Data were employed from the National Health and Nutrition Examination Survey (NHANES) database, a program administered by the Centers for Disease Control and Prevention and the National Center for Health Statistics in the US. The National Center for Health Statistics Ethics Review Board approved the study, which was conducted with the explicit written consent of all participants.

CRM disease is a constellation of conditions that includes cardiovascular disease (CVD), chronic kidney disease (CKD), and diabetes mellitus (DM) (14), and patients with CRM disease during 1999–2018 were included in this research. Participants were excluded if they met any of the following conditions: (1) age under 20; (2) presence of malignant tumors; (3) pregnancy; (4) missing OBS data; or (5) missing information on mortality or survival time. Finally, 12,886 CRM patients were included in the final analysis (Supplementary Figure 1).

OBS is calculated using earlier established research (15, 16). It is calculated using data on 16 nutrient intakes obtained from the initial in-person dietary recall, along with four lifestyle-related variables. These include a total of 5 pro-oxidants and 15 antioxidants, selected based on established links between oxidative stress and these factors (17). OBS elements are grouped into four categories: (1) dietary antioxidants—such as fiber, β-carotene, riboflavin, niacin, vitamin B6, folate, vitamin B12, vitamin C, vitamin E, calcium, magnesium, zinc, copper, and selenium; (2) dietary pro-oxidants—namely total fat and iron; (3) lifestyle antioxidants—including physical activity; and (4) lifestyle pro-oxidants—including alcohol consumption, smoking, and body mass index (BMI). All variables were equally weighted in the final OBS calculation.

Supplementary Table 1 outlines the point allocation for the OBS components. For alcohol intake, points were assigned as follows: 2 points for non-drinkers, 1 point for moderate drinkers (0 to 15 g/d for females and 0 to 30 g/d for males), and 0 points for heavy drinkers (≥15 g/d for females and ≥30 g/d for males). Other components were divided into sex-specific tertiles. Antioxidant components were given 0 to 2 points across the tertiles from lowest to highest, while pro-oxidant components were scored in the reverse order, with 0 points for the highest tertile and 2 points for the lowest tertile.

We used the NHANES Public-Use Linked Mortality File through December 2019,¹ which was linked to the National Death Index (NDI) data using a probabilistic matching algorithm to determine mortality status. The cause-specific mortality data in the NDI have been shown to accurately classify deaths, with only a relatively slight possibility of misclassification. The underlying cause of death was identified according to the International Classification of Diseases, and cardiovascular mortality was defined as death due to heart diseases (codes I00–I09, I11, I13, I20–I51) and cerebrovascular diseases (codes I60-I69).

Baseline information on demographics (age, sex, race, and education), lifestyle factors (smoking, alcohol drinking, and BMI), medical conditions (diabetes, hypertension, CVD, CKD), and medication usage (glucose-lowering drugs, antihypertensive drugs, and lipid-lowering drugs) was collected through structured interviews administered by trained personnel following standardized NHANES protocols. Lifestyle factors and some medical history elements were assessed through these structured interviews, while objective clinical and laboratory data were incorporated to validate and supplement the definitions of chronic conditions wherever possible (e.g., diabetes defined using HbA1c or fasting glucose, and CKD defined by eGFR or uACR). Anthropometric measurements [height, weight, BMI, systolic blood pressure (SBP), and diastolic blood pressure (DBP)] are performed by an experienced physician. Blood samples were taken by the Mobile Examination Centers during the medical examination component of the NHANES survey, and laboratory indices were selected. Demographic data were further classified as alcohol drinking (heavy moderate; none) and smoking (current smoker; former smoker; never smoker). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation (18), and CKD was defined as eGFR ≤ 60 mL/min/1.73 m2 or uACR ≥30 mg/g or self-reported diagnosis history. Diabetes was defined as fasting glucose ≥ 7.0 mmol/L or Hemoglobin A1c (HbA1c) (%) ≥ 6.5 or self-reported diagnosis history of diabetes or use of any hypoglycemic medication (19). CVD was defined as being informed by a health professional of having congestive heart failure, angina, coronary heart disease, heart attack, or stroke.

Mean ± standard error (SE) was used to describe continuous variables, with ANOVA applied for group comparisons. Categorical variables were expressed as frequencies and proportions, analyzed using the chi-square test. The patients were divided into four groups based on OBS using the weighted quartiles method. Adjusted hazard ratios (aHR) and 95% confidence intervals (CIs) for all-cause and cardiovascular mortality were estimated through multivariate Cox regression. Kaplan–Meier curves were used for survival analysis, and differences between OBS groups were assessed using a stratified log-rank test. Model 1 adjusted for age, gender, and race. Model 2 was adjusted for age, gender, race, alcohol drinking, smoking, and body mass index (BMI). Model 3 extended Model 2 by additionally adjusting for DM, CVD, chronic kidney disease (CKD), uric acid, blood urea nitrogen, HbA1c, eGFR, hemoglobin, aspartate aminotransferase (AST), systemic immune-inflammation index (SII), high-density lipoprotein cholesterol (HDL-C), total cholesterol, lipid-lowering drug, antihypertensive drug, and glucose-lowering drug. To examine the dose–response relationships between OBS and mortality in CRM patients, restricted cubic spline regression analysis was performed, and the analysis was adjusted for the variables in Model 3. Stratified analyses were performed to assess the associations between quartiles of OBS levels and all-cause as well as cardiovascular death, stratified by age, and the presence of CVD, CKD, and DM. To further investigate the potential heterogeneity of the relationship between OBS and mortality outcomes in CRM patients, we performed additional subgroup analyses based on the number of CRM conditions (1, 2, or 3). Participants who experienced mortality within the first 2 years of follow-up were excluded to minimize potential bias from reverse causation.

NHANES survey weights were applied in all analyses to accommodate its complex, stratified cluster sampling design and to produce estimates representative of the U. S. Population. R software (v4.2.1) was used for statistical analysis, and significance was defined as a two-tailed p-value less than 0.05.

A total of 12,886 participants from NHANES were included in our analysis. Among the 12,886 included participants, 8,599 (66.7%) had only one CRM condition, 3,377 (26.2%) had two CRM conditions, 910 (7.1%) had all three conditions (CVD, CKD, and DM) (Supplementary Figure 2). Participants averaged 57.37 ± 0.25 years in age, with females comprising nearly 52.66%. Participants were categorized into four groups according to OBS quartile. Patients with reduced OBS levels were more likely to be smokers and to have a higher prevalence of CVD and hypertension. They show lower levels of eGFR, albumin, HDL-C, but had higher levels of total cholesterol, SII. Moreover, the incidence of all-cause and cardiovascular mortality was higher among patients with lower OBS levels (Table 1).

Over a median follow-up of 7.9 years (interquartile range: 4.0–12.5 years), a total of 3,838 deaths from all causes and 1,412 cardiovascular deaths were recorded. Multivariate analyses using various models were performed to estimate the aHR for all-cause and cardiovascular mortality, considering OBS as both a continuous and categorical variable. When analyzed as a continuous variable, elevated OBS levels were independently associated with reduced risks of all-cause and cardiovascular death after adjustment for confounders in Model 2 [all-cause mortality: adjusted hazard ratio (aHR), 0.98, 95% CI: 0.97–0.99, p < 0.001; cardiovascular mortality: aHR, 0.98, 95%CI: 0.97–0.99, p = 0.003] (Table 2). In addition, restricted cubic splines also indicated a linear and negative correlation between OBS and all-cause mortality (P for nonlinear = 0.301) and CVD mortality (P for nonlinear = 0.863) (Figure 1).

To further clarify the association between OBS levels and the risk of all-cause and cardiovascular mortality, OBS was analyzed as a categorical variable in both univariate and multivariate models. After adjusting for confounders, higher OBS levels were independently linked to a diminished risk of all-cause mortality [model 3: Q2, Q3, Q4: aHR (95CI%) = 0.85 (0.75–0.96), 0.87 (0.77–0.98), 0.74 (0.62–0.88), respectively; P for trend<0.001]. An inverse association was found between elevated OBS quartiles and cardiovascular mortality, while lower SIRI quartiles showed no statistically significant relationship with cardiovascular death [model 3: Q2, Q3, Q4: aHR (95CI%) = 0.96 (0.77–1.19), 0.78 (0.63–0.97), 0.70 (0.53–0.93), respectively; P for trend = 0.003] (Table 2). Consistently, Kaplan–Meier survival curves also indicated a significantly higher survival probability associated with higher OBS levels (Figure 2). Overall, a linear inverse association between OBS levels and all-cause as well as cardiovascular mortality was supported by these findings.

We conducted a sensitivity analysis by removing participants who died within 2 years of follow-up. The findings indicated that the association of OBS level with all-cause and cardiovascular mortality remained significant (Supplementary Table 2). To further explore heterogeneity in subgroup analyses, stratified analyses were performed according to age, CVD, CKD, and DM status. The findings were consistent across subgroup analyses (non-DM, CVD, and regardless of age, as well as CKD) for all-cause mortality. A similar trend in cardiovascular mortality was observed across OBS quartiles in subgroups (age≥65, non-CKD, non-DM, and CVD). Interactive analysis revealed no significant interactions between all variables for both all-cause and cardiovascular mortality (Supplementary Figures 3, 4).

Furthermore, a higher OBS was significantly associated with a lower risk of both all-cause and cardiovascular mortality in patients with only one CRM condition. In contrast, in patients with two or three co-existing CRM conditions, the inverse association between OBS and mortality was attenuated and did not reach statistical significance. These findings suggest that the protective effect of a higher OBS may be more evident in patients with a less complex disease burden (Supplementary Table 3).

Despite its strengths, this study has several limitations. Firstly, OBS was assessed solely at baseline, which limits our ability to evaluate changes or fluctuations over time, and consequently, the need for longitudinal studies with repeated measures of oxidative balance to better establish temporality and causal relationships. Secondly, OBS was based on self-reported dietary and lifestyle data, which may be prone to recall bias. Thirdly, the study’s observational nature restricts causal conclusions, and residual confounding from unmeasured variables remains a possibility. Fourth, NHANES data do not include direct biomarkers of oxidative stress, such as ROS levels or antioxidant enzyme activity, which could further validate the relationship between OBS and mortality. Moreover, the absence of information on genetic susceptibility and medication adherence may further constrain the ability to fully elucidate individual variability in oxidative balance and its impact on clinical outcomes. Finally, since the NHANES study was conducted exclusively in the United States, the applicability of our findings to other regions remains uncertain. Therefore, further high-quality research in different populations is required to confirm these findings.