The National Health and Nutrition Examination Survey (NHANES) was designed to evaluate the health and nutritional status of non-hospitalized Americans using a stratified, multistage approach. The NHANES received approval from the Ethics Review Committee of the National Center for Health Statistics (NCHS), and all participants provided written informed consent prior to participation. This is a retrospective study based on the NHANES database, which contains data on over 31,126 patients from 1999 to 2004. In the study, 9,970 were adults aged 40 years or older who completed the interview and underwent MEC screening. After excluding 8,297 participants who did not have diabetes (n=8160) and those with missing data on diabetes foot ulcers (n=3) and hemoglobin (n=188), the remaining 1,673 participants were included in the analysis ( Figure 1 ).

The primary outcome variable was the status of diabetic lower extremity ulcers (DLEU), defined by the patient’s self-reported answer to the question in Question Data, ‘Have you had an ulcer or sore on your leg or foot that took more than four weeks to heal?’ Type 2 diabetes mellitus (T2DM) was identified based on the American Diabetes Association criteria and a self-report questionnaire. Participants were considered to have T2DM if they met any of the following criteria (12) (1): Glycated hemoglobin (HbA1c) levels of ≥6.5% (2), Fasting plasma glucose (FPG) levels of ≥126 mg/dL (3), 75 g oral glucose tolerance test (OGTT) levels of ≥11.1mmol/L (4), self-reported physician diagnosis of diabetes, or (5) receipt of oral glucose-lowering medicines or insulin.

The NHANES Laboratory/Medical Technologists Procedures Manual (LPM) provides detailed instructions for sample collection and processing. The study employed the Beckman Coulter method for counting and sizing, combined with an automated diluter and mixer for sample processing and a single-beam photometer for hemoglobinometry to derive complete blood count (CBC) parameters. (https://www.cdc.gov/nchs/nhanes/). Anemia was defined by World Health Organization (WHO) as hemoglobin (Hb) levels <13g/dL for males and <12 g/dL for females (7).

Based on the literature, several potential covariates were included in the analysis, such as age, sex, race/ethnicity, education level, marital status, PIR, smoking status, alcohol status, body mass index (BMI), laboratory parameters (total cholesterol and C-reactive protein [CRP], glycosylated hemoglobin [HbA1c], and serum ferritin, and comorbidities (13–16). The comorbidities included family history of diabetes, stroke, coronary heart disease, hyperlipidemia. Marital status was classified as living with a partner, or living alone (15). Family income was divided into three groups according to the poverty income ratio (PIR) as defined by a U.S. government report: low (PIR ≤ 1.3), medium (PIR > 1.3 to 3.5), and high (PIR > 3.5). Alcohol consumption was classified as never (< 12 drinks in lifetime), former (≥12 drinks in 1 year and no drinks in the last year, or no drinks in the previous year but≥12 drinks in lifetime), and current (≥12 drinks and currently drinking). Smoking status was categorized as never (<100 cigarettes in a lifetime), former (≥100 cigarettes but not currently smoking), and current (≥100 cigarettes and currently smoking) (16). Serum ferritin levels were classified as either <100 ng/mL or ≥100 ng/mL, according to previously reported classifications (17). The determination of previous disease (family history of diabetes, stroke, hyperlipidemia, and coronary heart disease) was based on the inquiry in the questionnaire of whether the doctor had been informed of the condition in the past.

The statistical analyses were conducted using R Statistical Software (Version 4.2.2, http://www.R-project.org, The R Foundation) and Free Statistics analysis platform (Version 1.9, Beijing, China, http://www.clinicalscientists.cn/freestatistics). The software is intended for reproducible analysis and interactive computing. A two-sided P value < 0.05 was considered statistically significant.

Normally distributed continuous variables were presented as mean ± SD, and skewed continuous variables were presented as median (interquartile range [IQR]). Categorical variables were expressed as frequencies (%). The Student’s t-test or Mann-Whitney U-test was used to compare continuous variables between groups, depending on the normality of the distribution, and categorical data were compared using the chi-squared or Fisher’s exact test, as appropriate.

Crude model was an unadjusted model. Model 1 was adjusted for age, sex, race and ethnicity, educational level, family income and marital status. Model 2 was developed using model 1 and additional factor such as BMI, coronary heart disease, stroke, family history of diabetes, hyperlipidemia, alcohol and smoking status. Model 3 was then developed using model 2 and additional factor such as HbA1c, CRP, total cholesterol, and serum ferritin. Subgroup analysis was conducted to investigate the correlation between anemia and diabetic lower extremity ulcers based on age, sex, BMI, and HbA1C category (<6.5, ≥6.5) as well as serum ferritin category (<100ng/mL, ≥100ng/mL). The percentage of missing values exceeded 20%. To address this issue, missing data for the covariates were imputed using multiple imputation.

Table 1 displays demographic, socioeconomic, comorbidity, and baseline characteristics by anemia status. The study included 1673 adults aged 40 years and older, with a mean age of 64.7 ± 11.8 years, of whom 52.6% were male. Anemia was found in 239 participants, accounting for 14% of the study group, with a prevalence of 57.3% in women. The prevalence of diabetic lower extremity ulcers was 8.1% (136 participants). The prevalence of diabetic foot ulcers was 12.7% among patients with anemia.

The univariate ordinal regression analysis results indicated that marital status, BMI, coronary heart disease, family history of diabetes, and hyperlipidemia. (P < 0.1; Table 2 ).

Table 3 presents the odds ratios (OR) and 95% confidence intervals (CI) for the presence of diabetic lower extremity ulcers (DLEU) determined by hemoglobin levels and anemia. When hemoglobin was analyzed as a continuous variable, a significant independent negative association was found between hemoglobin and the risk of DLEU. In the unadjusted model, each 1 unit increase in hemoglobin was associated with a 16% decrease in the presence of DLEU [OR=0.84, 95% CI: (0.75-0.993); p=0.001]. In model 1, 2 and 3, the association between hemoglobin (Hb) and diabetic lower extremity ulcers (DLEU) was marginally significant [OR: 0.74, 95% CI: (0.65-0.84); p<0.001] [OR: 0.76, 95% CI: (0.67-0.86); p<0.001] [OR: 0.76, 95% CI: (0.66-0.86); p<0.001], respectively.

The anemia group had a significantly higher risk of DLEU compared to the non-anemic group [OR: 1.79, 95% CI:(1.11-22.87)]. In Table 3 , when hemoglobin levels were categorized as anemic versus non-anemia, anemia was found to be positively associated with the risk of diabetic lower extremity ulcers. The odds ratios (OR) for anemia were calculated for Model 1, Model 2, and Model 3, with the crude model as the reference, using multivariable-adjusted regression and 95% confidence intervals (CIs). The odds ratio (OR) for anemia in Model 1 was [OR=2.02,95% CI:(1.28-185 3.19), P=0.002]. In Model 2, the OR for anemia was [OR=1.8,95% CI:(1.13-2.87), P=0.014] and in Model 3, it was [OR=1.79, 95% CI:(1.11-2.87),p=0.016] ( Table 2 ). Model 3 exhibited the lowest odds ratio (OR) compared to Model 1, which had the highest OR. This suggests a decreasing trend in the risk of diabetic lower extremity ulcers (DLEU). After conducting multivariate logistic regression analysis and smooth curve fitting, it was found that there is a negative association between hemoglobin levels and DLEU incidence when all potential confounders were taken into account (non-linearity: p=0.572).

Stratified analysis was performed in several subgroups to determine the potential effect modifications on the relationship between anemia and DLEU. No significant interactions were found in any subgroup after stratification by sex, age, HbA1c level, and BMI (all P for interaction >0.05). However, results differed between serum ferritin groups for diabetic lower extremity ulcers (P = 0.015 for interaction) ( Figure 2 ).