The distribution of the baseline characteristics and biochemical measures of the study subjects across the OBS quartiles range

The distribution of baseline characteristics and biochemical measures among the OBS quartiles are presented in Table 2. The mean age was 45.19 ± 0.26 years, comprising 7862 (46.66%) females and 8988 (53.34%) males. Predominantly, subjects were non-Hispanic white (54.59%). All participants were categorized into four groups according to OBS quartiles: Q1 (OBS, 3 to 15), Q2 (OBS, 16 to 21), Q3 (22 to 26), and Q4 (27 to 37). Compared to the lowest OBS quartile, individuals in the highest quartile tended to have higher income levels, be more married, highly educated, elevated total energy intake and lower concentrations of CRP. Notably, gender distribution did not exhibit significant variance across OBS quartiles, thereby indicating a balanced representation of both males and females within each quartile. Moreover, differences in age distribution across OBS groups did not achieve statistical significance. As for MetS traits, with the exception of diastolic blood pressure (DBP), the distribution of other cardiometabolic characteristics (WC, FBG, TG, HDL-C and SBP) was statistically different across OBS quartiles.

Table 2 The distribution of baseline characteristics and metabolic-related factors among OBS quantiles for 16,850 participantsAssociation between OBS and MetS and its components

Table 3 shows the association between OBS as a categorical variable and a continuous variable and MetS traits. When OBS was used as a continuous variable, we found higher OBS was negatively associated with MetS both in crude model and adjusted model. After adjusted all covariables, each one SD increase in the OBS was found to be associated with an 4% decrease in the risk of MetS (OR = 0.96, 95%CI = 0.95–0.97). When OBS was used as a categorical variable and compared with the lowest quantile as a reference, the OR values of MetS in the Q4 group (the OBS group with the strongest antioxidant properties) were 0.55 (95%CI: 0.47–0.64, P for trend < 0.0001). Participants in the other three groups were less likely to be at risk for MetS than participants in the lowest quartile.

Table 3 Association between OBS and MetS and its components in US adult population

The results of the multivariable logistic regressions showed that OBS was significantly associated with MetS components. When OBS was treated as a continuous variable, it was significantly and negatively associated with high WC (OR = 0.96), elevated TG (OR = 0.99), low HDL-C(OR = 0.97), hypertension (OR = 0.98) and elevated FBG levels (OR = 0.98). After full adjustment, the results remained significant, the estimates of OR were 0.95, 0.98, 0.97, 0.98 and 0.97, respectively. Participants in the highest quantile of OBS had lower odds of abdominal obesity, hypertension, elevated TG, low HDL-C, elevated FBG levels, respectively(WC: OR = 0.61, 95%CI = 0.54–0.69, P < 0.0001; hypertension: OR = 0.69, 95%CI = 0.58–0.83, P < 0.0001; elevated TG: OR = 0.68, 95%CI = 0.57–0.82, P < 0.0001; low HDL-C: OR = 0.60, 95%CI = 0.50–0.70, P < 0.0001: elevated FBG: OR = 0.74, 95%CI = 0.62–0.88, P < 0.0001). Our results indicated that the removal of any single OBS component did not significantly affect the results for subjects, and the removal of the OBS components brought OR estimates within 5% of the original model results (Table S1). When BMI was removed from OBS, the associations of OBS with the increase of TG and FBG levels were no longer significant, and other OBS components did not alter the associations between OBS and MetS and its components (Pinteraction > 0.05). When OBS were categorized into four intervals, the significant associations were no longer observed (Table S2).

Considering that the diagnostic criteria for MetS are slightly different, WC, HDL-C, TG, FBG levels, and blood pressure were used as continuous variables as components of MetS to further analyze the effect of OBS on them (Table S3). We found that higher level of OBS were associated with lower WC (β = -5.99; 95% CI: -6.86,-5.11; P < 0.001), TG (β = -23.1; 95% CI: -34.6, -11.7; P < 0.001), SBP (β = -3.25; 95% CI:-4.41, -2.08; P < 0.001) and DBP (β = -2.33; 95% CI: -3.32, -1.34; P < 0.001), while it was directly proportional to HDL-C levels in the multivariate model (β = 3.75; 95% CI: 2.70, 4.81; P < 0.001). However, the effect of OBS on FBG levels was not statistically significant (P = 0.228).

Association between dietary OBS/lifestyle OBS and MetS traits

Figure 1 lists the results of the multivariable logistic regression analysis used to assess the association of dietary and lifestyle OBS with MetS and its components. When dietary OBS/lifestyle OBS was treated as a continuous variable, an increased dietary OBS was negatively related to the risks of MetS (per 1SD, OR = 0.89, 95%CI = 0.84–0.95) and its components (ORs ranging from 0.86 to 0.90), but not of elevated FBG levels and hypertension (Fig. 1A). For lifestyle OBS, it was significantly and negatively associated with each trait of MetS (Fig. 1A). When dietary OBS and lifestyle OBS were analyzed as categorical variables, the associations of the two OBS with the risk of MetS and its components are shown in Fig. 1B and C, respectively. Dietary OBS showed significant negative associations with MetS (Q4 vs. Q1), HDL-C (Q4 vs. Q1), TG (Q4 vs. Q1), and WC (Q4 vs. Q1) with ORs of 0.75,0.67, 0.74, and 0.72, respectively (Fig. 1B). Lifestyle OBS showed similar significant negative associations with MetS and all its components, with ORs of 0.16, 0.52, 0.43, 0.44, 0.12, and 0.38 for MetS (Q4 vs. Q1), FBG (Q4 vs. Q1), HDL-C (Q4 vs. Q1), TG (Q4 vs. Q1), WC (Q4 vs. Q1), and hypertension (Q4 vs. Q1), respectively (Fig. 1C). This result shows that adherence to antioxidant dietary nutrients has a limited protective effect on MetS, and it is more necessary to adhere to healthy lifestyle behaviors.

Fig. 1

Association of the dietary/lifestyle OBS as a continuous per SD (A) and a categorical variable (B, C) with MetS and its components risk. Adjusted for age, sex, race, marital status, education level, energy intake and CRP level. Asterisk refers to significant difference for MetS and its components risk at different dietary/lifestyle OBS interval. OBS indicates oxidative balance score; MetS, metabolic syndrome; WC, waist circumference; HDL-C, high-density lipoprotein cholesterol; FBG, fasting blood glucose; TG, triglycerides; OR, odds ratio; CI, confidence interval

Association of different OBS with MetS traits stratified by sex

The study conducted subgroup analyses and interaction tests, stratified by sex, to assess the consistency of the relationship between OBS and MetS in the general population. Additionally, the aim was to identify potential variations in different population settings. Table S4 shows the relationship between total OBS, dietary OBS and lifestyle OBS and MetS traits discovered by multiple logistic regression in different sex subgroups. In the unadjusted model, there was a significant linear trend for total OBS, dietary OBS and lifestyle OBS and MetS traits in all subgroups (P for trend < 0.0001). After adjusting for all confounding factors, we observed a stable and significant negative association between different OBS and MetS risks in both male and female subgroups. Higher total OBS, dietary OBS, and lifestyle OBS in the highest quartile relative to the lowest quartile, were associated with a notably diminished risk of abdominal obesity, elevated TG and low HDL-C levels. The OBS was negatively associated with the odds of hypertension, while the subgroup analysis by sex unveiled no statistically significant relationship between dietary OBS and hypertension. Intriguingly, an analysis of the association between elevated blood glucose, OBS, and dietary OBS interval odds ratios revealed non-statistically significant results and a lack of dose-response effect among men (Figure S1). In stark contrast, among women, the odds of elevated FBG in the highest OBS interval were statistically significantly lower 34%, and the OBS-gender interaction was statistically significant (Pinteraction=0.03).

Analysis of restricted cubic spline regression stratified by sex

Subsequently, the relationships between OBS and MetS traits in both men and women were further assessed through the utilization of RCS curves and multivariable logistic regression (model 3, Fig. 2). Figure 2A illustrates a distinct trend of decreasing odds ratio (OR) for MetS with rising OBS, which was consistent across sex subgroups (males: P for nonlinear < 0.0001, females: P for nonlinear = 0.0054). Interestingly, the OBS was negatively associated with the FBG levels in a linear manner in both males and females. When the study endpoints were WC and TG levels, the non-linearity trend of decreasing OR with increasing OBS was noticed, which remained across sex subgroups. Significant nonlinear relationships were identified between OBS and hypertension (males: P for nonlinear < 0.0001, females: P for nonlinear = 0.2760) and HDL-C levels (males: P for nonlinear < 0.0001, females: P for nonlinear = 0.0558) in males and not in female. Despite slight variations observed in the results of the nonlinear analysis of the restricted cubic splines, the overall trends of the MetS traits and OBS remained generally consistent across the plots.

Fig. 2

Dose-response associations between OBS and risk of MetS traits stratified by sex. (A) MetS; (B) Elevated FBG; (C) Low HDL-C; (D) Elevated TG; (E) High WC; (F) Hypertension. The solid lines and shaded areas represent the central risk estimates and 95% CIs. OBS indicates oxidative balance score; MetS, metabolic syndrome; WC, waist circumference; HDL-C, high-density lipoprotein cholesterol; FBG, fasting blood glucose; TG, triglycerides; OR, odds ratio; CI, confidence interval