Association between ethylene oxide exposure and periodontitis: a cross-sectional study from NHANES 2013–2014

The data used in this study were downloaded from the NHANES 2013–2014 [4, 11]. A representative population sample was acquired based on a cluster, stratified, multistage sampling method and cross-sectional study design. NHANES participants are chosen on an annual basis by considering various factors such as counties, blocks, households, and individuals residing within households. To ensure accurate estimation, certain groups, including Mexican Americans and Non-Hispanic Blacks, are intentionally oversampled. The National Center for Health Statistics, which is affiliated with the Centers for Disease Control and Prevention, approved the NHANES protocol, and all participants signed an informed consent form. Individuals without complete full-mouth periodontal examination (FMPE), missing HbEO levels, and participants younger than 30 years were excluded. Finally, a total of 1497 participants were included in our study.

The washed, packed red blood cell samples underwent proper processing, storage, and shipment to the Division of Laboratory Sciences for thorough analysis. The main objective was to determine the total level of hemoglobin in order to assess the presence of hemoglobin adducts. To achieve this, the reaction mixture was carefully combined with the sample, followed by the isolation of Edman degradation products. Subsequently, the products were identified and detected using high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC-MS/MS). Results were quantified and expressed as pmol adducts per gram of hemoglobin. The detection of hemoglobin was carried out using a well-established, commercially available assay kit. Blood HbEO is expressed in pmol/g hemoglobin with a detection limit of 12.90 pmol/g hemoglobin.

The outcome of this study was moderate or severe periodontitis. The dental examiners underwent an extensive period of training and calibration to ensure the accuracy and quality of periodontal health data. This process included continuous monitoring and recalibration to maintain the required standards [12]. Periodontal examination contained attachment loss (AL) and probing pocket depth (PPD) at six sites per tooth without third molars based on the FMPE protocol. A maximum of 168 sites and 28 teeth per subject could be examined to assess periodontal status. CDC/AAP definitions were used for the classification of periodontitis [13]. No/mild periodontitis was characterized as no evidence of moderate/severe periodontitis; moderate periodontitis: ≥ 2 interproximal sites with PD ≥ 5 mm not on the same tooth, or ≥ 2 interproximal sites with CAL ≥ 4 mm not on the same tooth; severe periodontitis: ≥ 2 interproximal sites with CAL ≥ 6 mm not on the same tooth and ≥ 1 interproximal sites with PD ≥ 5 mm.

Covariates were collected from previous studies, including age, gender, race, education level, poverty index (PI), marital status, smoking status, alcohol consumption, obesity, diabetes mellitus and hypertension. The detail of covariates collect could obtain from the NHANES database [14]. We classified race into two categories: non-Hispanic white and others. Education level was categorized as follows: < high school (less than 9th grade or 9th to 11th grade), high school (high school grade/GED or equivalent), and > high school (some college/AA degree or college graduate or above). The poverty index (PI) was determined by calculating the ratio of family income to the poverty level, as defined by the Department of Health and Human Services poverty guidelines. We categorized PI into three groups: ≤ 1.3, 1.3–3.5, and > 3.5. Marital status was grouped as follows: married/living as married, never married, and separated/divorced/widowed. Smoking status was divided based on whether individuals had smoked less than 100 cigarettes in their lifetime or not. Alcohol consumption was grouped based on whether individuals had consumed at least 12 drinks or not [15]. Body mass index (BMI) was calculated by dividing weight (in kilograms) by the square of height (in meters, m^2). Obesity was defined as a BMI ≥ 30 kg/m^2, while a BMI < 30 kg/m^2 was classified as non-obesity [16]. The diagnosis of diabetes mellitus was based on self-reported physician diagnosis or a glycosylated hemoglobin level (HbA1c) ≥ 6.5% or a fasting blood glucose ≥ 7.0 mmol/L or current use of hypoglycemic drugs. The diagnosis of hypertension was based on self-reported physician diagnosis or current use of antihypertensive medications or a systolic blood pressure ≥ 130 mmHg or a diastolic blood pressure ≥ 80 mmHg [17].

Weights were considered in our study based on the NHANES analysis guide. Categorical variables were presented as percentage. Baseline characteristics were compared across the tertiles of HbEO levels using Chi-square test for categorical variables. We performed log2-transformation of HbEO levels owing to the skewed distribution found [18]. Natural cubic splines was used to identify any non-linear relationship of HbEO levels with moderate/severe periodontitis. We evaluated the relationship between HbEO levels and periodontitis using weighted multivariable logistic regression models. The analysis was performed using the ‘svyglm’ function with family = binomial in the R software. The outcome variable of this study was periodontitis, which was categorized into two groups: no/mild, and moderate/heavy. The exposure variable was the grouping of HbEO levels into tertiles. Model I was adjusted for age, gender and race, and model II was adjusted for age, gender, race, education level, marital status, PI, obesity, smoking status, alcohol consumption, diabetes and hypertension. In addition, we performed stratified and interaction analyses to assess whether the association between the HbEO levels tertiles and periodontitis differed by all variables in Table 1. We created a separate category for covariates with missing observations and conducted regression analysis to account for missing observations and their potential effect on the outcome [19]. All the analyses were performed using R software (version 4.1.2). A P-value less than 0.05 was considered significant.