Association of Milk Consumption and Vitamin D Status in the US Population by Ethnicity: NHANES 2001–2010 Analysis

Vitamin D has been identified as a nutrient of public health concern, and higher intake of natural or fortified food sources of vitamin D, such as milk, are encouraged by the 2015–2020 Dietary Guidelines for Americans. We, therefore, examined the association of milk consumption and vitamin D status in the United States (US) population. Twenty-four-hour dietary recall data and serum 25(OH)D concentrations were obtained from the National Health and Nutrition Examination Survey 2001–2010 and were analyzed by linear and logistic regression after adjusting for anthropometric and demographic variables. Significance was set at p < 0.05. Approximately 57–80% children and 42–60% adults were milk consumers. Milk intake (especially reduced-fat, low fat and no-fat milk) was positively associated (p linear trend < 0.05) with serum vitamin D status and with a 31–42% higher probability of meeting recommended serum vitamin D (>50 nmol/L) levels among all age groups. Serum vitamin D status was also associated with both type and amount of milk intake depending upon the age and ethnicity. In conclusion, the results indicate that milk consumers consistently have higher serum vitamin D levels and higher probability of meeting recommended levels. Therefore, increasing milk intake may be an effective strategy to improve the vitamin D status of the US population.


Introduction
Vitamin D (calciferol) is a fat-soluble vitamin, photosynthesized in the skin by the action of solar ultraviolet (UV) B radiation. It is naturally found in only a few foods, such as fish-liver oils, fatty fishes, mushrooms, egg yolks, and liver [1]. Vitamin D is known to regulate calcium and phosphorus absorption and, therefore, it has been traditionally associated with skeletal health, and its deficiency increases the risk of rickets in children, and osteoporosis, fractures and falls in adults [1][2][3][4][5]. Emerging evidence suggests that vitamin D deficiency may also be linked to the development of cardiovascular disease, hypertension, diabetes, metabolic syndrome, cancer, multiple sclerosis, rheumatoid arthritis, and sarcopenia [6][7][8][9][10][11][12][13].
The United States (US) Centers for Disease Control and Prevention (CDC) Second National Report on Biochemical Indicators of Diet and Nutrition in the US population reported that, in [2003][2004][2005][2006], approximately 8% of the population aged 1 year and older were at risk for vitamin D deficiency (VDD), which varied by age, gender, or race/ethnicity, and was as high as 31% in non-Hispanic blacks [14]. More recent analysis of NHANES 2011-2014 data, which oversampled Asian, non-Hispanic black,

Database & Subjects
Data from five separate cycles of What We Eat In America (WWEIA), the dietary intake component of the National Health and Nutrition Examination Survey (NHANES), a continuous survey conducted by the National Center for Health Statistics (NCHS), were used (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010). NHANES data are collected using a complex stratified multistage cluster sampling probability design. A detailed description of the subject recruitment, survey design, and data collection procedures are available online [22] and all data obtained for this study are publicly available at: http://www.cdc.gov/nchs/nhanes/. Dietary intake data with reliable 24-h recall dietary interviews (day 1 data only) from 33,672 participants (17,132 male and 16,540 female; 4061 aged 2-8 years, 8700 aged 9-18 years, 17,457 aged 19-70 years and 3454 aged 71-99 years; 7827 Mexican American, 2094 Other Hispanic, 14,525 non-Hispanic White, 7739 non-Hispanic Black, 1487 of other ethnicity) were used after with exclusions for unreliable data (n = 5690), aged < 2 years (n = 3079), pregnant or lactating females (n = 1272), missing serum vitamin D data (n = 5733), missing Poverty Index Ratio (PIR; n = 2808), missing Body Mass Index (BMI; n = 836) or zero calorie intake (n = 2). All participants or proxies (i.e., parents or guardians) provided written informed consent and the Research Ethics Review Board at the NCHS approved the survey protocol. This study was a secondary data analysis which lacked personal identifiers and, therefore, did not require Institutional Review Board review.

Estimation of Dietary Intake
Intake of milk was assessed using the sum of Food Patterns Equivalents Database (FPED) variable "D_Milk" as cup equivalents/day from associated WWEIA categories: All Milk-sum of whole, reduced-fat, low-fat and non-fat milks Non-consumers were defined as subjects not consuming any specific type of milk during the 24-h recall. Consumers of a specific type of milk were defined as subjects consuming that type of milk and no other milk during the 24-h recall. Subjects consuming more than one type of milk during the 24-h recall were designated as "Mixed Milk-Consumers". Consumer intake tertiles were calculated within each age/gender group (Table 1).

Statistics
All analyses were performed using SAS 9.2, 9.4 and SUDAAN 11. Day 1 weights were used all analyses and the data were adjusted for the complex sampling design of NHANES, using appropriate survey weights, strata, and primary sampling units. Separate analyses were conducted for the ages 2-8, [9][10][11][12][13][14][15][16][17][18], and 71+ years. Least Square Means (LSM) were generated from models for each age/gender/ethnic group using linear regression across tertiles of milk intake and different types of milk intake after adjusting for age, gender, ethnicity, poverty income ratio and BMI or BMI Z-score when the population being analyzed was <19 years. Significance was set at p < 0.05. Logistic regression analysis was used to assess odds ratios (OR) and 95% confidence limits (Lower confidence limit (LCL); Upper confidence limit (UCL)) of meeting recommended levels of serum vitamin D (>50 nmol/L) associated with milk intake with non-consumers as reference group after adjusting for age, gender, ethnicity, poverty income ratio and BMI or BMI Z-score when the population being analyzed was <19 years old. Additionally, vitamin D from dietary supplements, seafood, and other non-milk vitamin D sources were also subsequently added to models to assess whether the intake of these variables impacted the association of milk with serum vitamin D (serum 25(OH)D) levels.
Serum vitamin D (serum 25(OH)D) was positively associated with milk intake in children aged 9-18 years and adults aged 19+ years (p linear trend < 0.05). The association was also significant for both males and females of age 9+ years (p linear trend < 0.05), except for 9-18-year-old males and for 71+ yearold females (p linear trend > 0.05) ( Table 4). In children aged 2-8 years, consumers of whole milk, reduced fat milk, and low-fat milk had significantly higher (p < 0.05) serum vitamin D (serum 25(OH)D) levels than non-consumers. Similarly, in children aged 9-18 years and adults aged 19+ years, consumers of reduced-fat milk, low-fat milk, and non-fat milk had higher (p < 0.05) serum vitamin D (serum 25(OH)D) levels than non-consumers of same age group. Children (aged 9-18 years) and adult (aged 19-70 years) consumers of mixed milk also had higher (p < 0.05) serum vitamin D (serum 25(OH)D) than respective non-consumers (Table 5). Adjusting for data in Tables 2 and 3 for dietary supplements, seafood, and other non-milk vitamin D sources did not change these results (Tables 4 and 5).   Mean serum vitamin D concentration ± standard error (SE). Least Square Means (LSM) were modeled for each age/gender group using linear regression across tertiles after adjusting the data for age, gender, ethnicity, poverty income ratio and BMI or BMI Z-score when the population being analyzed was < 19 years. * Significant difference from non-consumer at p < 0.05. # Significant difference from non-consumer at p < 0.05 after additionally adjusting data for vitamin D intake from dietary supplements, sea food, and other non-milk dietary sources. a p linear trend after additionally adjusting data for vitamin D intake from non-milk sources, supplements and sea food. N = 4061 aged 2-8 years; 8700 aged 9-18 years; 17,457 aged 19-70 years; 3454 aged 71-99 years. Consumption of all milk (for all ages-i.e., 2+ years), whole milk (for ages 19-70 years), reduced-fat milk (for ages 9-70 years), low-fat milk (for those aged 19+ years), and non-fat milk (for those aged 9+ years) was associated with significantly higher probability of meeting serum vitamin D recommendations when the analysis was conducted by milk amount ( Table 6). Consumers of whole milk (of aged 2-8 years), reduced-fat milk (of all ages), low-fat milk (of all ages), non-fat milk (of aged 19+ years) and mixed milk (of aged 2-70 years) had a higher probability of meeting serum vitamin D recommendations in the analysis by milk type (Table 6). Table 7 shows the serum vitamin D (serum 25(OH)D) status by tertiles of all milk intake among different ethnic populations. Serum vitamin D (serum 25(OH)D) was positively associated with milk intake among children aged 9-18 years and adults aged 19+ years of Mexican American and of "other" ethnicity (p linear trend < 0.05). The increase in serum vitamin D (serum 25(OH)D) status with increasing milk intake was also significant (p linear trend < 0.05) among non-Hispanic Whites aged 19+ years, and non-Hispanic Blacks aged 2-18 years. Additional adjustment for dietary supplements, seafood, and other non-milk vitamin D sources did not change the results (Table 7). Table 8 presents the data on vitamin D status by milk type across different ethnic groups. Significantly (p < 0.05) higher levels of serum vitamin D (serum 25(OH)D) levels were observed for consumers of whole milk, reduced fat milk, low-fat milk, non-fat milk and mixed milk among the different age and ethnic groups examined (Table 6). Adjusting the data for dietary supplements, seafood, and other non-milk vitamin D sources did not modify the results (Table 8). Table 6. Odds ratios (OR), 95% confidence limits (Lower confidence limit (LCL)/Upper confidence limit (UCL)) for meeting the recommended serum vitamin D (>50 nmol/L), NHANES 2001-2010.

Discussion
The current cross-sectional analysis of data from the NHANES 2001-2010 demonstrated a significant association between milk consumption and serum vitamin D (serum 25(OH)D) status. Additionally, the results showed that the probability of meeting the vitamin D recommendations was greater in milk consumers vs. non-consumers. To the best of our knowledge, this is first analysis of nationally representative, non-institutionalized population of US children and adults examining the association of milk intake with vitamin D levels.
Poor vitamin D status (low serum 25(OH)D levels) is a global public health concern as over 50% of population has less than adequate serum vitamin D status [24]. In the US, about 18% of the US population aged 1+ years had insufficient serum vitamin D levels and were at risk of inadequacy (less than 50 nmol/L) according a recent analysis of NHANES 2011-2014 [15]. Liu et al. [25] estimated the prevalence of inadequate serum vitamin D levels in US adults to be 28.9% for VDD and 41.4% for VDI, from analysis of NHANES 2001-2010 and using the criteria recommended by the Endocrinology Society to define VDD as 25(OH)D <50 nmol/L and VDI as 25(OH)D <75 nmol/L [26]. The present analysis showed that the average serum vitamin D levels ranged from 64 to 75 nmol/L depending on age and gender in representative population of US children and adults aged 2+ years. These average serum vitamin D levels are well with in the 50-75 nmol/L range and are considered sufficient by IOM definition [1]. Vitamin D is a "nutrient of public health concern" as it is under-consumed to an extent that may lead to adverse health outcomes and higher intake of food sources are encouraged by DGA [17].
In the present analysis, serum Vitamin D levels were significantly associated with the intake of milk, depending on the type. Milk consumers, especially those of low fat and reduced fat milk, had higher probability of meeting >50 nmol/L serum vitamin D level benchmark set by IOM [1] than non-consumers; however, the mean serum vitamin D levels were always higher than 50 nmol/L. Although milk contains a low amount of naturally occurring vitamin D, almost all milk in the US is fortified with 100 IU/cup vitamin D irrespective of the type of milk [1, 19,21]. Effectiveness of milk and other fortified foods in improving serum vitamin D status has been demonstrated in both clinical and observational studies (see [27,28] for reviews). A recent systematic review and meta-analysis of randomized controlled trials showed that vitamin D fortified foods (mostly milk and dairy products) increased serum vitamin D levels by 1.2 nmol/L for each 1 µg/d increased intake of vitamin D [27]. A cup of milk/d provides ∼2.5 µg of vitamin D. A review of observational studies also concluded that the intake of vitamin D fortified milk products was positively associated with vitamin D intake and serum vitamin D status and the association was stronger in countries with a national vitamin D fortification policy [28]. However, this review included only five studies from US which had small sample sizes and included only certain population groups. In our present analysis, the intake of certain milk types (especially of whole milk) was not associated with an increase in serum vitamin D levels in all population sub-groups, which is not immediately understood.
Age and ethnicity have been shown to affect serum vitamin D status [14,15,29,30]. Vitamin D serum levels generally decrease with age, and non-Hispanic Blacks have the lowest vitamin D levels or highest prevalence of VDD, followed by Hispanics and Asians [14,15,29,30]. In contrast, non-Hispanic Whites have the highest vitamin D levels or lowest prevalence of VDD [14,15,29,30]. Lower intake of milk with age, which was also observed in our study, could potentially explain the inverse association of age with vitamin D. In regard to ethnicity and vitamin D status, differences in milk and overall vitamin D intake as well as skin pigmentation and other factors are potentially responsible for the ethnic differences in vitamin D status [31]. For instance, studies have shown lower milk intake among non-Hispanic Blacks compared to non-Hispanic Whites [32]. In the present analysis, serum vitamin D status was associated with both type and amount of milk intake depending upon the age and ethnicity. However, the effect of gender on serum vitamin D status has been reported to be insignificant or inconsistent [14,15,30], but the associations between amount of milk intake and serum vitamin D status were mostly significant for both males and females in the present analysis. Therefore, continuing to encourage an increase in milk intake, especially among populations with VDD or VDI, could be an effective strategy to improve vitamin D status.
In addition to providing vitamin D, milk and dairy products, make significant nutrient contributions including nutrients under-consumed by most Americans-calcium and potassium-as well as magnesium, phosphorus, zinc, vitamin A, vitamin B12, riboflavin (B2), choline, high-quality protein and saturated fat; as such, the inclusion of dairy foods into healthy dietary patterns is associated with improving diet quality and reducing risk of obesity and chronic diseases [17,18,[33][34][35][36].
A major limitation of our study is the inability to determine a cause-effect relationship due to the cross-sectional design of NHANES. Additionally, as with any study based on self-reported data, underor over-reporting cannot be ruled out. Additionally, the results from this study may not specifically reflect the effect of milk consumption on vitamin D status, although we used vitamin D from dietary supplements, seafood, and other non-milk sources as covariates to adjust some of our results. Strengths of this study included the use of a large nationally representative sample achieved through combining several sets of NHANES data releases and adjusting for numerous covariates, but even with these covariates, some residual confounding may still exist.

Conclusions
In conclusion, the results of this study indicate that milk consumers consistently have higher serum vitamin D levels and higher probability of meeting the recommended levels. Vitamin D has been identified as a "nutrient of public health concern in the US" and, therefore, increasing the intake of milk (especially low-fat and reduced-fat) should be encouraged. Other sources of vitamin D may also help in improving vitamin D status.