All data of participants in this cross-sectional study were extracted from the National Health and Nutrition Examination Survey (NHANES) database (2011–2012 and 2015–2016), which was a program of studies to evaluate the health and nutritional status of the civilian and non-institutionalized populations in the United States. The NHANES is a multipurpose research program conducted by the National Center of Health Statistics (NCHS) and the Centers for Disease Control and Prevention (CDC) (28). The information collection was carried out through a combination of questionnaires, physical examinations, and laboratory tests (29). Additional information was available at: https://wwwn.cdc.gov/nchs/nhanes/tutorials/module2.aspx. NHANES is a publicly available dataset and was approved by the NCHS Ethics Review Board, and all patients/participants provided their written informed consent. The hospital ethics committee exempted the study from ethical review. The inclusion criteria of participants were as follows: (1) age > 20 and ≤ 69 years old; (2) without hearing-related medical conditions [ear tubes, abnormal otoscopy, impacted cerumen and abnormal tympanometry (peak pressure ≤ −150 daPa; compliance ≤0.3 mL) at either ear] (30). The exclusion criteria were: (1) missing the complete audiometric data, (2) missing the complete information for OBS assessment, (3) abnormal total energy intake levels (<500 kcal/day or > 5,000 kcal/day for female, <500 kcal/day or > 8,000 kcal/day for male), and (4) missing important co-variables [marital status, sedentary time, information of tinnitus, recreational and firearm noise exposure, body mass index (BMI)].

The OBS was calculated on the basis of 4 lifestyles and 16 nutrients [4], which contains 5 pro-oxidants (smoking, alcohol consumption, BMI, iron and total fat) and 15 antioxidants (dietary fiber, selenium, copper, zinc, magnesium, calcium, calcium, vitamin E, vitamin C, vitamin B12, total folate, vitamin B6, niacin, riboflavin, and carotene). According to the OBS calculation reported by Zhang et al., the alcohol consumption was classified as non-drinkers, non-heavy drinkers (0–30 g/d for men and 0–15 g/d for women) and heavy drinkers (≥30 g/d for men and ≥ 15 g/d for women), and assigned as 2, 1 and 0 points, respectively (31). Other components were grouped first by sex and then by tertiles into 3 groups, with antioxidants scoring 0 to 2 points in groups 1 to 3 and pro-oxidants scoring 2 to 0 points in groups 1 to 3. Subsequently, factors other than alcohol consumption were grouped by sex and then by their tertiles into 3 groups, with antioxidants scoring 0 to 2 points in groups 1 to 3 and pro-oxidants scoring 2 to 0 points in groups 1 to 3 (31). Higher OBS scores indicate more significant antioxidant exposure. In this study, OBS were divided into three levels according to the tertiles, including Q1: OBS <16, Q2: 16 ≤ OBS <24, and Q3: OBS ≥24.

The potential covariates were age, gender (female and male), race (non-Hispanic White, other Hispanic, Mexican American, non-Hispanic Black, and other race-including multiracial), educational level (less than 9th grade, 9-11th grade, high school grad/general educational development (GED) or equivalent, some college or AA degree, and college graduate or above), marital status (married/living with partner, and never married/divorced/separated/widowed), poverty income ratio (PIR) (<1.0, ≥1.0, and unknown), sedentary time, height, weight, BMI, waist circumference, and total energy. Dyslipidemia was defined as triglyceride (TG) ≥150 mg/dL (1.7 mmol/L) or total cholesterol (TC) ≥200 mg/dL (5.2 mmol/L) or low-density lipoprotein cholesterol (LDL-C) ≥130 mg/dL (3.4 mmol/L) or high-density lipoprotein cholesterol (HDL-C) ≤40 mg/dL (1.0 mmol/L) or self-reported physician diagnosis or taking drug for cholesterol or taking lipid-lowering medications. Hypertension was defined as the doctor told you to have hypertension or systolic blood pressure (SBP) ≥ 130 mmHg or diastolic blood pressure (DBP) ≥ 80 mmHg, or taking blood pressure medication. Diabetes was defined as fasting glucose ≥7.0 mmol/L or glycosylated hemoglobin level (HbA1c) ≥6.5% or self-reported physician diagnosis or receiving hypoglycemic therapy. Data from the Medical Conditions Questionnaire (MCQ) were applied to identify rheumatoid arthritis; the survey question was “Has a doctor or other health professional ever told you that you had arthritis?” if your answer was yes, a follow-up question was asked “Which type of arthritis was it?” your answer was rheumatoid arthritis. Tinnitus was defined based on a response of “yes” to the question “In the past 12 months, have you been bothered by ringing, roaring, or buzzing in your ears or head that lasts for 5 min or more?” Physical activity was expressed as the metabolic equivalent task (MET) and calculated as follows: physical activity (met·min/week) = recommended MET×exercise time for corresponding activities (min/day) × the number of exercise days per week (day) (32). Carotene includes Alpha-carotene and Beta-carotene. Occupational noise exposure was evaluated with the question “Have you ever had a job, or combination of jobs where you were exposed to loud sounds or noise for 4 or more hours a day, several days a week?” (Yes/No). Firearm noise exposure was assessed by the question “Have you ever used firearms for any reason?” (Yes/No). Recreational noise exposure was evaluated by the question “Outside of a job, have you ever been exposed to very loud noise or music for 10 or more hours a week?” (Yes/No). Veteran status was evaluated as having a positive response to any of the following statements: “Have you ever served on active duty in the U.S. armed forces, military reserves, or national guard?” or “Did you ever serve in a foreign country during a time of armed conflict or on a humanitarian or peace-keeping mission?”

According to the World Health Organization (WHO) standard, the average hearing threshold of 0.5, 1, 2, and 4 kHz was used to classify the severity of hearing loss in either ear as follows: mild (25–40 dB HL), moderate (40–60 dB HL), severe (60–80 dB HL), and very severe (≥80 dB HL) (33, 34). In this study, hearing loss was defined as hearing threshold >25 dB in either ear.

The normal distribution of variables was evaluated with the Kolmogorov–Smirnov test, and the homogeneity of variables was tested with Levene’s test. Continuous data were expressed as mean ± standard deviation (mean ± SD), and the Student’s t test and Satterthwaite t test were used for comparison between two groups. Enumeration data were expressed as numbers and percentage [n (%)], and the Chi-square test was used for comparison between groups.

The covariates were screened via the step-based regression method in the weighted multivariate logistic model. Weighted univariate and multivariate logistic regression models were utilized to explore the association between OBS and hearing loss. Odds ratio (OR) and 95% confidence interval (CI) were used to assess the association. Model 1 was the univariate model. Model 2 adjusted for age, gender, race, hypertension, tinnitus, and recreational noise exposure. Model 3 adjusted for age, gender, race, hypertension, tinnitus, recreational noise exposure, and occupational noise exposure. The associations were further explored in different subgroups of age, gender, occupational noise exposure, recreational noise exposure, firearm noise exposure, and veteran status. A weighted random forest model was constructed to further explored the importance of each component in OBS.

All statistical analyses were carried out R version 4.2.3(Institute for Statistics and Mathematics, Vienna, Austria) and all data were weighted by WTMEC2YR, SDMVPSU and SDMVSTRA. p-value < 0.05 was considered statistically significant.

Figure 1 shows the screening process of the participants. A total of 5,131 subjects aged 20–69 years were screened. Among them, 1,574 subjects were excluded, including 61 without the complete audiometric data, 1,465 without complete OBS assessment, 18 with abnormal total energy intake (<500 kcal/day or > 5,000 kcal/day for female, <500 kcal/day or > 8,000 kcal/day for male), and 30 without the data of marital status, sedentary time, information of tinnitus, information of recreational noise exposure, information of firearm noise exposure, and BMI. The mean age of all participants was 41.38 (13.72) years. Table 1 demonstrates the characteristics of the adults with or without hearing loss. There were statistical differences in age, gender, race, education level, hypertension, diabetes, dyslipidemia, rheumatoid arthritis, tinnitus, occupational noise exposure, recreational noise exposure, firearm noise exposure, veteran status, weight, and waist circumference (all p < 0.05).

The relationships between OBS levels and hearing loss in adults were shown in Table 2. Compared with OBS <16, the OBS ≥24 was associated with lower odds of hearing loss in adults (OR = 0.58, 95%CI: 0.41–0.82, p = 0.005), after adjustments for age, gender, race, hypertension, tinnitus, recreational noise exposure, and occupational noise exposure.

Further analyses were conducted to explore the relationship between OBS and hearing loss in different age, gender, occupational noise exposure, recreational noise exposure, firearm noise exposure, and veteran status groups. High OBS levels were associated with lower odds of hearing loss in participants aged 50–59 years old (OR = 0.47, 95%CI: 0.24–0.93), aged 60–69 years old (OR = 0.31, 95%CI: 0.16–0.61), with female (OR = 0.44, 95%CI: 0.20–0.96), without occupational noise exposure [Q2: (OR = 0.60, 95%CI: 0.37–0.99); Q3: (OR = 0.31, 95%CI: 0.16–0.62)], without recreational noise exposure (OR = 0.48, 95%CI: 0.30–0.76), without firearm noise exposure (OR = 0.38, 95%CI: 0.19–0.77), and without veteran status (OR = 0.57, 95%CI: 0.39–0.82). Detailed results are shown in Figure 2.

The weighted random forest model was used to further examine the importance of each component in OBS. Figure 3 shows that the Mean Decrease Gini of vitamin B12, total fat and physical activity were 43.10, 40.68 and 33.78, respectively, suggesting that vitamin B12, total fat and physical activity were important for hearing loss.