The National Health and Nutrition Examination Survey (NHANES) is a comprehensive cross-sectional survey conducted biennially to assess the health and nutritional status of the civilian, non-institutionalized population in the United States. This survey received approval from the Research Ethics Review Board of the National Center for Health Statistics (NCHS), and written informed consent was obtained from all participants. The NHANES data used in this study are publicly available and can be accessed through the NHANES website (https://www.cdc.gov/nchs/nhanes/).

The present study utilized data from three NHANES cycles spanning from 1999 to 2004, encompassing a total of 31,126 participants. After excluding individuals without diabetes, 2034 subjects remained in the analysis. Further exclusions due to missing data on DFU (n = 176) and lead/cadmium exposure (n = 194) resulted in a final sample size of 1664 participants included in the analysis (see Fig. 1).

Flowchart of participant selection for the National Health and Nutrition Examination Survey (NHANES). HbA1c glycated hemoglobin, Pb blood lead levels, Cd blood cadmium levels

This study was secondary research using NHANES data, which is a publicly available de-identified data source. As such, institutional review board review or formal consent from participants to release or publish the information is not applicable. Ethical approval for the NHANES study was obtained from the NCHS Research Ethics Review Board (NHANES–NCHS Research Ethics Review Board Approval; https://www.cdc.gov/). This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments.

In the National Health and Nutrition Examination Survey (NHANES), medical professionals collect whole blood samples via venipuncture and store them in EDTA-treated tubes. The samples are centrifuged on-site and subsequently stored at – 30 °C before being sent to the California Department of Public Health Laboratory. To ensure a uniform distribution of cellular elements, the samples are thoroughly mixed, with those containing microclots excluded. The samples are then diluted by mixing with water and a diluent; the reagents in the diluent facilitate the release of metals from red blood cells and reduce ionization suppression. This diluted mixture is introduced into an inductively coupled plasma (ICP) mass spectrometer for analysis. During this process, the liquid sample is converted into an aerosol and undergoes vaporization, atomization, and ionization before the ions are introduced into the mass spectrometer. A dynamic reaction cell (DRC) is employed to eliminate interference and enhance the signal of specific elements. Final element concentrations are determined by converting electrical signals into digital data. Concentrations below the limit of detection (LOD) are expressed as LOD/√2. The quality assessment and quality control procedures of NHANES adhere to the standards set by the Clinical Laboratory Improvement Amendments of 1988 [11].

Participants are classified as having diabetes if they meet any of the following criteria [12]: (1) hemoglobin A1c (HbA1c) level ≥ 6.5%, (2) random blood glucose ≥ 11.1 mmol/l, (3) fasting blood glucose(FPG) level ≥ 7.0 mmol/l, (4) use of any antidiabetic medication, or (5) a previous diagnosis of diabetes by a healthcare professional. DFU are identified based on participants’ responses to the survey question, "Do you have any sores or ulcers on your legs or feet that take more than four weeks to heal?" Additionally, potential covariates such as gender, age, race/ethnicity, education level, marital status, poverty income ratio (PIR), smoking status, body mass index (BMI), and laboratory parameters (total cholesterol [TC], C-reactive protein [CRP], glycated hemoglobin [HbA1c], hemoglobin [HGB], white blood cells [WBC], and high-density lipoprotein [HDL]) were analyzed. Specific baseline characteristics are presented in Table 1.

Data for this study were collected following standardized procedures outlined on the NHANES website. The covariates considered included age, gender (male or female), race, marital status (married or living with a partner, widowed, divorced or separated, never married), education level (below high school, high school graduate/GED or equivalent, college or above), PIR (low < 1.3, medium 1.5–3.5, high ≥ 3.5), BMI (< 25, 25–30, ≥ 30 kg/m2), smoking status (determined by whether the individual has smoked more than 100 cigarettes in their lifetime), TC, HGB, HbA1c (< 6.5%, ≥ 6.5%), HDL, WBC, and CRP.

Statistical analyses were performed using the R software package (http://www.r-project.org) and Empower (R) (www.empowerstats.com, X&Y Solutions Inc., Boston, MA). These tools support reproducible analysis and interactive computing. A two-tailed p value of < 0.05 was considered statistically significant.

For data representation, continuous variables with a normal distribution are presented as mean ± standard deviation (SD), while continuous variables with a skewed distribution are presented as median and interquartile range (IQR). Categorical variables are presented as frequency (percent). Depending on the distribution characteristics of the data, the Student’s t test or Mann–Whitney U test was used to compare continuous variables among groups, while the Chi-square test or Fisher’s exact test was applied for categorical variables.

To further explore the relationships between multiple variables, we conducted binary logistic regression analysis and constructed four multivariate logistic regression models to examine the correlation between blood heavy metals and DFU. Model 1 is the unadjusted model. Model 2 adjusts for sociodemographic variables (age, sex, marital status, race/ethnicity, PIR, education level). Model 3 adjusts for sociodemographic variables and additional factors, including BMI and smoking status. Model 4 includes adjustments for sociodemographic variables, BMI, smoking status, TC, HGB, HbA1c, HDL, WBC, and CRP.

Simultaneous subgroup analyses were conducted based on age, sex, HGB category (dichotomous), HbA1c category (< 6.5%, ≥ 6.5%), and glucose level (< 7 mmol/l, ≥ 7 mmol/l). The association between blood heavy metals and DFU was assessed using multivariate logistic regression models. Additionally, smooth curve fitting was applied to examine the association between blood heavy metals and diabetic lower limb ulcers. For missing values, if the proportion was less than 10%, multiple imputation was used to interpolate the missing covariate data.