Association between diabetes mellitus and anemia among Korean adults according to sex: a cross-sectional analysis of data from the Korea National Health and Nutrition Examination Survey (2010–2016)

Diabetes mellitus (DM) is a chronic disease that causes hyperglycemia because of impaired insulin secretion by pancreatic beta cells and peripheral insulin resistance, which may lead to serious complications [1]. As of 2019, the number of patients with DM worldwide was approximately 422 million, with an estimated prevalence of 8.5% [2]. The incidence of DM is increasing most rapidly among middle-to-low income countries [3], and it is predicted that one in 10 adults aged ≥20 years will have DM by 2045 [4]. Such a marked increase in the prevalence of DM could pose a significant economic burden on a country’s healthcare system. According to data from the Health Insurance Review and Assessment Service [5], the total cost of care for patients diagnosed with DM in the US was 573.7 billion won in 2019, an increase of 8.3% from that in the previous year [6].

Approximately 50–80% of patients with cardiovascular disease die, and DM may increase the risk of cardiovascular disease by increasing blood cholesterol and triglyceride levels. Accordingly, cardiovascular disease is considered a major cause of premature death, and its incidence among patients with DM reaches 20% after approximately 7 years [7]. Moreover, DM is often accompanied by acute or chronic complications, and because chronic complications are directly associated with patient death, prevention of such complications is of utmost importance.

Anemia, which is a major global public health issue, is a disease caused by the reduced ability to transport oxygen to tissues due to the lack of hemoglobin (Hb) [8]. Anemia is a common disease with a prevalence of nearly one-third of the world’s population, including that in Korea [9, 10]. It is more common among women than among men and in the older population than in the younger population [11]. The causes of anemia are complex and multifactorial; as it is a potential risk factor for cardiovascular disease, it may originate from complications related to various chronic diseases. However, there are many conflicting opinions about its association with DM.

Numerous studies on the association between DM and anemia have been published. Studies have demonstrated that anemia occurs in 14–45% of patients with DM [12, 13]. A study from the United States that analyzed adults aged ≥20 years reported that there is no association between DM and the onset of anemia among men or women [14]. This review of previous studies indicates that there are conflicting views on the association between anemia and DM, and the mechanism by which DM influences anemia remains uncertain.

Accordingly, we aimed to investigate the association between DM and anemia among Korean adults using nationally representative data from the Korea National Health and Nutrition Examination Survey (KNHANES) V, VI, and VII to analyze risk factors that could cause anemia in patients with DM. In particular, we sought to analyze sex-based differences in specific disease factors that could affect DM and anemia among Korean adults.

Table 1 shows the characteristics of the study population depending on DM. There were 2903 participants (11.3%) in the DM group and 22,694 participants (88.7%) in the non-DM group and more women (n = 1486) than men (n = 1417; 48.8%). Mean patient age in the DM group was 60.09 ± 0.31 years, which was higher than that in the non-DM group. With regard to education level, “below primary school” was the most common response (n = 1299; 44.8%) in the DM group, whereas “high school” was the most common response (n = 7668; 33.8%) in the non-DM group (p < 0.0001). Mean BMI (kg/m²) was higher in the DM group (25.3 ± 0.1) than in the non-DM group (23.6 ± 0.03) (p < 0.0001); there were 1490 (51.3%) participants with a normal weight in the DM group and 14,766 (65.1%) in the non-DM group, showing higher percentages than both the underweight and overweight and obesity subgroups (p < 0.0001). There were more non−/ex-smokers in the DM group (n = 18,983; 83.7%) than in the non-DM group (n = 2352; 81%) (p = 0.0076). With regard to alcohol consumption, the number of participants who responded “yes” was 2284 (78.7%) in the DM group and 19,781 (87.2%) in the non-DM group, which was much higher than the number of participants who responded “no” (p < 0.0001).

Table 2 shows the characteristics of the study population depending on diabetes and anemia. There were 2258 participants (8.9%) in the anemia group and 23,339 (91.1%) in the non-anemia group, and there were more women (n = 1796, 79.5%) than men in the anemia group (n = 462, 20.5%) (p < 0.0001). Mean patient age in the anemia group with DM was 65.93 ± 0.86 years, which was higher than that in the non-anemia group in DM and all groups in non-DM (p < 0.0001). Regarding household income, the prevalence of anemia tended to be higher among the lower income quartiles in the DM group, whereas it tended to increase with an increasing income level in the non-DM group.

With regard to education level in the non-DM group, “high school” was the most common response in both the anemia (n = 678, 35.1%) and non-anemia (n = 6990, 33.7%) groups (p < 0.0001). In contrast, the most common education level in the DM group was “≤Elementary school” in both the anemia (n = 181, 55.2%) and non-anemia (n = 1118, 43.4%) groups (p < 0.0001). Mean BMI (kg/m²) was higher in the non-anemia group than in the anemia group in both the non-DM and DM groups (p < 0.0001). In the anemia group, the proportion of current smokers in the DM group (12.8%) was relatively higher than that in the non-DM group (5.1%) (p < 0.0001). Regarding clinical factors that influence anemia, there were significant differences in the daily iron intake (mg/day), Hb level (g/dL), serum creatinine level (mg/dL), waist circumference, high waist circumference, and fasting glucose level (mg/dL) between the two groups.

Table 3 shows the characteristics of the study population depending on anemia by sex in DM. Among men, there were significant differences between the anemia and non-anemia groups except in alcohol consumption and low HDL cholesterol factors. Among women, there were significant differences between the groups in age, mean BMI, daily iron intake, Hb level, serum creatinine level, waist circumference, high waist circumference, fasting glucose level, and HbA1c level.

Table 4 shows the association between DM and anemia analyzed using logistic regression analyses by sex; the results were significant in both the unadjusted model (Model 1) and adjusted models (Models 2 and 3) for men. In particular, the ORs for anemia in the DM group were higher than those in the non-DM group by 3.49-fold (95% CI 2.69–4.53, p < 0.0001), 2.18-fold (95% CI 1.67–2.85, p < 0.0001), and 1.87-fold (95% CI 1.42–2.50, p < 0.0001) in Models 1, 2, and 3, respectively. For women, the results were not significant in the unadjusted model (Model 1), although they were significant in the adjusted models (Models 2 and 3). In particular, the ORs for anemia were higher in the DM group than those in the non-DM group by 1.31-fold (95% CI 1.07–1.60, p = 0.0093) and 1.33-fold (95% CI 1.09–1.63, p = 0.0067) in Models 2 and 3, respectively. Compared to the group without a history of DM, OR for the incident DM was significantly higher in an unadjusted model (OR: 3.49; 95% CI: 2.69–4.53); the association observed remained significant in Model 2 (OR: 3.10; 95% CI: 2.38–4.04) and Model 3 (OR: 2.97; 95% CI: 2.27–3.89). Among men (the reference category for sex), compared to the group without a history of DM, OR for the incident DM was significantly higher in an unadjusted model (OR: 3.49; 95% CI: 2.69–4.53); the association observed remained significant in Model 2 (OR: 3.10; 95% CI: 2.38–4.04) and Model 3 (OR: 2.97; 95% CI: 2.27–3.89).

Moreover, among those with no DM (the reference category for diabetes), men had a 5.52-fold higher OR for anemia (95% CI 4.73–6.45, p < 0.0001) in Model 1, 4.36-fold higher OR (95% CI 3.70–5.14, p < 0.0001) in Model 2, 5.66-fold higher OR (95% CI 4.61–6.95, p < 0.0001) in Model 3 than women.

Lastly, our study demonstrated a significant association in the interaction between sex and DM on the development of anemia in Model 1 (OR: 0.35; 95% CI: 0.25–0.49, p < 0.0001), Model 2 (OR: 0.37; 95% CI: 0.27–0.52, p < 0.0001), and Model 3 (OR: 0.39; 95% CI: 0.27–0.54, p < 0.0001) in Additional file 1. This can be interpreted as the odds for anemia for those with DM versus no DM being higher for men (the reference category for sex) than for women.

Table 5 shows the results of logistic regression analyses on regarding the influence of risk factors on the prevalence of anemia among patients with DM as the change in the odds ratio for a unit change in the continuous variables. The results were significant for daily iron intake, serum creatinine level, high waist circumference, and hypertriglyceridemia in Models 1, 2, and 3. In particular, the serum creatinine level showed ORs of 6.28 (95% CI 3.95–9.99, p < 0.0001), 14.28 (95% CI 8.15–25.01, p < 0.0001), and 13.40 (95% CI 7.64–23.51, p < 0.0001) in Models 1, 2, and 3, respectively, which indicates that the serum creatinine level is a highly influential risk factor for anemia in patients with DM.

A high waist circumference showed significant ORs of 0.59 (95% CI 0.44–0.79, p = 0.0004), 0.52 (95% CI 0.31–0.88, p = 0.015), and 0.47 (95% CI 0.26–0.83, p = 0.0097) in Models 1, 2, and 3, respectively. Hypertriglyceridemia also showed significant ORs of 0.49 (95% CI 0.34–0.70, p < 0.0001), 0.69 (95% CI 0.42–0.90, p = 0.0129), and 0.62 (95% CI 0.41–0.92, p = 0.0179) in Models 1, 2, and 3, respectively.

When the influence of risk factors on the prevalence of anemia among patients with DM was examined by sex, the results showed slight differences between men and women. Among men, the serum creatinine level, hypertension, and hypertriglyceridemia were significant factors in both the unadjusted (Model 1) and adjusted (Models 2 and 3) models. The serum creatinine level showed significant ORs of 10.31 (95% CI 5.40–19.70, p < 0.0001), 8.00 (95% CI 3.84–16.65, p < 0.0001), and 6.30 (95% CI 3.08–12.90, p < 0.0001) in Models 1, 2, and 3, respectively. Hypertension also showed significant ORs of 2.49 (95% CI 1.61–3.85, p < 0.0001), 2.14 (95% CI 1.31–3.49, p = 0.0025), and 1.86 (95% CI 1.18–3.13, p = 0.0191) in Models 1, 2, and 3, respectively. Additionally, hypertriglyceridemia showed significant ORs of 0.24 (95% CI 0.13–0.43, p < 0.0001), 0.38 (95% CI 0.20–0.72, p = 0.003), and 0.40 (95% CI 0.21–0.77, p = 0.0065) in Models 1, 2, and 3, respectively.

Among women, the serum creatinine level and a high waist circumference showed significant ORs in Models 1, 2, and 3 (OR = 34.50, 95% CI 13.98–85.12, p < 0.0001 in Model 1; OR = 37.12, 95% CI 15.35–89.76, p < 0.0001 in Model 2; and OR = 42.63, 95% CI 17.25–105.36, p < 0.0001 in Model 3). A high waist circumference alone showed significant ORs of 0.61 (95% CI 0.42–0.89, p = 0.01), 0.49 (95% CI 0.28–0.87, p = 0.0149), and 0.38 (95% CI 0.21–0.71, p = 0.0022) in Models 1, 2, and 3, respectively.

With regard to previous studies on DM and anemia, a study conducted among adults from the United States reported no association between DM and anemia, regardless of sex, which was in line with the conclusion drawn in a study from Spain that analyzed patients on nephrology dialysis also reported no association between DM and anemia [14, 21]. However, findings of the present study indicated an association between DM and anemia by sex, which is supported by findings from other studies on the association between iron-deficiency anemia and HbA1c level in DM^, and the association between anemia and kidney function in patients with DM.^, However, those studies used different diagnostic criteria for anemia and included participants of different ages and ethnic groups; thus, it is difficult to make direct comparisons between studies.

A previous Korean study [22] investigated the influence of DM on the onset of anemia among Korean adults, but the study found that after adjusting for other risk factors for anemia, DM was not an independent risk factor for anemia; this is contrary to the findings of the present study. Even when examining by sex, another Korean study [18] analyzed the OR of DM on the onset of anemia among Korean adults and found that DM increased the OR for anemia by 2.6-fold among men after adjusting for age, education level, monthly household income, smoking status, alcohol consumption, iron intake, and disease history, whereas DM did not influence the onset of anemia among women.

The present study also investigated the influence of risk factors on the prevalence of anemia among patients with DM by sex. Our study found that men had a 5.66-fold higher OR for anemia than women, and there was a significant association in interaction between sex and DM on the development of anemia (ORs of 0.39 in Model 3).

Recent studies have identified the associations of chronic kidney failure and elevated creatinine levels with metabolic syndrome, including DM [23]. The results showed that the serum creatinine level was a significant risk factor for anemia in both men and women. In particular, the serum creatinine level showed ORs of 6.30 (p < 0.0001) in men and 42.63 (p < 0.0001) in women after adjusting for demographic, lifestyle, disease, and health-related factors, which indicates that the serum creatinine level greatly influences the prevalence of anemia, especially among women with DM.

The association between serum creatinine level and anemia in patients with DM was a noteworthy finding in the present study. A previous study [24] found no significant association between serum creatinine level (mg/dL) and anemia among patients with DM, as well as no significant association with GFR (mL/min). In the present study, however, the serum creatinine level showed a strong association with anemia in women with DM. Such a difference could be attributed to the fact that the existing report was a case study that recruited patients with DM and had a sample size of only 142, whereas the present study analyzed data from 25,597 people. In addition, the existing study particularly recruited patients with type 2 DM, whereas the KNHANES included individuals with type 1 and type 2 DM. Additional studies are needed to investigate the mechanisms involved in DM, anemia, and kidney disease by sex.

The present study had some limitations because it used KNHANES data [25]. First, as the KNHANES was a cross-sectional survey study, the association between DM and anemia could be examined, although the causal relationship between DM and anemia must be interpreted with caution. Second, a major limitation in studying anemia is that the onset involves diverse and complex mechanisms; however, the KNHANES defined anemia simply based on the Hb level. Therefore, it was difficult to identify other risk factors for anemia such as vitamin B₁₂ level. Moreover, chronic anemia and iron-deficiency anemia cannot be differentiated by the present study alone. Third, when defining the covariates hypertension and hyperlipidemia, this study defined only the numbers without considering whether participants were taking medication during the study period to lower their levels. Hypertension and hyperlipidemia are covariates rather than direct outcomes in this paper, and the two diseases are chronic diseases and belong to the related cardiovascular system. If each disease included people on medication, then it would be difficult to distinguish each group, which may have affected the significance of the results; therefore, people on medication were not considered. Another limitation of this study is that people whose diseases were being numerically controlled through drugs were not considered. Fourth, the number of patients with diabetic nephropathy among the included patients with DM could not be identified using the KNHANES data, making it difficult to accurately identify the association. Although the KNHANES provides nationally representative data, the survey itself did not focus on DM and anemia. As such, there is a lack of data on biological values such as those on the GFR and blood erythropoietin level, which could be used to determine kidney health [26].

While it is unclear whether correcting anemia is helpful for patients with DM, DM is known to cause various complications, such as diabetic nephropathy and cardiovascular disease, and increase hospitalization and mortality rates. Therefore, continued management and treatment of DM are important. Future studies should compare patients with kidney disease caused by DM and patients with DM who do not have kidney disease.

Our findings showed that, after adjusting for all other risk factors for anemia, DM was an independent risk factor for anemia among Korean men. Therefore, anemia should be monitored not only in women and malnourished patients but also in patients with DM. The management and treatment of anemia are important for preventing DM-related complications such as kidney disease and cardiovascular disease. Despite the limitations, the present study contributes to the existing literature in that it investigated the association between DM and anemia within the Korean population. This study is the first to investigate sex-based differences in factors that influence DM and anemia. Large-scale studies that overcome the limitations of the present study are needed in the future, and the findings of the present study may serve as baseline data.

Our findings indicate that there is an association between DM and anemia among Korean adults, with a strong association between DM and anemia among men. In addition, the serum creatinine level was identified as a factor that has a significant influence on DM and anemia among women. However, considering that the survey did not particularly investigate anemia, the mechanism could not be identified. As anemia may influence DM-related complications such as kidney disease and other cardiovascular diseases, management and treatment of anemia in patients with DM are considered highly important.