Introduction

Osteoporosis is a common metabolic osteopathy, with characteristics of decreased bone mineral density (BMD), which can cause bone fragility and an increased risk of fracture [1, 2]. As the population ages, approximately 200 million individuals worldwide suffer from osteoporosis. It is estimated that 13.5 million Americans will have osteoporosis by 2030, an increase from 10.2 million in 2010 [3, 4]. In terms of finance, as the global population ages, the direct cost of osteoporosis is expected to reach around $25.3 billion per year by 2025 [5], imposing a heavy burden on the health care system and social economy [6]. Furthermore, osteoporosis has been globally recognized as a public health concern due to its rising morbidity, mortality, and healthcare costs.

Current osteoporosis treatment is aimed at correcting the imbalance between bone resorption and bone formation [7]. It has recently become clear that oxidative stress contributes significantly to bone resorption and osteogenesis imbalances [8]. As a result of oxidative stress, osteoblasts are more likely to die, undergo bone remodeling, and lose bone mass as a result of increased inflammation. Several studies have revealed that antioxidants reduce free radicals to prevent bone loss caused by oxidative stress [9]. Thus, dietary antioxidants probably contribute to osteoporosis prevention and treatment.

There has been a lot of research on osteoporosis and individual antioxidants in the past. Carotenoids, particularly lycopene, may protect older people's BMD from bone loss, according to Sahni et al. [10]. In addition, Hyun et al. also found that dietary zinc intake and plasma zinc were positively correlated with BMD in men [11]. Another meta-analysis observed a negative association between vitamin C intake and osteoporosis risk [12]. In addition to individual nutrition, food effects on health were also affected by the interaction between them. For this reason, we used the Compound Dietary Antioxidant Index (CDAI). Developed by Wright et al. to quantify dietary antioxidant capacity, CDAI measures the potential antioxidant capacity of a diet [13]. Current research has shown that the composite antioxidant index is negatively linked with cancer, cardiovascular disease, and diabetes [14,15,16]. However, there are few studies on the relationship between CDAI and BMD. In this study, we used data from the National Health and Nutrition Examination Survey (NHANES) 2005–2010 to investigate the relationship between CDAI and BMD.

Materials and methods

Study population

To obtain nationally representative health and nutrition data for the noninstitutionalized population, the NHANES employs a complex, multistage, and stratified sampling method. In addition, an ethics review board of the National Center for Health Statistics Research approved all NHANES protocols, and all participants provided informed consent [17].

The NHANES dataset was utilized for this investigation from 2005 to 2010. A nutrition investigation, medical examination, and medical condition questionnaire were completed by all participants (n = 31,034). These criteria were used for exclusion: (1) missing BMD data (n = 14,344); (2) missing antioxidant data (n = 435); (3) patients with a history of cancer diseases, thyroid diseases or diabetes (2186); therefore, the final number of subjects included in this study was 14,069 (Fig. 1).

Fig. 1
figure 1

Study flowchart. NHANES, National Health and Nutrition Examination. Survey; BMD, bone mineral density

Full size image

Calculation of the composite dietary antioxidant index

A 24-h dietary recall interview is offered to all NHANES participants. Using Survey Net, which was developed for computer-assisted food coding and data management, interview data files were electronically sent from the field and imported into the system. The NHANES website contains two types of dietary data: Individual Foods and Total Nutrient Intakes. Based on the Total Nutrient Intake Files, this study analyzed participants’ daily total nutrient and energy intakes. For vitamin A, NHANES provides information on vitamin A intake (mg) on its website. Six dietary antioxidant micronutrients were considered in calculating the DAI, including vitamins A, C, E, zinc, magnesium, and selenium.

In a previous report, CDAI development was described and validated [13, 18]. To estimate CDAI, we subtracted the mean and divided by the SD from each of the above antioxidant micronutrients. We then calculated the CDAI by adding up the standardized intakes of the micronutrients, as follows:

$$\mathrm{CDAI}={\sum }_{i=1}^{6}\frac{Individual\, Intake - Mean}{SD}$$

BMD measurement

Bone health is often assessed using BMD. As a standard method of measuring BMD and identifying osteoporosis risks, dual-energy X-ray absorptiometry (DXA) is conducted [19]. DXA with a fan-beam densitometer was used to measure BMD in this study (Hologic Inc., Bedford, MA, USA [20], which is an internationally accepted standard-of-care screening tool used to assess fragility-fracture risk. Certified and trained radiology technologists administered the DXA examinations. Participants aged 8 and older who qualified for the survey received DXA scans. A DXA examination was not available for pregnant women, patients with a recent history of radiographic contrast material, and patients weighing more than 300 pounds.

Baseline data collection

Data on demographics, diet, examination, laboratory, and questionnaires are the covariates. Demographics data covered age (years, range 8 to 85, average:30), gender (male and female), race, level of education (under high school graduate, high school, some college or over, missing), family income-poverty ratio (%) and marital status (separated/divorced/widowed/never married, married/living with partner, missing/refused/don’t know). Dietary data included Vitamin E (mg), Vitamin A RAE (mg), Vitamin C (mg), Magnesium (mg), Zinc (mg), Selenium (mg), Calcium (mg); Energy (kcal); Phosphorus (mg). Examination data included weight (kg), height (m), body mass index (BMI, kg/m2). And, laboratory data consisted of white blood cell (1000 cells/μL), lymphocyte (1000 cells/μL), segmented neutrophils (1000 cells/μL), C-reactive protein (mg/dL). Finally, questionnaire data covered alcohol consumption (yes/no), smoking behavior (now, former, never), arthritis (yes/no) and hypertension (yes/no). Smoking was diagnosed according to the following criteria: The current active smoker has smoked more than 100 cigarettes throughout life, and they smoke occasionally or every day now; Former smokers have smoked more than 100 cigarettes throughout their lives and have given up smoking; A non-smoker has smoked fewer than 100 cigarettes throughout their lifetime. On the other hand, the diagnostic criteria for the consumption of alcohol were as follows: drunk, had at least 12 alcoholic drinks of any type in your lifetime; never drunk, who does not meet the above criteria; missing drinking data [16].

Statistical analysis

According to the National Center for Health Statistics (NCHS) normative analysis criteria, the NHANES' complicated survey design factors—including weighting, clustering, and stratification—were taken into account. A percentage was used for categorical variables, while a mean and standard deviation were used for continuous variables. Univariate and multivariate linear regression models were used to evaluate the relationship between CDAI and BMD levels. To further investigate the correlation between independent variables and dependent variables, we carried out multiple regressions. In the multiple linear regression models, an unadjusted model (Model 1) was first established, followed by an adjusted model (Model 2) taking into account age, gender and race. Then, using variables like age, gender, race/ethnicity, education level, marital status, smoking status, drinking, white blood cells, lymphocyte number, segmented neutrophils, ratio of family income to poverty, C-reactive protein, arthritis, BMI, hypertension, calcium, energy intake, phosphorus, fully adjusted models (Model 3) were calculated. Previous studies have found a correlation between white blood cells, lymphocytes, neutrophils, and BMD, so we adjusted these factors in Model 3 [21, 22]. Furthermore, to increase confidence in the results, we stratified the data by gender (male and female) and BMI (≤ 18.5, 18.5–24.9, 25–29.9, ≥ 30). As a further method to deal with non-linearity, we also employed weighted generalized additive models. All analyses were conducted using R software (version 4.0.3; easily accessible at https://www.R-project.org) and Empower Stats 2.0 (available at http://www.empowerstats.com). We set the significance probability at P < 0.05.

Results

Baseline characteristics of the participants

Using CDAI quartiles as a classification method, the weighted characteristics of 14,069 participants were analyzed. This study's participants are grouped by CDAI quartile, so Table 1 provides some weighted baseline info. There were younger, non-Hispanic white people, more married, less smokers, and poorer households in the fourth quartile. More notably, significantly higher CDAI was associated with significantly higher femoral neck, trochanter, and total spine BMD (P < 0.001).

Table 1 Weighted characteristics of the study participants by the quartiles of composite dietary antioxidant index
Full size table

The relationship between CDAI and BMD

An analysis of the association between CDAI and BMD was conducted using a multiple linear regression model. Table 2 shows the results for Model 1 (unadjusted), Model 2, and Model 3 (fully adjusted model). A significant correlation was found in Table 2 between CDAI and BMD in Model 1 (β = 0.004 (0.003, 0.004) P < 0.00001). We detected this significant correlation in Model 3 (β = 0.003 (0.003, 0.004) P < 0.00001) after fully adjusting for covariates. In addition, we evaluated the association between CDAI and trochanter BMD (P < 0.00001) and total spine BMD (P = 0.00039) and discovered a strong positive association in the fully adjusted model. Furthermore, for the sensitivity analysis, CDAI was also treated as a categorical variable (quartiles). In all quartile categories, CDAI was correlated significantly with BMD, and BMD level increased as CDAI extent increased. It was found that the femoral neck BMD of the Q4 group was greater than that of the Q1 group by 0.034 g/cm2, the trochanter by 0.039 g/cm2, and the total spine BMD by 0.009 g/cm2. And the trend remained significant among different CDAI quartile groups (P for trend < 0.01).

Table 2 Associations between CDAI and BMD in different models
Full size table

Subgroup analyses stratified by gender and BMI

To further confirm the robustness of the results observed in Table 2 to potential confounders, we conducted a stratified analysis by subgroups defined by the major covariates known to influence CDAI. As shown in Table 3, in a subgroup analysis stratified by sex, CDAI in males (P < 0.00001) and females (P < 0.00001) were positively correlated with femoral neck and trochanter BMD in a fully adjusted model. Besides, in the gender subgroup analysis between CDAI and total spine BMD, there was a positive association between men [0.006 (0.005, 0.007)] and women [0.005 (0.004, 0.006)] when not adjusted, and this positive association remained significant in men after full adjustment.

Table 3 Subgroup analyses stratified by gender and race
Full size table

Furthermore, in the BMI subgroup analysis, CDAI and femoral neck (P < 0.01) and trochanter BMD (P < 0.01) still showed significant correlation in the four BMI groups after completely adjusting for interference factors. Nevertheless, the relationship between total spine BMD (P = 0.008) and CDAI remained significantly positive only in BMI ≥ 30 group.

In addition, we used a generalized additive model and smooth curve fitting to define a relationship between CDAI and BMD. There was a positive correlation between their relationship in Figs. 2, 3 and 4.

Fig. 2
figure 2

The association between CDAI and total spine BMD. a Each black point represents a sample. Solid red line represents the sample trend. b Solid red line represents the smooth curve fit between variables. Blue bands represent the 95% of confidence interval from the fit. gender, age, race, education level, marital status, ratio of family income to poverty, white blood cell, lymphocyte number, Segmented neutrophils; C-reactive protein; arthritis; drinks; smoke; BMI; Hypertension; calcium; energy; phosphorus were adjusted

Full size image
Fig. 3
figure 3

The association between CDAI and trochanter BMD. a Each black point represents a sample. Solid red line represents the sample trend. b Solid red line represents the smooth curve fit between variables. Blue bands represent the 95% of confidence interval from the fit. gender, age, race, education level, marital status, ratio of family income to poverty, white blood cell, lymphocyte number, Segmented neutrophils; C-reactive protein; arthritis; drinks; smoke; BMI; Hypertension; calcium; energy; phosphorus were adjusted

Full size image
Fig. 4
figure 4

The association between CDAI and femoral neck BMD. a Each black point represents a sample. Solid red line represents the sample trend. b Solid red line represents the smooth curve fit between variables. Blue bands represent the 95% of confidence interval from the fit. gender, age, race, education level, marital status, ratio of family income to poverty, white blood cell, lymphocyte number, Segmented neutrophils; C-reactive protein; arthritis; drinks; smoke; BMI; Hypertension; calcium; energy; phosphorus were adjusted

Full size image

Discussion

To our knowledge, this is the first study to explore the relationship between CDAI and BMD based on the NHANES database. After adjusting for all potential covariates, it was found that CDAI was positively correlated with BMD, suggesting that CDAI is a protective factor for bone mass development. The increasing trend in CDAI was associated with higher BMD levels. Additionally, those in the highest CDAI quartile had greater BMD than those in the lowest quartile.

Most of the studies observed the relationship between individual antioxidants and BMD. The link between certain antioxidant micronutrients and BMD has been examined in earlier research. For instance, a cross-sectional study found that selenium intake was associated with BMD in postmenopausal women, and women with higher selenium levels were less likely to develop osteoporosis [23]. Furthermore, increased dietary consumption of vitamin C was related to higher lumbar and femoral neck BMD, according to a comprehensive review and meta-analysis of observational studies [12]. In addition, a protective effect of carotenoids on BMD in older individuals has also been observed. Carotenoids, especially lycopene, may have a protective effect against bone loss in older adults [10]. Previous cohort studies have shown that both higher blood alpha-tocopherol levels and higher dietary alpha-tocopherol intake are related to higher BMD and reduced osteoporosis and fracture risk, including hip fractures [24,25,26]. A correlation between antioxidants and BMD might be due to an oxidative stress mechanism.

The science of human nutrition has moved over the last several decades from emphasizing particular nutrients to stressing overall dietary quality. Studies that concentrate on the health consequences of a particular nutrient intake have a fundamental flaw in that they neglect the possibility of interactions between other dietary intakes. Because the foods people eat contain a variety of antioxidants, recent studies focused on the total dietary antioxidant index, which is a more comprehensive and accurate approach. In a recent study, the relationship between BMD and dietary intake of antioxidant vitamins and carotenoids was demonstrated [27]. The authors of the study emphasized the need to study combinations of antioxidants, rather than a single intake of antioxidants. In view of the above, we further considered examining the association between the antioxidant dietary quality index and BMD. The results showed that the dietary antioxidant mass fraction was significantly associated with BMD among all study participants. Our findings support the hypothesis that a high intake of antioxidants may be beneficial for bone health. According to our findings, we observed that higher overall intakes of the antioxidant vitamins A, C, E, zinc, selenium, and magnesium, corresponding to higher BMD.

Recent studies have discovered that oxidative stress is a pathogenic factor for many disease states. There is a strong link between oxidative stress and osteoporosis, according to some studies [28]. Oxidative stress affects bone metabolism from bone resorption and bone formation and then affects BMD change. Oxidative stress affects osteoblast activity and osteoblast number, which accelerates bone loss. In addition, the inactivation of β-catenin will lead to delayed bone mineralization. Thus, the Wnt/β-catenin signaling pathway is essential for bone formation and maintenance [29]. In addition, oxidative stress is related to the activation of NADPH oxidase and decreased levels of antioxidant enzymes and glutathione, which can enhance bone resorption [30], and other studies have shown that reactive oxygen species can regulate bone resorption by promoting osteoclast formation and increasing osteoclast activity [31]. Furthermore, the body's antioxidant system protects the body from free radical damage and can prevent the occurrence of oxidative stress. For the loss of bone mass caused by oxidative stress, the intake of antioxidants can be used to protect bone health and prevent osteoporosis [32]. Similarly, dietary antioxidants, such as vitamin E (also known as alpha-tocopherol) and C, beta-carotene and lycopene, are also significant in this context [33], so we explore the relationship between CDAI and BMD, increasing the corresponding supporting evidence.

We conducted subgroup analyses by gender to examine how sex affects bone metabolism. It has been observed that CDAI and BMD are significantly correlated in both genders. The association remained significant after adjusting for multiple confounding factors, suggesting that the relationship was independent of sex. Furthermore, in a subgroup analysis stratified by BMI, we found that CDAI was significantly associated with femoral neck, trochanter, and total spine BMD in all four BMI groups in Model 1. However, after the complete adjustment of confounding factors, the correlation between CDAI and femoral neck and trochanter BMD remained positive, and compared to those with low BMI, those with high BMI had higher BMD. It is worth noting that CDAI and total spine BMD were only found in the BMI ≥ 30 kg/m2 group in Model 3. There was no association between the other BMI groups. Similarly, a cohort study of 845 people reported higher BMD among older subjects in patients with high BMI. This was in addition to the protective effect of high BMI on femoral neck BMD [34]. Similarly, Saarelainen and colleagues found that low BMI increases the risk of osteoporosis and fractures [35].

In our understanding, this is the first study involving a relatively large sample size that examines the association between CDAI and BMD. Food items were evaluated for their overall antioxidant capacity rather than just on single antioxidants, and the accuracy of the assessments was increased. This study has several limitations, including the following: (1) the cross-sectional design of the original survey made it impossible to establish a causal relationship between CDAI and BMD, (2) it is possible that residual confounding exists, (3) food intake can be inaccurately assessed since participants’ memories were used to record food intake for one day.

In conclusion, in this cross-sectional study, we discovered that CDAI were positively correlated with BMD in the general American population, suggesting that people can reduce the risk of low BMD and further prevent the development of osteoporosis by taking antioxidant supplements. However, further prospective and mechanistic studies are needed to confirm these findings.