Abstract
Objective
To evaluate temporal associations of cancer with subsequent incident atrial fibrillation (AF) and temporal associations of AF with subsequent incident cancer, within 3, 3 to 12, and >12 months after index diagnosis.
Patients and Methods
We included 13,748 community-dwelling adults (mean age, 54 years) in the Atherosclerosis Risk in Communities study without cancer or AF histories at baseline (follow-up between January 1, 1987, and December 31, 2019). Atrial fibrillation was ascertained from electrocardiograms at study visits and health records. Cancer was ascertained via linkage with state registries and health records. We estimated associations of cancer with AF risk and AF with cancer risk by time since diagnosis using Cox regression, adjusting for shared risk factors and other cardiovascular diseases.
Results
In 3909 adults, cancer was diagnosed before AF. Atrial fibrillation risk was the highest within 3 months after cancer diagnosis (hazard ratio [HR], 11.71; 95% CI, 9.52-14.41), followed by 3 to 12 months (HR, 2.07; 95% CI, 1.54-2.80) and >12 months (HR, 1.46; 95% CI, 1.29-1.64). In 1973 adults, AF was diagnosed before cancer. Cancer risk was the highest within 3 months of AF diagnosis (HR, 2.24; 95% CI, 1.47-3.41), followed by 3 to 12 months (HR, 1.28; 95% CI, 0.91-1.80) and >12 months (HR, 1.09; 95% CI, 0.91-1.29).
Conclusion
In adult cancer patients, AF risk is the highest within 3 months after diagnosis and remains significantly elevated throughout survivorship but could be due to detection bias. Cancer risk is strongest within 3 months of AF diagnosis but significantly attenuated over time, suggesting detection bias and reverse causation.
Cancer is associated with a higher risk of atrial fibrillation (AF).1 Both, cancer and AF, share several common etiologic pathways and common risk factors.1, 2, 3 Systemic inflammation and oxidative stress can promote the development of both AF and cancer.3 Several markers of inflammation such as cytokines, chemokines, and systemic acute phase proteins such as C-reactive protein have been associated with an increased risk of both AF and cancer.2,3 Cardiometabolic and lifestyle risk factors such as obesity, smoking, malnutrition, and diabetes mellitus are also associated with greater incidence of both AF and cancer.2,3 Several treatment options for cancer might portend greater risk of AF.2,3 Chemotherapeutic agents, targeted therapy (such as immune checkpoint inhibitors), and radiation therapy are cardiotoxic and might contribute to a higher AF risk in cancer patients.2,3 Surgical cancer management could lead to increased physical and emotional stress, autonomic dysregulation, and electrolyte/fluid imbalance, which might precipitate AF.3 Greater incidence of AF in cancer patients could also be due to detection bias from increased contact with the health care system after a cancer diagnosis.
There is also emerging evidence of a higher risk of cancer in patients with AF.4, 5, 6, 7 However, previous epidemiologic studies reporting an association of AF with higher cancer incidence have inadequately addressed the potential for detection bias or reverse causation by design, inadequately controlled for confounders, focused on all cancers rather than specific cancers, or focused on a specific population limiting generalizability.4, 5, 6, 7, 8 There is a paucity of rigorous, prospective evidence of time-varying nature of associations between cancer and AF. To address this gap, we evaluated temporal associations between cancer and AF over 3 prespecified periods after index diagnosis; namely, within 3, 3 to 12, and >12 months after index diagnosis, in a community-dwelling cohort of White or Black adults with >25 years of surveillance data from the Atherosclerosis Risk in Communities (ARIC) study. Additionally, we evaluated the associations of specific cancers with incident AF and, conversely, the association of AF with the risk of specific cancers.
Patients and Methods
Study Population
The ARIC study (RRID:SCR_021769) is a prospective, observational cohort study dedicated to cardiovascular research. The study rationale and design have been described in detail previously.9 The ARIC study enrolled 15,792 participants from 4 communities in the United States including Jackson, MS; Minneapolis suburbs, MN; Forsyth County, NC; and Washington County, MD, at the baseline visit in 1987-1989. The institutional review board at each participating center reviewed and approved study protocols. Informed consent was obtained from all participants at the time of enrollment.9 Since the baseline examination, participants completed up to 8 additional visits, the most recent being in 2022 (visit 9). Follow-up also occurred annually or semiannually by telephone to maintain contact and to assess health status of the cohort.
At baseline, 15,653 ARIC study participants consented to participating in chronic disease research, including on cancer.10 From these, we excluded individuals who did not identify as Black or White, Blacks from Minneapolis suburbs or Washington County (due to small numbers), those with missing or prevalent cancer or AF at baseline, and those with missing covariate data. Our final analytic cohort included 13,748 individuals (follow-up between January 1, 1987, and December 31, 2019) (Figure).
Figure.
Flow diagram for study participant selection: the ARIC Study, 1987-2018. ECG, electrocardiogram.
Ascertainment of AF
Incident AF cases were ascertained at study visits and by surveillance during continuous follow-up through 2019. As previously described, at each study visit, a supine 12-lead resting electrocardiogram (ECG) was recorded using MAC PC Personal Cardiographs (Marquette Electronics). The ECGs were interpreted by ARIC-certified cardiologists at a single reading center at Wake Forest University, to confirm the AF diagnosis.11 During follow-up, new hospitalizations or deaths were identified by annual or semiannual telephone interviews and by searching local hospital records and the National Death Index. We defined AF as presence of International Classification of Diseases (ICD)-9 code 427.31 or ICD-10 code I48 in any position. For any hospitalization with open heart surgery, an AF discharge code was not considered an event. A validation study comprising a physician review of hospital discharge summaries in a random subset of the ARIC participants found high sensitivity and specificity of 84% and 98%, respectively, for specified ICD codes.12
Ascertainment of Cancer
All cases included in the main analyses were first primary invasive cancers (nonmelanoma skin cancer was not included). Cancer events were ascertained from baseline through 2015 using several sources, including linkage with cancer registries in the states from which the ARIC participants were recruited, interviews with participants or family members through follow-up calls or at study visits, and review of medical records, hospital discharge summary codes, and death certificates where cancer was listed as the underlying cause of death.10 For this study, we considered total cancer as well as the most common cancers, including postmenopausal breast, colorectal, lung, and prostate cancers, all cases of which were adjudicated by an expert panel. More details regarding the ARIC Cancer study and the high completeness of the data have been described previously.10
Measurement of Covariates
During study visits and follow-up phone interviews, trained study personnel obtained information on the covariates of interest through history, physical examination, and laboratory measurements.9 Age, race, smoking, and alcohol drinking status were self-reported. Body mass index (BMI) was calculated by dividing weight in kilograms by height in meters squared. Blood pressure was measured 3 times after 5 minutes of rest and reported as the average of the last 2 measurements. Antihypertensive medication use was self-reported and ascertained by medication reconciliation. Diabetes was defined as fasting glucose ≥126 mg/dL, nonfasting glucose ≥200 mg/dL, self-reported use of oral hypoglycemic agents or insulin, or self-reported physician diagnosis of diabetes. Cardiovascular disease (CVD) events, including myocardial infarction, heart failure, and stroke, were ascertained at study visits or by active follow-up through phone interviews, surveillance of medical records, and death certificates.9
Statistical Analyses
Demographics and age-adjusted, sex-adjusted, and race-adjusted cardiovascular risk factors at baseline (visit 1, 1987-1989) were compared between participants with and without incident cancer and participants with and without incident AF. We used t test and the χ2 test to compare continuous and categorical variables, respectively.
We constructed 3 serial, nested, multivariable Cox proportional hazards regression models to examine the following associations: (1) cancer (exposure) with incident AF (outcome) and (2) AF (exposure) with incident cancer (outcome). For the analysis of the association between cancer and subsequent AF, cancer was ascertained through 2015 and AF was ascertained through 2019; participants contributed person time at risk to the no cancer group from the date of visit 1 until an AF diagnosis, a cancer diagnosis, death, or the end of analytic follow-up. Participants diagnosed with cancer contributed person time at risk to the cancer group from the date of cancer diagnosis until an AF diagnosis, death, or the end of follow-up. The same strategy was used for the analysis of the association between AF and subsequent cancer, in which AF was ascertained through 2014 and cancer was ascertained through 2015. For both sets of analyses, the 3 models had similar set of covariates, which were selected a priori, based on a review of the existing literature. Model 1 was adjusted for age, sex, and race-center (categorical variable with 5 levels: Whites in Minneapolis suburbs, Washington county, and Forsyth field centers, and Blacks in Forsyth and Jackson centers). Model 2 was adjusted for the variables in model 1 plus shared risk factors between cancer and AF: education level, BMI, total cholesterol, high-density lipoprotein cholesterol, triglycerides, cholesterol-lowering medications, diabetes, systolic blood pressure, alcohol drinking status, hypertension-lowering medications, smoking status, pack-years smoked, and physical activity, all as time-varying covariates. Model 3 included model 2 covariates with additional adjustment for cardiovascular events (prevalent or incident myocardial infarction, heart failure, and stroke) as time-varying covariates over follow-up.
To study temporal associations, we performed analyses stratified over 3 prespecified periods over follow-up, after an index diagnosis of AF or cancer (<3, 3-12, and >12 months), following methodology from previous studies.4,5 We repeated primary analyses for the 4 most common cancer sites—lung (n=637), prostate (n=835), postmenopausal breast (n=557), and colorectal (n=411) cancers.
To address the possibility of detection bias (ie, greater surveillance of cancer patients than those without cancer leading to a greater opportunity to detect AF), we evaluated the association of specific cancer types with AF. If detection bias were acting, we would expect many cancers to be associated with a higher risk of AF, irrespective of whether there’s biological plausibility. For the analysis of the association between a specific cancer and subsequent AF, participants contributed person time at risk to the no cancer group from the date of visit 1 until an AF diagnosis, any cancer diagnosis, death, or the end of analytic follow-up. Participants diagnosed with a cancer at the specific site of interest contributed person time at risk to the specific cancer group from the date of that cancer diagnosis until an AF diagnosis, death, or the end of follow-up. A similar strategy was used for the analyses of association of AF with subsequent cancer at a specific site, wherein person time after diagnosis of cancer at other sites was administratively censored. Owing to lack of power, we were unable to study temporal associations by stratification of follow-up time for the 4 cancer sites.
To account for potential bias from excluding 577 participants (Figure) with missing covariate data, we repeated primary analyses using multiple imputation by chained equations, thereby increasing our analytic sample to 14,325 participants in this sensitivity analysis. We imputed 20 data sets and computed pooled hazard ratios (HR) and 95% CIs. Proportional hazards assumption was assessed with scaled Schoenfeld residuals. Statistical analyses were performed using SAS 9.4 (RRID:SCR_008567). All P values reported are 2-sided, and .05 was selected as the threshold for statistical significance.
Results
Among the 13,748 participants included in the study, mean age was 54 years, 26% were Black, and 45% were male. At baseline, burden of shared cardiometabolic risk factors was higher among individuals who were diagnosed with cancer over follow-up (Table 1) as well as individuals who diagnosed with AF over follow-up (Supplemental Table 1, available online at http://www.mcpiqojournal.org).
Table 1.
Participant Characteristics at Baseline (Visit 1, 1987-1989) Stratified by Incident Cancer Diagnosis Over Follow-up Through 2015 in the ARIC Study
| Risk factor | Incident cancer |
No cancer |
Pa |
|---|---|---|---|
| n=4319 | n=9429 | ||
| Age (y), mean ± SD | 54.8 ± 5.7 | 53.6 ± 5.7 | <.001 |
| Race, n (%) | .04 | ||
| Black | 24.9 | 26.6 | |
| White | 75.1 | 73.4 | |
| Sex, n (%) | <.001 | ||
| Female | 46.1 | 58.4 | |
| Male | 53.9 | 41.6 | |
| Clinical covariatesb | |||
| Body mass index (kg/m2) | 27.8 (0.1) | 27.7 (0.1) | .26 |
| Total cholesterol (mmol/L) | 5.5 (0.0) | 5.6 (0.0) | .04 |
| LDL cholesterol (mmol/L) | 3.5 (0.0) | 3.6 (0.0) | .01 |
| HDL cholesterol (mmol/L) | 1.3 (0.0) | 1.3 (0.0) | .12 |
| Triglycerides (mmol/L) | 1.5 (0.0) | 1.6 (0.0) | .23 |
| Cholesterol-lowering medications | 1.3 | 1.2 | .29 |
| Diabetes | 4.5 | 5.5 | <.001 |
| Systolic blood pressure (mm Hg) | 121.0 (0.3) | 121.2 (0.2) | .42 |
| Diastolic blood pressure (mm Hg) | 73.7 (0.1) | 73.8 (0.2) | .62 |
| Hypertension-lowering medications | 30.3 | 30.1 | .79 |
| Alcohol drinking statusb | <.001 | ||
| Nondrinker | 57.2 | 62.7 | |
| <2 Drinks/d | 34.3 | 31.2 | |
| ≥2 Drinks/d | 8.5 | 6.1 | |
| Education levelb | .46 | ||
| <High school graduate | 23.2 | 22.6 | |
| High school graduate/vocational school | 40.3 | 41.4 | |
| College/graduate school | 36.5 | 36.0 | |
| Physical activity levelb, n (%) | .21 | ||
| Ideal | 37.3 | 37.0 | |
| Intermediate | 23.4 | 25.1 | |
| Poor | 39.3 | 37.9 | |
| Smoking statusb, n (%) | <.001 | ||
| Current | 31.2 | 23.7 | |
| Former | 34.0 | 31.6 | |
| Never | 34.8 | 44.7 | |
| Pack-years of smoking in ever smokersb, median (IQR) | 30.0 (0.4) | 26.3 (0.3) | <.001 |
| Prevalent myocardial infarction | 3.9 | 4.1 | .62 |
| Prevalent heart failure | 4.5 | 4.5 | .92 |
| Prevalent stroke | 1.6 | 1.7 | .53 |
| Prevalent chronic kidney disease | 0.9 | 1.3 | .13 |
HDL, high-density lipoprotein; LDL, low-density lipoprotein.
P value for difference comparing those with incident cancer with those without cancer.
Age-adjusted, sex-adjusted, and race- adjusted estimates are shown as estimated % for categorical variables and estimated mean (SE) or median (IQR) for continuous variables.
Associations of Cancer With Incident AF
Over follow-up, 3909 adults were diagnosed with cancer before AF. After adjustment for potential confounders, participants with cancer were at a significantly higher AF risk (HR, 1.51; 95% CI, 1.38-1.64) compared with participants without cancer (Table 2). After stratifying follow-up time into 3 prespecified periods, AF risk was the highest within 3 months of a cancer diagnosis (HR, 11.71; 95% CI, 9.52-14.41), followed by 3 to 12 months (HR, 2.07; 95% CI, 1.54-2.80) and >12 months (HR, 1.46; 95% CI, 1.29-1.64), after adjustment for confounders (Table 3). In the full cohort, among cancer cases within the stratum with >12 months follow-up, median follow-up time (IQR) was 8.89 years (4.28-14.7 years). We found no significant differences in the adjusted associations stratified by race or sex.
Table 2.
Association of a First Primary Cancer With Incident AF: The ARIC Study, 1987-2019
| Estimand | Full cohort | White | Black | Female | Male |
|---|---|---|---|---|---|
| N | 13,748 | 10,172 | 3576 | 7494 | 6254 |
| Participants with a first primary cancer diagnosed before AF (n) | 3909 | 2907 | 1002 | 1834 | 2075 |
| AF incidence in participants with cancer | |||||
| Cases (n) | 734 | 614 | 120 | 319 | 415 |
| Total person-y | 30,489.9 | 23,706.3 | 6783.7 | 14,974.4 | 15,515.5 |
| Incidence rate (per 100,000 person-y) | 2407.4 | 2590.0 | 1769.0 | 2130.3 | 2674.7 |
| AF incidence in participants without cancer | |||||
| Cases (n) | 2384 | 1973 | 411 | 1180 | 1204 |
| Total person-y | 229,562.0 | 172,834.8 | 56,727.1 | 138,987.1 | 90,574.8 |
| Incidence rate (per 100,000 person-y) | 1038.5 | 1141.6 | 724.5 | 849.0 | 1329.3 |
| Model 1, HR (95% CI) | 2.09 (1.91-2.27) | 2.12 (1.89-2.30) | 2.01 (1.50-2.48) | 2.39 (2.06-2.75) | 1.97 (1.78-2.19) |
| P | <.001 | <.001 | <.001 | <.001 | <.001 |
| Model 2, HR (95% CI) | 1.84 (1.67-2.07) | 1.95 (1.68-2.23) | 1.88 (1.36-2.15) | 2.01 (1.82-2.48) | 1.72 (1.49-2.01) |
| P | <.001 | <.001 | <.001 | <.001 | <.001 |
| Model 3, HR (95% CI) | 1.51 (1.38-1.64) | 1.51 (1.37-1.66) | 1.51 (1.22-1.87) | 1.58 (1.39-1.79) | 1.46 (1.30-1.64) |
| P | <.001 | <.001 | <.001 | <.001 | <.001 |
Model 1: Cox regression adjusted for age (time-varying), sex, and race/center. Model 2: Additional adjustment for body mass index, total cholesterol, high-density lipoprotein cholesterol, triglycerides, cholesterol-lowering medications, diabetes, systolic blood pressure, hypertension-lowering medications, alcohol drinking status, education level, smoking status, pack-years smoked, and physical activity as time-varying covariates. Model 3: Additional adjustment for incident cardiovascular events (myocardial infarction, heart failure, and stroke) as time-varying covariates. Participants contributed person time at risk to the no cancer group from the date of visit 1 until an AF diagnosis, a cancer diagnosis, death, or the end of analytic follow-up. Participants diagnosed with cancer contributed person time at risk to the cancer group from the date of cancer diagnosis until an AF diagnosis, death, or the end of follow-up.
AF, atrial fibrillation; HR, hazard ratio.
Table 3.
Association of a First Primary Cancer With Incident AF Over 3 Periods After Cancer Diagnosis (<3, 3-12, and >12 Mo): The ARIC Study, 1987-2019
| Estimand | <3 Mo | 3-12 Mo | >12 Mo |
|---|---|---|---|
| Full cohort (N=13,748) | |||
| First primary cancer (n) | 3909 | 3412 | 2940 |
| AF incidence in participants with cancer | |||
| Cases (n) | 111 | 57 | 566 |
| Total person-years | 39.6 | 265.3 | 30,185.0 |
| Incidence rate (per 100,000 person-y) | 280,165.5 | 21,485.1 | 1875.1 |
| Model 1, HR (95% CI) | 13.79 (11.21-16.95) | 2.42 (1.79-3.26) | 1.65 (1.46-1.86) |
| P | <.001 | <.001 | <.001 |
| Model 2, HR (95% CI) | 12.54 (10.19-15.42) | 2.24 (1.66-3.03) | 1.63 (1.44-1.84) |
| P | <.001 | <.001 | <.001 |
| Model 3, HR (95% CI) | 11.71 (9.52-14.41) | 2.07 (1.54-2.80) | 1.46 (1.29-1.64) |
| P | <.001 | <.001 | <.001 |
| Whites (n=10,172) | |||
| First primary cancer (n) | 2907 | 2553 | 2208 |
| AF incidence in participants with cancer | |||
| Cases (n) | 97 | 45 | 472 |
| Total person-years | 30.1 | 195.7 | 23,480.5 |
| Incidence rate (per 100,000 person-y) | 322,582.6 | 22,991.0 | 2010.2 |
| Model 1, HR (95% CI) | 14.92 (11.95-18.62) | 2.32 (1.65-3.26) | 1.66 (1.45-1.90) |
| P | <.001 | <.001 | <.001 |
| Model 2, HR (95% CI) | 13.43 (10.75-16.77) | 2.13 (1.52-3.00) | 1.63 (1.42-1.86) |
| P | <.001 | <.001 | <.001 |
| Model 3, HR (95% CI) | 12.68 (10.15-15.85) | 1.98 (1.40-2.78) | 1.47 (1.28-1.68) |
| P | <.001 | <.001 | <.001 |
| Black (n=3576) | |||
| First primary cancer (n) | 1002 | 859 | 732 |
| AF incidence in participants with cancer | |||
| Cases (n) | 14 | 12 | 94 |
| Total person-y | 9.5 | 69.6 | 6704.5 |
| Incidence rate (per 100,000 person-y) | NAa | NAa | 1402.0 |
| Model 1, HR (95% CI) | 12.65 (5.03-16.04) | 2.91 (1.55-5.48) | 1.96 (1.25-2.21) |
| P | <.001 | .001 | .001 |
| Model 2, HR (95% CI) | 11.95 (4.68-14.92) | 2.88 (1.94-5.12) | 1.78 (1.28-2.26) |
| P | <.001 | <.001 | <.001 |
| Model 3, HR (95% CI) | 11.22 (9.25-13.58) | 2.22 (1.71-2.90) | 1.54 (1.40-1.70) |
| P | <.001 | <.001 | <.001 |
| Females (n=7494) | |||
| First primary cancer (n) | 1834 | 1612 | 1402 |
| AF incidence in participants with cancer | |||
| Cases (n) | 44 | 19 | 256 |
| Total person-y | 17.7 | 117.1 | 14,839.6 |
| Incidence rate (per 100,000 person-y) | 248,661.6 | 16,221.6 | 1725.1 |
| Model 1, HR (95% CI) | 15.34 (10.97-21.46) | 2.44 (1.32-3.80) | 2.00 (1.66-2.40) |
| P | <.001 | .003 | <.001 |
| Model 2, HR (95% CI) | 13.18 (9.41-18.45) | 2.01 (1.15-3.33) | 1.91 (1.59-2.29) |
| P | <.001 | .01 | <.001 |
| Model 3, HR (95% CI) | 11.33 (8.07-15.91) | 1.89 (1.02-2.87) | 1.73 (1.44-2.08) |
| P | <.001 | .03 | <.001 |
| Males (n=6254) | |||
| First primary cancer (n) | 2075 | 1800 | 1538 |
| AF incidence in participants with cancer | |||
| Cases (n) | 67 | 38 | 310 |
| Total person-years | 21.9 | 148.2 | 15,345.4 |
| Incidence rate (per 100,000 person-y) | 305,591.3 | 25,645.8 | 2020.1 |
| Model 1, HR (95% CI) | 13.04 (10.03-16.96) | 2.52 (1.75-3.63) | 1.48 (1.26-1.74) |
| P | <.001 | <.001 | <.001 |
| Model 2, HR (95% CI) | 12.19 (9.37-15.87) | 2.39 (1.66-3.45) | 1.48 (1.25-1.74) |
| P | <.001 | <.001 | <.001 |
| Model 3, HR (95% CI) | 11.85 (9.10-15.44) | 2.29 (1.59-3.30) | 1.32 (1.12-1.55) |
| P | <.001 | <.001 | .001 |
Model 1: Cox regression adjusted for age (time-varying), sex, and race/center. Model 2: Additional adjustment for body mass index, total cholesterol, high-density lipoprotein cholesterol, triglycerides, cholesterol-lowering medications, diabetes, systolic blood pressure, hypertension-lowering medications, alcohol drinking status, education level, smoking status, pack-years smoked, and physical activity as time-varying covariates. Model 3: Additional adjustment for incident cardiovascular events (myocardial infarction, heart failure, and stroke) as time-varying covariates. In the full cohort, among cancer cases within the stratum with >12-mo follow-up, median follow-up time (IQR) was 8.89 y (4.28-14.7 y).
AF, atrial fibrillation; HR, hazard ratio.
As per data user agreement, numbers cannot be displayed for n<6 and rates cannot be displayed for n<16.
In analyses for specific cancer sites, over the total length of follow-up between 1987 and 2019, compared with participants without cancer, AF risk was significantly higher in those with lung cancer (HR, 3.75; 95% CI, 2.83-4.54) (Supplemental Table 2, available online at http://www.mcpiqojournal.org) and postmenopausal breast cancer (HR, 1.66; 95% CI, 1.30-2.12) (Supplemental Table 3, available online at http://www.mcpiqojournal.org), but not in those with colorectal cancer (HR, 1.08; 95% CI, 0.78-1.44) (Supplemental Table 4, available online at http://www.mcpiqojournal.org) and prostate cancer (HR, 1.08; 95% CI, 0.67-1.53) (Supplemental Table 5, available online at http://www.mcpiqojournal.org).
Associations of AF With Incident Cancer
Over follow-up, 1973 adults were diagnosed with AF before cancer. Their unadjusted HR of cancer was 2777.3 per 100,000 person-years. After adjustment for potential confounders, participants with AF were at a significantly higher cancer risk (HR, 1.24; 95% CI, 1.10-1.41) compared with individuals without AF (Table 4). After stratifying follow-up time into 3 prespecified periods, cancer risk was the highest within 3 months of an AF diagnosis (HR, 2.24; 95% CI, 1.47-3.41), followed by 3 to 12 months (HR, 1.28; 95% CI, 0.91-1.80) and >12 months (HR, 1.09; 95% CI, 0.91-1.29), after adjustment for confounders (Table 5). In the full cohort, among AF cases within the stratum with >12 months follow-up, median follow-up time (IQR) was 5.59 years (2.97-10.26 years). We found no significant differences in adjusted associations stratified by race or sex.
Table 4.
Association of AF With Incident Cancer: The ARIC Study, 1987-2015
| Estimand | Full cohort | White | Black | Female | Male |
|---|---|---|---|---|---|
| N | 13,748 | 10,172 | 3576 | 7494 | 6254 |
| Participants with AF (n) | 1973 | 1605 | 368 | 950 | 1023 |
| First primary cancer incidence in participants with AF | |||||
| Cases (n) | 311 | 259 | 52 | 107 | 204 |
| Total person-years | 11,197.8 | 9606.6 | 1591.2 | 5205.2 | 5992.6 |
| Incidence rate (per 100,000 person-y) | 2777.3 | 2696.1 | 3268.0 | 2055.6 | 3404.2 |
| First primary cancer incidence in participants without AF | |||||
| Cases (n) | 4008 | 2986 | 1022 | 1885 | 2123 |
| Total person-years | 241,229.7 | 178,527.8 | 62,701.8 | 141,764.0 | 99,465.7 |
| Incidence rate (per 100,000 person-y) | 1661.5 | 1672.6 | 1629.9 | 1329.7 | 2134.4 |
| Model 1, HR (95% CI) | 1.48 (1.32-1.67) | 1.29 (1.11-1.49) | 1.65 (1.20-2.25) | 1.30 (1.03-1.65) | 1.54 (1.15-1.59) |
| P | <.001 | .001 | .002 | .03 | <.001 |
| Model 2, HR (95% CI) | 1.37 (1.22-1.55) | 1.35 (1.09-1.48) | 1.55 (1.13-2.13) | 1.27 (0.93-1.49) | 1.48 (1.08-1.51) |
| P | <.001 | .01 | .01 | .19 | .004 |
| Model 3, HR (95% CI) | 1.24 (1.10-1.41) | 1.21 (1.05-1.39) | 1.45 (1.08-1.96) | 1.11 (0.90-1.37) | 1.34 (1.16-1.59) |
| P | <.001 | .01 | .02 | .31 | <.001 |
Model 1: Cox regression adjusted for age (time-varying), sex, and race/center. Model 2: Additional adjustment for body mass index, total cholesterol, high-density lipoprotein cholesterol, triglycerides, cholesterol-lowering medications, diabetes, systolic blood pressure, hypertension-lowering medications, alcohol drinking status, education level, smoking status, pack-years smoked, and physical activity as time-varying covariates. Model 3: Additional adjustment for incident cardiovascular events (myocardial infarction, heart failure, and stroke) as time-varying covariates. Participants contributed person time at risk to the no AF group from the date of visit 1 until an AF diagnosis, a cancer diagnosis, death, or the end of analytic follow-up. Participants diagnosed with AF contributed person time at risk to the AF group from the date of AF diagnosis until a cancer diagnosis, death, or the end of follow-up.
AF, atrial fibrillation; HR, hazard ratio.
Table 5.
Association of AF With Incident Cancer over 3 Periods After AF Diagnosis (<3, 3-12, and >12 Mo): The Atherosclerosis Risk In Communities Study, 1987-2015
| Estimand | <3 Mo | 3-12 Mo | >12 Mo |
|---|---|---|---|
| Full cohort (N=13,748) | |||
| Incident AF (n) | 1973 | 1704 | 1507 |
| First primary cancer incidence in participants with AF (n) | |||
| Cases (n) | 22 | 34 | 255 |
| Total person-years | 13.7 | 122.2 | 11,061.9 |
| Incidence rate (per 100,000 person-y) | 160,293.2 | 27,829.8 | 2305.2 |
| Model 1 HR (95% CI) | 2.57 (1.69-3.91) | 1.47 (1.05-2.06) | 1.35 (1.16-1.51) |
| P | <.001 | .03 | <.001 |
| Model 2 HR (95% CI) | 2.38 (1.56-3.61) | 1.36 (0.97-1.71) | 1.22 (1.07-1.42) |
| P | <.001 | .06 | .007 |
| Model 3 HR (95% CI) | 2.24 (1.47-3.41) | 1.28 (0.91-1.80) | 1.09 (0.91-1.29) |
| P | <.001 | .25 | .19 |
| White (n=10,172) | |||
| Incident AF (n) | 1605 | 1407 | 1252 |
| First primary cancer incidence in participants with AF (n) | |||
| Cases (n) | 16 | 28 | 215 |
| Total person-y | 9.8 | 94.7 | 9502.0 |
| Incidence rate (per 100,000 person-y) | 162,468.7 | 29,551.8 | 2262.7 |
| Model 1 HR (95% CI) | 1.76 (0.95-3.28) | 1.43 (0.94-2.18) | 1.25 (1.06-1.47) |
| P | .07 | .09 | .01 |
| Model 2 HR (95% CI) | 1.59 (0.86-2.97) | 1.30 (0.85-1.98) | 1.15 (0.98-1.36) |
| P | .14 | .22 | .09 |
| Model 3 HR (95% CI) | 1.50 (0.80-2.80) | 1.22 (0.80-1.87) | 1.08 (0.91-1.28) |
| P | .20 | .35 | .37 |
| Black (n=3576) | |||
| Incident AF (n) | 368 | 297 | 255 |
| First primary cancer incidence in participants with AF (n) | |||
| Cases (n) | 6 | 6 | 40 |
| Total person-years | 3.9 | 27.4 | 1559.9 |
| Incidence rate (per 100,000 person-y) | NAa | NAa | 2564.3 |
| Model 1 HR (95% CI) | 3.73 (1.55-9.00) | 1.40 (0.58-3.38) | 1.55 (1.09-2.22) |
| P | .003 | .45 | .02 |
| Model 2 HR (95% CI) | 3.52 (1.46-8.50) | 1.33 (0.55-3.20) | 1.46 (1.02-2.10) |
| P | .01 | .53 | .04 |
| Model 3 HR (95% CI) | 3.23 (1.33-7.82) | 1.21 (0.50-2.93) | 1.36 (0.94-1.96) |
| P | .01 | .67 | .11 |
| Females (n=7,494) | |||
| Incident AF (n) | 950 | 821 | 717 |
| First primary cancer incidence in participants with AF (n) | |||
| Cases (n) | NAa | 13 | 89 |
| Total person-y | NAa | 65.1 | 5133.4 |
| Incidence rate (per 100,000 person-y) | NA∗ | NAa | 1733.8 |
| Model 1 HR (95% CI) | 1.30 (0.42-4.05) | 1.58 (0.85-2.94) | 1.27 (0.98-1.64) |
| P | .65 | .15 | .08 |
| Model 2 HR (95% CI) | 1.15 (0.37-3.56) | 1.40 (0.75-2.62) | 1.14 (0.88-1.48) |
| P | .82 | .29 | .31 |
| Model 3 HR (95% CI) | 1.03 (0.33-3.19) | 1.26 (0.67-2.35) | 1.03 (0.79-1.34) |
| P | .96 | .47 | .84 |
| Males (n=6254) | |||
| Incident AF (n) | 1023 | 883 | 790 |
| First primary cancer incidence in participants with AF (n) | |||
| Cases (n) | 17 | 21 | 166 |
| Total person-y | 7.0 | 57.0 | 5928.5 |
| Incidence rate (per 100,000 person-y) | 241,605.1 | 36,821.3 | 2800.0 |
| Model 1 HR (95% CI) | 2.55 (1.45-4.51) | 1.36 (0.84-2.19) | 1.29 (1.08-1.55) |
| P | .001 | .21 | .01 |
| Model 2 HR (95% CI) | 2.41 (1.36-4.25) | 1.28 (0.79-2.06) | 1.23 (1.02-1.47) |
| P | .002 | .32 | .03 |
| Model 3 HR (95% CI) | 2.33 (1.32-4.13) | 1.24 (0.76-2.00) | 1.19 (0.98-1.44) |
| P | .004 | .39 | .08 |
Model 1: Cox regression adjusted for age (time-varying), sex, and race/center. Model 2: Additional adjustment for body mass index, total cholesterol, high-density lipoprotein cholesterol, triglycerides, cholesterol-lowering medications, diabetes, systolic blood pressure, hypertension-lowering medications, alcohol drinking status, education level, smoking status, pack-years smoked, and physical activity as time-varying covariates. Model 3: Additional adjustment for incident cardiovascular events (myocardial infarction, heart failure, and stroke) as time-varying covariates. In the full cohort, among atrial fibrillation cases within the stratum with >12-mo follow-up, median follow-up time (IQR) was 5.59 years (2.97-10.26 years).
AF, atrial fibrillation; HR, hazard ratio.
As per data user agreement, numbers cannot be displayed for n<6 and rates cannot be displayed for n<16.
Over the total length of follow-up between 1987 and 2015, compared with participants without AF, participants with AF were not at a higher risk of lung (HR, 1.21; 95% CI, 0.85-1.72), breast (HR, 1.02; 95% CI, 0.62-1.69), colorectal (HR, 0.95; 95% CI, 0.59-1.54), or prostate (HR, 1.01; 95% CI, 0.72-1.44) cancers (Supplemental Tables 6-9, available online at http://www.mcpiqojournal.org). We found no significant differences in adjusted associations stratified by race or sex.
In sensitivity analyses accounting for missing data using multiple imputations, our findings were similar to findings from primary analyses in magnitude and direction (Supplemental Tables 10-12, available online at http://www.mcpiqojournal.org).
Discussion
In a community-dwelling cohort of Black or White adults with over 30 years of follow-up, participants diagnosed with a first primary cancer had a significantly higher risk of AF independent of demographics, shared cardiometabolic and lifestyle risk factors, and other CVD. This elevated AF risk was the highest within 3 months of cancer diagnosis but persisted over a longer follow-up period, beyond 12 months of cancer survivorship. In this cohort, AF risk was significantly higher in participants with lung and postmenopausal breast cancer compared with participants without cancer; AF risk was not higher in participants with colorectal or prostate cancer, suggesting that the elevated AF risk in patients with cancer is not spurious and due to detection bias. Over the total follow-up period, patients with AF had a significantly higher risk of cancer, after adjustment for demographics and shared risk factors. This elevated risk was significant within 3 months after diagnosing AF but greatly diminished over a longer follow-up period as well as after adjustment for shared cardiometabolic and lifestyle risk factors. These findings suggest that the association of AF with greater risk of subsequent cancer is likely explained by occult cancer that developed before AF, that is, reverse causation.
Previous studies have reported an association between cancer and a higher risk for AF, independent of shared cardiometabolic and lifestyle risk factors.1 In a large Korean administrative data study, the risk of AF was the highest during the first 90 days after a cancer diagnosis and persisted over 5 years of follow-up.13 That study, however, was unable to control for several relevant cardiometabolic risk factors, which were unavailable in administrative data.13 Similarly, in 2 Danish studies, AF risk was higher in cancer patients, but adjustment for cardiometabolic risk factors could not be made due to the administrative nature of their data source.14,15 In the Women’s Health Study, in a large cohort of White women, AF risk was higher within 3 months of a new cancer diagnosis, but not beyond 3 months. However, in this study, incidence of AF in cancer patients was lower than expected.4 In our study with rigorously collected risk factor data and large number of AF and cancer events, we confirmed that the greater risk of AF persists beyond 12 months of a cancer diagnosis independent of shared lifestyle and cardiometabolic risk factors. This association could partially be explained by detection bias due to AF screening. Several biological mechanisms, including shared risk factors between AF and cancer, inflammation, and cardiotoxic cancer management strategies, might also partially explain this association.1, 2, 3,16, 17, 18 However, the clinical net benefit of screening and treating AF in new cancer patients remains ambiguous.3 Previous studies have reported a significantly elevated risk of stroke, heart failure, and pulmonary embolism in cancer patients, which are common outcomes of AF.19, 20, 21 More research is needed to evaluate whether AF diagnosed within 90 days to 1 year of a cancer diagnosis may be associated with a higher risk of CVD in new cancer patients.3 Future research should identify opportunities for long-term AF risk mitigation in cancer survivors.
Previous studies have reported a significantly higher risk of cancer in patients with AF.4,5 In the Women’s Health Study, a large cohort of predominantly White women, new-onset AF was associated with a significantly higher risk of developing cancer, independent of shared risk factors.4 This association was greatest within 3 months of diagnosing AF, reduced by half over a longer follow-up period but remained statistically significant. However, when considering specific cancer sites, the association between AF and incident cancer persisted only for colorectal cancer and not lung or breast cancer, suggesting the potential for detection bias.4 In a large Danish cohort, there was a similar trend of associations.5 The risk of cancer was the highest within 90 days of diagnosing AF and decreased with time, but remained significant. However, in this study, long-term risk of cancer might be influenced by a higher rate of occult cancer detection immediately (within 90 days) following a diagnosis of AF.5,6 In our study, we observed a similar trend of higher rate of cancer diagnosis within 3 months of an AF diagnosis, which decreased over time, and was no longer significant beyond 1 year after an AF diagnosis. This suggests that the association of AF and higher incidence of cancer is likely due to underlying, undiagnosed cancer or residual confounding by shared risk factors.
In our study, lung cancer (vs no cancer) in the entire cohort and postmenopausal breast cancer (vs no cancer) in women were independently associated with higher AF risk. These findings align with previous administrative data studies unable to rigorously adjust for shared risk factors as potential confounders.13,15 These associations may be partially explained by inflammation, oxidative stress, specific cancer treatment strategies, and lower lung function.3,16, 17, 18, 19,22, 23, 24, 25 In our study, colorectal cancer was not significantly associated with higher AF risk after adjustment for confounders. Previous administrative studies have reported a small yet significant association of colorectal cancer with higher AF risk; however, these studies did not adjust for several shared risk factors.13,15 Associations of AF (as an exposure) with the risk of lung, breast, colorectal, or prostate cancer were greatly attenuated and nonsignificant. In the Women’s Health Study, AF was not associated with a higher risk of lung or breast cancer; however, AF was associated with a higher risk of colorectal cancer, which might be explained by detection bias.4
Strengths of this study include a large, prospective cohort of community-dwelling adults with rigorous event ascertainment and surveillance. There are several limitations. First, despite a thorough follow-up, AF ascertainment relied on study visit ECG screening and medical records, which could lead to under diagnosis of AF. Second, there may be selection bias due to missing data, although missing data was < 5%. Third, this study included Black or White adults; so, findings may not be generalizable to other race and ethnic groups. Fourth, in analyses for specific cancer sites, we did not have adequate power to study the temporal difference in associations by varying lengths of follow-up. Future studies should attempt to validate our findings across the 3 follow-up periods (up to 3, 3-12, and >12 months after index diagnosis) for different cancer sites. Fifth, cancer patients may have an increased contact with the health care system, which may lead to detection bias; however, we found that lung and breast cancers, but not colorectal and prostate cancers, were significantly associated with incident AF, suggesting that our findings are not spurious and due to detection bias. Sixth, residual confounding might still exist because of potential unmeasured confounders such as obstructive sleep apnea, a family history, other lifestyle factors, and genetic factors.
Conclusion
Risk of AF in cancer patients is high throughout survivorship but markedly high within 3 months of a new diagnosis of primary cancer. Future research should identify optimal AF screening and treatment strategies in new cancer patients and evaluate prevention strategies for long-term AF risk mitigation in middle-aged and older cancer survivors. Risk of cancer in patients with AF is the strongest within 3 months of AF diagnosis and is significantly attenuated over time, which suggests that this association is likely explained by reverse causation due to occult cancer.
Potential Competing Interests
Dr Joshu reports grants from National Cancer Institute, Genentech, American Cancer Society, and The Ralph Lauren Corporate Foundation. The other authors report no competing interests.
Ethics Statement
The institutional review board at each participating center reviewed and approved study protocols. Informed consent was obtained from all participants at the time of enrollment.
Acknowledgments
The authors thank the staff and participants of the ARIC study for their important contributions. Cancer data were provided by the Maryland Cancer Registry, Center for Cancer Prevention and Control, and Maryland Department of Health, with funding from the State of Maryland and the Maryland Cigarette Restitution Fund. The collection and availability of cancer registry data are also supported by the Cooperative Agreement NU58DP006333, funded by the Centers for Disease Control and Prevention. This article contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Department of Health and Human Services.
Footnotes
Grant Support: The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services (grant numbers 75N92022D00001, 75N92022D00002, 75N92022D00003, 75N92022D00004, and 75N92022D00005). Studies on cancer in ARIC are also supported by the National Cancer Institute (grant number U01CA164975). Dr Shenoy is supported by grant K23HL132011. Dr Alonso is supported by grant K24HL148521. Dr Chen is supported by grant R01HL141288, R01HL126637, RF1NS127266, R01HL158022, R01AG075883, and K24HL155813. Dr Platz and Dr Joshu are supported by grant U01CA164975.
Supplemental material can be found online at http://www.mcpiqojournal.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data.
Supplemental Online Material
References
- 1.Fradley M.G., Beckie T.M., Brown S.A., et al. Recognition, prevention, and management of arrhythmias and autonomic disorders in cardio-oncology: a scientific statement from the American Heart Association. Circulation. 2021;144(3):e41–e55. doi: 10.1161/CIR.0000000000000986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lateef N., Kapoor V., Ahsan M.J., et al. Atrial fibrillation and cancer; understanding the mysterious relationship through a systematic review. J Community Hosp Intern Med Perspect. 2020;10(2):127–132. doi: 10.1080/20009666.2020.1726571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Siontis K.C., Noseworthy P.A., Ruddy K.J. Atrial fibrillation and cancer: where biology and epidemiology intertwine. JACC CardioOncol. 2021;3(2):233–235. doi: 10.1016/j.jaccao.2021.04.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Conen D., Wong J.A., Sandhu R.K., et al. Risk of malignant cancer among women with new-onset atrial fibrillation. JAMA Cardiol. 2016;1(4):389–396. doi: 10.1001/jamacardio.2016.0280. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Vinter N., Christesen A.M.S., Fenger-Grøn M., Tjønneland A., Frost L. Atrial fibrillation and risk of cancer: a Danish population-based cohort study. J Am Heart Assoc. 2018;7(17) doi: 10.1161/JAHA.118.009543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.O’Neal W.T., Lakoski S.G., Qureshi W., et al. Relation between cancer and atrial fibrillation (from the Reasons for Geographic and Racial Differences in Stroke Study) Am J Cardiol. 2015;115(8):1090–1094. doi: 10.1016/j.amjcard.2015.01.540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Nouraie M., Kansal V., Belfonte C., et al. Atrial fibrillation and colonic neoplasia in African Americans. PLoS One. 2015;10(8) doi: 10.1371/journal.pone.0135609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Saliba W., Rennert H.S., Gronich N., Gruber S.B., Rennert G. Association of atrial fibrillation and cancer: Analysis from two large population-based case-control studies. PLoS One. 2018;13(1) doi: 10.1371/journal.pone.0190324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wright J.D., Folsom A.R., Coresh J., et al. The ARIC (Atherosclerosis Risk in Communities) study: JACC Focus Seminar 3/8. J Am Coll Cardiol. 2021;77(23):2939–2959. doi: 10.1016/j.jacc.2021.04.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Joshu C.E., Barber J.R., Coresh J., et al. Enhancing the Infrastructure of the Atherosclerosis Risk in Communities (ARIC) study for cancer epidemiology research: ARIC cancer. Cancer Epidemiol Biomarkers Prev. 2018;27(3):295–305. doi: 10.1158/1055-9965.EPI-17-0696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Soliman E.Z., Prineas R.J., Case L.D., Zhang Z.M., Goff D.C., Jr. Ethnic distribution of ECG predictors of atrial fibrillation and its impact on understanding the ethnic distribution of ischemic stroke in the Atherosclerosis Risk in Communities (ARIC) study. Stroke. 2009;40(4):1204–1211. doi: 10.1161/STROKEAHA.108.534735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Alonso A., Agarwal S.K., Soliman E.Z., et al. Incidence of atrial fibrillation in whites and African-Americans: the Atherosclerosis Risk in Communities (ARIC) study. Am Heart J. 2009;158(1):111–117. doi: 10.1016/j.ahj.2009.05.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yun J.P., Choi E.K., Han K.D., et al. Risk of atrial fibrillation according to cancer type: a nationwide population-based study. JACC CardioOncol. 2021;3(2):221–232. doi: 10.1016/j.jaccao.2021.03.00. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Erichsen R., Christiansen C.F., Mehnert F., Weiss N.S., Baron J.A., Sørensen H.T. Colorectal cancer and risk of atrial fibrillation and flutter: a population-based case-control study. Intern Emerg Med. 2012;7(5):431–438. doi: 10.1007/s11739-011-0701-9. [DOI] [PubMed] [Google Scholar]
- 15.Jakobsen C.B., Lamberts M., Carlson N., et al. Incidence of atrial fibrillation in different major cancer subtypes: a nationwide population-based 12 year follow up study. BMC Cancer. 2019;19(1):1105. doi: 10.1186/s12885-019-6314-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Yang X., Li X., Yuan M., et al. Anticancer therapy-induced atrial fibrillation: electrophysiology and related mechanisms. Front Pharmacol. 2018;9:1058. doi: 10.3389/fphar.2018.01058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Apte N., Dherange P., Mustafa U., et al. Cancer radiation therapy may be associated with atrial fibrillation. Front Cardiovasc Med. 2021;8 doi: 10.3389/fcvm.2021.610915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Siontis K.C., Gersh B.J., Weston S.A., et al. Association of new-onset atrial fibrillation after noncardiac surgery with subsequent stroke and transient ischemic attack. JAMA. 2020;324(9):871–878. doi: 10.1001/jama.2020.12518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Farmakis D., Parissis J., Filippatos G. Insights into onco-cardiology: atrial fibrillation in cancer. J Am Coll Cardiol. 2014;63(10):945–953. doi: 10.1016/j.jacc.2013.11.026. [DOI] [PubMed] [Google Scholar]
- 20.Florido R., Daya N.R., Ndumele C.E., et al. Cardiovascular disease risk among cancer survivors: the Atherosclerosis Risk in Communities (ARIC) study. J Am Coll Cardiol. 2022;80(1):22–32. doi: 10.1016/j.jacc.2022.04.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Paterson D.I., Wiebe N., Cheung W.Y., et al. Incident cardiovascular disease among adults with cancer: a population-based cohort study. JACC CardioOncol. 2022;4(1):85–94. doi: 10.1016/j.jaccao.2022.01.100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Koene R.J., Prizment A.E., Blaes A., Konety S.H. Shared risk factors in cardiovascular disease and cancer. Circulation. 2016;133(11):1104–1114. doi: 10.1161/CIRCULATIONAHA.115.020406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kumar M., Lopetegui-Lia N., Malouf C.A., Almnajam M., Coll P.P., Kim A.S. Atrial fibrillation in older adults with cancer. J Geriatr Cardiol. 2022;19(1):1–8. doi: 10.11909/j.issn.1671-5411.2022.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Li J., Agarwal S.K., Alonso A., et al. Airflow obstruction, lung function, and incidence of atrial fibrillation: the Atherosclerosis Risk in Communities (ARIC) study. Circulation. 2014;129(9):971–980. doi: 10.1161/CIRCULATIONAHA.113.004050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Lee S.N., Ko S.H., Her S.H., et al. Association between lung function and the risk of atrial fibrillation in a nationwide population cohort study. Sci Rep. 2022;12(1):4007. doi: 10.1038/s41598-022-07534-4. [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.