Atrial fibrillation (AF) secondary to seizure has been described in case reports, but the association between AF and risk of seizure has never been evaluated in longitudinal studies. The objectives of this study were to investigate the role of AF on the risk of development of seizure and the usefulness of CHADS 2 score for predicting the risk of seizure. Our analyses were conducted using information from a random sample of 1 million subjects enrolled in Taiwan National Health Insurance Research Database. A total of 11,552 subjects aged ≥18 years, comprising 5,776 subjects diagnosed with AF during the study period and 5,776 age and sex-matched subjects without AF were enrolled in our study. During the mean follow-up of 6.7 ± 3.3 years, seizure events occurred in 235 patients. In comparison, the AF group had a higher incidence rate of seizure occurrence (4.17 vs 1.90 per 1,000 person-years). Cox proportional hazard regression model analysis showed that development of AF was independently associated with a higher risk of developing future seizure (adjusted HR 2.30; 95% confidence interval 1.73 to 3.05). In multivariate Cox regression analysis adjusted for potentially confounding variables, a higher CHADS 2 score was associated with a higher risk of seizure in a dose-dependent manner. AF may cause an ischemic stroke that subsequently leads to seizure, and present study further demonstrates that AF patients are associated with higher rate of subsequent seizure, even after adjusting for stroke. The CHADS 2 score was found to be a useful scheme for predicting the risk of seizure occurrence.
In addition to stroke, recent evidence has shown that atrial fibrillation (AF) is associated with an increased risk of cognitive impairment and development of dementia, which is probably because of reduced cardiac output and cerebral blood flow resulting from irregular heartbeat and lack of atrial contractions. Moreover, the relation between AF and other neurologic complications, such as seizure disorders, has also been reported in the literature, and AF secondary to prolonged epileptic seizures has been described in previous case reports. It has been postulated that an increase in sympathetic tone and release of catecholamines contributes to the development of AF following generalized tonic-clonic seizures. However, no previous studies to date have looked for a relation in the other direction, that is, examining the influence of AF on seizure incidence in a longitudinal manner. The aim of the present study was to investigate whether AF is associated with a higher risk of new-onset seizure using a nationwide, population-based, longitudinal cohort database in Taiwan. We also incorporated the CHADS 2 score to test the hypothesis that this stroke prediction scoring system could also be useful in predicting the risk of seizure occurrence.
Method
This study used data from the National Health Insurance Research Database (NHIRD) released by the Taiwan National Health Research Institutes (NHRI). The Taiwan NHRI program, established in 1995, has enrolled nearly all (99%) the inhabitants of Taiwan. This cohort data set comprises 1,000,000 randomly sampled beneficiaries enrolled in the NHI program during 2000 and is a collection of all available records on these subjects for the years 1996 to 2011. Each patient’s original identification number is encrypted to protect individual privacy in a consistent manner, allowing the linkage of claims belonging to a particular patient within the database. Diseases are coded according to the International Classification of Disease , Ninth Revision, Clinical Modification ( ICD-9-CM ) diagnosis codes, 2001 edition. The accuracy of diagnoses in the NHIRD has been validated previously, and numerous high-quality scientific research reports have been published using data from NHIRD.
From January 1, 2000, to December 31, 2010, a total of 5,776 patients with AF aged ≥18 years without history of seizure ( ICD-9-CM code: 345.xx) preceding study enrollment were retrieved from the NHIRD because the record of NHIRD can be traced back to 1996. AF was diagnosed using the ICD-9-CM codes 427.31. To ensure accuracy of diagnosis, we defined patients with AF only when it was listed as a discharge diagnosis or confirmed more than twice by an outpatient department. The diagnostic accuracy of AF using this definition in NHIRD has been validated previously. The time of AF diagnosis was defined as the time of enrollment. The age- and sex-matched (1:1) control cohort was randomly identified from among 1,000,000 subjects after eliminating those who had been diagnosed with AF at any time and those with previous incident of seizure before enrollment. The diagnosis of seizure was based on the ICD-9-CM codes (345.xx) registered by the physicians responsible for the treatments of patients. Information about important co-morbid conditions of each individual was retrieved from the medical claims based on the ICD-9-CM codes. We defined patients with a certain disease only when it was a discharge diagnosis or repeatedly confirmed (more than twice) by an outpatient department. Similar methods for identifying patients with seizure have also been applied previously. The study end point was the new occurrence of seizure ( ICD-9-CM code: 345.xx) before the end of 2011. The onset of seizure after the diagnosis of AF was defined as the date on which the diagnostic code for seizure was noted for the first time in the records.
Sensitivity analysis (test 1) was performed using different diagnostic criteria for seizure with the ICD-9-CM codes 345.xx given by the neurologists, neurosurgeons, or emergency room doctors after the electroencephalographic study. We also performed a separate sensitivity analysis (test 2) to investigate the aforementioned associations after excluding the first year of observation.
Descriptive statistics were used to describe the baseline characteristics of our cohort. Baseline characteristics of the 2 groups were compared using Pearson chi-square tests for categorical variables and the independent t test and Mann-Whitney U test for parametric and nonparametric continuous variables, respectively. The incidence rate of seizure between the 2 groups was calculated by Poisson distribution. The risk of seizure between groups was calculated according to the hazard ratio from Cox regression analysis after adjusting for demographic data and medical co-morbidities. The cumulative incidence curve of seizure was plotted by way of the Kaplan–Meier method, with statistical significance examined by the log-rank test. Furthermore, we used Cox regression analysis to calculate the predictive power of CHADS 2 score with the risk of subsequent seizure. A p value of <0.05 was considered statistically significant. All data processing and statistical analyses were performed using Statistical Package for Social Science (SPSS), version 17.0, software (SPSS Inc.) and Statistical Analysis Software (SAS), version 9.1 (SAS Institute, Cary, North Carolina).
Results
In total, 5,776 patients with a diagnosis of AF who met the inclusion criteria from January 2000 to December 2010 were identified. The baseline characteristics of the study subjects are listed in Table 1 . The AF group exhibited more co-morbidities with hypertension, coronary artery disease, and heart failure than the non-AF group. More patients with AF were taking antiplatelet drugs, oral anticoagulants, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, calcium channel blockers, thiazides, and statins compared with those in the non-AF group.
| Variable | Atrial Fibrillation | P value | |
|---|---|---|---|
| No (n = 5,776) | Yes (n = 5,776) | ||
| Age (years) | 73.6 ± 12.7 | 73.4 ± 12.9 | 0.350 |
| Age ≥ 75 years | 1,640 (28.4%) | 1,659 (28.7%) | 0.711 |
| Men | 3,260 (56.4%) | 3,260 (56.4%) | >0.99 |
| Income-related insured amount ∗ /months | 0.084 | ||
| NT$ ≤ 15,840 | 3,395 (58.8%) | 3,315 (57.4%) | |
| NT$ 15,841-25,000 | 1,817 (31.5%) | 1,828 (31.6%) | |
| > NT$ 25,000 | 564 (9.8%) | 633 (11.0%) | |
| Urbanization | 0.696 | ||
| Level 1 (most urbanized) | 1,475 (25.5%) | 1,533 (26.5%) | |
| Level 2 | 1,708 (29.6%) | 1,664 (28.8%) | |
| Level 3 | 870 (15.1%) | 852 (14.8%) | |
| Level 4 | 912 (15.8%) | 931 (16.1%) | |
| Level 5 (most rural) | 811 (14.0%) | 796 (13.8%) | |
| Underlying diseases | |||
| Hypertension | 2,940 (50.9%) | 3,048 (52.8%) | 0.044 |
| Diabetes mellitus | 1,062 (18.4%) | 1,002 (17.3%) | 0.145 |
| Coronary artery disease | 1,554 (26.9%) | 1,685 (29.2%) | 0.007 |
| Heart failure | 482 (8.3%) | 678 (11.7%) | <0.001 |
| Myocardial infarction | 31 (0.5%) | 31 (0.5%) | >0.99 |
| Stroke | 1,251 (21.7%) | 1,309 (22.7%) | 0.194 |
| Chronic kidney disease | 262 (4.5%) | 291 (5.0%) | 0.207 |
| Chronic obstructive pulmonary disease | 1,287 (22.3%) | 1,352 (23.4%) | 0.150 |
| Valvular heart disease | 175 (3.0%) | 187 (3.2%) | 0.557 |
| Peripheral artery disease | 194 (3.4%) | 209 (3.6%) | 0.478 |
| Head injury | 196 (3.4%) | 211 (3.7%) | 0.480 |
| Central nervous system infection | 18 (0.3%) | 26 (0.5%) | 0.290 |
| CHADS 2 score | 1.49 ± 1.52 | 1.56 ± 1.46 | 0.018 |
| Medication use, n (%) | |||
| Antiplatelet agents | 2,100 (36.4%) | 2,960 (51.2%) | <0.001 |
| Oral anticoagulants | 62 (1.1%) | 407 (7.0%) | <0.001 |
| ACEIs/ ARBs | 3,025 (52.4%) | 3,897 (67.5%) | <0.001 |
| Calcium channel blockers | 2,895 (50.1%) | 3,543 (61.3%) | <0.001 |
| Thiazides | 1,626 (28.2%) | 1,975 (34.2%) | <0.001 |
| Statins | 1,017 (17.6%) | 957 (16.6%) | 0.145 |
During the follow-up period, there were 235 newly diagnosed seizure events: 160 in the AF cohort and 75 in the matched cohort, respectively. The incidence rate of seizure was 4.17 per 1,000 person-years in the AF group and 1.90 per 1,000 person-years in the non-AF group ( Supplementary Table 1 ). Compared with the matched cohort, the AF group was associated with an increased risk of seizure. The presence of AF was significantly associated with an increased risk of seizure, with a hazard ratio of 2.19 (95% confidence interval [CI] 1.67 to 2.88) in the univariate Cox regression analysis and 2.30 (95% CI 1.73 to 3.05) after adjusting for baseline characteristics, medication use, and socioeconomic status ( Table 2 ). Patients with AF are associated with higher rate of subsequent seizure, even after adjusting for time-updated incident stroke events during the follow-up period. The result of Kaplan-Meier survival analysis log-rank test for seizure is shown in Figure 1 . During the follow-up period, the patients with AF had significantly higher risk of seizure than those in the non-AF group.
| Models | Hazard ratio ∗ | 95% CI | P value |
|---|---|---|---|
| Model 1: unadjusted regression analysis | 2.19 | 1.67-2.88 | <0.001 |
| Model 2: adjusted for hypertension, DM, heart failure, stroke, myocardial infarction, CAD, COPD, CKD, VHD, PAD, head injury and CNS infection | 2.19 | 1.67-2.89 | <0.001 |
| Model 3: adjusted for all variables in model 2, and the use of antiplatelet agents, OACs, ACEIs/ARBs, CCBs and statins | 2.28 | 1.72-3.03 | <0.001 |
| Model 4: adjusted for all variables in model 3, and income level | 2.30 | 1.73-3.05 | <0.001 |
In multivariate Cox regression analysis adjusted for potentially confounding variables, a higher CHADS 2 score was associated with a higher risk of seizure occurrence in a dose-dependent manner ( Table 3 ). The Kaplan-Meier survival analysis showed that patients with a higher CHADS 2 score were associated with a higher risk of seizure during the follow-up period (log-rank p <0.0001, trend p <0.0001; Figure 2 ). The incidence of seizure augmented gradually with each point increase in the CHADS 2 score. To further evaluate individual risk factor’s contribution to the predictive power of CHADS 2 score, we performed univariate Cox regression analysis for age, hypertension, diabetes, heart failure, and stroke, respectively. The results are listed in Supplementary Table 2 .
| CHADS 2 score | Number of patients | Number of new-onset seizure | Person-years | Incidence ∗ | Hazard ratio (95% CI) | P value |
|---|---|---|---|---|---|---|
| 0 | 3,748 | 48 | 29,592 | 1.62 | As Reference | |
| 1 | 2,933 | 50 | 20,058 | 1.70 | 1.56 (1.05-2.32) | 0.029 |
| 2 | 1,961 | 43 | 12,031 | 3.57 | 2.26 (1.49-3.42) | <0.001 |
| 3 | 1,425 | 40 | 8,531 | 4.69 | 2.95 (1.94-4.51) | <0.001 |
| ≥ 4 | 1,485 | 54 | 7,648 | 7.06 | 4.47 (3.01-6.64) | <0.001 |
| Total | 11,552 | 235 | 77,860 | 3.02 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
