Study population
The NHANES, conducted by the National Center for Health Statistics (NCHS) at the Centers for Disease Control and Prevention (CDC), is a large-scale, multistage, ongoing, nationally representative health survey to represent the civilian noninstitutionalized population in the U.S. We used publicly available data on the blood ethylene oxide concentrations of 20,146 individuals from the 2013–2016 NHANES cycles (n=5,446). We excluded individuals who lacked a vitamin D measurement (n=63). Given that the metabolism of ethylene oxide may be abnormal in teenagers and pregnant women due to hormonal alterations, we excluded individuals aged < 18 years (n=1,216) and pregnant women (n=25). Finally, a total of 4,125 individuals were left for our analysis (Fig. 1). The NCHS Research Ethics Review Board approved the study protocol, and written informed consent was obtained from each participant.
Ethylene oxide exposure assessment
Red blood cell specimens were processed and stored until shipped to the CDC’s National Center for Environmental Health for laboratory testing. Ethylene oxide was detected by high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC–MS/MS) as described in detail elsewhere (Zeng et al. 2021). Undetectable concentrations were substituted with the respective detection limit divided by the square root of two.
Vitamin D measurements
The standardized liquid chromatography-tandem mass spectrometry (LC–MS/MS) method was used to measure serum 25-hydroxyvitamin D (25(OH)D) in serum sample data from the NHANES. In addition to the quantitative analyses of vitamin D as a continuous variable, we also considered individuals with a measured 25(OH)D level of ≥75.0 nmol/L as having vitamin D sufficiency, individuals with a measured 25(OH)D level of 50-74.9 nmol/L as having vitamin D insufficiency, and individuals with a measured 25(OH)D level of <50.0 nmol/L as having vitamin D deficiency (Liu et al. 2018).
Assessment of covariates
Information on age, sex, race/ethnicity, family income, education level, and physical activity was obtained from the household interview using demographic questionnaires. Race/ethnicity was recorded as non-Hispanic white, non-Hispanic black, Mexican American, other Hispanic, or other. The family income-to-poverty ratio was positively correlated with family income status. The self-reported education level was segmented into below high school, high school, and above high school. Physical activity, described as leisure time moderate-to-vigorous exercise, comprised the inactive group (indicating fewer than 150 minutes per week), insufficiently active group (indicating 150 to 300 minutes per week), and active group (indicating more than 300 minutes per week). Data on smoking status and alcohol intake were collected from the health questionnaire. In compliance with the NCHS classifications, the participants were categorized as never smokers and smokers based on their responses about smoking more than 100 cigarettes. Alcohol intake was grouped into drinkers or nondrinkers by having at least 12 alcoholic drinks per year or not. Body weight and height were measured by certified examiners to calculate the body mass index (BMI) as the ratio of weight in kilograms to height in meters squared (kg/m2). Additionally, chronic health conditions (hypertension, diabetes) were defined using the participants’ self-reported answers to whether a doctor or other health professional had ever informed that they had a certain chronic condition.
Statistical analysis
The baseline characteristics were described across three levels of vitamin D. Log2 transformation was performed due to the skewed distribution of ethylene oxide levels. We modeled differences in blood ethylene oxide levels across the three levels of vitamin D using linear regression. We developed three models by sequential adjustment for three groups of covariates. Model 1 included age and sex; Model 2 was further adjusted for education level, race, smoking status, alcohol use, diabetes, hypertension and body mass index; and Model 3 additionally included the poverty-income ratio, energy intake, and physical activity. Our results are also presented as the average percent difference and 95% confidence interval (CI). The interquartile range (IQR) of the vitamin D concentration was calculated. Then, the formula % = (IQR ^ beta-1)*100% was calculated. In addition, we assessed the potential nonlinear relationships between serum vitamin D concentrations and blood ethylene oxide levels using restricted cubic splines with three knots located at the 10th, 50th, and 90th percentiles. All analyses were conducted using R software version 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided P value < 0.05 was considered statistically significant.