Association of Estimated Glomerular Filtration Rate with Hemoglobin Level in Korean Adults: The 2010–2012 Korea National Health and Nutrition Examination Survey

Purpose Little is known about anemia in patients with early renal dysfunction. We aimed to investigate the association of hemoglobin level and anemia prevalence with estimated glomerular filtration rate (eGFR) decline using a nation-wide representative sample of the adult Korean population. Methods In total, 17,373 participants (7,296 men; weighted n = 18,330,187; mean age, 44.2±0.3 years; 9,886 women, weighted n = 18,317,454; mean age, 46.9±0.3 years) were included. eGFR was divided into 5 groups: Group 1, ≥105; Group 2, 90–104; 75–89; Group 4, 60–74; and Group 5, <60 mL/min/1.73m2. Results The weighted anemia prevalence rates were 2.6% in men and 12.8% in women. In men, the weighted hemoglobin level increased with a decrease in eGFR; this value peaked at an eGFR of 60–89 mL/min/1.73m2 and decreased thereafter at an eGFR of <60 mL/min/1.73m2 (15.19±0.03, 15.35±0.03, 15.53±0.03, 15.52±0.06, and 14.90±0.12 g/dL from Groups 1 to 5) after adjustment for age, college graduation, cancer history, current smoking, waist circumference, serum cholesterol level, serum triglyceride level, and diastolic blood pressure. In women, the weighted hemoglobin level increased with a decrease in eGFR; this value peaked with an eGFR of 75–89 mL/min/1.73m2 and decreased thereafter (12.90±0.03, 13.08±0.02, 13.20±0.04, 13.14±0.05, and 12.47±0.11 g/dL from Groups 1 to 5) after adjustment for menstruation, pregnancy, estrogen replacement, and the above-mentioned variables. In both sexes, the weighted prevalence of anemia with an eGFR of 60–104 mL/min/1.73m2 was significantly lower than that with an eGFR of ≥105 mL/min/1.73m2 (men, 3.2±0.4%, 1.9±0.3%, 1.8±0.3%, 2.0±0.9%, and 18.1±3.1%; women, 14.0±0.8%, 11.2±0.7%, 10.5±1.0%, 13.2±1.6%, and 32.3±3.2% from Groups 1 to 5). Conclusions We noted a compensatory increase in the hemoglobin level with a minor decline in kidney function (in the range of eGFR ≥60 mL/min/1.73m2) prior to a marked decrease in hemoglobin level with severe renal dysfunction.


Introduction
Anemia is a common complication of chronic kidney disease (CKD), and is associated with an increased risk of cardiovascular disease (CVD) and mortality, particularly in high-risk populations [1,2]. The National Health and Nutrition Examinational Survey (NHANES) III in the USA reported that the prevalence of anemia is increased in subjects with a glomerular filtration rate (GFR) of <60 mL/min/1.73m 2 [3]. Anemia associated with decreased renal function is attributed to the reduced production of erythropoietin. However, as renal function deteriorates, anemia may be attributed to other factors such as iron deficiency, blood loss, and inflammation [4,5].
Although it is well known that the prevalence of anemia is higher among cases with an eGFR of <60 mL/min/1.73m 2 , information on this association during early renal dysfunction is scarce, and only few studies have reported on this issue. Furthermore, the findings of most studies are not applicable to the general population, and most studies did not consider the factors that may affect the hemoglobin level [6]. In the present study, we aimed to investigate the association of hemoglobin level and prevalence of anemia with the decline in eGFR using a nation-wide representative sample of the adult Korean population.

Study population and data collection
This study is based on data from the 2010-2012 Korea National Health and Nutrition Examination Survey (KNHANES), which is a cross-sectional and nationally representative survey conducted by the Korean Center for Disease Control and Prevention. The following information is reproduced from our previous work [7]. The KNHANES has been conducted periodically since 1998 to assess the health and nutritional status of the civilian non-institutionalized population of Korea. A representative population was recruited using population-allocationsystematic sampling with multistage stratification. The study was approved by the Institutional Review Board of Ilsan-Paik Hospital (IB-1412-056). The KNHANES dataset was made available at the request of the investigator after approval. Since the dataset did not include any personal information and the participants in KNHANES had already consented, our study was exempted by the board from the necessity to obtain participant consent. capacity levels were measured by using the direct bathophenanthroline method (Hitachi Automatic Analyzer 7600, Hitachi, Japan), whereas serum ferritin levels were measured by using an immunoradiometric assay (1470 WIZARD gamma-counter, PerkinElmer, Finland).
The eGFR was determined using the Chronic Kidney Disease Epidemiology Collaboration formula: eGFR (ml/min/1.73 m 2 ) = 141 × min(SCr/k,1) a × max(SCr/k,1) - dL], k: 0.7 for females and 0.9 for males, a: -0.329 for females and -0.411 for males, min: minimum of SCr/k or 1, and max: maximum of SCr/k or 1) [8]. The results of the detailed health interviews, physical examinations, and other laboratory procedures have been published elsewhere [9].
Anemia was defined as a hemoglobin level of less than 13 g/dL in men and less than 12 g/dL in women according to the criteria of the World Health Organization [10]. Iron deficiency anemia (IDA) was defined as a transferrin saturation of less than 10% or ferritin levels of less than 15 ug/L. Obesity was defined as a BMI of 25 kg/m 2 [11]. The eGFR was classified into 5 groups: Group 1, 105; Group 2, 90-104; Group 3, 75-89; Group 4, 60-74; Group 5, <60 mL/ min/1.73m 2 .

Statistical analyses
The following information is reproduced from our previous work [7]. The KNHANES participants were not randomly sampled. The survey was designed using a complex, stratified, multistage probability sampling model; thus, individual participants were not equally representative of the Korean population. To obtain representative prevalence rates from the dataset, it was necessary to consider the power of each participant (sample weight) as representative of the Korean population. After approval was obtained from the Korea Centers for Disease Control and Prevention, we received a survey dataset that included information about the survey location; strata by age, sex, and various other factors; and the sample weight for each participant. The survey sample weights, calculated while considering the sampling rate, response rate, and age/sex proportions of the reference population (2005 Korean National Census Registry), were used in all analyses to provide representative estimates of the non-institutionalized Korean civilian population.
Statistical analyses were performed using SPSS software (ver. 21.0 for Windows; SPSS, Chicago, IL, USA). To compare the weighted prevalence of anemia according to age and sex, the chi-square test was performed. We compared age-and sex-adjusted clinical characteristics according to the presence of anemia for each sex using a general linear model (GLM). Weighted and adjusted hemoglobin level, weighted and adjusted prevalence of anemia, and weighted and adjusted prevalence of IDA and non-IDA were compared among the 5 eGFR groups using the GLM. All tests were two-sided, and P-values of < .05 were considered to indicate statistical significance.

Demographics and clinical characteristics of the study population
A total of 25,534 people participated in the KNHANES V-2, 2010-2012. Of those, 19,599 participants were aged 19 years and older. Among them, a total of 17,373 participants (weighted number, 36,647,641) who completed the laboratory examination were finally included in the analysis. The overall mean age was 45.0±0.2 years (95% confidence interval [CI], 44.7-46.5). The weighted mean age of the men (unweighted n = 7,296/weighted n = 18,330,187) and women (unweighted n = 9,886/weighted n = 18,317,454) were 44.2±0.3 and 46.9±0.3 years, respectively. The demographics and clinical characteristics of the study population are presented in Table 1.

Demographic and clinical characteristics according to the presence of anemia
In weighted and age-adjusted comparisons, we noted that although anemic individuals were more likely to have a history of cancer, they were less likely to be heavy alcohol drinkers, current smokers, obese, and college graduates, and were also less likely to have lower diastolic blood pressure, serum total cholesterol levels, and serum triglyceride levels (Table 3). Moreover, we found that anemic women were likely to have higher a rate of menstruation and pregnancy, and were more likely to be college graduates. However, they were less likely to be heavy alcohol drinkers, current smokers, and obese, and were also less likely to have lower systolic and diastolic BP, serum total cholesterol levels, and serum triglyceride levels.
Prevalence of anemia, IDA, and non-IDA according to eGFR In both sexes, the weighted and adjusted prevalence of anemia at an eGFR of 60-104 mL/min/ 1.73m 2 was significantly lower than that at an eGFR of 105 mL/min/1.73m 2 , whereas the weighted prevalence of anemia at an eGFR of <60 mL/min/1.73m 2 was markedly higher than The weighted prevalence of IDA did not differ according to the eGFR in men but was likely to decrease with a decline in eGFR in women (Table 5, Fig 2).

Discussion
In this nation-wide representative population-based analysis, the prevalence of anemia was 2.6% (IDA: 0.7%) in men and 12.8% (IDA: 8.0%) in women. The study showed that the estimated hemoglobin level increased with a decrease in the eGFR; this value then peaked at an eGFR of 60-89 mL/min/1.73m 2 and decreased thereafter, following the adjustment for several factors; the estimated prevalence of anemia at an eGFR of 60-104 mL/min/1.73m 2 was significantly lower than that at an eGFR of 105 mL/min/1.73m 2 ; and the weighted prevalence of anemia at an eGFR of <60 mL/min/1.73 m 2 was markedly higher than that at an eGFR of 60-104 mL/min/1.73m 2 .
Only a few reports have described the association between hemoglobin concentration and a minor decline in eGFR in patients with CKD stage 1 and 2. In the Kidney Early Evaluation Program (KEEP) study, the serum hemoglobin levels were reported to be 13.5, 13.7, and 13.5 g/dL in CKD Stages 1,2, and 3, respectively [1]. NHANES III showed a slight elevation in the hemoglobin level at an eGFR of 60-89 mL/min/1.73m 2 , through a graph of hemoglobin percentile versus eGFR [3]. The Atherosclerosis Risk in Communities (ARIC) study showed a subtle In both sex, the prevalence of anemia in the group of an eGFR>105-89 mL/min/1.73m 2 and <59 mL/min/1.73m 2 were significantly higher than that in other groups after adjustment of various factors. In men, the prevalence of non-iron deficiency anemia showed similar pattern with that of anemia, but there was no difference in the prevalence of iron deficiency anemia. In women, the prevalence of iron deficiency anemia was highest in the group of an eGFR105 mL/min/1.73m 2 . The prevalence of non-iron deficiency anemia was significantly higher in the group less than eGFR 75 mL/min/1.73m 2 than other groups. After adjustment, the hemoglobin level was the highest in the group of an eGFR of 75-89 mL/min/1.73m 2 in both sex. P value means P for difference.  [6]. All these findings were comparable with our results. However, the above-mentioned studies have certain limitations that preclude the drawing of conclusions regarding this issue. First, the main purpose of the analyses was not the investigation of the association between hemoglobin concentration and minor decline in eGFR in population with eGFR greater than 60 mL/min/1.73m 2 , except for the study of Oh et al. For example, the main aim of the ARIC study was to investigate the association of kidney function and hemoglobin level with left ventricular morphology. Second, the KEEP, ARIC, and Oh et al studies were not representative of the general population. As the KEEP and ARIC studies involved a screening program that targeted the high-risk population, the prevalence of obesity, diabetes, hypertension, and cardiovascular risk factors would be much higher in the samples of these studies than in the general population. Moreover, the study of Oh et al analyzed participants who visited a health promotion center for a general check-up. Third, to examine the changes in hemoglobin level according to the decline of renal function alone, we need to adjust for several variables that could potentially affect the hemoglobin level. The adjustment of several parameters (age, college graduation, smoking history, cancer history, degree of obesity, serum lipid level, blood pressure, and menstruation history of women) in the present study facilitated the observation of a more direct association between the hemoglobin level and mild decline in renal function, with minimal concerns regarding confounding effects. Furthermore, as the prevalence of anemia is markedly higher in women than in men, it would be desirable to perform analyses after stratification according to sex. A major salient feature of this study is that it is the first report regarding the association of hemoglobin concentration and anemia prevalence with a minor decline in eGFR at earlier CKD stages from a nation-wide representative population after stratification according to sex and adjustment for several confounding factors.
The main causes of anemia in advanced CKD may be multifactorial, and include erythropoietin deficiency, uremia-induced inhibition of erythropoiesis, shorter red blood cell survival, and iron metabolism disorders [13][14][15]. Although the pathogenesis of anemia during early CKD is unclear, angiotensin II is suggested as a possible cause of tissue hypoxia during early CKD. Initial hypoxia in a remnant kidney model is dependent on the activation of the reninangiotensin system and hemodynamic alterations after nephron loss [16]. The activated reninangiotensin system induces tubular sodium reabsorption and vasoconstriction of the afferent arteriole, resulting in higher oxygen consumption and relative tubular hypoxia [17][18][19]. Peritubular cells sense low oxygen tension, produce hypoxia-inducible factor and erythropoietin, thus increasing hemoglobin synthesis [20][21][22]. However, it is conflicting whether serum erythropoietin level is increased or not in patients with eGFR greater than 60 mL/min/1.73m 2 [23,24]. In the present study, the prevalence of IDA did not differ according to eGFR in men, and but decreased according to eGFR in women. Therefore, the influence of certain causes of anemia, other than IDA, should be clarified in patients with early CKD.
Anemia is a well-known risk factor for cardiovascular morbidity and all-cause mortality in patients with CKD [25,26]. Anemia is also important in patients with an eGFR of >60 mL/ min/1.73m 2 . In particular, among non-CKD patients with heart failure, a hemoglobin level of <13.0 g/dL is an independent risk factor of death [27]. However, patients with a hemoglobin level of <14 g/dL also show a higher mortality rate as compared to those with hemoglobin levels of 14-14.9 or >15 g/dL [28]. These results indicate that survival could be affected by subtle changes in the hemoglobin concentration, or even at a normal hemoglobin level, in patients with CKD stage 1 or 2.
This study has several limitations. The main limitation is that it is performed on a limited cross-section of the population. To clarify the causal relationship between early renal dysfunction and hemoglobin level, a longitudinal prospective study is needed. Second, we used the change in GFR based on serum creatinine level as a marker of renal dysfunction. However, the use of creatinine to estimate GFR has certain limitations in terms of tubular secretion of creatinine and the variability in creatinine generation between individuals and for the same individual [29]. Although serum cystatin C or 51 Cr-EDTA GFR can be used to validate an eGFR, these tools are not available in our study. Third, the history of specific drug use, which might affect the hemoglobin level or eGFR, was not available. Fourth, no information on iron supplementation was available.
In conclusion, the estimated prevalence of anemia in Korean adults was approximately 2.6% in men and 12.8% in women. The data indicate a compensatory increase in hemoglobin level with a minor decline in kidney function, prior to a marked decrease in hemoglobin level with severe renal dysfunction (eGFR < 60 mL/min/1.73m 2 ). However, the lower prevalence of anemia with an eGFR of 60-104 mL/min/1.73m 2 could not be explained by the difference in IDA prevalence. Nevertheless, larger prospective studies of the association between hemoglobin level and early renal dysfunction, as well as re-analysis of the datasets of previous cohort studies, are required.