Participants in NHANES (2007-2018) were employed. Participants without even one valid diet recall, participants without information on OBS components, participants with tumors and cardiovascular disease, pregnant women, participants with implausible energy intakes (<500 kcal/day or >3,500 kcal/day for females and <800 kcal/day or >4,200 kcal/day for males) (23, 24, 28), participants without information on covariates (educational level and poverty income ratio (PIR)), and participants without information components of MetS were excluded. Eventually, 11,171 subjects were included ( Figure 1 ).

Utilizing a sophisticated sampling framework, the NHANES allows for the extrapolation of estimations to the broader spectrum of the U.S. populace. The NHANES includes interviews and examinations (29, 30). Ethical clearance for the execution of NHANES was obtained from the Ethics Review Board of the National Center for Health Statistics Moreover, comprehensive written informed consent was diligently obtained from each participant.

The main exposure was the OBS. Four lifestyle variables and sixteen nutrients were combined to create the OBS, and the information was described in Table S1 . The 20 components in this study-five prooxidants and fifteen antioxidants (23, 24)-represented the pertinent variables identified in the NHANE. It is acknowledged that divergent datasets utilized in alternative studies may encompass a slightly disparate array of factors (27, 31) ( Table S2 ). Nonetheless, given due consideration to the distinctive attributes of NHANES and a thorough review of antecedent investigations, our selection of factors is deemed comprehensive and reflective of the prevailing research landscape. Variables were categorically scored from 0 to 2 based on the sex-specific tertile of each component, and the point assignment for antioxidants and prooxidants was inverse. To enable generalization of the results, the calculation of the tertiles for each component was weighted. Then, the total score was calculated by the sum of the points for each of the 20 OBS components, and the possible OBS values were on a scale of 0 to 40. Moreover, the higher OBS indicated predominantly antioxidant exposures and lower OBS represented predominantly pro-oxidant exposures.

The sixteen nutrients were derived from the average nutrients’ intake during the 2 days. The two dietary interviews were conducted sequentially, in person and by telephone, respectively. We could obtain at least one 24-hour dietary recall of the participant and the participants’ dietary information from the U.S. Department of Agriculture Food and Nutrient Database for Dietary Studies.

The metabolic equivalent (MET) scores, which were calculated from data collected by the Physical Activity Questionnaire (PAQ), were assigned for physical activity. MET serves to quantify the relative energy expenditure associated with various activities. Within the PAQ survey, the MET values corresponding to distinct categories of activities, encompassing vigorous work-related takes, moderate work-related takes, walking or bicycling for transportation, vigorous leisure-time physical activities, and moderate leisure-time physical activities, were meticulously documented. Moreover, the quantification of physical activity was ascertained through the computation of the product derived by multiplying the MET value by the weekly frequency and duration of each specific physical activity (32).

Alcohol consumption was collected from the Alcohol Use Questionnaire, the average amount (drinks) of alcoholic beverages on those days when alcohol was consumed in the past 12 months was treated as sex-specific tertiles and subsequently assigned points. Body mass index (BMI) was from NHANES body measures collected by trained health technicians. Cotinine is generally regarded as the marker of active smoking and secondhand smoke exposure. This is owing to its distinct advantages, including a high concentration and a prolonged half-life. Moreover, the method of measurement of serum cotinine was described in detail in the NHANES website.

The outcomes were the diagnosis of MetS and each of its components. The criteria of MetS are based on the National Cholesterol Education Program Adult Treatment Panel III (33, 34). Participants considered to be with MetS required the presence of at least three of five components (33). Table S3 showed the definition of MetS.

The data on blood pressure was collected from the acquisition of a minimum of three successive readings during blood pressure measurements. These readings were subsequently computed to derive the arithmetic mean. Details concerning the anti-hypertensive treatment and the diagnosis of hypertension were collected from the blood pressure/cholesterol section in the interview data. Similarly, information regarding anti-diabetic drugs, taking insulin, and the diagnosis of diabetes were meticulously collated from the interview data’s diabetes section.

However, although previous studies have reported several methods to quantify MetS, there is still no validated tool. We chose a continuous MetS severity score developed by Gurka and colleagues (35), because it has been applied in other NHANES studies (36). The MetS severity score weighs the five factors in a way that takes into account gender and race/ethnicity differences in risk status. Operationalized definitions in NHANES have been documented in other studies (36). Moreover, the Homeostatic Model Assessment for insulin resistance (HOMA-IR) was applied to reflect the status of patients.

Furthermore, mortality follow-up data spanning from the date of survey participation through December 31, 2019, were compiled by leveraging information from the National Death Index data (https://www.cdc.gov/nchs/data-linkage/mortality-public.htm).

Potential covariates included demographic data, such as age (years), gender (male, female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American, and other), educational level (<high school, high school/general educational development, and >high school), PIR, and dietary energy intake.

Data on demographic characteristics, dietary intake, and outcomes (continuous variables) were shown as the median (P25, P75), while categorical variables were represented as numbers (percentages). The Kruskal-Wallis test was employed for the comparison of continuous baseline characteristics, as the continuous variables analyzed in this study were determined to exhibit non-normal distribution, and the Rao-Scott chi-squared test was adopted to test for categorical variables.

Survey weights from full sample 2-year interview weights (WTINT2YR) were used as recommended by NHANES (1/6 * WTINT2YR), and weighted models were conducted in the following regression analyses. Logistic models were conducted to estimate associations of OBS with the presence of MetS (dichotomous outcomes), the OBS was also categorized into quartiles and served as a categorical variable. Furthermore, linear models were performed to estimate the relationships of OBS with continuous outcomes (HOMA-IR and MetS severity score). Three models were conducted: Model one was the crude model; Model two adjusted for age, gender, race/ethnicity, educational level, and PIR; Model three further adjusted for dietary energy intake based on Model two. These covariates were chosen because they were regarded as clinically relevant confounders of the relationship between OBS and MetS. Moreover, weighted Cox proportional hazards regression models were employed to ascertain and quantify the potential associations of OBS with all-cause mortality among MetS participants.

Analysis was performed in R 4.1.1. P for trend was calculated by converting categorical variables to continuous variables. P for interaction was also calculated. Statistical significance was defined at a two-sided P value < 0.05.

To estimate the robustness of the results, some sensitivity analyses were undertaken. 1) Stratified analyses by stratifying factors including gender, age, education level, and PIR were performed. 2) The P for interaction was carried out. 3) Examination of the relationships based on dietary OBS and lifestyle OBS. 4) The analysis of removing one component from the total score at one time was conducted.

Among 11,171 participants, a total of 3,621 subjects were identified as having MetS. The distribution of OBS exhibited statistically significant variations across demographic parameters such as age, gender, race, and education level (P<0.05), as delineated in Table S4 . Notably, individuals afflicted with MetS demonstrated comparatively lower OBS scores (19 [14, 25]) in contrast to their counterparts without MetS (21 [15, 26]). Furthermore, an elevated OBS was discerned among survivors in comparison to individuals who succumbed, underscoring a potential association between OBS and survival outcomes. According to the results, the variations in the studied variables across different quartiles of OBS were revealed in Table 1 . The prevalence of MetS was observed to be comparatively lower among those in the 4^th OBS quartile (OBS ≥26) in contrast to in the other three quartiles. Moreover, individuals exhibiting lower OBS values displayed elevated levels of both the HOMA-IR and MetS severity scores in comparison to those with higher OBS values. Notably, statistically significant differences were detected between the OBS quartile groups for each component of MetS (P < 0.05). Persons in a higher quartile were more prone to have higher age, PIR, energy intake, and educational level. No significant differences in gender across OBS quartiles were observed.

Table 2 and Table S5 summarized the results of correlations between OBS and MetS. The OBS was significantly negatively related to MetS (adjusted odds ratio (OR): 0.95; 95%CI: 0.94-0.96) after full adjustment. Every one-SD increase in OBS was linked with 32% decreased odds of MetS after full adjustment. When OBS was deemed as a categorical variable, there was also a negative correlation between the highest quartile OBS and MetS (adjusted OR: 0.42; 95%CI: 0.33-0.53), after full adjustment, compared with the lowest quartile OBS. The relationship was also observed in Model 1 and Model 2. Moreover, significant associations were also discovered in the 3^rd quartile OBS (Model 2 & Model 3) and the 2^nd quartile OBS (Model 3). Importantly, the P for trend was <0.001 in all models.

Regarding the MetS components, significant negative associations were observed between OBS with abdominal obesity (adjusted OR: 0.93; 95%CI: 0.92-0.93), hypertriglyceridemia (adjusted OR: 0.96; 95%CI: 0.95-0.97), low high-density cholesterols (adjusted OR: 0.96; 95%CI: 0.95-0.97), hypertension (adjusted OR: 0.96; 95%CI: 0.95-0.97), and hyperglycemia (adjusted OR: 0.98; 95%CI: 0.97-0.99). Similarly, the relationships between OBS quartiles and each MetS component were consistently apparent. Remarkably, these relationships were found to be statistically significant within the highest OBS quartile group (all P values for trend < 0.05), as illustrated in Table S5 .

The continuous MetS severity score, quantified by the Mets Z-score, is exclusively accessible for adult individuals belonging to Hispanic, white, and black adults (35). The adjusted formula can also be applied to Asian populations (37). Importantly, the score was applied in other NHANES studies (36). However, no formula is currently available for the Mexican-American population. Therefore, we excluded Mexican Americans and some incomplete information, and eventually, 4,380 of 11,171 individuals were calculated for the MetS Z-score.

The adjusted linear regression model indicated that, even after accounting for potential confounders, for each one-unit increment in OBS, a person’s MetS Z-score decreased by 3%. The adjusted mean differences (MDs) for MetS were as follows: -0.17 (-0.29, -0.06) for the 2^nd OBS quartile, -0.29 (-0.39, -0.20) for the 3^rd OBS quartile, and -0.56 (-0.68, -0.44) for the 4^th OBS quartile, in comparison to this reference group (participants in the lowest OBS quartile as the reference). These findings suggested a consistent trend of decreasing MetS severity scores with higher OBS quartiles, as supported by a statistically significant trend (P for trend < 0.05) ( Table 3 ).

Furthermore, the degree of insulin resistance was also evaluated through the HOMA-IR, with calculations available for 5,378 out of the total 11,171 individuals possessing complete data. Increased OBS was related to lower HOMA-IR (adjusted MD, -0.53; 95% CI: -0.71, -0.35). The adjusted MD for the highest OBS quartile was -8.62 (-12.64, -4.60) (P for trend < 0.001) ( Table S6 ).

Survival status was recorded for 3,622 participants out of 3,621 MetS patients, with 181 dead cases and 3,440 alive cases. Due to the low prevalence of deaths among MetS patients (45 cases of coronary heart disease-related dead cases and 49 cases of cancer-related dead cases), we will not explore the correlation of OBS with cause-specific mortality. The weighting algorithm enabled the sampling sample to reflect the overall sample.

For MetS patients, a one-unit OBS increase was related to a significant decrease in the risk of all-cause mortality (adjusted hazard ratio (HR) 0.96; 95% CI; 0.92-0.9995). When the subjects in the lowest quartile of the OBS were served as a reference, the participants in the high quartiles had a decreased risk of all-cause mortality [3^rd quartile: 0.66 (0.18-0.82); 4^th quartile: 0.38 (0.18-0.82)] after full adjustment ( Table 4 ). Both dietary OBS and lifestyle OBS played an important role, and they were both negatively related to the risk of all-cause mortality in subjects with MetS. The strength of the association between OBS and the risk of all-cause mortality among males with MetS might be pronounced, but the association was statistically significant for male participants solely in Model 1 and Model 2, but not in Model 3 ( Table 4 ).

The relationships of dietary/lifestyle OBS with the susceptibility of MetS were also estimated ( Table 5 ). Notably, both dietary OBS and lifestyle OBS exhibited inverse associations with the risk of MetS, alongside reductions in MetS severity score (P for interaction <0.05). Additionally, the negative correlations between OBS and the prevalence of MetS, as well as MetS severity, persisted even after conducting stratified analyses based on gender, age, education, and PIR ( Figure 2 and Table S7 ). Furthermore, the exclusion of any individual OBS component did not yield a substantial impact on the association between OBS and the risk of MetS, nor on the severity of MetS or all-cause mortality of MetS patients, as elucidated in Table S8 .