are presented in Table 1.

Table 1.

Baseline characteristics of participants according to social jet lag categories.

Variable	<1 h	1–2 h	≥2 h	p value
Participants, N	41,101	8436	2025	–
CVD events, n	3151	545	157	–
Follow‐up, years (median [IQR])	8.32 [7.74–8.84]	8.34 [7.79–8.87]	8.34 [7.78–8.83]	–
Age, years (mean ± SD)	56.61 ± 7.65	52.83 ± 7.67	50.96 ± 7.45	<0.001
Male sex, n (%)	16,572 (40.3)	3701 (43.9)	1030 (50.9)	<0.001
Ethnicity				<0.001
White ethnicity, n (%)	40,114 (97.8)	8099 (96.2)	1899 (94.0)	–
Asian ethnicity, n (%)	405 (1.0)	123 (1.5)	47 (2.3)	–
Black ethnicity, n (%)	210 (0.5)	108 (1.3)	55 (2.7)	–
Mixed ethnicity, n (%)	88 (0.2)	28 (0.3)	8 (0.4)	–
Other ethnicity, n (%)	182 (0.4)	60 (0.7)	12 (0.6)	–
Body mass index, kg/m2 (median [IQR])	25.86 [23.46–28.73]	26.27 [23.71–29.41]	26.71 [24.09–30.27]	<0.001
Systolic BP, mmHg (mean ± SD)	138.76 ± 19.32	136.11 ± 18.43	135.68 ± 18.54	<0.001
Diastolic BP, mmHg (mean ± SD)	81.59 ± 10.51	81.77 ± 10.52	81.95 ± 10.57	0.144
Total cholesterol, mmol/L (mean ± SD)	5.80 ± 1.10	5.70 ± 1.08	5.64 ± 1.07	<0.001
HbA1c, mmol/mol (median [IQR])	34.80 [32.50–37.20]	34.50 [32.10–37.10]	34.40 [32.00–37.20]	<0.001
Diabetes, n (%)	1162 (2.8)	282 (3.3)	81 (4.0)	<0.001
Current smoker, n (%)	2463 (6.0)	724 (8.6)	277 (13.7)	<0.001
Former smoker, n (%)	14,713 (35.8)	2972 (35.2)	688 (34.0)	0.173
Never smoker, n (%)	23,835 (58.0)	4728 (56.0)	1055 (52.1)	<0.001
Current alcohol intake, n (%)	38,893 (94.6)	7987 (94.7)	1891 (93.4)	0.050
Former alcohol intake, n (%)	1032 (2.5)	219 (2.6)	71 (3.5)	0.021
Never alcohol intake, n (%)	1156 (2.8)	227 (2.7)	63 (3.1)	0.575
Townsend deprivation index (median [IQR])	−2.58 [−3.87 to −0.47]	−2.20 [−3.68 to 0.15]	−1.62 [−3.35 to 1.13]	<0.001
Sleep duration, h/day (median [IQR])	8.74 [8.09–9.50]	8.67 [7.92–9.50]	8.67 [7.82–9.84]	<0.001
Sleep variability, h (median [IQR])	1.01 [0.74–1.35]	1.27 [0.94–1.68]	1.73 [1.26–2.43]	<0.001
Insomnia (yes), n (%)	11,038 (26.9)	2087 (24.7)	492 (24.3)	<0.001
Snoring (yes), n (%)	14,303 (34.8)	3116 (36.9)	759 (37.5)	<0.001
Evening chronotype, n (%)	14,404 (35.0)	3394 (40.2)	967 (47.8)	<0.001

Note: Values are presented as mean ± SD, median [IQR], or n (%), as appropriate. Social jet lag was categorized as <1 h, 1–2 h, and ≥2 h based on the absolute difference between sleep midpoint on weekend and weekdays derived from accelerometer data. p values represent differences across social jet lag categories (Kruskal–Wallis test for continuous variables and χ ² test for categorical variables).

Abbreviation: CVD, cardiovascular disease.

Participants with higher levels of social jet lag were younger and more likely to be male. Mean age decreased progressively across categories, whereas the proportion of men increased from 40.3% in the <1 h group to 50.9% in the ≥2 h group (p < 0.001). Body mass index increased across categories (p < 0.001).

Blood pressure profiles differed modestly. Mean systolic blood pressure was lower in higher social jet lag groups (p < 0.001), whereas diastolic blood pressure did not differ significantly across categories (p = 0.144). Total cholesterol levels were slightly lower with increasing social jet lag (p < 0.001), whereas the prevalence of diabetes increased from 2.8% to 4.0% (p < 0.001).

Socioeconomic and lifestyle factors varied substantially across categories. Townsend deprivation index increased progressively with social jet lag (p < 0.001), indicating greater socioeconomic disadvantage. Current smoking prevalence rose from 6.0% to 13.7% across increasing social jet lag categories (p < 0.001). Differences in alcohol intake were modest.

Sleep characteristics differed markedly by social jet lag. Sleep variability increased substantially across categories (p < 0.001), whereas sleep duration showed smaller differences. Individuals with higher social jet lag were more likely to report snoring, insomnia, and evening chronotype (all p < 0.001).

During follow‐up, 3151, 545, and 157 incident cardiovascular events were observed in the <1 h, 1–2 h, and ≥2 h groups, respectively. Median follow‐up duration was similar across categories.

Study population and distribution of social jet lag

The analytic cohort comprised 51,562 participants with complete accelerometer‐derived sleep timing data, covariates, and longitudinal cardiovascular follow‐up, during which 3853 incident cardiovascular events were recorded. Social jet lag demonstrated substantial inter‐individual variability, with a right‐skewed distribution indicating that most participants had low levels of circadian misalignment, while a smaller proportion exhibited pronounced shifts in sleep timing (Fig. 2a).

Fig. 2.

Distribution of social jet lag and association with cardiovascular risk. (a) Distribution of accelerometer‐derived social jet lag in the analytic cohort, with dashed lines indicating thresholds at 1 h and 2 h. (b) Participant counts across social jet lag categories (<1 h, 1–2 h, and ≥2 h). (c) Dose–response association between social jet lag and incident composite cardiovascular disease estimated using restricted cubic spline models, adjusted for clinical and lifestyle covariates. The shaded region represents 95% confidence intervals. (d) Kaplan–Meier curves showing cardiovascular event–free survival according to social jet lag category (log‐rank p = 0.0004). CVD, cardiovascular disease.

When categorized using clinically interpretable thresholds, the majority of participants had <1 h of social jet lag (n = 41,101), followed by 1–2 h (n = 8436), and ≥2 h (n = 2025) (Fig. 2b). Approximately 4% of the cohort experienced social jet lag ≥2 h, consistent with levels considered to reflect substantial circadian misalignment.

To further characterize exposure patterns, social jet lag was examined across finer intervals. Most individuals clustered below 1 h, with progressively fewer participants at higher levels of misalignment, supporting the use of both continuous and categorical exposure definitions. This distribution enabled assessment of both graded and threshold relationships between social jet lag and cardiovascular risk.

Dose–response analysis using restricted cubic spline models demonstrated a nonlinear association between social jet lag and incident CVD (Fig. 2c). Cardiovascular risk remained relatively stable at lower levels of social jet lag but increased progressively beyond approximately 2 h, suggesting a threshold effect at higher levels of circadian misalignment.

Consistent with these findings, Kaplan–Meier analyses showed progressively lower cardiovascular event–free survival as social jet lag increased (Fig. 2d). Individuals with ≥2 h of social jet lag exhibited the lowest survival probabilities, with significant differences across groups (log‐rank p = 0.0004).

Association between social jet lag and incident cardiovascular disease

Multivariable Cox proportional hazards models were used to examine the association between social jet lag and incident CVD after adjustment for demographic, cardiometabolic, lifestyle, sleep duration, chronotype, insomnia symptoms, snoring, and shift‐work covariates.

Continuous association and overall dose–response

When modeled as a continuous exposure, higher social jet lag was associated with increased cardiovascular risk. Each one‐standard‐deviation increase in social jet lag was associated with a hazard ratio (HR) of 1.052 (95% confidence interval [CI] 1.020–1.086, p = 0.0015) for composite CVD. Similarly, each 1‐h increase corresponded to an HR of 1.076 (95% CI 1.028–1.125, p = 0.002) (Table 2; Fig. 3a).

Table 2.

Association of social jet lag with incident cardiovascular disease.

Exposure	HR (95% CI)	p value
Continuous social jet lag
Per SD increase	1.052 (1.020–1.086)	0.002
Per 1‐h increase	1.076 (1.028–1.125)	0.002
Standard categories
<1 h	Reference	–
1–2 h vs. <1 h	1.025 (0.935–1.124)	0.599
≥2 h vs. <1 h	1.305 (1.109–1.536)	0.001
30 min categories
0–0.5 h	Reference	–
0.5–1 h vs. 0–0.5 h	1.074 (0.998–1.157)	0.058
1–1.5 h vs. 0–0.5 h	1.017 (0.913–1.135)	0.755
1.5–2 h vs. 0–0.5 h	1.154 (0.976–1.364)	0.094
2–2.5 h vs. 0–0.5 h	1.358 (1.079–1.708)	0.009
≥2.5 h vs. 0–0.5 h	1.326 (1.058–1.662)	0.014

Note: Models were adjusted for age, sex, body mass index, systolic blood pressure, diabetes status, Townsend deprivation index, smoking status, cholesterol, HbA1c, alcohol intake, insomnia, snoring, chronotype, and mean sleep duration. Analytic cohort N = 51,562 with 3853 incident cardiovascular events.

Abbreviation: CI, confidence interval.

Fig. 3.

Associations of social jet lag with cardiovascular outcomes: (a) hazard ratios for incident cardiovascular outcomes per standard deviation increase in social jet lag; (b) hazard ratios for cardiovascular outcomes according to standard social jet lag categories (1–2 h and ≥2 h vs. <1 h); and (c) hazard ratios for cardiovascular outcomes across finer social jet lag categories in 30 min increments, with 0–0.5 h as the reference group. Models were adjusted for demographic, clinical, and sleep‐related covariates. Error bars represent 95% confidence intervals. CVD, cardiovascular disease, SJL, social jet lag.

Categorical analyses and threshold effects

Using clinically defined categories, individuals with ≥2 h of social jet lag had a significantly higher risk of CVD compared with those with <1 h (HR 1.305, 95% CI 1.109–1.536, p = 0.001), whereas the 1–2 h category was not associated with risk (HR 1.025, p = 0.599) (Table 2; Fig. 3b).

To further delineate exposure–response relationships, analyses using finer 30‐min categories with 0–0.5 h as the reference group demonstrated a clear threshold effect (Fig. 3c). Cardiovascular risk remained largely unchanged across lower exposure categories but increased at higher levels of social jet lag. Specifically, HRs were 1.358 (95% CI 1.079–1.708, p = 0.0091) for 2–2.5 h and 1.326 (95% CI 1.058–1.662, p = 0.0144) for ≥2.5 h. These findings indicate that cardiovascular risk becomes evident once social jet lag exceeds approximately 2 h.

Associations with individual cardiovascular outcomes

Associations between social jet lag and CVD varied across outcome types (Fig. 3). For myocardial infarction, higher social jet lag was associated with an increased risk, with an HR of 1.086 per standard deviation increase (95% CI 1.016–1.161, p = 0.0157). In categorical analyses, individuals with ≥2.5 h of social jet lag had an HR of 1.741 (95% CI 1.113–2.725, p = 0.0152). For ischemic heart disease, continuous social jet lag was associated with an increased risk (HR 1.066, 95% CI 1.025–1.109, p = 0.0016). Categorical analyses showed consistent increases in risk, and higher exposure categories (≥2 h) were associated with significantly elevated risk (Fig. 3b,c). For heart failure, social jet lag was also associated with increased risk (HR 1.093, 95% CI 1.025–1.166, p = 0.0066), with the strongest association observed at ≥2.5 h (HR 1.844, 95% CI 1.196–2.843, p = 0.0056).

In contrast, associations were weaker for cerebrovascular outcomes. For stroke, the association with continuous social jet lag did not reach statistical significance (HR 1.076, 95% CI 0.989–1.170, p = 0.0902), and categorical estimates showed wide CIs. No meaningful association was observed for atrial fibrillation (HR 0.999, 95% CI 0.932–1.071, p = 0.978).

Survival analysis

Kaplan–Meier curves demonstrated progressively lower cardiovascular event–free survival with increasing social jet lag (Fig. 2d). The separation of survival curves became more apparent at higher levels of social jet lag, with individuals in the ≥2 h group exhibiting the lowest event‐free survival. The overall log‐rank test indicated significant differences across groups (p = 0.0004).

Clinical impact of social jet lag and interaction with cardiac genetic risk

Absolute cardiovascular risk across social jet lag categories

Absolute risk estimates demonstrated a graded increase in cardiovascular risk with higher levels of social jet lag (Fig. 4a). The predicted 5‐year risk of composite CVD increased from 3.14% in individuals with <1 h of social jet lag to 3.22% in those with 1–2 h and 4.08% in those with ≥2 h. At 10 years, corresponding risks increased from 8.47% to 8.68% and 10.91%, respectively.

Fig. 4.

Clinical impact of severe social jet lag and interaction with cardiac genetic risk: (a) predicted 5‐ and 10‐year absolute risk of incident cardiovascular disease according to social jet lag category (<1 h, 1–2 h, and ≥2 h); (b) mean age at cardiovascular disease onset across social jet lag categories, with error bars representing standard deviation; (c) joint associations of cardiac genetic risk and social jet lag with incident cardiovascular outcomes, using low cardiac genetic risk and low social jet lag as the reference group; and (d) predicted 5‐ and 10‐year absolute cardiovascular risk across joint categories of cardiac genetic risk and severe social jet lag, showing the highest risk among individuals with both high cardiac genetic risk and severe social jet lag. Hazard ratios are shown with 95% confidence intervals. CVD, cardiovascular disease.

This corresponds to an absolute risk increase of approximately 0.94 percentage points at 5 years and 2.44 percentage points at 10 years between the lowest and highest exposure groups, indicating that pronounced social jet lag is associated with clinically meaningful increases in long‐term cardiovascular risk.

Age at cardiovascular disease onset

Higher levels of social jet lag were associated with earlier onset of CVD (Fig. 4b). Mean age at first cardiovascular event decreased from 65.3 years in individuals with <1 h of social jet lag to 62.3 years in those with 1–2 h and 61.2 years in those with ≥2 h.

These findings indicate that individuals with severe social jet lag developed CVD approximately 4 years earlier than those with minimal circadian misalignment.

Joint associations of social jet lag and cardiac genetic risk

After adjustment for cardiac genetic risk score, severe social jet lag remained independently associated with cardiovascular outcomes. In joint analyses, individuals with both high genetic risk and severe social jet lag exhibited the highest risk across outcomes (Fig. 4c).

For composite CVD, the HR for individuals with both exposures was 2.34 (95% CI 1.84–2.99) compared with those with low genetic risk and low social jet lag. Similar patterns were observed for myocardial infarction (HR 5.98, 95% CI 3.74–9.56) and ischemic heart disease (HR 3.35, 95% CI 2.51–4.46), indicating substantial amplification of risk when genetic susceptibility and circadian misalignment coexist.

No significant multiplicative interaction was observed, suggesting additive rather than synergistic effects of social jet lag and cardiac genetic risk.

Absolute risk stratification across cardiac genetic risk and social jet lag

Absolute risk stratification further highlighted the combined impact of genetic susceptibility and social jet lag (Fig. 4d). At 5 years, predicted cardiovascular risk ranged from 2.63% in individuals with low genetic risk and low social jet lag to 4.85% in those with both high genetic risk and high social jet lag. At 10 years, corresponding risks ranged from 7.20% to 13.00%.

These findings demonstrate that social jet lag meaningfully stratifies cardiovascular risk within genetic risk groups, identifying individuals at particularly high absolute risk when both exposures are present.

Population attributable fraction

The prevalence of severe social jet lag (≥2 h) in the cohort was 3.93%. Based on the adjusted HR of 1.305, the estimated population attributable fraction was 1.18% (95% CI 0.42–2.06), indicating that approximately 1% of cardiovascular events in the population may be attributable to severe social jet lag.

Sensitivity analyses

Multiple sensitivity analyses were performed to evaluate the robustness of the association between severe social jet lag (≥2 h) and incident composite CVD (Table S1). In the primary fully adjusted model, severe social jet lag was associated with higher cardiovascular risk compared with <1 h (HR 1.305, 95% CI 1.109–1.536, p = 0.0014), whereas the 1–2 h category was not associated with risk.

To address potential reverse causation, events occurring within the first 2 years of follow‐up were excluded. The association remained materially unchanged (HR 1.304, 95% CI 1.087–1.566, p = 0.0044), indicating that the findings were not driven by early events.

Analyses excluding major baseline cardiometabolic conditions demonstrated consistent directionality with partial attenuation. After excluding participants with baseline diabetes, the association remained significant (HR 1.254, 95% CI 1.052–1.494, p = 0.0114). In contrast, among individuals without baseline hypertension (systolic blood pressure <140 mmHg), the association was attenuated and no longer statistically significant (HR 1.093, 95% CI 0.834–1.432, p = 0.5206).

In stratified analyses, the association was evident in men (HR 1.278, 95% CI 1.052–1.552, p = 0.0133), whereas estimates in women were directionally similar but did not reach statistical significance (HR 1.299, 95% CI 0.961–1.756, p = 0.0893). A stronger association was observed among individuals with an evening chronotype (HR 1.459, 95% CI 1.144–1.861, p = 0.0023), whereas estimates in non‐evening chronotypes were weaker and not statistically significant (HR 1.207, 95% CI 0.968–1.505, p = 0.0952). However, formal interaction testing between social jet lag and chronotype was not statistically significant (p‐interaction = 0.3719 for severe social jet lag; p‐interaction = 0.6136 for continuous social jet lag), suggesting that the observed differences across chronotype strata may reflect variation in effect magnitude rather than significant effect modification.

Analyses by occupational schedule demonstrated that the association was most pronounced among non‐shift workers (HR 1.495, 95% CI 1.200–1.862, p = 0.0003). No association was observed among shift workers (HR 0.833, 95% CI 0.526–1.320, p = 0.4371), although this subgroup was substantially smaller with fewer events, limiting precision.

To further address the concern regarding potential misclassification of workdays and free days, we additionally performed sensitivity analyses after excluding occupations commonly associated with irregular or nonstandard work schedules, including shift‐based healthcare, transport, protective service, and hospitality occupations. In this restricted cohort (n = 47,405; 3588 events), severe social jet lag remained associated with higher cardiovascular risk (HR 1.290, 95% CI 1.082–1.538, p = 0.0045). The continuous association also remained significant (HR 1.043 per standard deviation increase, 95% CI 1.009–1.078, p = 0.0128). These findings reduce the likelihood that the observed associations were driven by participants with irregular occupational schedules or misclassification of work and free days.

Additional adjustment for accelerometer‐derived overall physical activity levels did not materially alter the association between severe social jet lag and incident CVD. In models additionally adjusted for overall acceleration average, severe social jet lag remained associated with higher cardiovascular risk (HR 1.294, 95% CI 1.099–1.523, p = 0.0019), whereas the continuous association also remained significant (HR 1.050 per standard deviation increase, 95% CI 1.017–1.083, p = 0.0026). These findings indicate that the observed associations were independent of overall physical activity levels.

Across all sensitivity analyses, the intermediate category of social jet lag (1 H–2 h) was not associated with cardiovascular risk. Overall, these findings support the robustness of the primary association and reinforce a threshold effect, whereby cardiovascular risk becomes evident primarily at higher levels of social jet lag.

Exploratory attenuation analyses were performed to evaluate whether sleep‐related variables contributed to the association between social jet lag and CVD. Adjustment for sleep variability produced the greatest attenuation of the association, reducing the HR for severe social jet lag by approximately 20%, although the association remained statistically significant (HR 1.236, 95% CI 1.045–1.462, p = 0.0134). In contrast, adjustment for sleep duration, insomnia symptoms, snoring, or chronotype produced comparatively smaller changes in effect estimates. Simultaneous adjustment for all sleep‐related variables attenuated but did not abolish the association (HR 1.231, 95% CI 1.041–1.457, p = 0.0152), suggesting that broader sleep instability contributed to the observed association, while social jet lag remained significantly associated with CVD.

Additional adjustment for ethnicity did not materially alter the association between severe social jet lag and incident CVD. After inclusion of ethnicity categories in the multivariable model, severe social jet lag remained associated with higher cardiovascular risk (HR 1.311, 95% CI 1.114–1.543, p = 0.0011), whereas the continuous association also remained significant (HR 1.053 per standard deviation increase, 95% CI 1.020–1.087, p = 0.0013).

Additional analyses examined whether the association between severe social jet lag and cardiovascular outcomes differed across strata of baseline clinical risk, defined using predicted risk from a multivariable clinical model (Table S2). In stratified analyses, severe social jet lag was associated with a higher risk of composite CVD among individuals in the higher‐risk stratum (HR 1.271, 95% CI 1.018–1.587, p = 0.034), whereas the corresponding estimate in the lower‐risk stratum was not statistically significant. Similar patterns were observed for myocardial infarction, ischemic heart disease, and heart failure, whereas no consistent associations were observed for stroke or atrial fibrillation. Joint analyses showed that individuals with both high clinical risk and severe social jet lag had the highest overall risk. These findings suggest that the association of social jet lag with cardiovascular outcomes may be more evident in individuals with greater underlying clinical risk.

Circadian misalignment phenotypes

To examine the combined impact of sleep timing misalignment and sleep regularity, participants were classified into four circadian phenotypes defined by social jet lag (<2 h vs. ≥2 h) and sleep variability (regular vs. irregular) (Fig. 5a). The majority of participants were distributed across low social jet lag groups, with smaller numbers in the high social jet lag categories. Baseline demographic, cardiometabolic, lifestyle, and sleep‐related characteristics across circadian phenotypes are summarized in Table S3.

Fig. 5.

Circadian phenotypes defined by social jet lag and sleep regularity: (a) definition of four circadian phenotypes based on social jet lag (<2 h vs. ≥2 h) and sleep timing variability (regular vs. irregular); (b) hazard ratios for incident cardiovascular outcomes across circadian phenotypes using low social jet lag with regular sleep as the reference group; (c) predicted 5‐ and 10‐year absolute cardiovascular risk across circadian phenotypes; and (d) mean age at cardiovascular disease onset by circadian phenotype, with error bars indicating standard deviation. CVD, cardiovascular disease, SJL, social jet lag.

Association of circadian phenotypes with cardiovascular outcomes

Compared with the reference phenotype of low social jet lag and regular sleep, combined circadian misalignment and sleep irregularity were associated with higher cardiovascular risk (Fig. 5b). Individuals with high social jet lag and irregular sleep exhibited the highest risk of composite CVD (HR 1.36, 95% CI 1.14–1.62).

A similar pattern was observed for individual outcomes. The combined phenotype was associated with increased risk of myocardial infarction (HR 1.72, 95% CI 1.20–2.47), ischemic heart disease (HR 1.41, 95% CI 1.13–1.75), and heart failure (HR 1.74, 95% CI 1.22–2.48). In contrast, associations with stroke and atrial fibrillation were not statistically significant.

Notably, intermediate phenotypes (low social jet lag with irregular sleep or high social jet lag with regular sleep) showed modest or inconsistent associations, indicating that coexistence of both circadian misalignment and irregularity confers the greatest cardiovascular risk.

Additional analyses examined whether the association between social jet lag and CVD differed according to habitual sleep duration. In models jointly evaluating social jet lag and sleep duration categories, severe social jet lag remained associated with increased cardiovascular risk even among individuals with normal sleep duration (7–9 h) (HR 1.29, 95% CI 1.01–1.65, p = 0.039). In contrast, neither short sleep duration (<7 h) nor long sleep duration (>9 h) alone was independently associated with cardiovascular risk after multivariable adjustment. Similarly, in models simultaneously adjusting for both social jet lag and sleep duration categories, severe social jet lag remained independently associated with CVD (HR 1.29, 95% CI 1.10–1.52, p = 0.0019), whereas sleep duration categories were not significantly associated with risk. These findings support the importance of circadian alignment beyond sleep duration alone.

Absolute cardiovascular risk across circadian phenotypes

Absolute risk estimates further illustrated the clinical relevance of circadian phenotypes (Fig. 5c). At 5 years, predicted cardiovascular risk increased from 3.10% in individuals with low social jet lag and regular sleep to 4.18% in those with both high social jet lag and irregular sleep. At 10 years, corresponding risks increased from 8.34% to 11.14%.

These findings indicate a stepwise increase in absolute cardiovascular risk with increasing circadian disruption, with the highest risk observed when both misalignment and irregularity are present.

Age at cardiovascular disease onset across circadian phenotypes

Circadian phenotype was also associated with differences in the timing of disease onset (Fig. 5d). Individuals with low social jet lag and regular sleep developed CVD at a mean age of 65.8 years, whereas those with combined circadian disruption (high social jet lag and irregular sleep) developed disease earlier at 61.1 years.

Intermediate phenotypes showed graded differences, with low social jet lag and irregular sleep associated with onset at 64.2 years and high social jet lag with regular sleep showing earlier onset (57.9 years), although this subgroup was relatively small. Interestingly, earlier CVD onset was also observed among individuals with high social jet lag despite relatively regular sleep timing, suggesting that circadian misalignment itself may contribute to cardiovascular vulnerability independent of generalized sleep irregularity.

Overall, these findings indicate that combined circadian disruption is associated not only with increased cardiovascular risk but also with earlier disease onset.

Discussion

In this large population‐based study with objective assessment of habitual sleep timing patterns, we found that severe social jet lag was associated with a higher risk of incident CVD independent of sleep duration and cardiac genetic risk. Individuals with social jet lag ≥2 h had a significantly increased risk of composite cardiovascular events as well as myocardial infarction, ischemic heart disease, and heart failure, whereas more modest levels of social jet lag were not associated with risk, indicating a threshold pattern. These associations remained consistent across sensitivity, stratified, and multivariable analyses and were observed after adjustment for demographic, cardiometabolic, lifestyle, and sleep‐related factors. Importantly, social jet lag was derived from accelerometer‐based measurements, providing objective characterization of habitual sleep timing behavior. In addition to relative risk estimates, severe social jet lag was associated with higher absolute cardiovascular risk and earlier age at disease onset, highlighting its potential clinical relevance. These findings suggest that pronounced social jet lag may contribute to increased cardiovascular vulnerability in the general population.

Severe social jet lag was associated with a higher risk of both composite and individual cardiovascular outcomes, particularly ischemic cardiovascular endpoints, indicating that circadian misalignment imposed by everyday social schedules relates to long‐term cardiovascular vulnerability. Notably, this association persisted after adjustment for conventional sleep characteristics and cardiometabolic risk factors and accelerometer‐derived overall physical activity levels, suggesting that social jet lag captures a dimension of risk not reflected by traditional sleep metrics. Exploratory attenuation analyses demonstrated that adjustment for sleep duration, insomnia symptoms, snoring, and chronotype produced comparatively small changes in effect estimates, suggesting that the association between social jet lag and CVD was not fully explained by conventional sleep‐related characteristics. In contrast, adjustment for sleep variability, representing a broader measure of day‐to‐day sleep instability, produced greater attenuation of the association. However, severe social jet lag remained significantly associated with CVD even after simultaneous adjustment for sleep duration, sleep variability, chronotype, insomnia symptoms, and snoring, suggesting that social jet lag captures aspects of circadian disruption beyond generalized sleep instability alone. Notably, social jet lag identified increased cardiovascular risk even in individuals without overt sleep deprivation or diagnosed sleep disorders, highlighting that adverse circadian exposures may occur despite apparently normal sleep duration. These findings suggest that persistent misalignment between biological and social time may represent a distinct and underrecognized behavioral risk factor that accumulates over time. This interpretation is consistent with previous population‐based studies showing that irregular sleep timing, variability in sleep patterns, and circadian misalignment are associated with higher cardiovascular risk independent of sleep duration and traditional risk factors [29, 30, 31, 32, 33]. In this context, social jet lag can be viewed as a chronic distortion of biological timing arising from everyday social constraints, rather than a manifestation of acute sleep loss alone. The use of absolute differences enabled capture of both delayed and advanced shifts in free‐day sleep timing relative to weekdays, recognizing that both directions may reflect behavioral circadian instability.

Building on these observations, a central finding of the present study is the nonlinear association between social jet lag and cardiovascular risk, characterized by a clear threshold effect. Cardiovascular risk remained largely stable across lower levels of social jet lag but increased beyond approximately 2 h, as consistently demonstrated across categorical analyses, finer exposure stratification, and spline modeling. Individuals with 1–2 h of social jet lag did not exhibit increased risk, whereas those with ≥2 h showed significantly elevated risk across multiple cardiovascular outcomes. This pattern was further supported in analyses using finer exposure categories in which risk estimates remained close to unity at lower levels but increased at higher levels of circadian misalignment.

These findings suggest that clinically meaningful circadian disruption may emerge only beyond a critical magnitude of weekday–weekend sleep timing displacement, rather than increasing linearly across the exposure range. Such a threshold relationship strengthens the robustness of the association by arguing against a simple linear effect driven by residual confounding or generalized lifestyle differences. Consistent with this interpretation, prior studies have linked higher levels of social jet lag to adverse cardiometabolic outcomes [10, 11, 33, 34], whereas experimental studies of circadian misalignment demonstrate disproportionate physiological disruption. Although previous evidence, including that of Gamboa Madeira et al. (2021), has largely been cross‐sectional [12], our findings extend this literature by demonstrating that severe social jet lag is prospectively associated with future cardiovascular risk once circadian misalignment exceeds a critical level.

Importantly, social jet lag in the present study was derived using weekday and weekend sleep timing as proxies for socially constrained and freer sleep schedules because participant‐specific work and nonwork schedules during accelerometer assessment were unavailable in the UK Biobank. Although this may have introduced greater exposure misclassification among unemployed, retired, or nonstandard schedule participants, the association remained materially unchanged in sensitivity analyses restricted to participants with conventional daytime occupational profiles while excluding occupations commonly associated with irregular schedules. These findings reduce the likelihood that the observed associations were driven primarily by atypical work–rest patterns.

Beyond relative risk estimates, social jet lag was associated with clinically meaningful differences in absolute cardiovascular risk and age at disease onset. Individuals with severe social jet lag had higher predicted 5‐ and 10‐year risk and developed CVD several years earlier than those with minimal misalignment, indicating potential relevance for long‐term risk stratification. Notably, these associations persisted after adjustment for cardiac genetic risk, and joint analyses showed that individuals with both high cardiac genetic susceptibility and severe social jet lag had the highest risk across outcomes, supporting an additive contribution of behavioral circadian disruption to inherited risk. This observation aligns with prior evidence demonstrating that behavioral and lifestyle factors can meaningfully modify genetic cardiovascular risk and improve risk stratification [35, 36, 37]. These findings further suggest that incorporating circadian behavior into risk models may enhance identification of individuals at elevated risk beyond traditional clinical and genetic markers.

Furthermore, combined circadian phenotypes integrating social jet lag and sleep irregularity identified a subgroup with the greatest risk, highest absolute event rates, and earliest disease onset, suggesting that multidimensional circadian disruption may provide additional resolution beyond single metrics. Similar approaches integrating multiple sleep characteristics have been shown to improve cardiovascular risk prediction in population‐based studies [38]. Recent work has highlighted that accelerometer‐derived sleep regularity metrics, including the sleep regularity index, may vary depending on calculation methodology and preprocessing approaches, potentially influencing associations observed in prospective analyses. Emerging efforts to standardize reporting practices, such as the proposed Reporting Items for Regularity Indices framework, may improve methodological transparency, reproducibility, and comparability across circadian epidemiology studies by encouraging clearer reporting of preprocessing methods, scoring approaches, and sleep regularity definitions [39]. Accordingly, the interpretation of sleep irregularity‐related findings in the present study should be considered within the context of the specific variability‐based measures used. Our findings indicate that social jet lag not only relates to incident CVD but also refines risk stratification when considered alongside genetic susceptibility and broader patterns of circadian disruption. Notably, the association between severe social jet lag and CVD persisted even among individuals with habitual sleep durations within the conventional 7–9 h range, whereas sleep duration itself was not independently associated with cardiovascular risk after adjustment. These findings suggest that circadian alignment may represent a biologically distinct dimension of cardiovascular vulnerability beyond sleep duration alone. This observation supports emerging concepts that the temporal organization of sleep behavior may have independent cardiometabolic relevance even in individuals achieving conventionally adequate sleep duration.

The robustness of associations across extensive sensitivity analyses further supports the interpretation that the observed risk is not driven by occupational extremes, early reverse causation, or residual confounding by lifestyle factors. Associations remained materially unchanged after exclusion of early events, reducing the likelihood of reverse causation [40]. Furthermore, our findings suggest that socially imposed circadian misalignment represents a pervasive temporal stressor with the capacity to shape cardiovascular risk trajectories over time [41, 42]. The absence of effect modification by chronotype reinforces this interpretation, indicating that the risk associated with social jet lag reflects external misalignment pressures rather than intrinsic circadian preference.

The direction of effect was consistent across subgroups, including sex, chronotype, and cardiometabolic status, with somewhat stronger associations observed among individuals with an evening chronotype and those not engaged in shift work. Stratified analyses by baseline clinical risk indicated that the association of severe social jet lag was more evident among individuals with higher underlying risk, consistent with a model in which circadian misalignment may amplify existing cardiovascular vulnerability [43].

This study has several strengths, including the use of accelerometer‐derived sleep timing to objectively quantify social jet lag, a large population‐based cohort with substantial numbers of incident cardiovascular events, and comprehensive adjustment for demographic, cardiometabolic, lifestyle, and sleep‐related factors. The availability of cardiac genetic risk data and detailed circadian phenotyping further enhances the translational relevance of the findings. However, several limitations should be considered. The observational design precludes causal inference, and residual confounding from unmeasured behavioral or environmental factors cannot be excluded. Social jet lag may also partially reflect broader lifestyle and social behavioral patterns, including later meal timing, altered weekend schedules, or differences in alcohol consumption that frequently accompany delayed sleep timing on free days. Although alcohol intake was adjusted for, these measures were self‐reported and may therefore remain subject to measurement error and residual confounding. Similarly, unmeasured social and behavioral factors, including marital or relationship status, may also have influenced the observed associations. In addition, because sleep‐related variables and social jet lag were assessed contemporaneously, formal causal mediation relationships could not be definitively established. Accelerometry captures sleep timing behavior but does not directly measure endogenous circadian phase, and exposure was assessed over a limited recording period, which may not fully capture long‐term circadian behavior. In addition, participant‐specific work schedules, holidays, or vacation periods during accelerometer assessment were unavailable in UK Biobank. Therefore, weekdays and weekends were used as proxies for socially constrained and freer sleep schedules rather than true workdays and free days, which may have introduced non‐differential exposure misclassification, predominantly among individuals with nonstandard occupational schedules. In addition, several covariates were derived from baseline assessments conducted prior to accelerometer acquisition, whereas social jet lag was derived from later accelerometer recordings. Consequently, some lifestyle and clinical variables may have changed between assessments, potentially introducing time‐varying misclassification. Additionally, generalizability to other populations may be limited.

In conclusion, severe social jet lag is associated with incident CVD, particularly ischemic outcomes, with a clear threshold effect and clinically meaningful differences in absolute risk and age at onset. Chronotype has a recognized genetic basis, with genome‐wide association studies identifying multiple loci and polygenic influences associated with morningness–eveningness preference [44, 45]. Future studies incorporating chronotype‐specific polygenic risk scores and clock gene variants may, therefore, help disentangle inherited circadian preference from environmental, occupational, and behavioral contributors to social jet lag, thereby improving understanding of the relative contributions of biological circadian predisposition and socially imposed sleep schedules to cardiovascular risk. These findings highlight social jet lag as a measurable behavioral marker of cardiovascular risk and support its potential relevance in digital health–based risk assessment and prevention strategies.

Conclusion

In this large prospective study using objectively measured sleep timing, severe social jet lag was associated with increased risk of incident CVD, particularly ischemic outcomes, with evidence of a threshold effect beyond approximately 2 h of misalignment. These associations were independent of conventional cardiometabolic risk factors, sleep duration, and cardiac genetic susceptibility and were accompanied by higher absolute risk and earlier age at disease onset. Social jet lag also provided additional cardiovascular risk information when considered alongside cardiac genetic susceptibility and sleep irregularity. Together, these findings identify social jet lag as a measurable behavioral marker of circadian disruption with potential relevance for cardiovascular risk assessment. Future studies are needed to determine whether reducing social jet lag improves long‐term cardiovascular outcomes.

Author contributions

Neeraj Kumar: Conceptualization; methodology; data curation; formal analysis; investigation; visualization; writing—original draft; and writing—review and editing. Sairam Krishnamurthy: Conceptualization; methodology; supervision; project administration; validation; interpretation of findings; and writing—review and editing.

Conflict of interest statement

The authors declare no conflicts of interest.

Funding information

This work was supported by the University Grants Commission (UGC), New Delhi, India (Grant No. F. 82‐44/2020 (SA‐III)).

Ethics statement

UK Biobank received ethical approval from the North West Multi‐Centre Research Ethics Committee, and all participants provided written informed consent at recruitment. The present analyses were conducted under an approved UK Biobank research application and complied with UK Biobank governance and data access policies.

Supporting information

Table S1. Sensitivity analysis.

Table S2. Association of severe social jet lag with cardiovascular outcomes stratified by baseline clinical risk.

Table S3. Baseline demographic, cardiometabolic, lifestyle, and sleep‐related characteristics across circadian phenotypes defined by social jet lag and sleep regularity.

Data availability statement

UK Biobank data are accessible to qualified researchers through application to the UK Biobank resource. Individual‐level participant data cannot be publicly shared due to UK Biobank governance policies. Code used for data preparation and statistical analyses is available from the corresponding author upon reasonable request and subject to UK Biobank data access restrictions.