Coronary artery calcium score and the long-term risk of atrial fibrillation in patients undergoing non-contrast cardiac computed tomography for suspected coronary artery disease: a Danish registry-based cohort study

Coronary artery calcium score and the long-term risk of atrial fibrillation in patients... Abstract Aims To examine the association between coronary artery calcium score (CACS) and risk of future atrial fibrillation (AF), and to estimate the predictive accuracy of CACS for AF development in patients undergoing non-contrast cardiac computed tomography (nCCT). Methods and results We conducted a registry-based cohort study of 27 962 patients suspected of having coronary artery disease and without history of AF who were identified in the Western Denmark Heart Registry. The patients underwent nCCT between 2010 and 2015 and were followed until 2016 (median 2.9 years). CACSs were determined using nCCT. We used Cox proportional hazards models to estimate hazard ratios (HR) with 95% confidence intervals (CI). A receiver operating characteristic (ROC) curve for AF was used to assess the predictive accuracy of CACS. Among the patients, 52% had a CACS of 0, 26% of 1–99, 13% of 100–399, 6% of 400–999, and 4% of ≥ 1000. AF occurred in 622 patients after nCCT, corresponding to an overall incidence rate of 7.5 (95% CI: 6.9–8.1) per 1000 person-years. After multivariable adjustment, the HRs (95% CIs) were (ref. CACS 0) CACS 1–99: 1.00 (0.80–1.25); CACS 100–399: 1.36 (1.06–1.74); CACS 400–999: 1.76 (1.33–2.35); and CACS ≥ 1000: 1.67 (1.20–2.34). An ROC curve showed an area under the curve of 0.68 (0.65–0.71) for the prediction of AF within one year after nCCT. Conclusion A high CACS is associated with a high risk of subsequent AF development and may have potential to guide future follow-ups for AF detection after CACS measurement in order to identify AF patients earlier. coronary artery calcium score , atrial fibrillation , cardiac computed tomography , coronary arteries , Western Denmark Heart Registry Introduction Atrial fibrillation (AF) is one of the most frequently occurring arrhythmias, and projections have shown that the number of adults with AF will increase. The prevalence of AF in Europe may reach 17.9 million people in 20601 and the prevalence in the USA may reach 15.9 million people in 2050.2 Several risk factors for AF have been identified, many of which are also risk factors for coronary calcification, including increasing age, sex, systolic blood pressure, body mass index (BMI), hypertension, and diabetes mellitus.3–5 Use of non-contrast cardiac computed tomography (nCCT) and the Agatston score method6 for measuring the coronary artery calcium score (CACS) has identified CACS as a compelling predictor of future cardiovascular events, including cardiovascular death, and all-cause mortality.7–9 CACS also has a potential as a biomarker for the development of non-cardiovascular events such as cancer, chronic kidney disease, chronic obstructive pulmonary disease, and hip fractures.10 Recently, CACS was introduced as a novel independent risk factor for AF based on results reported by the Multi-Ethnic Study of Atherosclerosis.11 Gaining more insight into the role of CACS in relation to the risk of future AF is of major clinical relevance. First, more accurate identification of individuals at high risk for AF may facilitate a stronger focus on disease prevention and, in particular, interventions indented to address modifiable risk factors. Secondly, stroke is a well-known complication of AF that can be effectively prevented using oral anticoagulant therapy.12 However, as many as 40% of patients with AF-related strokes are not diagnosed with AF until they have a stroke.13 Thus, earlier detection of AF may lead to improvements in AF care and contribute to reductions in the number of AF-related stroke episodes. Identifying CACS as a biomarker for the development of AF may consequently have important clinical and public health implications. Therefore, in this study with a longitudinal design, we examined the association between CACS and development of AF and the predictive accuracy of CACS in a large cohort of patients who underwent nCCT for the indication of suspected coronary artery disease (CAD). Methods Design and population We conducted a registry-based cohort study involving patients in Western Denmark who were suspected of CAD. We defined baseline as the time at which these patients had their CACS measured using nCCT between January 2010 and September 2015. Patients were identified from the Western Denmark Heart Registry, a validated clinical quality database covering Western Denmark, whose population comprises approximately 3.3 million people.14,15 It is an encrypted online data management system that uses the Civil Registration number as the key identifier. Staff members examining patients for CAD enter data into the system from each individual cardiac centre. Information regarding referral, medical history, and procedure are reported for each procedure. We used information from the first procedure for patients who underwent multiple procedures during the study period. All Danish citizens are assigned a unique 10-digit Civil Registration number enabling unambiguous linkages between public registers. The Civil Registration System contains individual information pertaining to sex and place of residence, as well as information on vital statistics and migration that is updated daily.16 The following conditions led to exclusion before the nCCT: invalid Civil Registration number, invalid municipality, and residency in a municipality in Greenland. Furthermore, we excluded individuals with a history of valvular disease (mitral stenosis or mechanical prosthetic heart valve),17 acute myocardial infarction, percutaneous coronary intervention (PCI), heart surgeries such as coronary artery bypass grafting (CABG), CAD, and AF (including atrial flutter), as well as patients with missing CACS. The final study population consisted of 27 962 patients. Figure 1 shows the full flowchart. Figure 1 View largeDownload slide The flowchart depicts the total number of procedures in the Western Denmark Heart Registry (WDHR) and the exclusion of patients; NPR, National Patient Registry; CAD, coronary artery disease. PCI, percutaneous coronary intervention; AF, atrial fibrillation; CACS, coronary artery calcium score; CABG, coronary artery bypass grafting. Figure 1 View largeDownload slide The flowchart depicts the total number of procedures in the Western Denmark Heart Registry (WDHR) and the exclusion of patients; NPR, National Patient Registry; CAD, coronary artery disease. PCI, percutaneous coronary intervention; AF, atrial fibrillation; CACS, coronary artery calcium score; CABG, coronary artery bypass grafting. Measurement of CACS CACSs were measured using electrocardiography-triggered nCCT and the Agatston score method.6 Coronary calcium lesions were defined as having a threshold ≥ 130 Hounsfield units (HUs) and an area ≥ 1 mm2. The products of the area of each calcified plaque and peak HU, defined as 1 (130–199 HUs), 2 (200–299 HUs), 3 (300–399 HUs), and 4 (≥400 HUs), were summed for the left main coronary artery, left anterior descending coronary artery, left circumflex coronary artery, and right coronary artery to determine the total CACS. During the study period, the different centres used the following scanner types: Dual source Siemens Somatom Sensation 64 CT Scanner, Philips Brilliance 64 CT scanner, Philips Brilliance 256 CT, GE LightSpeed VCT 64, Toshiba Aquilion 64, Toshiba Aquilion ONE 320, Siemens Somatom Definition Flash, and GE Discovery CT750 HD. Inter- and intrascan reproducibility of CACS measurement18,19 and intraobserver or interobserver variation have been reported to be excellent.20 Ascertainment of new-onset AF The patients were followed up for development of AF until February 2016 using the National Patient Registry. The Danish National Patient Registry was established in 1977 and contains prospectively registered data on all patients in Danish hospitals, including inpatients and outpatients, and is considered to be the most comprehensive registry of its kind.21 Diagnoses were coded according to the International Classification of Diseases 10th Revision (ICD-10). We identified all patients with an incident AF (or atrial flutter) diagnosis, irrespective of the type of AF, after nCCT, and AF (or atrial flutter) could not influence CACS information. The ICD-10 I48 diagnosis code has been found to have a high positive predictive value in the National Patient Registry,22,23 corresponding to a high probability of having AF or atrial flutter according to information recorded in the medical record. Only approximately 5% of ICD-10 I48 diagnoses correspond to atrial flutter.23 Death and migration led to censoring but death was also considered a competing risk. Covariates We identified prior acute myocardial infarction, prior PCI, prior heart surgery, BMI (computed from height and weight), diabetes mellitus, heart failure (ejection fraction < 40%), hypertension (defined as receiving antihypertensive medication), smoking status, systolic and diastolic blood pressure, lipid-lowering treatment reception, and CACS from the Western Denmark Heart Registry. We identified valvular disease (ICD-10: I050, I052, I342, procedure code for mistral stenosis surgery, or mechanical valve implantation), CAD (ICD-10: I21-I25), AF (and atrial flutter) (ICD-10: I48), stroke/transient ischaemic attack/systemic embolism (ICD-10: I60-I64, I74, G45 without G45.3), and CABG from the Danish National Patient Registry. We identified acute myocardial infarction (ICD-10: I21-I22), PCI, heart surgery, hypertension (I10-I13, I15), diabetes mellitus (E10-E11), and heart failure (I50) from the National Patient Registry when information was missing from the Western Danish Heart Registry. Age at nCCT and sex were determined using the Civil Registration System. Ethics The Danish Data Protection Agency approved the study (1-16-02-77-13). Registry-based studies do not require approval from an ethics committee in Denmark according to Danish law. Statistical analysis CACS was examined both as a categorical (0, 1–99, 100–399, 400–999, and ≥1000) and as a continuous variable (log2[CACS + 1]). Log2(CACS + 1) increases by one when CACS + 1 doubles. We used the Aalen–Johansen estimator to compute adjusted cumulative incidence of AF by CACS group using time from nCCT as the timescale, with death as a competing risk. Missing data were considered missing at random. We used multiple imputations by chained equations to replace missing values for BMI, smoking status, systolic blood pressure, diastolic blood pressure, and lipid-lowering treatment reception. The imputed values were based on information pertaining to all variables included in the analysis model and a censoring/event indicator. Imputed values were used to complete Models 2 and 3. We created 20 imputed datasets to reduce sampling variability from the imputation process. Hazard ratios (HRs) for CACS were estimated using Cox proportional hazards regression, using age as the underlying timescale, corresponding to an adjustment for age in all analyses. The proportionality assumption of the Cox model was evaluated using log–log plots for survival and the assumption was not violated. HRs were estimated with 95% confidence intervals (CI). CACSs of 0 were defined as the reference group when CACSs were analysed as a categorical variable. We applied four models in the regression analysis, including an unadjusted model that was adjusted only for age (timescale). In Model 1, we adjusted for sex. In Model 2, we also adjusted for BMI, systolic, and diastolic blood pressure as continuous variables; smoking status as a categorical variable; and diabetes mellitus, heart failure, lipid-lowering medication reception, antihypertensive medication reception, and prior stroke/transient ischaemic attack/systemic embolism as dichotomous variables. Model 3 included Model 2 and CABG as a time-dependent covariate. We tested for interactions between sex and CACS as a categorical and continuous variable. All analyses were performed using STATA version 14.2, College Station, TX, USA. Results Baseline characteristics Table 1 presents the baseline characteristics of the entire study cohort. The mean age was 56.9 years and 44.6% of the patients were males. Of the 27 962 patients, 52.2% had a CACS of 0; 25.7% had a CACS of 1–99; 12.6% had a CACS of 100–399; 5.9% had a CACS of 400–999; and 3.7% had a CACS ≥ 1000. Table 1 Baseline characteristics Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Continuous variables are presented as mean (standard deviation) and categorical variables are presented as numbers (percentage). SD, standard deviation; CACS, coronary artery calcium score. Table 1 Baseline characteristics Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Continuous variables are presented as mean (standard deviation) and categorical variables are presented as numbers (percentage). SD, standard deviation; CACS, coronary artery calcium score. Table 2 shows the baseline characteristics of the patients stratified by CACS group after multiple imputations. We found that a high CACS was associated with high age, a high proportion of men, and a high proportion of former or current smokers. Consistent with this finding, less severe coronary calcification was observed among non-smokers. A high CACS was significantly more likely to be found among patients with diabetes mellitus, heart failure, hypertension, hyperlipidaemia, and prior stroke/transient ischaemic attack/systemic embolism. Table 2 Baseline characteristics stratified by CACS group Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Missing data were imputed. P-values show test for linear trend using regression. SD, standard deviation; CACS, coronary artery calcium score. Table 2 Baseline characteristics stratified by CACS group Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Missing data were imputed. P-values show test for linear trend using regression. SD, standard deviation; CACS, coronary artery calcium score. The risk of AF During a median follow-up of 2.9 (interquartile range 1.6–4.2) years and a total time at risk of 83 001 years, 622 patients were diagnosed with AF, corresponding to an overall incidence of 7.5 (95% CI: 6.9–8.1) per 1000 person-years. When comparing the incidence rates of AF according to CACS (Table 3), we found that the incidence increased from 4.5 (95% CI: 3.9–5.2) among patients with a CACS of 0 to 23.8 (95% CI: 18.7–30.2) among patients with a CACS ≥ 1000 per 1000 person-years. Figure 2 shows the cumulative incidence of AF, and we noted the following cumulative incidence of AF 4 years after nCCT, according to CACS group: CACSs of 0: 1.77% (95% CI: 1.52–2.05), CACSs of 1–99: 2.54% (95% CI: 2.09–3.06), CACSs of 100–399: 4.48% (95% CI: 3.66–5.42), CACSs of 400–999: 6.92% (95% CI: 5.46–8.60), and CACSs ≥ 1000: 7.76% (95% CI: 5.96–9.86). For CACSs ≥ 1000, we observed a very sharp increase in the cumulative incidence of AF during first three months after nCCT. Table 3 AF incidence stratified by CACS group   CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)    CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)  CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals. Table 3 AF incidence stratified by CACS group   CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)    CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)  CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals. Figure 2 View largeDownload slide Aalen–Johansen estimator showing the cumulative incidence of AF according to CACS group adjusted for the competing risk of death. CACS, coronary artery calcium score. Figure 2 View largeDownload slide Aalen–Johansen estimator showing the cumulative incidence of AF according to CACS group adjusted for the competing risk of death. CACS, coronary artery calcium score. We performed an unadjusted analysis and noted that the risk of AF increased as the CACS category increased (Table 4). When the CACS doubled, the HR was 1.07 (95% CI: 1.05–1.10). When adjusting for sex in Model 1, the overall association between CACS and the risk of AF remained virtually unchanged. In the fully adjusted Model 3, the HRs remained unchanged and thus appeared robust under the influence of confounding factors. The HRs for each CACS category were as follows: CACSs of 1–99: 1.00 (95% CI: 0.80–1.25); CACSs of 100–399: 1.36 (95% CI: 1.06–1.74); CACSs of 400–900: 1.76 (95% CI: 1.33–2.35); CACSs ≥ 1000: 1.67 (95% CI: 1.20–2.34). In the fully adjusted Model 3, the HR was 1.05 (95% CI: 1.03–1.08) when the CACS doubled. Table 4 The baseline CACS and AF risk CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  We adjusted for age (as timescale) in all analyses. Model 1: Adjusted for sex. Model 2: Adjusted as in Model 1 and for BMI, smoking status, diabetes mellitus, systolic blood pressure, diastolic blood pressure, heart failure, lipid-lowering medical treatment, medical treatment for hypertension, and prior stroke/transient ischaemic attack/systemic embolism. Model 3: Adjusted as in Model 2 and for CABG (n = 508) during follow-up (time-dependent exposure). P-values for CACS categories show test for exposure effect of CACS using Wald test. CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals; HR, hazard ratio; BMI, body mass index; CABG, coronary artery bypass grafting. Table 4 The baseline CACS and AF risk CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  We adjusted for age (as timescale) in all analyses. Model 1: Adjusted for sex. Model 2: Adjusted as in Model 1 and for BMI, smoking status, diabetes mellitus, systolic blood pressure, diastolic blood pressure, heart failure, lipid-lowering medical treatment, medical treatment for hypertension, and prior stroke/transient ischaemic attack/systemic embolism. Model 3: Adjusted as in Model 2 and for CABG (n = 508) during follow-up (time-dependent exposure). P-values for CACS categories show test for exposure effect of CACS using Wald test. CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals; HR, hazard ratio; BMI, body mass index; CABG, coronary artery bypass grafting. In a subanalysis, we excluded patients with a history of stroke/transient ischaemic attack/systemic embolism before baseline. However, the results showed no substantial difference (results not shown). Using Model 3, we tested for interactions between CACS group and sex (P = 0.59) and log2(CACS + 1) and sex (P = 0.82) and noted no interactions. Predictive accuracy of CACS We examined the post-test probability of developing AF within 1 year after nCCT according to CACS. Figure 3 presents an ROC curve for the post-test probability of developing AF according to log2(CACS + 1). We found an area under the curve (AUC) of 0.68 (95% CI: 0.65–0.71), which was not substantially different between men and women. Figure 3 View largeDownload slide ROC curve showing the predictive accuracy of CACS for predicting the development of AF within 1 year after nCCT. ROC, receiver operating characteristic. Figure 3 View largeDownload slide ROC curve showing the predictive accuracy of CACS for predicting the development of AF within 1 year after nCCT. ROC, receiver operating characteristic. We applied cut-off values according to the CACS groups (Table 5). When using a CACS of 0 as cut-off, we found that 73.1% of patients who developed AF during the first year after nCCT had a CACS > 0, and that 52.7% of patients who did not develop AF during first year after nCCT had a CACS of 0. These data indicate that 1.6% of patients with CACS > 0 developed AF during the first year after nCCT and that 99.4% of patients with a CACS of 0 did not develop AF during the first year. Table 5 shows the results for all of the defined cut-offs. Table 5 Predictive accuracy of CACS for AF development within 1 year after nCCT CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS, coronary artery calcium score. Table 5 Predictive accuracy of CACS for AF development within 1 year after nCCT CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS, coronary artery calcium score. Discussion In this large registry-based cohort study of patients suspected of having CAD who underwent nCCT, we demonstrated that a high CACS was associated with a high risk of AF and that the relationship between the two parameters was such that an individual’s risk of AF increased as the CACS increased, even after multivariable adjustment. We also demonstrated that CACS had moderate predictive accuracy with respect to identifying individuals at risk for developing AF within one year after nCCT. We found that the relationship between CACS and AF is not attributable to the most common risk factors for AF because our models included the most relevant and accessible risk factors and potential confounders. Our study results are consistent with findings from the Multi-Ethnic Study of Atherosclerosis by O’Neal et al.,11 which, to our knowledge, is the only other study on this specific topic to date. In the Multi-Ethnic Study of Atherosclerosis, O’Neal et al.11 noted that the HRs for each CACS category were as follows: (CACS of 0 as reference) CACSs of 1–100: HR 1.4 (95% CI: 1.01–2.0); CACSs of 101–300: HR 1.6 (95% CI: 1.1–2.4); CACSs > 300: HR 2.1 (95% CI: 1.4–2.9). However, our study is larger and it extends the association between CACS and AF from an American to a European population. Our results are suggestive of a weaker association between CACS and risk of AF than that reported by the Multi-Ethnic Study of Atherosclerosis.11 Unequal confounder distribution may partially account for the discrepancy between the studies. For instance, the present study cohort was younger and consisted of fewer men than the aforementioned study. We based our AF ascertainments on registries containing complete follow-up information, independent of the facilities to which patients were admitted. Furthermore, we used age as the underlying timescale in our analyses because of the strong association between age and AF. In contrast to O’Neal et al.,11 we noted a sharp increase in the cumulative incidence in patients with CACSs ≥ 1000, probably because a high CACS is associated with a high revascularization rate, and CABG, which is performed for revascularization, is known to be associated with a high risk of AF. The exact pathogenic processes explaining how patients progress from coronary calcification to AF are not clear. Several mechanisms underlying the relationship between coronary obstruction/occlusion and risk of AF have been suggested. CAD, which is correlated with the CACS,24 is associated with low-grade subclinical inflammation, and ischaemia or atrial myocardial infarction is associated with a healing process that triggers myocardial damage and atrial inflammation.25,26 Inflammatory mediators seem to alter electrical and structural remodelling and thus trigger AF. Additionally, inflammation also seems to play a role in post-operative AF, for example, AF after CABG.26 Furthermore, it has been shown that the extent of CACS progression may be an important factor in the relationship between CACS and AF.27 The relationship between CACS and AF may also be explained by the fact that a high CACS is associated with dilated pulmonary veins and a larger left atrium, which have the potential to initiate and perpetuate atrial re-entrant circuits.28 Our results may have long-term clinical implications. We noted a moderate AUC when using CACS as a tool for predicting AF. Hence, CACSs should be interpreted in connection with patient risk profiles. Adding CACS to risk scores from the Framingham Heart Study and Aging Research in Genomic Epidemiology (CHARGE)-AF improves the AUC with respect to predicting AF.11 Using CACS in combination with knowledge of other risk factors makes it possible to identify patients at high risk for AF, in whom the tolerance of modifiable AF risk factors should be set at an even lower level than in patients not at high risk of AF, and disease prevention efforts should be more aggressive. Accordingly, using CACS in combination with knowledge regarding other AF risk factors will enable AF prevention and earlier treatment start-up in patients who most likely develop AF and thus reduce the number of incident strokes among these patients. Given that the number of patients with AF is increasing, this potential improvement in early AF detection may ultimately benefit more patients and have important public health consequences. However, additional studies are needed to examine the usefulness of CACS for stroke reduction in AF patients. The main strengths of our study were its use of a population-based method, its large population size, and its application of prospectively collection of data. Additionally, this study was strengthened by the fact that it utilized the Western Denmark Heart Registry, whose data are characterized by a high degree of completeness and validity,14 and that it uses Danish registers to ensure that patient follow-up was complete. Migration bias was unlikely because patient migration status is recorded in the Civil Registration System. The diagnosis of AF in the National Patient Registry has been found to have a high positive predictive value.23 Patients enrolled in this study do not represent the general population, as they were suspected of having CAD. According to a report from the Western Denmark Heart Registry 2014, 58% of patients undergoing nCCT do not have diseased vessels, and nCCT does not lead to consequences in terms of treatment in 63% of patients.29 Using data on antiarrhythmic medication as exclusion may have excluded patients with AF before baseline more accurately; unfortunately, those data were unavailable. Differential misclassification may occur if patients with a high CACS are followed more carefully than patients with a CACS of 0, which may lead to a higher likelihood of being diagnosed with AF. It is possible that differences may exist in the way the CACSs were measured, which could have introduced some misclassification, most likely of a non-differential nature. It should also be stressed that our finding of CACS being a predictor of AF does not imply causality. We were not able to distinguish between persistent, permanent, and paroxysmal AF; therefore, paroxysmal AF might be underreported. Furthermore, residual confounding is possible, for instance, with respect to categorizing smoking, and a lack of information on potential confounders, such as ethnicity and alcohol consumption, may lead to imprecision. Conclusions A high CACS is associated with a high risk of AF. CACS combined with knowledge regarding other AF risk factors may help clinicians to identify patients at high risk for AF, which may facilitate both disease prevention and earlier treatment of patients who most likely develop AF and thus reduce their risk of stroke. Acknowledgements We are grateful for the grants received by Regional Hospital Central Jutland, Viborg, Denmark; Aarhus University, Aarhus, Denmark; and Grosserer A. V. Lykfeldt og Hustrus Legat, Copenhagen, Denmark. Funding Grants received from Regional Hospital Central Jutland, Viborg, Denmark; Aarhus University, Aarhus, Denmark; and Grosserer A. V. Lykfeldt og Hustrus Legat, Copenhagen, Denmark. The funders had no role in the design or conduction of the study; the management, analysis, and interpretation of the data; the preparation, review, or approval of the article; or the decision to submit the manuscript for publication. Conflict of interest: None declared. References 1 Krijthe BP, Kunst A, Benjamin EJ, Lip GY, Franco OH, Hofman A et al.   Projections on the number of individuals with atrial fibrillation in the European Union, from 2000 to 2060. Eur Heart J  2013; 34: 2746– 51. Google Scholar CrossRef Search ADS PubMed  2 Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Abhayaratna WP et al.   Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence. Circulation  2006; 114: 119– 25. Google Scholar CrossRef Search ADS PubMed  3 Gassett AJ, Sheppard L, McClelland RL, Olives C, Kronmal R, Blaha MJ et al.   Risk factors for long-term coronary artery calcium progression in the multi-ethnic study of atherosclerosis. J Am Heart Assoc  2015; 4: e001726. Google Scholar CrossRef Search ADS PubMed  4 Oei HH, Vliegenthart R, Hofman A, Oudkerk M, Witteman JC. Risk factors for coronary calcification in older subjects. The Rotterdam Coronary Calcification Study. Eur Heart J  2004; 25: 48– 55. Google Scholar CrossRef Search ADS PubMed  5 Schnabel RB, Yin X, Gona P, Larson MG, Beiser AS, McManus DD et al.   50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: a cohort study. Lancet  2015; 386: 154– 62. Google Scholar CrossRef Search ADS PubMed  6 Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte MJr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol  1990; 15: 827– 32. Google Scholar CrossRef Search ADS PubMed  7 Shemesh J, Henschke CI, Shaham D, Yip R, Farooqi AO, Cham MD et al.   Ordinal scoring of coronary artery calcifications on low-dose CT scans of the chest is predictive of death from cardiovascular disease. Radiology  2010; 257: 541– 8. Google Scholar CrossRef Search ADS PubMed  8 Raggi P, Gongora MC, Gopal A, Callister TQ, Budoff M, Shaw LJ. Coronary artery calcium to predict all-cause mortality in elderly men and women. J Am Coll Cardiol  2008; 52: 17– 23. Google Scholar CrossRef Search ADS PubMed  9 Chang SM, Nabi F, Xu J, Peterson LE, Achari A, Pratt CM et al.   The coronary artery calcium score and stress myocardial perfusion imaging provide independent and complementary prediction of cardiac risk. J Am Coll Cardiol  2009; 54: 1872– 82. Google Scholar CrossRef Search ADS PubMed  10 Handy CE, Desai CS, Dardari ZA, Al-Mallah MH, Miedema MD, Ouyang P et al.   The association of coronary artery calcium with noncardiovascular disease: the multi-ethnic study of atherosclerosis. JACC Cardiovasc Imaging  2016; 9: 568– 76. Google Scholar CrossRef Search ADS PubMed  11 O'Neal WT, Efird JT, Dawood FZ, Yeboah J, Alonso A, Heckbert SR et al.   Coronary artery calcium and risk of atrial fibrillation (from the multi-ethnic study of atherosclerosis). Am J Cardiol  2014; 114: 1707– 12. Google Scholar CrossRef Search ADS PubMed  12 Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke  1991; 22: 983– 8. Google Scholar CrossRef Search ADS PubMed  13 Johnsen SP, Svendsen ML, Hansen ML, Brandes A, Mehnert F, Husted SE. Preadmission oral anticoagulant treatment and clinical outcome among patients hospitalized with acute stroke and atrial fibrillation: a nationwide study. Stroke  2014; 45: 168– 75. Google Scholar CrossRef Search ADS PubMed  14 Nielsen LH, Norgaard BL, Tilsted HH, Sand NP, Jensen JM, Bottcher M et al.   The Western Denmark Cardiac Computed Tomography Registry: a review and validation study. Clin Epidemiol  2015; 7: 53– 64. Google Scholar PubMed  15 Schmidt M, Maeng M, Jakobsen CJ, Madsen M, Thuesen L, Nielsen PH et al.   Existing data sources for clinical epidemiology: the Western Denmark Heart Registry. Clin Epidemiol  2010; 2: 137– 44. Google Scholar CrossRef Search ADS PubMed  16 Pedersen CB, Gotzsche H, Moller JO, Mortensen PB. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull  2006; 53: 441– 9. Google Scholar PubMed  17 Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B et al.   2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J  2016; 37: 2893– 962. Google Scholar CrossRef Search ADS PubMed  18 Sabour S, Rutten A, van der Schouw YT, Atsma F, Grobbee DE, Mali WP et al.   Inter-scan reproducibility of coronary calcium measurement using Multi Detector-Row Computed Tomography (MDCT). Eur J Epidemiol  2007; 22: 235– 43. Google Scholar CrossRef Search ADS PubMed  19 Ghadri JR, Goetti R, Fiechter M, Pazhenkottil AP, Kuest SM, Nkoulou RN et al.   Inter-scan variability of coronary artery calcium scoring assessed on 64-multidetector computed tomography vs. dual-source computed tomography: a head-to-head comparison. Eur Heart J  2011; 32: 1865– 74. Google Scholar CrossRef Search ADS PubMed  20 Williams MC, Golay SK, Hunter A, Weir-McCall JR, Mlynska L, Dweck MR et al.   Observer variability in the assessment of CT coronary angiography and coronary artery calcium score: substudy of the Scottish COmputed Tomography of the HEART (SCOT-HEART) trial. Open Heart  2015; 2: e000234. Google Scholar CrossRef Search ADS PubMed  21 Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health  2011; 39: 30– 3. Google Scholar CrossRef Search ADS PubMed  22 Rix TA, Riahi S, Overvad K, Lundbye-Christensen S, Schmidt EB, Joensen AM. Validity of the diagnoses atrial fibrillation and atrial flutter in a Danish patient registry. Scand Cardiovasc J  2012; 46: 149– 53. Google Scholar CrossRef Search ADS PubMed  23 Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol  2015; 7: 449– 90. Google Scholar CrossRef Search ADS PubMed  24 Cheng YJ, Church TS, Kimball TE, Nichaman MZ, Levine BD, McGuire DK et al.   Comparison of coronary artery calcium detected by electron beam tomography in patients with to those without symptomatic coronary heart disease. Am J Cardiol  2003; 92: 498– 503. Google Scholar CrossRef Search ADS PubMed  25 Dzeshka MS, Lip GY, Snezhitskiy V, Shantsila E. Cardiac fibrosis in patients with atrial fibrillation: mechanisms and clinical implications. J Am Coll Cardiol  2015; 66: 943– 59. Google Scholar CrossRef Search ADS PubMed  26 Hu YF, Chen YJ, Lin YJ, Chen SA. Inflammation and the pathogenesis of atrial fibrillation. Nat Rev Cardiol  2015; 12: 230– 43. Google Scholar CrossRef Search ADS PubMed  27 O'Neal WT, Efird JT, Qureshi WT, Yeboah J, Alonso A, Heckbert SR et al.   Coronary artery calcium progression and atrial fibrillation: the multi-ethnic study of atherosclerosis. Circ Cardiovasc Imaging  2015; 8: e003786. Google Scholar CrossRef Search ADS PubMed  28 Pan NH, Tsao HM, Chang NC, Lee CM, Chen YJ, Chen SA. Dilated left atrium and pulmonary veins in patients with calcified coronary artery: a potential contributor to the genesis of atrial fibrillation. J Cardiovasc Electrophysiol  2009; 20: 153– 8. Google Scholar CrossRef Search ADS PubMed  29 Secretariat. Western Denmark Heart Registry: Year Report 2015. Activities and Results. http://vdhd.dk/wp-content/uploads/2014/10/Årsrapport-2014.pdf (31 August, 2016, date last accessed). Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2017. For permissions, please email: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Heart Journal – Cardiovascular Imaging Oxford University Press

Coronary artery calcium score and the long-term risk of atrial fibrillation in patients undergoing non-contrast cardiac computed tomography for suspected coronary artery disease: a Danish registry-based cohort study

Loading next page...
 
/lp/ou_press/coronary-artery-calcium-score-and-the-long-term-risk-of-atrial-2jJut6TwQS
Publisher
Oxford University Press
Copyright
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2017. For permissions, please email: journals.permissions@oup.com.
ISSN
2047-2404
D.O.I.
10.1093/ehjci/jex201
Publisher site
See Article on Publisher Site

Abstract

Abstract Aims To examine the association between coronary artery calcium score (CACS) and risk of future atrial fibrillation (AF), and to estimate the predictive accuracy of CACS for AF development in patients undergoing non-contrast cardiac computed tomography (nCCT). Methods and results We conducted a registry-based cohort study of 27 962 patients suspected of having coronary artery disease and without history of AF who were identified in the Western Denmark Heart Registry. The patients underwent nCCT between 2010 and 2015 and were followed until 2016 (median 2.9 years). CACSs were determined using nCCT. We used Cox proportional hazards models to estimate hazard ratios (HR) with 95% confidence intervals (CI). A receiver operating characteristic (ROC) curve for AF was used to assess the predictive accuracy of CACS. Among the patients, 52% had a CACS of 0, 26% of 1–99, 13% of 100–399, 6% of 400–999, and 4% of ≥ 1000. AF occurred in 622 patients after nCCT, corresponding to an overall incidence rate of 7.5 (95% CI: 6.9–8.1) per 1000 person-years. After multivariable adjustment, the HRs (95% CIs) were (ref. CACS 0) CACS 1–99: 1.00 (0.80–1.25); CACS 100–399: 1.36 (1.06–1.74); CACS 400–999: 1.76 (1.33–2.35); and CACS ≥ 1000: 1.67 (1.20–2.34). An ROC curve showed an area under the curve of 0.68 (0.65–0.71) for the prediction of AF within one year after nCCT. Conclusion A high CACS is associated with a high risk of subsequent AF development and may have potential to guide future follow-ups for AF detection after CACS measurement in order to identify AF patients earlier. coronary artery calcium score , atrial fibrillation , cardiac computed tomography , coronary arteries , Western Denmark Heart Registry Introduction Atrial fibrillation (AF) is one of the most frequently occurring arrhythmias, and projections have shown that the number of adults with AF will increase. The prevalence of AF in Europe may reach 17.9 million people in 20601 and the prevalence in the USA may reach 15.9 million people in 2050.2 Several risk factors for AF have been identified, many of which are also risk factors for coronary calcification, including increasing age, sex, systolic blood pressure, body mass index (BMI), hypertension, and diabetes mellitus.3–5 Use of non-contrast cardiac computed tomography (nCCT) and the Agatston score method6 for measuring the coronary artery calcium score (CACS) has identified CACS as a compelling predictor of future cardiovascular events, including cardiovascular death, and all-cause mortality.7–9 CACS also has a potential as a biomarker for the development of non-cardiovascular events such as cancer, chronic kidney disease, chronic obstructive pulmonary disease, and hip fractures.10 Recently, CACS was introduced as a novel independent risk factor for AF based on results reported by the Multi-Ethnic Study of Atherosclerosis.11 Gaining more insight into the role of CACS in relation to the risk of future AF is of major clinical relevance. First, more accurate identification of individuals at high risk for AF may facilitate a stronger focus on disease prevention and, in particular, interventions indented to address modifiable risk factors. Secondly, stroke is a well-known complication of AF that can be effectively prevented using oral anticoagulant therapy.12 However, as many as 40% of patients with AF-related strokes are not diagnosed with AF until they have a stroke.13 Thus, earlier detection of AF may lead to improvements in AF care and contribute to reductions in the number of AF-related stroke episodes. Identifying CACS as a biomarker for the development of AF may consequently have important clinical and public health implications. Therefore, in this study with a longitudinal design, we examined the association between CACS and development of AF and the predictive accuracy of CACS in a large cohort of patients who underwent nCCT for the indication of suspected coronary artery disease (CAD). Methods Design and population We conducted a registry-based cohort study involving patients in Western Denmark who were suspected of CAD. We defined baseline as the time at which these patients had their CACS measured using nCCT between January 2010 and September 2015. Patients were identified from the Western Denmark Heart Registry, a validated clinical quality database covering Western Denmark, whose population comprises approximately 3.3 million people.14,15 It is an encrypted online data management system that uses the Civil Registration number as the key identifier. Staff members examining patients for CAD enter data into the system from each individual cardiac centre. Information regarding referral, medical history, and procedure are reported for each procedure. We used information from the first procedure for patients who underwent multiple procedures during the study period. All Danish citizens are assigned a unique 10-digit Civil Registration number enabling unambiguous linkages between public registers. The Civil Registration System contains individual information pertaining to sex and place of residence, as well as information on vital statistics and migration that is updated daily.16 The following conditions led to exclusion before the nCCT: invalid Civil Registration number, invalid municipality, and residency in a municipality in Greenland. Furthermore, we excluded individuals with a history of valvular disease (mitral stenosis or mechanical prosthetic heart valve),17 acute myocardial infarction, percutaneous coronary intervention (PCI), heart surgeries such as coronary artery bypass grafting (CABG), CAD, and AF (including atrial flutter), as well as patients with missing CACS. The final study population consisted of 27 962 patients. Figure 1 shows the full flowchart. Figure 1 View largeDownload slide The flowchart depicts the total number of procedures in the Western Denmark Heart Registry (WDHR) and the exclusion of patients; NPR, National Patient Registry; CAD, coronary artery disease. PCI, percutaneous coronary intervention; AF, atrial fibrillation; CACS, coronary artery calcium score; CABG, coronary artery bypass grafting. Figure 1 View largeDownload slide The flowchart depicts the total number of procedures in the Western Denmark Heart Registry (WDHR) and the exclusion of patients; NPR, National Patient Registry; CAD, coronary artery disease. PCI, percutaneous coronary intervention; AF, atrial fibrillation; CACS, coronary artery calcium score; CABG, coronary artery bypass grafting. Measurement of CACS CACSs were measured using electrocardiography-triggered nCCT and the Agatston score method.6 Coronary calcium lesions were defined as having a threshold ≥ 130 Hounsfield units (HUs) and an area ≥ 1 mm2. The products of the area of each calcified plaque and peak HU, defined as 1 (130–199 HUs), 2 (200–299 HUs), 3 (300–399 HUs), and 4 (≥400 HUs), were summed for the left main coronary artery, left anterior descending coronary artery, left circumflex coronary artery, and right coronary artery to determine the total CACS. During the study period, the different centres used the following scanner types: Dual source Siemens Somatom Sensation 64 CT Scanner, Philips Brilliance 64 CT scanner, Philips Brilliance 256 CT, GE LightSpeed VCT 64, Toshiba Aquilion 64, Toshiba Aquilion ONE 320, Siemens Somatom Definition Flash, and GE Discovery CT750 HD. Inter- and intrascan reproducibility of CACS measurement18,19 and intraobserver or interobserver variation have been reported to be excellent.20 Ascertainment of new-onset AF The patients were followed up for development of AF until February 2016 using the National Patient Registry. The Danish National Patient Registry was established in 1977 and contains prospectively registered data on all patients in Danish hospitals, including inpatients and outpatients, and is considered to be the most comprehensive registry of its kind.21 Diagnoses were coded according to the International Classification of Diseases 10th Revision (ICD-10). We identified all patients with an incident AF (or atrial flutter) diagnosis, irrespective of the type of AF, after nCCT, and AF (or atrial flutter) could not influence CACS information. The ICD-10 I48 diagnosis code has been found to have a high positive predictive value in the National Patient Registry,22,23 corresponding to a high probability of having AF or atrial flutter according to information recorded in the medical record. Only approximately 5% of ICD-10 I48 diagnoses correspond to atrial flutter.23 Death and migration led to censoring but death was also considered a competing risk. Covariates We identified prior acute myocardial infarction, prior PCI, prior heart surgery, BMI (computed from height and weight), diabetes mellitus, heart failure (ejection fraction < 40%), hypertension (defined as receiving antihypertensive medication), smoking status, systolic and diastolic blood pressure, lipid-lowering treatment reception, and CACS from the Western Denmark Heart Registry. We identified valvular disease (ICD-10: I050, I052, I342, procedure code for mistral stenosis surgery, or mechanical valve implantation), CAD (ICD-10: I21-I25), AF (and atrial flutter) (ICD-10: I48), stroke/transient ischaemic attack/systemic embolism (ICD-10: I60-I64, I74, G45 without G45.3), and CABG from the Danish National Patient Registry. We identified acute myocardial infarction (ICD-10: I21-I22), PCI, heart surgery, hypertension (I10-I13, I15), diabetes mellitus (E10-E11), and heart failure (I50) from the National Patient Registry when information was missing from the Western Danish Heart Registry. Age at nCCT and sex were determined using the Civil Registration System. Ethics The Danish Data Protection Agency approved the study (1-16-02-77-13). Registry-based studies do not require approval from an ethics committee in Denmark according to Danish law. Statistical analysis CACS was examined both as a categorical (0, 1–99, 100–399, 400–999, and ≥1000) and as a continuous variable (log2[CACS + 1]). Log2(CACS + 1) increases by one when CACS + 1 doubles. We used the Aalen–Johansen estimator to compute adjusted cumulative incidence of AF by CACS group using time from nCCT as the timescale, with death as a competing risk. Missing data were considered missing at random. We used multiple imputations by chained equations to replace missing values for BMI, smoking status, systolic blood pressure, diastolic blood pressure, and lipid-lowering treatment reception. The imputed values were based on information pertaining to all variables included in the analysis model and a censoring/event indicator. Imputed values were used to complete Models 2 and 3. We created 20 imputed datasets to reduce sampling variability from the imputation process. Hazard ratios (HRs) for CACS were estimated using Cox proportional hazards regression, using age as the underlying timescale, corresponding to an adjustment for age in all analyses. The proportionality assumption of the Cox model was evaluated using log–log plots for survival and the assumption was not violated. HRs were estimated with 95% confidence intervals (CI). CACSs of 0 were defined as the reference group when CACSs were analysed as a categorical variable. We applied four models in the regression analysis, including an unadjusted model that was adjusted only for age (timescale). In Model 1, we adjusted for sex. In Model 2, we also adjusted for BMI, systolic, and diastolic blood pressure as continuous variables; smoking status as a categorical variable; and diabetes mellitus, heart failure, lipid-lowering medication reception, antihypertensive medication reception, and prior stroke/transient ischaemic attack/systemic embolism as dichotomous variables. Model 3 included Model 2 and CABG as a time-dependent covariate. We tested for interactions between sex and CACS as a categorical and continuous variable. All analyses were performed using STATA version 14.2, College Station, TX, USA. Results Baseline characteristics Table 1 presents the baseline characteristics of the entire study cohort. The mean age was 56.9 years and 44.6% of the patients were males. Of the 27 962 patients, 52.2% had a CACS of 0; 25.7% had a CACS of 1–99; 12.6% had a CACS of 100–399; 5.9% had a CACS of 400–999; and 3.7% had a CACS ≥ 1000. Table 1 Baseline characteristics Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Continuous variables are presented as mean (standard deviation) and categorical variables are presented as numbers (percentage). SD, standard deviation; CACS, coronary artery calcium score. Table 1 Baseline characteristics Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Characteristics  All patients n = 27 962  Missing data (%)  Age, mean (SD), years  56.9 (11.1)    Male, n (%)  12 468 (44.6)    Body mass index, mean (SD), kg/m2  26.5 (4.3)    Smoking, n (%)    2290 (8.2)   Never  10 632 (41.4)   Former  9033 (35.2)   Current  6007 (23.4)  Diabetes mellitus, n (%)  1851 (6.6)    Systolic blood pressure, mean (SD), mmHg  139 (19.2)  4664 (16.7)  Diastolic blood pressure, mean (SD) mmHg  82 (10.5)  4674 (16.7)  Heart failure, n (%)  160 (0.6)    Current lipid-lowering medical treatment, n (%)  8436 (32.5)  2015 (7.2)  Current medical treatment for hypertension, n (%)  10 174 (36.4)    Prior stroke/transient ischaemic attack/systemic embolism, n (%)  1350 (4.8)    CACS, n (%)       0  14 596 (52.2)   1–99  7173 (25.7)   100–399  3515 (12.6)   400–999  1654 (5.9)   ≥1000  1024 (3.7)  log2(CACS+1), mean (SD) for a CACS > 0  3.02 (3.6)  Continuous variables are presented as mean (standard deviation) and categorical variables are presented as numbers (percentage). SD, standard deviation; CACS, coronary artery calcium score. Table 2 shows the baseline characteristics of the patients stratified by CACS group after multiple imputations. We found that a high CACS was associated with high age, a high proportion of men, and a high proportion of former or current smokers. Consistent with this finding, less severe coronary calcification was observed among non-smokers. A high CACS was significantly more likely to be found among patients with diabetes mellitus, heart failure, hypertension, hyperlipidaemia, and prior stroke/transient ischaemic attack/systemic embolism. Table 2 Baseline characteristics stratified by CACS group Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Missing data were imputed. P-values show test for linear trend using regression. SD, standard deviation; CACS, coronary artery calcium score. Table 2 Baseline characteristics stratified by CACS group Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Characteristics  CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  P-value  Age, mean (SD), year  52.3 (10.6)  59.7 (9.2)  63.2 (8.6)  65.3 (8.0)  67.9 (7.8)  <0.001  Male, n (%)  5398 (37.0)  3430 (47.8)  1935 (55.1)  1020 (61.7)  685 (66.9)  <0.001  Body mass index, mean, kg/m2  26.4  26.6  26.6  26.7  26.5  0.001  Smoking, %            <0.001   Never  46.8  40.0  32.7  27.4  26.1   Former  30.5  36.6  42.5  45.9  48.1   Current  22.7  23.4  24.7  26.6  25.7  Diabetes mellitus, n (%)  651 (4.5)  491 (6.9)  345 (9.8)  207 (12.5)  157 (15.3)  <0.001  Systolic blood pressure, mean, mmHg  136  141  144  146  148  <0.001  Diastolic blood pressure, mean, mmHg  81  82  83  83  82  <0.001  Heart failure, n (%)  62 (0.4)  40 (0.6)  28 (0.8)  13 (0.8)  17 (1.7)  <0.001  Current lipid-lowering medical treatment, %  23.5  36.6  45.3  51.1  54.9  <0.001  Current medical treatment for hypertension, n (%)  3939 (27.0)  2889 (40.3)  1745 (49.6)  942 (57.0)  659 (64.4)  <0.001  Prior stroke/transient ischaemic attack/systemic embolism, n (%)  512 (3.5)  354 (4.9)  218 (6.2)  142 (8.6)  124 (12.1)  <0.001  Missing data were imputed. P-values show test for linear trend using regression. SD, standard deviation; CACS, coronary artery calcium score. The risk of AF During a median follow-up of 2.9 (interquartile range 1.6–4.2) years and a total time at risk of 83 001 years, 622 patients were diagnosed with AF, corresponding to an overall incidence of 7.5 (95% CI: 6.9–8.1) per 1000 person-years. When comparing the incidence rates of AF according to CACS (Table 3), we found that the incidence increased from 4.5 (95% CI: 3.9–5.2) among patients with a CACS of 0 to 23.8 (95% CI: 18.7–30.2) among patients with a CACS ≥ 1000 per 1000 person-years. Figure 2 shows the cumulative incidence of AF, and we noted the following cumulative incidence of AF 4 years after nCCT, according to CACS group: CACSs of 0: 1.77% (95% CI: 1.52–2.05), CACSs of 1–99: 2.54% (95% CI: 2.09–3.06), CACSs of 100–399: 4.48% (95% CI: 3.66–5.42), CACSs of 400–999: 6.92% (95% CI: 5.46–8.60), and CACSs ≥ 1000: 7.76% (95% CI: 5.96–9.86). For CACSs ≥ 1000, we observed a very sharp increase in the cumulative incidence of AF during first three months after nCCT. Table 3 AF incidence stratified by CACS group   CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)    CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)  CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals. Table 3 AF incidence stratified by CACS group   CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)    CACS   0 (n = 14 596)  1–99 (n = 7173)  100–399 (n = 3515)  400–999 (n = 1654)  ≥1000 (n = 1024)  Number of AF cases, n (%)  199 (1.4)  143 (2.0)  125 (3.6)  88 (5.3)  67 (6.5)  Time at risk from baseline, years  44 272  20 744  10 424  4744  2817  Incidence per 1000 person-years (95% CI)  4.5 (3.9–5.2)  6.9 (5.9–8.1)  12.0 (10.1–14.3)  18.5 (15.1–22.9)  23.8 (18.7–30.2)  CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals. Figure 2 View largeDownload slide Aalen–Johansen estimator showing the cumulative incidence of AF according to CACS group adjusted for the competing risk of death. CACS, coronary artery calcium score. Figure 2 View largeDownload slide Aalen–Johansen estimator showing the cumulative incidence of AF according to CACS group adjusted for the competing risk of death. CACS, coronary artery calcium score. We performed an unadjusted analysis and noted that the risk of AF increased as the CACS category increased (Table 4). When the CACS doubled, the HR was 1.07 (95% CI: 1.05–1.10). When adjusting for sex in Model 1, the overall association between CACS and the risk of AF remained virtually unchanged. In the fully adjusted Model 3, the HRs remained unchanged and thus appeared robust under the influence of confounding factors. The HRs for each CACS category were as follows: CACSs of 1–99: 1.00 (95% CI: 0.80–1.25); CACSs of 100–399: 1.36 (95% CI: 1.06–1.74); CACSs of 400–900: 1.76 (95% CI: 1.33–2.35); CACSs ≥ 1000: 1.67 (95% CI: 1.20–2.34). In the fully adjusted Model 3, the HR was 1.05 (95% CI: 1.03–1.08) when the CACS doubled. Table 4 The baseline CACS and AF risk CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  We adjusted for age (as timescale) in all analyses. Model 1: Adjusted for sex. Model 2: Adjusted as in Model 1 and for BMI, smoking status, diabetes mellitus, systolic blood pressure, diastolic blood pressure, heart failure, lipid-lowering medical treatment, medical treatment for hypertension, and prior stroke/transient ischaemic attack/systemic embolism. Model 3: Adjusted as in Model 2 and for CABG (n = 508) during follow-up (time-dependent exposure). P-values for CACS categories show test for exposure effect of CACS using Wald test. CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals; HR, hazard ratio; BMI, body mass index; CABG, coronary artery bypass grafting. Table 4 The baseline CACS and AF risk CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  CACS  Unadjusted HR (95% CI)  P-value  Model 1 HR (95% CI)  P-value  Model 2 HR (95% CI)  P-value  Model 3 HR (95% CI)  P-value  0 (ref.)  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  =1.00  <0.001  1–99  1.03 (0.82–1.28)  0.97 (0.78–1.22)  1.00 (0.80–1.25)  1.00 (0.80–1.25)  100–399  1.45 (1.14–1.84)  1.32 (1.04–1.68)  1.38 (1.08–1.77)  1.36 (1.06–1.74)  400–999  2.03 (1.55–2.65)  1.80 (1.37–2.37)  1.87 (1.41–2.48)  1.76 (1.33–2.35)  ≥1000  2.20 (1.63–2.96)  1.90 (1.40–2.59)  1.97 (1.44–2.71)  1.67 (1.20–2.34)  Log2(CACS + 1)  1.07 (1.05–1.10)  <0.001  1.06 (1.04–1.09)  <0.001  1.06 (1.04–1.09)  <0.001  1.05 (1.03–1.08)  <0.001  We adjusted for age (as timescale) in all analyses. Model 1: Adjusted for sex. Model 2: Adjusted as in Model 1 and for BMI, smoking status, diabetes mellitus, systolic blood pressure, diastolic blood pressure, heart failure, lipid-lowering medical treatment, medical treatment for hypertension, and prior stroke/transient ischaemic attack/systemic embolism. Model 3: Adjusted as in Model 2 and for CABG (n = 508) during follow-up (time-dependent exposure). P-values for CACS categories show test for exposure effect of CACS using Wald test. CACS, coronary artery calcium score; AF, atrial fibrillation; CI, confidence intervals; HR, hazard ratio; BMI, body mass index; CABG, coronary artery bypass grafting. In a subanalysis, we excluded patients with a history of stroke/transient ischaemic attack/systemic embolism before baseline. However, the results showed no substantial difference (results not shown). Using Model 3, we tested for interactions between CACS group and sex (P = 0.59) and log2(CACS + 1) and sex (P = 0.82) and noted no interactions. Predictive accuracy of CACS We examined the post-test probability of developing AF within 1 year after nCCT according to CACS. Figure 3 presents an ROC curve for the post-test probability of developing AF according to log2(CACS + 1). We found an area under the curve (AUC) of 0.68 (95% CI: 0.65–0.71), which was not substantially different between men and women. Figure 3 View largeDownload slide ROC curve showing the predictive accuracy of CACS for predicting the development of AF within 1 year after nCCT. ROC, receiver operating characteristic. Figure 3 View largeDownload slide ROC curve showing the predictive accuracy of CACS for predicting the development of AF within 1 year after nCCT. ROC, receiver operating characteristic. We applied cut-off values according to the CACS groups (Table 5). When using a CACS of 0 as cut-off, we found that 73.1% of patients who developed AF during the first year after nCCT had a CACS > 0, and that 52.7% of patients who did not develop AF during first year after nCCT had a CACS of 0. These data indicate that 1.6% of patients with CACS > 0 developed AF during the first year after nCCT and that 99.4% of patients with a CACS of 0 did not develop AF during the first year. Table 5 shows the results for all of the defined cut-offs. Table 5 Predictive accuracy of CACS for AF development within 1 year after nCCT CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS, coronary artery calcium score. Table 5 Predictive accuracy of CACS for AF development within 1 year after nCCT CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS cut-offs  Sensitivity (%)  Specificity (%)  Positive predictive value (%)  Negative predictive value (%)  0  73.1  52.7  1.6  99.4  99  49.4  78.1  2.3  99.3  399  26.9  90.6  2.9  99.2  999  14.6  96.5  4.1  99.1  CACS, coronary artery calcium score. Discussion In this large registry-based cohort study of patients suspected of having CAD who underwent nCCT, we demonstrated that a high CACS was associated with a high risk of AF and that the relationship between the two parameters was such that an individual’s risk of AF increased as the CACS increased, even after multivariable adjustment. We also demonstrated that CACS had moderate predictive accuracy with respect to identifying individuals at risk for developing AF within one year after nCCT. We found that the relationship between CACS and AF is not attributable to the most common risk factors for AF because our models included the most relevant and accessible risk factors and potential confounders. Our study results are consistent with findings from the Multi-Ethnic Study of Atherosclerosis by O’Neal et al.,11 which, to our knowledge, is the only other study on this specific topic to date. In the Multi-Ethnic Study of Atherosclerosis, O’Neal et al.11 noted that the HRs for each CACS category were as follows: (CACS of 0 as reference) CACSs of 1–100: HR 1.4 (95% CI: 1.01–2.0); CACSs of 101–300: HR 1.6 (95% CI: 1.1–2.4); CACSs > 300: HR 2.1 (95% CI: 1.4–2.9). However, our study is larger and it extends the association between CACS and AF from an American to a European population. Our results are suggestive of a weaker association between CACS and risk of AF than that reported by the Multi-Ethnic Study of Atherosclerosis.11 Unequal confounder distribution may partially account for the discrepancy between the studies. For instance, the present study cohort was younger and consisted of fewer men than the aforementioned study. We based our AF ascertainments on registries containing complete follow-up information, independent of the facilities to which patients were admitted. Furthermore, we used age as the underlying timescale in our analyses because of the strong association between age and AF. In contrast to O’Neal et al.,11 we noted a sharp increase in the cumulative incidence in patients with CACSs ≥ 1000, probably because a high CACS is associated with a high revascularization rate, and CABG, which is performed for revascularization, is known to be associated with a high risk of AF. The exact pathogenic processes explaining how patients progress from coronary calcification to AF are not clear. Several mechanisms underlying the relationship between coronary obstruction/occlusion and risk of AF have been suggested. CAD, which is correlated with the CACS,24 is associated with low-grade subclinical inflammation, and ischaemia or atrial myocardial infarction is associated with a healing process that triggers myocardial damage and atrial inflammation.25,26 Inflammatory mediators seem to alter electrical and structural remodelling and thus trigger AF. Additionally, inflammation also seems to play a role in post-operative AF, for example, AF after CABG.26 Furthermore, it has been shown that the extent of CACS progression may be an important factor in the relationship between CACS and AF.27 The relationship between CACS and AF may also be explained by the fact that a high CACS is associated with dilated pulmonary veins and a larger left atrium, which have the potential to initiate and perpetuate atrial re-entrant circuits.28 Our results may have long-term clinical implications. We noted a moderate AUC when using CACS as a tool for predicting AF. Hence, CACSs should be interpreted in connection with patient risk profiles. Adding CACS to risk scores from the Framingham Heart Study and Aging Research in Genomic Epidemiology (CHARGE)-AF improves the AUC with respect to predicting AF.11 Using CACS in combination with knowledge of other risk factors makes it possible to identify patients at high risk for AF, in whom the tolerance of modifiable AF risk factors should be set at an even lower level than in patients not at high risk of AF, and disease prevention efforts should be more aggressive. Accordingly, using CACS in combination with knowledge regarding other AF risk factors will enable AF prevention and earlier treatment start-up in patients who most likely develop AF and thus reduce the number of incident strokes among these patients. Given that the number of patients with AF is increasing, this potential improvement in early AF detection may ultimately benefit more patients and have important public health consequences. However, additional studies are needed to examine the usefulness of CACS for stroke reduction in AF patients. The main strengths of our study were its use of a population-based method, its large population size, and its application of prospectively collection of data. Additionally, this study was strengthened by the fact that it utilized the Western Denmark Heart Registry, whose data are characterized by a high degree of completeness and validity,14 and that it uses Danish registers to ensure that patient follow-up was complete. Migration bias was unlikely because patient migration status is recorded in the Civil Registration System. The diagnosis of AF in the National Patient Registry has been found to have a high positive predictive value.23 Patients enrolled in this study do not represent the general population, as they were suspected of having CAD. According to a report from the Western Denmark Heart Registry 2014, 58% of patients undergoing nCCT do not have diseased vessels, and nCCT does not lead to consequences in terms of treatment in 63% of patients.29 Using data on antiarrhythmic medication as exclusion may have excluded patients with AF before baseline more accurately; unfortunately, those data were unavailable. Differential misclassification may occur if patients with a high CACS are followed more carefully than patients with a CACS of 0, which may lead to a higher likelihood of being diagnosed with AF. It is possible that differences may exist in the way the CACSs were measured, which could have introduced some misclassification, most likely of a non-differential nature. It should also be stressed that our finding of CACS being a predictor of AF does not imply causality. We were not able to distinguish between persistent, permanent, and paroxysmal AF; therefore, paroxysmal AF might be underreported. Furthermore, residual confounding is possible, for instance, with respect to categorizing smoking, and a lack of information on potential confounders, such as ethnicity and alcohol consumption, may lead to imprecision. Conclusions A high CACS is associated with a high risk of AF. CACS combined with knowledge regarding other AF risk factors may help clinicians to identify patients at high risk for AF, which may facilitate both disease prevention and earlier treatment of patients who most likely develop AF and thus reduce their risk of stroke. Acknowledgements We are grateful for the grants received by Regional Hospital Central Jutland, Viborg, Denmark; Aarhus University, Aarhus, Denmark; and Grosserer A. V. Lykfeldt og Hustrus Legat, Copenhagen, Denmark. Funding Grants received from Regional Hospital Central Jutland, Viborg, Denmark; Aarhus University, Aarhus, Denmark; and Grosserer A. V. Lykfeldt og Hustrus Legat, Copenhagen, Denmark. The funders had no role in the design or conduction of the study; the management, analysis, and interpretation of the data; the preparation, review, or approval of the article; or the decision to submit the manuscript for publication. Conflict of interest: None declared. References 1 Krijthe BP, Kunst A, Benjamin EJ, Lip GY, Franco OH, Hofman A et al.   Projections on the number of individuals with atrial fibrillation in the European Union, from 2000 to 2060. Eur Heart J  2013; 34: 2746– 51. Google Scholar CrossRef Search ADS PubMed  2 Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Abhayaratna WP et al.   Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence. Circulation  2006; 114: 119– 25. Google Scholar CrossRef Search ADS PubMed  3 Gassett AJ, Sheppard L, McClelland RL, Olives C, Kronmal R, Blaha MJ et al.   Risk factors for long-term coronary artery calcium progression in the multi-ethnic study of atherosclerosis. J Am Heart Assoc  2015; 4: e001726. Google Scholar CrossRef Search ADS PubMed  4 Oei HH, Vliegenthart R, Hofman A, Oudkerk M, Witteman JC. Risk factors for coronary calcification in older subjects. The Rotterdam Coronary Calcification Study. Eur Heart J  2004; 25: 48– 55. Google Scholar CrossRef Search ADS PubMed  5 Schnabel RB, Yin X, Gona P, Larson MG, Beiser AS, McManus DD et al.   50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: a cohort study. Lancet  2015; 386: 154– 62. Google Scholar CrossRef Search ADS PubMed  6 Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte MJr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol  1990; 15: 827– 32. Google Scholar CrossRef Search ADS PubMed  7 Shemesh J, Henschke CI, Shaham D, Yip R, Farooqi AO, Cham MD et al.   Ordinal scoring of coronary artery calcifications on low-dose CT scans of the chest is predictive of death from cardiovascular disease. Radiology  2010; 257: 541– 8. Google Scholar CrossRef Search ADS PubMed  8 Raggi P, Gongora MC, Gopal A, Callister TQ, Budoff M, Shaw LJ. Coronary artery calcium to predict all-cause mortality in elderly men and women. J Am Coll Cardiol  2008; 52: 17– 23. Google Scholar CrossRef Search ADS PubMed  9 Chang SM, Nabi F, Xu J, Peterson LE, Achari A, Pratt CM et al.   The coronary artery calcium score and stress myocardial perfusion imaging provide independent and complementary prediction of cardiac risk. J Am Coll Cardiol  2009; 54: 1872– 82. Google Scholar CrossRef Search ADS PubMed  10 Handy CE, Desai CS, Dardari ZA, Al-Mallah MH, Miedema MD, Ouyang P et al.   The association of coronary artery calcium with noncardiovascular disease: the multi-ethnic study of atherosclerosis. JACC Cardiovasc Imaging  2016; 9: 568– 76. Google Scholar CrossRef Search ADS PubMed  11 O'Neal WT, Efird JT, Dawood FZ, Yeboah J, Alonso A, Heckbert SR et al.   Coronary artery calcium and risk of atrial fibrillation (from the multi-ethnic study of atherosclerosis). Am J Cardiol  2014; 114: 1707– 12. Google Scholar CrossRef Search ADS PubMed  12 Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke  1991; 22: 983– 8. Google Scholar CrossRef Search ADS PubMed  13 Johnsen SP, Svendsen ML, Hansen ML, Brandes A, Mehnert F, Husted SE. Preadmission oral anticoagulant treatment and clinical outcome among patients hospitalized with acute stroke and atrial fibrillation: a nationwide study. Stroke  2014; 45: 168– 75. Google Scholar CrossRef Search ADS PubMed  14 Nielsen LH, Norgaard BL, Tilsted HH, Sand NP, Jensen JM, Bottcher M et al.   The Western Denmark Cardiac Computed Tomography Registry: a review and validation study. Clin Epidemiol  2015; 7: 53– 64. Google Scholar PubMed  15 Schmidt M, Maeng M, Jakobsen CJ, Madsen M, Thuesen L, Nielsen PH et al.   Existing data sources for clinical epidemiology: the Western Denmark Heart Registry. Clin Epidemiol  2010; 2: 137– 44. Google Scholar CrossRef Search ADS PubMed  16 Pedersen CB, Gotzsche H, Moller JO, Mortensen PB. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull  2006; 53: 441– 9. Google Scholar PubMed  17 Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B et al.   2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J  2016; 37: 2893– 962. Google Scholar CrossRef Search ADS PubMed  18 Sabour S, Rutten A, van der Schouw YT, Atsma F, Grobbee DE, Mali WP et al.   Inter-scan reproducibility of coronary calcium measurement using Multi Detector-Row Computed Tomography (MDCT). Eur J Epidemiol  2007; 22: 235– 43. Google Scholar CrossRef Search ADS PubMed  19 Ghadri JR, Goetti R, Fiechter M, Pazhenkottil AP, Kuest SM, Nkoulou RN et al.   Inter-scan variability of coronary artery calcium scoring assessed on 64-multidetector computed tomography vs. dual-source computed tomography: a head-to-head comparison. Eur Heart J  2011; 32: 1865– 74. Google Scholar CrossRef Search ADS PubMed  20 Williams MC, Golay SK, Hunter A, Weir-McCall JR, Mlynska L, Dweck MR et al.   Observer variability in the assessment of CT coronary angiography and coronary artery calcium score: substudy of the Scottish COmputed Tomography of the HEART (SCOT-HEART) trial. Open Heart  2015; 2: e000234. Google Scholar CrossRef Search ADS PubMed  21 Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health  2011; 39: 30– 3. Google Scholar CrossRef Search ADS PubMed  22 Rix TA, Riahi S, Overvad K, Lundbye-Christensen S, Schmidt EB, Joensen AM. Validity of the diagnoses atrial fibrillation and atrial flutter in a Danish patient registry. Scand Cardiovasc J  2012; 46: 149– 53. Google Scholar CrossRef Search ADS PubMed  23 Schmidt M, Schmidt SA, Sandegaard JL, Ehrenstein V, Pedersen L, Sorensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol  2015; 7: 449– 90. Google Scholar CrossRef Search ADS PubMed  24 Cheng YJ, Church TS, Kimball TE, Nichaman MZ, Levine BD, McGuire DK et al.   Comparison of coronary artery calcium detected by electron beam tomography in patients with to those without symptomatic coronary heart disease. Am J Cardiol  2003; 92: 498– 503. Google Scholar CrossRef Search ADS PubMed  25 Dzeshka MS, Lip GY, Snezhitskiy V, Shantsila E. Cardiac fibrosis in patients with atrial fibrillation: mechanisms and clinical implications. J Am Coll Cardiol  2015; 66: 943– 59. Google Scholar CrossRef Search ADS PubMed  26 Hu YF, Chen YJ, Lin YJ, Chen SA. Inflammation and the pathogenesis of atrial fibrillation. Nat Rev Cardiol  2015; 12: 230– 43. Google Scholar CrossRef Search ADS PubMed  27 O'Neal WT, Efird JT, Qureshi WT, Yeboah J, Alonso A, Heckbert SR et al.   Coronary artery calcium progression and atrial fibrillation: the multi-ethnic study of atherosclerosis. Circ Cardiovasc Imaging  2015; 8: e003786. Google Scholar CrossRef Search ADS PubMed  28 Pan NH, Tsao HM, Chang NC, Lee CM, Chen YJ, Chen SA. Dilated left atrium and pulmonary veins in patients with calcified coronary artery: a potential contributor to the genesis of atrial fibrillation. J Cardiovasc Electrophysiol  2009; 20: 153– 8. Google Scholar CrossRef Search ADS PubMed  29 Secretariat. Western Denmark Heart Registry: Year Report 2015. Activities and Results. http://vdhd.dk/wp-content/uploads/2014/10/Årsrapport-2014.pdf (31 August, 2016, date last accessed). Published on behalf of the European Society of Cardiology. All rights reserved. © The Author 2017. For permissions, please email: journals.permissions@oup.com.

Journal

European Heart Journal – Cardiovascular ImagingOxford University Press

Published: Aug 4, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off