Associations of New-Onset Atrial Fibrillation With Risks of Cardiovascular Disease, Chronic Kidney Disease, and Mortality Among Patients With Type 2 Diabetes

Type 2 diabetes (T2D) is a global public health crisis, and its related macro- and microvascular complications are major contributors to morbidity and mortality (1). The number of people living with T2D has been increasing and is projected to reach 783 million by 2045 (2). Patients with diabetes are at two- to fourfold greater risk of developing cardiovascular disease (CVD) and of premature death than those without diabetes (3). Further, it is estimated that ∼40% of patients with diabetes will develop chronic kidney disease (CKD) (4), which is in turn associated with higher risks of cardiovascular events and mortality (5). Therefore, CVD and CKD prevention among patients with T2D is of paramount importance.

Atrial fibrillation (AF), the most common arrhythmia, is a burgeoning health threat affecting >33 million people worldwide (6,7). Patients with T2D are at higher risks of developing AF (8,9), AF-related hospitalizations, and death (10). In the past decade, epidemiological evidence has linked new-onset AF with subsequent adverse cardiovascular events and death (11–17), as well as progression of kidney function decline among patients with CKD (18–20). However, the longitudinal impact of new-onset AF on the risk of adverse outcomes among patients with T2D is largely unknown, which may have substantial public health and clinical implications for the management of patients at high risk to reduce diabetes complications.

To address the research gaps, we examined the associations of incident AF with subsequent risks of atherosclerotic cardiovascular disease (ASCVD), heart failure, CKD, all-cause mortality, and CVD mortality among patients with T2D. We used data from the UK Biobank study, a large prospective cohort study including >0.5 million participants.

The UK Biobank study is a large population-based prospective cohort study for common diseases of middle-aged and older adults. The design of the UK Biobank study has previously been described (21,22). Briefly, >500,000 participants (aged 37–73 years) were recruited from 22 assessment centers across the U.K. between 13 March 2006 and 1 October 2010. Participants completed information on sociodemographics, habitual diet, lifestyle factors, medical history, and medication history through touch screen questionnaires at recruitment, had standardized physical measurements, and provided biological samples (blood, urine, and saliva).

Among 23,748 adults with T2D, we excluded individuals with preexisting coronary artery disease (CAD), stroke (ischemic and hemorrhagic), peripheral artery disease (PAD), heart failure, AF, and CKD, leaving a total of 16,551 patients with T2D in the current analysis. The flowchart for the selection of the study population is presented in Supplementary Fig. 1. The prevalent cases of T2D were identified using the algorithms method developed by the UK Biobank, which has been shown to be a reliable measurement with 96% accuracy (23).

The UK Biobank study was approved by the National Information Governance Board for Health and Social Care in England and Wales, the Community Health Index Advisory Group in Scotland, and the North West Multicentre Research Ethics Committee. All participants gave written informed consent.

The exposure, new-onset AF during follow-up, was identified through multiple sources including self-report, primary care records, hospital admissions, and death registries. ICD-10 codes used to define AF are shown in Supplementary Table 1.

Data on height and body weight were examined at baseline by trained nurses, and BMI was calculated as weight in kilograms divided by the square of height in meters. Information on age, sex, ethnicity, smoking status, and sleep straits including sleep hours, insomnia, and snoring was collected through interviews at recruitment. Townsend deprivation index (TDI) is a composite measure of area-level socioeconomic deprivation; a higher score indicates higher levels of socioeconomic deprivation (24). Information on habitual diet and alcohol intake was captured with a touch screen food-frequency questionnaire. A hypothesis-driven dietary pattern was generated to reflect the overall diet using five well-known heart health–related dietary components (25–28). Definitions and variables used for dietary components are shown in Supplementary Table 2. Physical activity (duration and intensity) was assessed with a short-form international physical activity questionnaire, and physically active was defined as ≥150 min/week moderate or ≥75 min/week vigorous or 150 min/week moderate/vigorous activities (29).

Prevalent hypertension cases were defined according to self-reported physician diagnosis, use of antihypertensive medications, identification of essential hypertension cases via linking the electronic health records, or blood pressure ≥140/90 mmHg. Medication history (e.g., aspirin and antihypertensive, cholesterol-lowering, and antidiabetes drugs) was self-reported. Participants were also asked to provide the medicines in the following visits if they were not certain about the types of the medications taken. Further, serum creatinine, triglycerides, total cholesterol, LDL cholesterol (LDL-C), C-reactive protein (CRP), and glycated hemoglobin A_1c (HbA_1c) were also measured at baseline via robust and reliable analytical methods (30). Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation was used to generate the estimated glomerular filtration rate (eGFR) (31).

The primary outcomes of the study were the occurrence of ASCVD, heart failure, and CKD. ASCVD was defined as a composite outcome of CAD, ischemic stroke, and PAD. The secondary outcomes were all-cause mortality and CVD mortality. Details on the ICD-10 codes are shown in Supplementary Table 1. The electronic health records were available up to 30 September 2020, 31 August 2020, and 28 February 2018 for centers in England, Wales, and Scotland, respectively. Death data were available up to 31 March 2020 for all participants. Patients were censored at occurrence of first end point, death, loss to follow-up, or end of follow-up—whichever occurred first.

The differences in baseline characteristics by individuals with and without new-onset AF were examined with Student t test for continuous variables and χ² test for categorical variables. We used time-varying Cox proportional hazards regression models to compute the hazard ratio (HR) and 95% CI for the associations of new-onset AF and subsequent risks of ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality. In the current analysis, AF was considered as a time-varying exposure. Thus, the date of AF and the date of outcomes were each recorded for AF cases. Individuals who developed AF during follow-up contributed person-time to the no AF exposure group before they were diagnosed as AF and thereafter to the AF exposure group.

In model 1, we adjusted for age at recruitment (years, continuous) and sex (men, women). In Model 2, we further adjusted for Townsend deprivation index (continuous), ethnicity (White, non-White), BMI (kg/m², continuous), alcohol intake (never or special occasions, monthly or weekly, daily), smoking status (never, past, current), healthy diet score (in quintiles), sleep duration (≤5, 6–8, ≥9 h/day), physical activity (yes, no), family history of CVD (yes, no), history of hypertension (yes, no), systolic blood pressure (mmHg, continuous), duration of diabetes (years, continuous), concentrations of HbA_1c (mmol/mol, continuous), medication for diabetes (none, oral medicine only, insulin and others), use of aspirin (yes, no), use of antihypertensive drugs (yes, no) and use of cholesterol-lowering drugs (yes, no).

We further evaluated the associations of new-onset AF with subtypes of ASCVD including CAD, ischemic stroke, and PAD. We also stratified the analyses by age (≤65, >65 years), sex (men, women), BMI (≤30, >30 kg/m²), duration of diabetes (≤5, >5 years), smoking status (never, ever), and baseline eGFR (<90, ≥90 mL/min/1.73 m²) and tested for interactions by including the multiplicative interactions between incident AF and the stratified factors for the outcomes of interest in model 2. In addition, we also assessed the associations between the prevalent AF cases and outcomes, which was conducted among 16,893 patients with T2D including 322 prevalent AF cases after exclusion of individuals with preexisting CAD, stroke, PAD, heart failure, and CKD at baseline.

To test the robustness of our findings, we conducted several sensitivity analyses. First, we repeated the main analyses after excluding the individuals with <1 year of follow-up to minimize the effect of subclinical conditions on the observed associations. Second, we performed the analyses in a 1:4 propensity score–matched cohort to minimize differences between incident AF cases and noncases. Propensity scores were calculated with use of a logistic regression model including all the covariates in model 2. Third, to account for competing risks of death, we assessed the associations of incident AF with risks of ASCVD, heart failure, and CKD using Fine-Gray subdistribution hazards models. Fourth, to assess whether the associations could be partially explained by blood lipids, systemic inflammation, or kidney function, we further adjusted for the circulating levels of lipids biomarkers (total cholesterol, triglycerides, and LDL cholesterol), inflammation biomarker (C-reactive protein), or eGFR on the basis of model 2. Fifth, in addition to sleep duration, insomnia and snoring were additionally adjusted for on the basis of model 2. Finally, waist circumference was adjusted for instead of BMI, as evidence showed that central obesity was associated with a higher risk of incident CVD and mortality independent of BMI (32).

All analyses were performed with Stata statistical software, release 15.1 (StataCorp, College Station, TX), and a two-sided P < 0.05 was set as the threshold for statistical significance.

The UK Biobank data are available on application to the UK Biobank (www.ukbiobank.ac.uk/).

Of 16,551 participants with T2D (mean [SD] age 59.3 [7.0] years, 60.0% men), 1,394 developed AF during follow-up. Distributions of the baseline characteristics among T2D patients with and without incident AF are shown in Table 1. Relative to participants who did not develop AF during follow-up, new-onset AF case subjects were more likely to be older, men, White, current smokers, daily drinkers, physically inactive, have higher BMI, have lower eGFR, and have a longer duration of diabetes. They also tended to have higher rates of antihypertensive medication use, cholesterol-lowering medication use, and aspirin use.

The median follow-up time for ASCVD, heart failure, CKD, and mortality was 10.9 years (interquartile range 8.6–11.9), 11.0 years (10.2–12.0), 10.9 years (9.7–11.9), and 10.7 years (9.8–11.6), respectively. Over the follow-up period, 2,872 participants had incident ASCVD, 852 had heart failure, 1,548 had CKD, and 1,776 died (409 from CVD). Incidence rates of all outcomes were substantially higher among participants who developed AF compared with those who did not develop AF. The incidence rates of ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality among T2D patients who did not develop AF were 16.6, 4.0, 8.6, 9.1, and 1.9 per 1,000 person-years, respectively, whereas the respective incident rates among those who developed AF were 36.9, 21.3, 17.3, 31.3, and 9.0 per 1,000 person-years (Table 2). Participants who developed AF during follow-up had higher HRs for all outcomes than those who did not develop AF (Table 2), and the corresponding HRs of ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality were 1.85 (95% CI 1.59–2.16), 4.40 (95% CI 3.67–5.28), 1.68 (95% CI 1.41–2.01), 2.91 (95% CI 2.53–3.34), and 3.75 (95% CI 2.93–4.80) in the multivariable-adjusted model 2.

In the analyses for different subtypes of ASCVD, incident AF was consistently associated with higher risk of CAD (HR 1.84; 95% CI 1.57–2.15), ischemic stroke (HR 1.74; 95% CI 1.27–2.37), and PAD (HR 2.12; 95% CI 1.59–2.83) (Table 3). In the stratified analyses, we observed that the AF-related CKD risk in patients with T2D was stronger in men than women (P_interaction = 0.004) (Fig. 1). However, we did not find significant heterogeneity in the risk estimates between any other stratified factors and incident AF for risks of outcomes (all P_interaction >0.05) (Fig. 1 and Supplementary Fig. 2).

In analyzing the association between prevalent AF at baseline with subsequent outcomes, compared with T2D patients without AF at recruitment, the HRs of ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality in prevalent AF patients were 2.01 (95% CI 1.76–2.55), 4.05 (95% CI 3.18–5.16), 1.23 (95% CI 0.91–1.65), 1.44 (95% CI 1.12–1.87), and 2.20 (95% CI 1.42–3.41), respectively (Table 4).

Our results were robust in a number of sensitivity analyses. Excluding the individuals with <1 year of follow-up yielded similar results (Supplementary Table 3). In addition, we performed the analysis in a 1:4 propensity score–matched cohort. The standardized difference for all matching variables from model 2 was <0.1, indicating balance between the incident AF group and no AF group (Supplementary Table 4). The risk estimates were modestly attenuated. In comparison of incident AF with no AF, the HRs of ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality were 1.51 (95% CI 1.30–1.76), 3.37 (95% CI 2.84–4.01), 1.55 (95% CI 1.30–1.85), 2.95 (95% CI 2.58–3.39), and 3.57 (95% CI 2.80–4.58), respectively (Supplementary Table 5). Further, the results were consistent when we used a competing risk model accounting for death (Supplementary Table 6), with additional adjustment for lipids, inflammation, kidney function biomarkers, sleep quality factors (insomnia and snoring), or waist circumference instead of BMI (Supplementary Table 7).

In this large prospective cohort study of patients with T2D without AF at recruitment, we found that patients with new-onset AF during follow-up had higher risks of developing subsequent ASCVD, heart failure, CKD, all-cause mortality, and CVD mortality compared with those who did not develop AF. The results were consistent in different subgroups including age, sex, BMI, smoking status, duration of diabetes, and baseline eGFR. In addition, our results were robust in propensity-matched score analyses.

Our results corroborate prior work suggesting that AF is a risk factor of cardiovascular event and mortality in patients with CVD or general populations and extend these findings to patients with T2D. Prior studies have addressed the AF-related cardiovascular morbidity and mortality in patients with established CVD. Results from a U.S. community-based cohort study including 3,220 patients with first-ever myocardial infarction showed that the occurrence of AF after myocardial infarction was associated with a substantially higher risk of all-cause mortality (HR 3.77; 95% CI 3.37–4.21) (14). Another study from the U.K. with 15,415 patients with heart failure showed that incident AF was associated with twofold increased risks of cardiovascular death or heart failure hospitalization. Further, among patients with a history of AF at baseline, paroxysmal AF posed a higher risk of heart failure hospitalization than persistent or permanent AF (11). In addition, a study from Taiwan showed that among patients with hypertrophic cardiomyopathy, patients with incident AF had greater risk of developing sudden cardiac arrest (HR 3.63; 95% CI 2.76–4.79) and stroke-related death (HR 6.61; 95% CI 3.79–9.73) (13). In addition, excess risk of cardiovascular event, progressive kidney function decline, and death with new-onset AF was also observed among patients with CKD. In patients with CKD, new-onset AF was consistently associated with a higher risk of major cardiovascular event (20,33,34) and progression to end-stage renal disease (18–20). In the Chronic Renal Insufficiency Cohort (CRIC) Study, incident AF was associated with two- to fivefold higher risk of developing subtypes of cardiovascular event and a threefold higher risk of developing end-stage renal disease among patients with CKD (19,33).

The associations between incident AF and cardiovascular outcomes and premature death were also demonstrated in general populations. A cohort in the Netherlands including 8,265 participants showed that incident AF during follow-up was associated with higher risk of total cardiovascular events (HR 2.24; 95% CI 1.06–4.75), heart failure (HR 4.52; 95% CI 2.02–10.1), and all-cause mortality (HR 3.02; 95% CI 1.73–5.27) (35). Similarly, findings from the Women’s Health Study showed that incident AF was significantly associated with higher risk of all-cause mortality (HR 2.14; 95% CI 1.64–2.77) and CVD mortality (HR 4.18; 95% CI 2.69–6.51) (36). However, some variations also existed. Results from the Atherosclerosis Risk in Communities (ARIC) study showed that the association between AF and myocardial infarction was limited to non-ST-segment-elevation myocardial infarction but not ST-segment-elevation myocardial infarction and was limited to women but not men (37,38).

Notably, our study extended the previous findings to patients with T2D, which may have significant clinical relevance and public health implications for the prevention and management of T2D-related macro- and microvascular complications. Both AF and T2D are rising public health crises and important contributors to high morbidity and mortality. It is predicted that there will be 783 million people living with diabetes by 2045 and 60 million people living with AF by 2050 (2,39). The results of our study provide novel insights into the limited published literature on the longitudinal associations of new-onset AF with adverse cardiovascular events, CKD, and premature death among individuals with T2D, suggesting that patients with T2D and incident AF are a group at high risk among whom CVD and kidney function prevention strategies should be optimized. Our findings underscore the importance of early prevention of AF among patients with T2D and will contribute toward the scientific basis for the development of intervention studies targeting AF prevention for T2D management in the future.

Several possible mechanisms may explain how incident AF could contribute to the risk of macro- and microvascular complications in T2D. First, both T2D and AF may promote systemic inflammation and oxidative stress, which could result in cardiac/renal fibrosis via cardiac structural and electrical remodeling and renal hemodynamic changes, and ultimately lead to CVD and CKD (40–42). In addition, chronic hyperglycemia and AF may alter cardiac fibroblast function and induce renal profibrotic pathways, which may lead to cardiac and renal fibrosis (43–45). Further, hyperglycemia and AF result in endothelial dysfunction affecting the regulation of the prothrombotic-related pathways, which may cause ASCVD and CKD (46,47). Finally, AF can cause heart failure through tachycardia-induced cardiomyopathy and further contribute to heart failure decompensation via ventricular dysfunction caused by elevated left atrial pressure, decreased stroke volume, and cardiac output (48,49).

To our best knowledge, this study is the first investigation of the longitudinal associations of new-onset AF with subsequent development of cardiovascular and kidney outcomes among patients with T2D. However, our results should be interpreted in the context of several potential limitations. First, due to lack of information on repeated measurements of the covariates, we were unable to address the time-varying covariates issue in the current analysis and only baseline covariates were used in the models. Hence, our results may be prone to misclassification bias. Second, due to the small number of cases of subtypes of AF (paroxysmal, persistent, and chronic AF), we were unable to further clarify the associations of different AF subtypes with health outcomes. Third, treatments of AF after diagnosis were not accounted for in the current analysis, as the information was not available. Further studies are warranted to investigate whether anticoagulant therapy/ablation could reduce the risk of macro- and microvascular complications in people with T2D. Fourth, as the majority of the study population was White, racial/ethnic differences in the associations cannot be addressed and generalizability of the results to other ethnic groups should be considered with caution. Fifth, as the AF cases were identified through electronic health records, some paroxysmal AF cases might not be captured. Sixth, self-reported data on lifestyle factors are vulnerable to measurement error. Finally, due to the nature of observational study design, residual confounding cannot be completely ruled out and causality remains to be further explored.

In conclusion, our findings suggest that new-onset AF is associated with substantially higher risks of macro- and microvascular complications among patients with T2D. The findings highlight that the strategies for AF prevention are of particular relevance, given that most countries are encumbered by a substantial burden of diabetes complications and would benefit from strategies to eliminate the modifiable risk factors of AF.

Acknowledgments. This research was conducted using the UK Biobank under application no. 68307. The authors thank all the participants of UK Biobank and all the people involved in building the UK Biobank study.

Funding. A.P. is funded by grants from the National Natural Science Foundation of China (81930124 and 82021005) and the Fundamental Research Funds for the Central Universities (2021GCRC075). G.L. is funded by grants from the National Nature Science Foundation of China (82073554), the Hubei Province Science Fund for Distinguished Young Scholars (2021CFA048), and the Fundamental Research Funds for the Central Universities (2021GCRC076). T.G. is funded by grants from the China Postdoctoral Science Foundation (2021M691129).

The funders had no role in the design or conduct of the study; collection, management, analysis, or interpretation of data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.G. and A.P. designed the research. T.G., Y.W., and Q.L. analyzed the data. T.G., Y.W., G.L., and A.P. interpreted the statistical analysis. T.G. wrote the manuscript with critical input from all authors. All authors approved the manuscript. G.L. and A.P. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.