Abstract
We explored the relationship between the oxidative balance score (OBS) and all-cause, as well as cause-specific mortality, within a large and nationally representative sample. A total of 30,074 adults participated in this prospective study, utilizing data from the National Health and Nutrition Examination Survey cycles spanning 1999 to 2018 in the United States. Mortality outcomes were determined through linkage to National Death Index records up to December 31, 2019. OBS was computed based on 16 dietary and 4 lifestyle components. Cox proportional hazards models and restricted cubic splines were employed to estimate associations between OBS and mortality across all-cause, cancer, cardiovascular disease (CVD), and respiratory disease. Over the mean 10.3 years of follow-up, we observed 3561 deaths. In comparison to the lowest OBS quartile, individuals in the highest quartile exhibited a significantly reduced multi-adjusted hazard ratio of 0.66 (95% confidence interval [CI]: 0.56–0.78), 0.56 (95% CI: 0.40–0.77), and 0.43 (95% CI: 0.22–0.83) for all-cause, cancer, and respiratory disease mortality, respectively. However, no significant association was found for CVD mortality (0.80 [95% CI: 0.59–1.09]). Restricted cubic splines demonstrated a linear association. Sensitivity analysis and subgroup analysis confirmed the stability of results for all-cause and cancer mortality, while respiratory disease mortality might be influenced by chronic disease. A robust negative correlation was identified between OBS and all-cause, cancer, and respiratory disease mortality, with no such association observed for CVD mortality. This study provides insights into enhancing overall antioxidant dietary and lifestyle practices among adults.
Keywords: mortality, NHANES, oxidative balance score, respiratory disease
1. Introduction
Oxidative stress is conceptualized as an imbalance between the production of reactive oxygen species within the body and the body’s capacity to neutralize these species through its antioxidant defense system.[1] It has been identified as one of pivotal mechanisms influencing various health outcomes.[2–4] The magnitude of oxidative stress is influenced by diverse factors, encompassing dietary components such as dietary fiber, vitamin C, glucosinolates, tocopherols, vitamins C and E, polyphenols, polyunsaturated fats, and specific minerals (zinc, selenium, and calcium).[5,6] Additionally, physical activity, obesity, smoking status, and other behavioral factors play a role.[7] Despite robust evidence from basic science and animal studies, observational and clinical studies focusing on individual factors have produced inconsistent results.[8,9] The intricate interplay and biological interactions of multiple pro-oxidant and antioxidant factors are increasingly acknowledged as challenging to comprehensively assess with a singular factor. To gain a more holistic understanding of the diverse effects of various dietary pattern and lifestyles on the overall oxidative/antioxidant balance, numerous studies have explored the impacts of different antioxidants and antioxidants as a collective entity, referred to as the oxidative balance score (OBS), on disease. Various iterations of OBS have demonstrated associations with diverse health outcomes, including aging and cardiovascular disease (CVD).[10] Furthermore, investigations into the associations between OBS and all-cause mortality, as well as cause-specific mortality, have been conducted. However, due to limitations in the populations selected for these studies, which limited in adults at high cardiovascular risk, older women, and middle-aged adults, and the considerable heterogeneity in OBS composition, lacking crucial antioxidant dietary factors such as dietary fiber, a new study with a larger sample size was deemed necessary to comprehensively examine the association of OBS with all-cause and cause-specific mortality.[11–13]
The principal aim of this study was to prospectively investigate the association between OBS (comprising 20 a priori OBS components, 16 dietary factors, and 4 lifestyles factors) and the risk of all-cause, cardiovascular, cancer, and respiratory disease mortality among participants in the National Health and Nutrition Examination Survey (NHANES) study, a nationwide, broad adult cohort.
2. Methods
2.1. Design
The study population were obtained from the NHANES cycles 1999 to 2018, and data were obtained by questionnaire and interview, mobile physical examination, and laboratory tests with a complex, multistage, and probability sampling method. Details of NHANES have been described on the web.[14] In this analysis, 101,316 participants from the continuous NHANES (1999–2018) data sets were first enrolled. After excluding those participants aged <20 years and those without mortality data. 54,945 participants remained. Furthermore, we excluded those without dietary data, physical activity, body mass index (BMI), and cotinine data. Finally, a total of 30,074 participants were retained in our cohort for analysis. The flowchart of the study is presented in Fig. 1.
Figure 1.
Participants flow chart.
2.2. Oxidative balance score
The OBS serves as a comprehensive metric, integrating assessments of 16 dietary and 4 lifestyle factors, delineating between 5 pro-oxidants and 15 antioxidants. This scoring system is rooted in existing knowledge elucidating the intricate interplay between oxidative stress and various dietary and lifestyle components.[15,16] The OBS is constructed by evaluating the dietary intake of 16 specific nutrients, including dietary fiber, carotene, riboflavin, niacin, vitamin B6, total folate, vitamin B12, vitamin C, vitamin E, calcium, magnesium, zinc, copper, selenium, total fat, and iron. This information is gathered through the first day of the two 24-hour dietary recalls.
Concurrently, 4 pivotal lifestyle factors (physical activity, BMI, alcohol consumption, and smoking) are integrated into the OBS calculation. These lifestyle factors are meticulously examined to assess their impact on oxidative balance. Among them, total fat, iron, BMI, alcohol consumption, and smoking are identified as pro-oxidants, while the remaining factors are categorized as antioxidants. For instance, alcohol consumption is stratified into 3 distinct groups: heavy drinkers (≥15 g/d for women and ≥30 g/d for men), non-heavy drinkers (0–15 g/d for women and 0–30 g/d for men), and nondrinkers. Following the methodology outlined by Cheng et al,[17] these groups are assigned scores of 0, 1, and 2 points, respectively. Subsequent to this initial categorization, the other OBS components are further stratified by gender and subsequently divided into 3 groups based on tertiles. In this classification, antioxidants receive scores of 0 to 2 in groups 1 to 3, while prooxidants are assigned scores of 2 to 0 in groups 1 to 3, as detailed in Tables S1 and S2, Supplemental Digital Content, https://links.lww.com/MD/P671. Importantly, a higher OBS score signifies a more pronounced exposure to antioxidants.
2.3. Death mortality
Death status was determined through probabilistic matching to the National Death Index up to December 31, 2019, utilizing a unique study identifier. Comprehensive details regarding the matching methodology can be referenced from the National Center for Health Statistics. Causes of death were categorized according to ICD-10 codes. The primary outcomes assessed in this study encompassed mortality from all causes, CVDs (codes I00-I09, I11, I13, I20-I51, and I60-I69), cancer (codes C00-C97), and respiratory diseases (codes J40-J47 and J09-J18).
2.4. Covariates
Potential confounding variables in the analysis was guided by established findings in prior literature, ensuring a comprehensive assessment when incorporated into our multivariate models. Among these covariates were socio-demographic characteristics, including sex, age, and race. Race categories were delineated as non-Hispanic white, non-Hispanic black, Mexican American, and others. Additionally, socio-economic status was considered through the poverty income ratio, categorized as low (≤1.3), middle (>1.3–≤3.5), and high (>3.5). Marital status, another vital socio-demographic variable, was classified into 3 groups: never married, married or living with a partner, and widowed, divorced, or separated. Home status was evaluated by considering individuals’ housing arrangements, including whether they rented, owned, were in the process of buying their homes, or had other housing arrangements. Furthermore, we accounted for specific chronic diseases, including hypertension (yes or no), diabetes mellitus (DM) (yes or no), CVD (encompassing congestive heart failure, coronary heart disease, angina, heart attack, or stroke), chronic respiratory disease (CRD) (encompassing chronic bronchitis, emphysema, or chronic obstructive pulmonary disease), and cancer (yes or no). Additionally, total energy intake (kcal) was factored in as relevant covariates to ensure a comprehensive and robust analysis.
2.5. Statistical analysis
All analyses adhere to NHANES guidelines, employing a nonrandom, stratified sampling design to effectively capture specific subgroups within the population. Sample weights are assigned to participants to address nonresponse and other complexities inherent in the survey’s design. Our analytical approach involves consolidating data from ten distinct survey cycles spanning the years 1999 to 2018 into a unified 20-year dataset.
To estimate standard errors for continuous variables, we utilized the Taylor Series Linearization method. Associations among categorical variables were assessed using Student t test. For categorical variables, we employed weighted percentages, calculated means with 95% confidence intervals (CI), and utilized survey-weighted chi-squared tests.
Mortality was assessed by age standardization. The Kaplan–Meier method was used to plot survival curves associated with the OBS. Multivariable cox regression models were employed to estimate hazard ratios (HRs) and 95% CI of OBS quartiles to all-cause and cause-specific mortalities. The same approach was used to estimate HRs and 95% CI for dietary and lifestyle components of OBS in relation to all-cause and cause-specific mortality. We further used competitive risk analysis to assess the relationship between OBS and cardiovascular, cancer, and respiratory disease mortality. Several sensitivity analyses were performed to test the robustness of the results. First, subjects with a history of chronic disease (hypertension, DM, CVD, CRD, and cancer) were excluded to minimize potential reverse causality due to chronic disease. Second, individuals with follow-up less than or equal to 3 years were excluded.
We further explored the linear relationship between OBS and all-cause and cause-specific mortalities within multivariable regression models using restricted cubic spline, employing 4 knots (at the 20th, 40th, 60th, and 80th percentiles) to flexibly model and visually represent the relationship. Finally, stratified analysis and interaction tests were performed by age group (<60 or ≥60 years), sex, and chronic disease (yes or no).
All data analyses were conducted using R software (version 4.3.1) in conjunction with the “survey” package.
3. Results
The baseline characteristics of the population are presented in Table 1. The average follow-up time for survivors was 10.3 years. Among the 30,074 individuals, the mean age was 45.9 years, and 49.7% were female. Compared to those in the lowest OBS quartile, individuals in the highest OBS quartile had higher education levels, greater wealth (indicated by poverty income ratio and home status), were married or living with a partner, predominantly female, and identified as non-Hispanic White. Notably, the prevalence of hypertension, DM, CVD, and CRD showed a gradual decrease as OBS values increased. During the follow-up, there were 3561 deaths: 1064 deaths from CVD, 877 deaths from cancer, and 237 deaths from respiratory disease. Figure 2 illustrates the age-standardized all-cause and cause-specific mortality across OBS quartiles.
Table 1.
Baseline characteristics of participants from the National Health and Nutrition Examination Survey (NHANES) 1999 to 2018 according to quartiles of the OBS.
| Variable | Quartiles of the OBS | P value | |||
|---|---|---|---|---|---|
| Q1 [3,15] | Q2 [15,21] | Q3 [21,26] | Q4 [26,37] | ||
| Age | 45.53 (0.28) | 46.63 (0.30) | 45.86 (0.29) | 45.58 (0.34) | .01 |
| Sex | .02 | ||||
| Female | 47.74 (0.73) | 49.20 (0.87) | 50.79 (0.83) | 50.90 (0.72) | |
| Male | 52.26 (0.73) | 50.80 (0.87) | 49.21 (0.83) | 49.10 (0.72) | |
| Race/ethnicity | <.001 | ||||
| Non-Hispanic White | 66.98 (1.29) | 70.40 (1.21) | 73.70 (1.10) | 76.31 (0.98) | |
| Non-Hispanic Black | 15.40 (0.81) | 10.36 (0.70) | 7.94 (0.49) | 5.68 (0.40) | |
| Mexican American | 6.92 (0.53) | 7.55 (0.60) | 7.37 (0.58) | 6.98 (0.55) | |
| Others | 10.70 (0.65) | 11.69 (0.62) | 11.00 (0.59) | 11.04 (0.58) | |
| Marital status | <.001 | ||||
| Never married | 20.82 (0.78) | 17.99 (0.76) | 18.07 (0.73) | 17.79 (0.82) | |
| Married/living with partner | 59.33 (0.86) | 64.66 (0.87) | 66.50 (0.88) | 68.19 (0.96) | |
| Widowed/divorced/separated | 19.85 (0.62) | 17.35 (0.58) | 15.43 (0.67) | 14.02 (0.61) | |
| PIR | <.001 | ||||
| Low | 26.56 (0.90) | 19.04 (0.72) | 17.22 (0.71) | 13.05 (0.61) | |
| Middle | 37.48 (1.01) | 36.51 (1.01) | 33.58 (0.91) | 30.77 (1.03) | |
| High | 35.96 (1.08) | 44.45 (1.13) | 49.20 (1.14) | 56.18 (1.23) | |
| Education | <.001 | ||||
| College or more | 51.69 (0.97) | 59.56 (0.89) | 66.46 (1.00) | 75.26 (0.89) | |
| Middle school or lower | 5.77 (0.33) | 4.29 (0.29) | 3.18 (0.27) | 2.32 (0.19) | |
| High school | 42.53 (0.94) | 36.15 (0.87) | 30.36 (0.93) | 22.42 (0.86) | |
| Home status | <.001 | ||||
| Rented or others | 33.86 (0.96) | 30.05 (0.92) | 27.64 (1.04) | 27.04 (1.08) | |
| Owned or being bought | 66.14 (0.96) | 69.95 (0.92) | 72.36 (1.04) | 72.96 (1.08) | |
| CVD | 8.60 (0.42) | 7.95 (0.38) | 6.45 (0.43) | 4.79 (0.33) | <.001 |
| CRD | 9.06 (0.47) | 8.08 (0.45) | 6.06 (0.45) | 5.09 (0.40) | <.001 |
| DM | <.001 | ||||
| Yes | 8.22 (0.38) | 8.19 (0.43) | 7.07 (0.40) | 5.09 (0.28) | |
| Borderline | 1.77 (0.20) | 1.81 (0.18) | 1.51 (0.19) | 1.36 (0.18) | |
| No | 90.01 (0.42) | 90.01 (0.47) | 91.42 (0.47) | 93.54 (0.34) | |
| Hypertension | 37.28 (0.76) | 37.14 (0.76) | 33.02 (0.89) | 29.51 (0.89) | <.001 |
| Cancer | 8.75 (0.43) | 8.92 (0.46) | 9.40 (0.49) | 8.97 (0.40) | .76 |
| Energy (kcal) | 1593.5 (10.7) | 2044.5 (12.7) | 2368.6 (15.7) | 2766.9 (19.1) | <.001 |
Weighted mean±SE and Student t test for continuous variables. Weighted %, mean (95% CI), and Cochran–Mantel–Haenszel Chi-square test for categorical variables.
OBS = oxidative balance score.
Figure 2.
Age-standardized all-cause and cause specific mortality by quartiles of the OBS (NHANES 1999–2018). NHANES = National Health and Nutrition Examination Survey, OBS = oxidative balance score.
Table 2 illustrates the associations within quartiles of the OBS and all-cause and cause-specific mortality. The findings suggest significant negative associations between OBS and all-cause, cancer, and respiratory diseases mortality, while no such association was observed for CVD mortality. In the context of the fully adjusted weighted model (model 3), the highest OBS quartile demonstrated a diminished risk of all-cause, cancer, and respiratory disease mortality in comparison to the lowest OBS quartile. The weighted multivariate HRs were 0.66 (95% CI: 0.56–0.78) with P for trends < .001, 0.56 (95% CI: 0.40–0.77) with P for trends < .001, and 0.43 (95% CI: 0.22–0.83) with P for trends = .003, respectively. Additionally, we conducted separate analyses of the dietary and lifestyle components of OBS to investigate their associations with all-cause and cause-specific mortality (Tables S3 and S4, Supplemental Digital Content, https://links.lww.com/MD/P671). Notably, the relationship between dietary OBS and mortality was similar to that of total OBS. After adjusted by all covariates, including dietary OBS, lifestyle OBS was found to be linked to all-cause, cancer, and CVD mortality, however not include respiratory disease mortality. Subsequent sensitivity analyses, excluding participants without chronic diseases, yielded consistent results in OBS and all-cause mortality and cancer mortality, but not for respiratory disease mortality (Table S5, Supplemental Digital Content, https://links.lww.com/MD/P671). While, next sensitivity analyses, excluding follow-up times less than or equal to 3 years, showed stable results in all-cause, cancer, and respiratory disease mortality (Table S6, Supplemental Digital Content, https://links.lww.com/MD/P671). Although, the highest OBS quartile was not significantly associated with CVD mortality, but P for trends = .036. This information is presented in Table 3. Further results using competing risk analysis showed negative associations between OBS and mortality from cancer, respiratory disease, and cardiovascular disease (Table S7, Supplemental Digital Content, https://links.lww.com/MD/P671). Moreover, Fig. 3 visually depicts a progressive reduction in cumulative hazard with OBS over the follow-up period (P < .001 o P < .001). Finally, the restricted cubic spline regression model (Fig. 4) indicates linear associations between OBS and all-cause and cause-specific mortality.
Table 2.
Associations of quartiles of the OBS and all-cause and specific cause mortality.
| Character | Quartiles of OBS | ||||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | P trends * | |
| All-causes | |||||
| Model 1 | ref | 0.90 (0.79,1.02) | 0.74 (0.66,0.84) | 0.60 (0.53,0.68) | <.001 |
| Model 2 | ref | 0.92 (0.81,1.05) | 0.81 (0.71,0.92) | 0.66 (0.58,0.75) | <.001 |
| Model 3 | ref | 0.91 (0.80,1.05) | 0.79 (0.69,0.90) | 0.66 (0.56,0.78) | <.001 |
| Cardiovascular | |||||
| Model 1 | ref | 1.02 (0.86,1.21) | 0.72 (0.58,0.90) | 0.63 (0.49,0.82) | <.001 |
| Model 2 | ref | 1.09 (0.91,1.32) | 0.82 (0.64,1.04) | 0.75 (0.56,0.99) | .013 |
| Model 3 | ref | 1.10 (0.92,1.32) | 0.81 (0.64,1.02) | 0.80 (0.59,1.09) | .051 |
| Cancer | |||||
| Model 1 | ref | 0.85 (0.65,1.11) | 0.73 (0.55,0.96) | 0.57 (0.44,0.73) | <.001 |
| Model 2 | ref | 0.81 (0.62,1.06) | 0.72 (0.55,0.93) | 0.58 (0.45,0.75) | <.001 |
| Model 3 | ref | 0.80 (0.61,1.07) | 0.71 (0.53,0.95) | 0.56 (0.40,0.77) | <.001 |
| Respiratory disease | |||||
| Model 1 | ref | 0.70 (0.44,1.12) | 0.41 (0.26,0.66) | 0.35 (0.21,0.60) | <.001 |
| Model 2 | ref | 0.71 (0.42,1.19) | 0.45 (0.26,0.80) | 0.39 (0.22,0.69) | <.001 |
| Model 3 | ref | 0.72 (0.45,1.16) | 0.50 (0.27,0.91) | 0.43 (0.22,0.83) | .003 |
Model 1: adjusted for age (continuous), sex, and race/ethnicity.
Model 2: model 1 + PIR, marital status, home status, and education.
Model 3: model 2 + hypertension, DM, CVD, CRD, cancer, and energy (kcal).
CRD = chronic respiratory disease, CVD = cardiovascular disease, DM = diabetes mellitus, OBS = oxidative balance score, PIR = poverty income ratio, Ref: reference.
Ptrends was calculated by using the median value of each quartile as a continuous variable in each model.
Table 3.
Stratified analyses on associations of quartiles of the OBS and all-causes and specific cause mortality by age and chronic disease groups.
| Character | Quartiles of OBS | P * | |||
|---|---|---|---|---|---|
| Q1 | Q2 | Q3 | Q4 | ||
| All-causes | |||||
| Age group | .32 | ||||
| <60 | ref | 1.06 (0.78,1.45) | 0.85 (0.64,1.13) | 0.73 (0.50,1.07) | |
| ≥60 | ref | 0.91 (0.78,1.05) | 0.90 (0.78,1.04) | 0.83 (0.70,0.99) | |
| Sex | .86 | ||||
| Female | ref | 0.90 (0.74,1.08) | 0.77 (0.62,0.95) | 0.70 (0.54,0.91) | |
| Male | ref | 0.94 (0.78,1.13) | 0.82 (0.68,0.97) | 0.65 (0.52,0.82) | |
| Chronic disease | .45 | ||||
| Yes | ref | 0.88 (0.76, 1.01) | 0.85 (0.74, 0.98) | 0.70 (0.61, 0.81) | |
| No | ref | 1.06 (0.77, 1.47) | 0.80 (0.57, 1.12) | 0.78 (0.56, 1.09) | |
| Cardiovascular | |||||
| Age group | .56 | ||||
| <60 | ref | 1.49 (0.84,2.63) | 0.90 (0.48,1.66) | 1.04 (0.42,2.58) | |
| ≥60 | ref | 1.06 (0.83,1.34) | 0.94 (0.71,1.25) | 0.96 (0.68,1.34) | |
| Sex | .11 | ||||
| Female | ref | 1.35 (0.97,1.87) | 1.29 (0.88,1.89) | 1.01 (0.64,1.59) | |
| Male | ref | 1.02 (0.80,1.30) | 0.61 (0.45,0.83) | 0.75 (0.49,1.15) | |
| Chronic disease | .19 | ||||
| Yes | ref | 0.99 (0.81, 1.22) | 0.87 (0.67, 1.13) | 0.76 (0.56, 1.04) | |
| No | ref | 1.84 (0.96, 3.53) | 0.85 (0.38, 1.89) | 1.35 (0.66, 2.77) | |
| Cancer | |||||
| Age group | .09 | ||||
| <60 | ref | 0.98 (0.58,1.65) | 0.65 (0.38,1.13) | 0.52 (0.28,0.96) | |
| ≥60 | ref | 0.78 (0.58,1.04) | 0.91 (0.67,1.24) | 0.77 (0.54,1.11) | |
| Sex | .41 | ||||
| Female | ref | 0.68 (0.39,1.19) | 0.53 (0.32,0.87) | 0.51 (0.27,0.95) | |
| Male | ref | 0.91 (0.66,1.25) | 0.86 (0.60,1.24) | 0.61 (0.41,0.91) | |
| Chronic disease | .75 | ||||
| Yes | ref | 0.80 (0.60, 1.07) | 0.70 (0.53, 0.93) | 0.64 (0.47, 0.86) | |
| No | ref | 0.82 (0.46, 1.45) | 0.84 (0.49, 1.43) | 0.54 (0.31, 0.94) | |
| Respiratory disease | |||||
| Age group | .84 | ||||
| <60 | ref | 0.88 (0.32, 2.45) | 0.79 (0.25, 2.49) | 0.42 (0.11, 1.66) | |
| ≥60 | ref | 0.81 (0.47,1.40) | 0.59 (0.29,1.20) | 0.71 (0.34,1.46) | |
| Sex | .96 | ||||
| Female | ref | 0.79 (0.46, 1.36) | 0.64 (0.28, 1.47) | 0.57 (0.23, 1.41) | |
| Male | ref | 0.70 (0.35,1.39) | 0.40 (0.19,0.84) | 0.34 (0.13,0.89) | |
| Chronic disease | .86 | ||||
| Yes | ref | 0.68 (0.40, 1.17) | 0.42 (0.23, 0.76) | 0.43 (0.24, 0.79) | |
| No | ref | 0.72 (0.21, 2.52) | 0.73 (0.23, 2.33) | 0.51 (0.13, 2.02) | |
Models were adjusted for age, sex, race/ethnicity, PIR, marital status, home status, education, hypertension, DM, CVD, CRD, cancer, and energy(kcal), but except the subgroup variables.
CRD = chronic respiratory disease CVD = cardiovascular disease, DM = diabetes mellitus, OBS = oxidative balance score, PIR = poverty income ratio.
P: P for interaction.
Figure 3.
Unadjusted Kaplan–Meier hazard curves: oxidative balance score (OBS) and all causes (A); OBS and cardiovascular diseases (B); OBS and cancer (C); OBS and respiratory disease (D). CI = confidence intervals, OBS = oxidative balance score.
Figure 4.
Dose–response relationship between OBS and all-cause mortality (A), cardiovascular disease mortality (B), cancer mortality (C), and respiratory disease mortality (D) in US adults, NHANES 1999 to 2018. Red solid lines and red dotted line represent restricted cubic spline models and 95% CI, respectively. Multivariable cox-regression model is used to estimate the fully adjusted HR in all-cause and cause specific mortality and corresponding 95% CI. Models were adjusted by age (continuous), sex, race/ethnicity, PIR, marital status, home status, education, hypertension, DM, CVD, CRD, cancer, and energy (kcal). CI = confidence intervals, CRD = chronic respiratory disease, CVD = cardiovascular disease, HR = hazard ratio, OBS = oxidative balance score, PIR = poverty income ratio, NHANES = National Health and Nutrition Examination Survey.
Subgroup analyses and interaction tests, stratified by age group, sex and chronic diseases, revealed predominantly negative associations between OBS and all-cause, cancer, and respiratory disease mortality across most populations. However, no significant differences between subgroups were observed. The negative association between OBS and cardiovascular mortality did not reach statistical significance, and no disparities were noted between subgroups.
4. Discussion
In this large prospective cohort study involving adults in the United States, our findings indicate a significant association between the OBS and a reduced risk of all-cause, cancer, and respiratory disease mortality. This association remained statistically significant even after adjusting for sociodemographic factors, chronic diseases, and other relevant variables, demonstrating a linear relationship. Consistent with prior studies reporting a link between OBS and the risk of all-cause and cancer mortality, our results remained robust across various sensitivity analyses. Specifically, the association persisted when considering dietary OBS, lifestyle OBS, and after excluding chronic patients or individuals with follow-up periods of 3 years or less. However, additional evidence is warranted to establish a clear association between OBS and the risk of respiratory diseases and cardiovascular mortality.
As the oxidative stress theory continues to evolve, assessing the impact of oxidation/antioxidation on overall health outcomes has become increasingly imperative.[18,19] A valuable concept in this regard is the OBS. OBS is predominantly comprised of 2 components: dietary OBS and lifestyle OBS, encompassing smoking status, physical activity, alcohol consumption, and BMI. While lifestyle OBS components have garnered relatively consistent recognition across studies,[17,20,21] the composition of dietary OBS exhibits considerable heterogeneity, ranging from more than 10 to over 40 items in some instances.[10,22,23] Prior to this study, 3 studies had explored the association between overall OBS and the risk of mortality, revealing significant variations in the OBS components included. The earliest population-based cohort study, Geographical and Ethnic Differentiated Causes of Stroke, incorporated 12 dietary and lifestyle factors to calculate total OBS, revealing associations with all-cause and cancer mortality.[11] However, significant associations with cardiac and heart failure mortality were not observed. Notably, this study oversampled individuals from the “stroke belt,” with covariates including cellulose, body mass index, and physical activity, now recognized as crucial components in oxidative balance evaluation. In another study within the Iowa Women’s Health Study, Mao et al reported similar associations.[13] Women with higher OBS scores (15 a priori factors: 11 dietary and 4 lifestyle) exhibited a significantly reduced risk of all-cause and cancer mortality by 34% and 29%, and a significant association with CVD by 39%, respectively. A recent study involving nearly 20,000 middle-aged Hispanics in the Mediterranean cohort revealed an association between overall oxidative balance and the risk of all-cause, CVD, cancer, and other mortality.[12] This study employed the same lifestyle scores as the Iowa Women’s Health Study and our study but with limited dietary choices. However, it is important to note the study population restrictions, as the participants were limited to women aged 55 to 69 years and middle-aged adults. Collectively, these studies consistently demonstrate a robust association between OBS and the risk of all-cause and cancer mortality, with varying associations with cardiovascular mortality observed in relatively elderly populations. Our results showed that the association between OBS and CVD mortality was not significant in COX analysis, while the results of competitive risk analysis showed a significant negative association. The first possible reason is that there are other causes of death, and competing events cause the COX model to ignore this part of the information. Second, OBS may affect outcomes indirectly, which COX models cannot detect by ignoring competing events. The significant negative association of competing risk models suggests that OBS may indirectly reduce the chance of cardiovascular death by increasing the risk of non-cardiovascular death.
Our study broadened the participant population, retained 4 lifestyle components in the OBS, and introduced additional oxidative and antioxidant dietary factors, such as fiber and calcium. We validated the preventive effect of OBS on all-cause and cause-specific mortality risk in a more extensive population using the NHANES database. This study boasted a larger sample size, a longer mean follow-up, a more pronounced response to OBS in the U.S. adult population, and robust results validated through several sensitivity analyses. Drawing from the theory of oxidative balance, our adjustment for numerous potential confounding factors further elucidated the beneficial effect of overall oxidative stress on mortality prevention. Sensitivity analyses, stratification analyses, and restricted cubic splines collectively underscored the robustness of the results concerning all-cause and tumor mortality risk, revealing the existence of linear associations. Additionally, we highlighted the potential preventive effects on specific causes of mortality, such as respiratory disease mortality risk, though acknowledging the need for further evidence. These findings underscore the importance of overall healthy behaviors, encompassing both dietary and lifestyle.
However, our research is not without limitations. Firstly, all measurements were taken only at baseline, and participants’ lifestyle and eating habits may change over time during long-term follow-up. Unaccounted-for changes could potentially impact the results, but given the database’s design, these alterations cannot be provided. Secondly, despite controlling for numerous covariates, we acknowledge the inability to completely eliminate residual confounding caused by other relevant variables, such as the effects of drugs and visits, which were omitted due to their complexity. Thirdly, recall bias may influence self-reported data, particularly regarding chronic disease.
In conclusion, our findings further support the notion that a more balanced antioxidant profile is associated with lower all-cause mortality and a reduced risk of specific causes of death, including cancer and respiratory diseases, within a broader U.S. adult population. Despite the heterogeneity in the composition of OBS, these findings underscore the efficacy of an antioxidant-rich diet and a healthy lifestyle in preventing premature death.
Acknowledgments
Data from NHANES collection was sponsored by the CDC.
Author contributions
Conceptualization: Weiliang Kong.
Data curation: Shanni Ma.
Formal analysis: Shanni Ma, Weiliang Kong.
Software: Shanni Ma, Weiliang Kong.
Writing – original draft: Shanni Ma, Weiliang Kong.
Writing – review & editing: Shanni Ma, Weiliang Kong.
Supplementary Material
Abbreviations:
- BMI
- body mass index
- CI
- confidence intervals
- CRD
- chronic respiratory disease
- CVD
- cardiovascular disease
- DM
- diabetes mellitus
- HR
- hazard ratio
- NHANES
- National Health and Nutrition Examination Survey
- OBS
- oxidative balance score
- PIR
- poverty income ratio
All the authors listed have approved the manuscript that is enclosed.
Detailed methods and protocols for the NHANES study were approved by the CDC/NCHS Research Ethics Review Board. They are publicly available through the CDC.gov website; this includes informed consent procedures for all participants. All methods in this study were performed by the relevant guidelines and regulations. This study was exempt from human subject ethical review as the data is freely available in the public domain.
The authors have no funding and conflicts of interest to disclose.
The datasets generated during and/or analyzed during the current study are publicly available.
Supplemental Digital Content is available for this article.
How to cite this article: Ma S, Kong W. Association between oxidative balance score and mortality: Prospective cohort study in a representative US population. Medicine 2025;104:33(e43932).
References
- [1].Sohal RS, Weindruch R. Oxidative stress, caloric restriction, and aging. Science (New York, NY). 1996;273:59–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Butterfield DA, Halliwell B. Oxidative stress, dysfunctional glucose metabolism and Alzheimer disease. Nat Rev Neurosci. 2019;20:148–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Benz CC, Yau C. Ageing, oxidative stress and cancer: paradigms in parallax. Nat Rev Cancer. 2008;8:875–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Givertz M, Colucci W. New targets for heart-failure therapy: endothelin, inflammatory cytokines, and oxidative stress. Lancet (London, England). 1998;352:SI34–8. [DOI] [PubMed] [Google Scholar]
- [5].Ciudad-Mulero M, Domínguez L, Morales P, Fernández-Ruiz V, Cámara M. A review of foods of plant origin as sources of vitamins with proven activity in oxidative stress prevention according to EFSA scientific evidence. Molecules (Basel, Switzerland). 2023;28:7269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Mangione CM, Barry MJ, Nicholson WK, et al. Vitamin, mineral, and multivitamin supplementation to prevent cardiovascular disease and cancer: US preventive services task force recommendation statement. JAMA. 2022;327:2326–33. [DOI] [PubMed] [Google Scholar]
- [7].Niemann B, Rohrbach S, Miller M, Newby D, Fuster V, Kovacic J. Oxidative Stress and cardiovascular risk: obesity, diabetes, smoking, and pollution: part 3 of a 3-part series. J Am Coll Cardiol. 2017;70:230–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Onishi S, Miyai N, Zhang Y, et al. Excessive alcohol intake and liver fibrosis are associated with skeletal muscle mass reduction in elderly men: the Wakayama study. Aging Clin Exp Res. 2022;34:185–92. [DOI] [PubMed] [Google Scholar]
- [9].Hong SH, Bae YJ. Association between alcohol consumption and the risk of sarcopenia: a systematic review and meta-analysis. Nutrients. 2022;14:3266. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Hernández-Ruiz A, García-Villanova B, Guerra-Hernández E, Amiano P, Ruiz-Canela M, Molina-Montes E. A review of A priori defined oxidative balance scores relative to their components and impact on health outcomes. Nutrients. 2019;11:774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Kong SY, Goodman M, Judd S, Bostick RM, Flanders WD, McClellan W. Oxidative balance score as predictor of all-cause, cancer, and noncancer mortality in a biracial US cohort. Ann Epidemiol. 2015;25:256–62.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Talavera-Rodriguez I, Fernandez-Lazaro CI, Hernández-Ruiz A, et al. Association between an oxidative balance score and mortality: a prospective analysis in the SUN cohort. Eur J Nutr. 2023;62:1667–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Mao Z, Prizment AE, Lazovich D, Bostick RM. Associations of dietary and lifestyle oxidative balance scores with mortality risk among older women: the Iowa Women’s Health Study. Eur J Nutr. 2021;60:3873–86. [DOI] [PubMed] [Google Scholar]
- [14].Prevention. CfDCa. National Health and Nutrition Examination Survey: questionnaires, datasets, and related documentation. Available at: https://www.cdc.gov/nchs/nhanes. Accessed June 1, 2024.
- [15].Qu H. The association between oxidative balance score and periodontitis in adults: a population-based study. Front Nutr. 2023;10:1138488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Li H, Song L, Cen M, et al. Oxidative balance scores and depressive symptoms: mediating effects of oxidative stress and inflammatory factors. J Affect Disord. 2023;334:205–12. [DOI] [PubMed] [Google Scholar]
- [17].Cheng S, Han Y, Jiang L, Lan Z, Liao H, Guo J. Associations of oxidative balance score and visceral adiposity index with risk of ischaemic heart disease: a cross-sectional study of NHANES, 2005-2018. BMJ Open. 2023;13:e072334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Gorrini C, Harris IS, Mak TW. Modulation of oxidative stress as an anticancer strategy. Nat Rev Drug Discovery. 2013;12:931–47. [DOI] [PubMed] [Google Scholar]
- [19].Forman HJ, Zhang H. Targeting oxidative stress in disease: promise and limitations of antioxidant therapy. Nat Rev Drug Discovery. 2021;20:689–709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Zhan F, Lin G, Duan K, Huang B, Chen L, Ni J. Higher oxidative balance score decreases risk of stroke in US adults: evidence from a cross-sectional study. Front Cardiovasc Med. 2023;10:1264923. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Son DH, Lee HS, Seol SY, Lee YJ, Lee JH. Association between the Oxidative Balance Score and Incident Chronic Kidney Disease in Adults. Antioxidants (Basel). 2023;12:335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Cho AR, Kwon YJ, Lee JH. Oxidative balance score is inversely associated with the incidence of non-alcoholic fatty liver disease. Clin Nutr. 2023;42:1292–300. [DOI] [PubMed] [Google Scholar]
- [23].Shahriarpour Z, Nasrabadi B, Hejri-Zarifi S, et al. Oxidative balance score and risk of osteoporosis among postmenopausal Iranian women. Arch Osteoporos. 2021;16:43. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.