Impact of Hemoglobin Levels and Their Dynamic Changes on the Risk of Atrial Fibrillation: A Nationwide Population-Based Study

Anemia is a risk factor for cardiovascular disease, but its impact on new-onset atrial fibrillation (AF) is unclear. In this study, we investigated the effect of hemoglobin (Hb) levels and their changes on the risk of AF development in the general population of Korea. We retrospectively analyzed a cohort from the Korean National Health Insurance Service database and identified 9,686,314 subjects (49.8% male) without a history of AF, aged ≥40 years, and with Hb levels available for both first (2009–2010) and second (2011–2012) health checkups. These subjects were followed up until 2017 to determine AF development. The presence of anemia (Hb level <13 g/dL in men and <12 g/dL in women) was a significant risk factor for AF development. However, Hb levels showed a U-shaped association with incident AF after adjustment for cardiovascular risk factors. AF incidence was lowest at Hb levels of 14–14.9 g/dL in men and 12–12.9 g/dL in women. Among individuals with Hb levels within normal ranges (13–15.9 g/dL in men and 12–14.9 g/dL in women), both decrease and increase in Hb levels at the second measurement outside the normal ranges showed an elevation of AF risk by 11% and 21% for men and 3% and 36% for women, respectively, compared with those who maintained normal Hb levels. In conclusion, low or high Hb levels are associated with an increased risk of incident AF. This study suggests that maintaining Hb levels within the normal ranges confers a low risk of AF development.


Methods
Data sources. We used data from the national health claims database provided by the National Health Insurance Service (NHIS). The NHIS is a mandatory health insurance program managed by the Korean government, which covers 97% of the Korean population (approximately 50 million people). The NHIS database includes each subject's demographic information, diagnoses, their use of inpatient and outpatient services, and prescription claims. Diagnoses are recorded using the International Classification of Disease, 10th Revision, Clinical Modification (ICD-10-CM) codes.
Individuals in the NHIS are recommended to receive standardized health checkups every 2 years. People born in even-numbered years undergo screening every even-numbered years, and those born in odd-numbered Kaplan-Meier curves for the incidence probability of AF between subjects with anemia and without anemia. Compared with individuals without anemia, the incidence of AF increased significantly in those with anemia (log-rank P < 0.001). After adjustment for age, sex, body mass index, smoking status, alcohol use, exercise, economic status, various cardiovascular comorbidities, and kidney function (Cox proportional hazards model 5), subjects with anemia have a 6% increase in the risk of AF compared with those without anemia ( Table 2). Table 3 presents the incidence and risk of AF based on Hb levels. The crude AF incidence in men was highest in the group with the lowest Hb level (<12 g/dL) and gradually decreased as Hb level increased up to 16 g/dL. In women, the crude AF incidence was high in both groups with the lowest and highest Hb levels (<11 and ≥15 g/ dL, respectively). After Cox proportional hazards analysis (model 5), both low and high Hb levels were associated with an increased risk of AF in both sexes. AF risk was lowest in Hb levels of 14-14.9 g/dL in men and 12-12.9 g/ dL in women (reference groups). In other words, Hb levels showed a U-shaped relationship with the risk of AF Follow-up duration (years) 5.9 ± 0.9 5.9 ± 0.8 Table 1. Baseline characteristics of the study population. Hb indicates hemoglobin. *All the P values were <0.001 in the comparison between men and women.
www.nature.com/scientificreports www.nature.com/scientificreports/ development, and individuals with higher Hb levels have a greater risk than those with lower Hb levels ( Figure 3). In men with Hb level ≥16 g/dL, the risk of AF increased by 22% compared with the reference group, and in women with Hb ≥15 g/dL, the risk increased by 29%. Table 4 demonstrates the association between dynamic change in Hb levels over a 2-year interval and risk of AF development. Among subjects with normal Hb levels (13-15.9 g/dL in men and 12-14.9 g/dL in women) at the first health checkup, both decrease and increase in Hb levels at the second measurement outside the normal ranges elevated AF risk by 11% and 21% for men and 3% and 27% for women, respectively, compared with those who maintained normal Hb levels (reference groups). Among those with anemia, increase in Hb levels above normal ranges (≥16 g/dL in men and ≥15 g/dL in women) at the second evaluation significantly increased AF risk by 37% for men and 9% for women, compared with the reference groups. Individuals maintaining high Hb levels above normal ranges in 2 years had a 21% higher risk of AF in men and 36% in women. Taken together, not only subjects with low or high Hb levels but also those with dynamic change in Hb levels outside the normal range showed a higher risk of AF compared with subjects maintaining normal Hb levels in 2 years.   www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
To the best of our knowledge, this is the largest population-based study to evaluate the impact of Hb levels on the risk of AF development. Our study showed that (1) anemia was a risk factor for AF development, (2) Hb levels showed a U-shaped relationship with the risk of incident AF after adjusting for various demographic and cardiovascular risk factors, and (3) maintaining Hb levels within normal range conferred the lowest risk of AF, while maintaining high Hb levels or increase to high Hb levels posed the highest risk of AF.
There are only a few previous studies on anemia and incident AF. In a retrospective cohort study of 2,873 elderly patients, anemia was not associated with new-onset AF during the 2-year follow-up 12 . However, the size of the cohort was small, and the follow-up duration was not long enough to determine the difference in incident AF. In a Japanese population-based cohort study of 132,250 subjects with a mean follow-up period of 13.8 years, lower Hb levels and chronic kidney disease were independently associated with an increased risk of incident AF 13 . The authors divided the total population into three groups according to Hb levels (<13, 13-14.9, 15-17.9 g/dL for men, <11, 11-12.9, 13-15.9 g/dL for women), and the third group was used as a reference. Consequently, the third reference group included people with both normal and high Hb levels. In contrast, we divided the total population into six Hb groups and found that both low and high Hb levels are independent risk factors for incident AF and high Hb levels increase AF risk more strongly. There are another two studies with a rather specific study population. One is a retrospective, nested case-control study that evaluated whether parameters of the blood count are associated with the development of new-onset AF in 198 patients with acute myocardial infarction 14 . Among subjects with acute myocardial infarction, those who developed AF showed a significantly higher hemoglobin level compared with those without AF (14.2 g/dl, IQR 12.4-15 vs. 12.9 g/dl, IQR 11.7-13.8; P < 0.001). The other is a nationwide cohort study of aplastic anemia in Taiwan 15 . Patients with aplastic anemia had a higher risk of incident AF compared with the general population after controlling for the competing risk of death (HR 1.21, 95% CI 0.97-1.50). However, the previous two studies do not represent the general population.
Since Hb values are variable with time, we analyzed the risk of AF by dynamic changes in Hb levels. Our study showed that subjects maintaining Hb levels within normal ranges had the lowest risk of new-onset AF. However, those who had maintained high Hb levels or whose Hb levels had increased to high levels in 2 years had the highest AF risk. Among those with high Hb levels, change in Hb levels to normal ranges resulted in a reduction in AF risk. These results suggest that AF risk may be lowered by controlling the conditions that can increase Hb levels, such as cigarette smoking or chronic obstructive pulmonary disease as well as by avoiding anemia.
Chronic anemia can cause many physiologic changes to the circulatory system. Low blood viscosity and hypoxic vasodilation contribute to low peripheral resistance 16 . Along with increased sympathetic activity, chronic anemia leads to an increment in cardiac output, leading to left ventricular (LV) remodeling 17 . On the contrary, an increased number of red blood cells is related to high blood viscosity and increased peripheral resistance, leading to elevated afterload 18 . Therefore, LV remodeling may also develop in patients with high hematocrit levels resulting from chronic hypoxic conditions such as chronic obstructive pulmonary disease or heavy smoking 19 . This relationship between hematocrit levels and LV remodeling is well demonstrated in the MONICA Augsburg substudy 20 . In this study, hematocrit levels showed a U-shape relationship with LV mass and wall thickness in the general population. Accordingly, LV remodeling is more likely to develop in subjects with low or high Hb levels. This might be the reason why low and high Hb levels increase the risk of AF, as LV hypertrophy itself is an independent risk factor for incident AF 21 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Structural and functional left atrial (LA) remodeling is usually a consequence of impaired LV function and elevated LV filling pressure. However, LA remodeling seems to begin at an early stage of LV remodeling or LV diastolic dysfunction 22,23 . Katayama et al. showed that increased LV mass index and low Hb concentration was independently associated with LA enlargement in patients with normal LV systolic function 24 . This result suggests that hemodynamic change by low or high Hb levels can affect LA remodeling and the development of AF before noticeable change such as LV hypertrophy or systolic dysfunction occurs.
In our study, the risk of AF increased in subjects with anemia and low-normal Hb levels (Hb level <13 and 13-13.9 g/dL in men, and <12 and 12-12.9 g/dL in women) compared with the reference groups. However, the absolute risk was higher in men than women (24% and 8% in men, and 12% and 4% in women, respectively). This difference may additionally explain the sex-difference in AF incidence. In Korea, the estimated incidence of AF is 1.17 times higher in men than in women 4 . The difference in AF risk between sex could be explained by differences in risk factors, genetics, cardiac structure, and hormones 25 . Study limitations. This study has several limitations. First, selection bias could exist, although the number of our study subjects was more than 9 million. The NHIS provides health checkups every 2 years to all health insurers, but only half of the subjects received this health checkup. Therefore, the study population is likely to include those who maintain a healthy lifestyle or are more concerned about their health. Second, the study population was not regularly followed during the study period, so there is a possibility of bias of lost to follow-up. Considering all the Korean population were covered by mandatory national health insurance, most of the subjects could be tracked in the Korean National Health Insurance Service database except for immigration. Third, the definitions of diseases including AF were manipulated using claims data from the NHIS database. The misclassification of diagnostic codes could be the cause of bias, resulting in an underestimation or overestimation.  ; (B) women. The X-axis represents hemoglobin (Hb) groups, while the Y-axis the hazard ratios with reference to Hb levels of 14-14.9 g/dL in men and 12-12.9 g/dL in women. Squares represent hazard ratios, and vertical lines 95% confidence intervals. Hazard ratios were calculated using a Cox proportional hazards model adjusted for age, body mass index, smoking, alcohol consumption, exercise frequency, low income, hypertension, diabetes, dyslipidemia, ischemic heart disease, congestive heart failure, stroke, chronic obstructive pulmonary disease, estimated creatinine clearance, obstructive sleep apnea, and thyroid disease (Model 5 in Table 3). (2020) 10:6762 | https://doi.org/10.1038/s41598-020-63878-9 www.nature.com/scientificreports www.nature.com/scientificreports/ Nevertheless, the diseases were well defined and repeatedly validated in our previous studies 10, 26,27 . Lastly, we were unable to identify the cause of anemia in subjects with anemia and the reasons for high Hb levels in some individuals. However, we adjusted for various factors that may cause low or high Hb levels, such as age, smoking status, ischemic heart disease, congestive heart failure, chronic obstructive pulmonary disease, and kidney function estimated by creatinine clearance.

conclusions
Anemia is a risk factor for incident AF. Both low and high Hb levels are associated with an increased AF risk. Our study suggests that maintaining Hb levels within normal ranges would lower the risk of AF development.  Table 4. Risk of atrial fibrillation according to the dynamic change in hemoglobin levels. AF indicates atrial fibrillation; CI, confidence interval; Hb, hemoglobin; HR, hazard ratio. *Incidence rates were calculated per 1,000 person-years. **HR was calculated by Cox proportional hazards model 5 as in Table 3. † P < 0.001, ‡ P < 0.05, § P ≥ 0.05.