Statistical analysis

Continuous variables are expressed as mean ± standard deviation (SD), or median and interquartile range. Categorical variables are presented as numbers and percentages. We compared categorical variables using the chi-square test when appropriate; otherwise, we used Fisher’s exact test. We compared continuous variables using Student’s t-test on the basis of the distribution. We stratified the entire cohort by OAC prescription at baseline, and compared the baseline characteristics between patients without OAC and those with it.

To investigate the predictors for death/stroke/SE, we analyzed the patients without OAC at baseline. First, we tested the predictive ability of the CHADS₂ and CHA₂DS₂-VASc scores on the composite endpoint of ‘death/stroke/SE’ in patients without OAC, using a receiver-operator characteristic curve analysis (as a measure of the C-index). Thereafter, we performed univariate and multivariate analysis using Cox proportional hazard models to explore the predictors for composite endpoint of ‘death/stroke/SE’. All clinically relevant potential risk factors for death/stroke/SE were included on multivariate analysis. The patients with at least 1 missing covariate were excluded on multivariate analysis. We also analyzed the patients without OAC throughout follow-up period.

We investigated whether the cumulative number of each risk factors which were significant on multivariate analysis could stratify the incidence of ‘death/stroke/SE’ and ‘stroke/SE’ during the follow-up period. The continuous distribution of cumulative risk factors was then stratified into 5 categories, and the cumulative incidence of death/stroke/SE across these 5 categories were estimated by Kaplan-Meier method, and compared by the log-rank test. We also investigated whether the cumulative number of risk factors could stratify the incidence of events in AF patients taking OAC. Finally, we evaluated the model predictive ability for the cumulative number of risk factors on the composite endpoint of ‘death/stroke/SE’ in patients without OAC and in those with OAC, again using a receiver-operator characteristic curve analysis. We used JMP version 10 (SAS Institute, Cary, NC) to perform all analyses. Two-sided P values less than 0.05 were considered as statistically significant.

Ethics

The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki, and was approved by the ethical committees of the National Hospital Organization Kyoto Medical Center and Ijinkai Takeda General Hospital. Since the present research involves an observational study not using human biological specimens, written informed consent was not obtained from each patient for their clinical records to be used in this study, according to the ethical guidelines for epidemiological research issued by the Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Health, Labour and Welfare, Japan. However, we have published all relevant details regarding this study to be carried out and provide each patient an opportunity to refuse inclusion in this research by posting the details at every participating clinic and at the homepages of our institutions. We also held a public meeting with citizens in Fushimi-ku to demonstrate outlines of the present study. As consent was not obtained, patient records and information was anonymized and de-identified prior to analysis.

CHADS₂/CHA₂DS₂-VASc scores and death/stroke/SE

In our registry, congestive heart failure, advanced age (≥75 years), and history of stroke/SE/TIA were independent predictors for the incidence of death/stroke/SE. Meanwhile, hypertension, diabetes mellitus, or female sex were not significantly associated with increased risk of death/stroke/SE. They were reported to be risk factors for stroke, but the association has been inconsistent [14, 15]. Well-controlled hypertension and diabetes mellitus may not be a risk factor for adverse events of AF [16]. Moreover, neither hypertension nor diabetes mellitus was associated with the stroke in Chinese AF patients [17], and neither was female sex in Japanese patients [18], suggesting that the risk factors for stroke/SE may possibly differ between Western and Asian countries.

Anemia and death/stroke/SE in AF patients

Anemia is associated with increased mortality and morbidity in various cardiovascular diseases [19–21]. Anemia is also an independent predictor of 1-year all-cause death and re-hospitalization in elderly patients hospitalized with AF [22], and has been associated with major adverse cardiac and cerebrovascular events in patients with AF undergoing percutaneous coronary intervention [23]. A plausible mechanism whereby anemia increases the mortality may be a high risk of coronary ischemic events [19, 20], but cardiac death was relatively small in our registry. Alternatively, the higher rate of all-cause death and cardiovascular events might be related to the higher risk profile in anemic patients and underlying diseases causing anemia. Despite some studies of increased mortality and morbidity in anemic patients with cardiovascular diseases, the results of therapeutic intervention trials targeting anemia have been inconsistent [23]. Although we cannot address whether anemia is a mediator and therefore potentially amenable to therapeutic intervention, or merely a marker of adverse events in AF patients, our data showed that anemia is an independent, and strongest predictor for death/stroke/SE on multivariate analysis.

CKD and death/stroke/SE in AF patients

CKD is considered to be an independent predictor for stroke/SE in AF [24–26], although it does not incrementally add to existing stroke risk scores [27]. CKD is also considered to be an independent predictor for composite endpoints of death/stroke/SE [28, 29]. Multiple possible mechanisms exist for the association between CKD and increased risks of death and cardiovascular diseases; reduced kidney function is associated with increased levels of inflammatory factors, enhanced coagulability, arterial calcification, arterial stiffness, and endothelial dysfunction [30]. Our data show that CKD was independently associated with increased risk of death/stroke/SE in AF patients, consistent with previous studies.

Underweight and death/stroke/SE in AF patients

Being overweight has been associated with a lower risk of all-cause mortality and cardiovascular mortality in AF [31–33], while another study suggested sex-related differences [34]. These studies excluded underweight patients with a BMI <18.5 kg/m², and the characteristics of underweight patients with AF are unknown. Japanese patients with AF are generally small and lean, and the Fushimi AF Registry contains many frail elderly patients who are underweight. We recently demonstrated that underweight was associated with a higher risk of stroke/SE, despite that the mechanisms remain speculative [35]. Our study showed the association between underweight and death/stroke/SE, but the impact of promoting weight gain in patients with AF is controversial. The recent trial suggests that weight reduction is actually associated with better outcomes in AF [36].

Study limitations

Our study has several limitations. First, our data are based on an observational study and provides only associative evidence, not causative. Second, OAC was prescribed to only 53% of all the participating patients, and 60% of the patients with CHADS₂ score ≥2 points in our registry, based on the responsible attending physician. However, OAC is often underused amongst AF patients in daily clinical practice, due to concerns about the bleeding complications [37, 38]. Our registry included many old and frail patients, who were perhaps considered unsuitable for OAC. Third, we did not collect the data of time in therapeutic range in patients with warfarin, and we only stratified entire cohort by OAC prescription at baseline. Indeed, there were some changes in the status of OAC prescription seen during follow-up period. However, our sensitivity analysis still shows that the 6 risk factors remained independently associated with the risk of death/stroke/SE in patients who had never taken any OAC throughout the follow-up period. Whilst exclusion of any OAC use during follow-up may result in conditioning for the future, but despite this limitation, our risk stratification is still predictive of the composite endpoint. Fourth, among 1,541 patients without OAC, 233 patients’ BMI data and 111 patients’ laboratory data of hemoglobin were missing, leading to selection bias. Fifth, the number of patients and incidence of endpoints might be underpowered to detect the impact of some risk factors, such as hypertension and diabetes mellitus. Finally, all enrolled patients were Japanese, thus the body compositions or other factors might differ from other populations. Our risk factors require external validation, with a view to development of risk stratification schemes that can be used in Asian AF patients.

Working group members

The following is a list of the institutions participating in the registry.

Chief investigator: Akao M (National Hospital Organization Kyoto Medical Center).

Vice-chief investigator: Chun YH (Ijinkai Takeda General Hospital).

Steering Committee: Esato M (Ijinkai Takeda General Hospital), Abe M (National Hospital Organization Kyoto Medical Center), Tsuji H (Tsuji Clinic), Furuke K (Furuke Clinic).

Statistical Analysis: Wada H (National Hospital Organization Kyoto Medical Center).

Coordinator: Ogawa T (Ogawa Medical Office), Tasato H (Tasato Clinic), Taniguchi Y (Taniguchi Clinic), Nishikawa M (Nishikawa Clinic), Furukawa K (Furukawa Medical Clinic), Kawai C, Hashimoto T, Kanda M (Ijinkai Takeda General Hospital), Tsukahara T, Fukuda S, Nakamura M, Ito T, Hasegawa K (National Hospital Organization Kyoto Medical Center).

Clinical Event Committee: Kawabata Y, Yasuda K, Kameyama K (National Hospital Organization Kyoto Medical Center).

Participating institutions: Department of Cardiology, National Hospital Organization Kyoto Medical Center (Akao M, Abe M, Ogawa H, Masunaga N, Iguchi M, Ishii M, Unoki T, Takabayashi K, Hamatani Y, Yamashita Y, Takagi D, Niki S, An Y, Tezuka Y, Osakada G, Nakashima Y, Kanasaki M, Nakano T, Funatsu J, Nishio M, Takenaka Y); Department of Arrhythmia, Ijinkai Takeda General Hospital (Chun YH, Esato M, Kida Y, Nishina N); Koujinkai Oshima Hospital (Terada K); Divison of Translational Research, National Hospital Organization Kyoto Medical Center (Hasegawa K, Wada H); Kanai Hospital (Nishio M, Kamiya Y, Abe M, Ishii M); Tsuji clinic (Tsuji H); Furukawa Medical Clinic (Furukawa K); Nishikawa Clinic (Nishikawa M); Taniguchi Clinic (Taniguchi Y); Gushiken Clinic (Gushiken T); Fushimi Shimizu Hospital (Hirata Y); Yoda Clinic (Yoda J); Tasato Clinic (Tasato H); Ogawa Medical Office (Ogawa T); Saiwai Hospital (Wakatsuki Y, Yahata M, Higashitani N); Itoh Hemodialysis Clinic (Itoh H); Itoh Clinic (Itoh H, Ohmori Y); Ryokuhoukai Tsuji Clinic (Tsuji K); Kitamura Clinic (Kitamura S); Izumikawa Clinic (Izumikawa F); Hirota Clinic (Hirota N); Kyomachi-Oota Clinic (Oota K); Kouseikai Rehabilitation Clinic (Kou K); Inariyama Hospital (Tanaka T, Iguchi M); Matsushita Clinic (Matsushita N); Kitani Clinic (Kitani K); Kimura Clinic (Kimura F); Hayashi Clinic (Hayashi S); Handa Clinic (Handa S); Soseikai General Hospital (Hasegawa S, Kono T, Otsuka K, Soyama A, Okamoto J, Nakai Y); Asamoto Clinic (Asamoto H); Sugano Clinic (Tanaka H, Murata T); Kyoto Ohashi General Hospital (Kayawake S, Kinoshita Y); Furuke Clinic (Furuke K); Kanehisa Clinic (Asano N); Tahara Clinic (Tahara K); Matsumoto Medical Office (Matsumoto K); Kuroda Clinic (Kuroda O); Ochiai Clinic (Ochiai K, Ochiai J); Fujii Clinic (Fujii M); Kurihara Clinic (Kurihara M); Kuzuyama Clinic (Ito A); Kenkokai Fushimi Clinic (Totsuzaki S); Nakayama Orthopedic Clinic (Nakayama H); Department of Cardiovascular Medicine, Ijinkai Takeda General Hospital (Kawai C, Hashimoto T, Kakio T, Watanabe C, Takeda S, Sasaki Y, Shirasawa K, Beppu K, Inoue T, Shirasaka A, Doi T); Tatsumi Clinic (Ueda T); Oishi Clinic (Oishi M); Koizumi Clinic (Kasahara A); Kishida Clinic (Kishida Su, Kishida Sa); Shibata Clinic (Shibata M); Shimizu Clinic (Shimizu J); Shirasu Clinic (Shirasu M); Fujinokai Clinic (Tateishi S); Tsukuda Clinic (Tsukuda N); Shinseikai Tsuji Clinic (Tsuji K); Nishi Clinic (Nishi T); Nishimura Clinic (Nishimura S); Haba Clinic (Haba T); Higashimae Clinic (Higashimae R); Fujimori Clinic (Fujimori C); Hotta Clinic (Hotta T); Matsui Clinic (Matsui H, Matsui H); Shadan Matsumoto Clinic (Matsumoto H); Maruo Clinic (Maruo N); Misu Clinic (Mikami M); Mekata Clinic (Mekata H); Mori Pediatric Clinic (Mori H); Wakabayashi Clinic (Wakabayashi M); Nakatsugawa Clinic (Sasaki Z); Shiseikai Nishimura Clinic (Nishimura S); Yuge Eye Clinic (Yuge K); Gokita Hospital (Haruta M); Soseikai Clinic (Tsuda E); Toujinkai Hospital (Nishimura M); Kouno Clinic (Kouno T); Matsumura Clinic (Matsumura S); Fujita Clinic (Fujita A); Takayasu Clinic (Takayasu F, Takayasu S), Yano Clinic (Yano Y), Niki Clinic (Niki S), Hasegawa Meiando Clinic (Hasegawa S), Watanabe Medical Clinic (Watanabe T).