Is the Level of Vitamin D in Children Correlated to Their Frequency of Exercise? 

Background

For children, vitamin D insufficiency is defined as having a level of serum 25(OH)D less than or equal to 50.0 nmol/L (ng/mL) (Misra et al., 2008). In recent years, conditions linked to vitamin D deficiency have been on the rise in children, like rickets, in which bones fail to mineralize fully. It is estimated that 15.7% of children are vitamin D-deficient globally, and 47.3% have insufficient vitamin D levels (Cui et al., 2023). Research has found that vitamin D deficiency can harm children’s health-related quality of life (HRQoL) by creating health problems such as rickets, hypocalcemic cardiomyopathy, hypocalcemic seizures, and an overall increase in feelings of tiredness (Aguiar et al., 2019). The low levels can lead to further negative impacts than just health issues, but can affect school attainment and the general well-being of the children (Aguiar et al., 2019). It has been shown that vitamin D plays a role in metabolizing adipose tissue, which is connective tissue that stores fat cells (Cheng, 2018). Similarly, vitamin D levels are positively correlated with healthier lipid profiles in children, indicating vitamin D’s positive effect on metabolism and overall health (Cheng, 2018). Thus, vitamin D is important in regulating metabolism by promoting calcium intake, promoting bone development and strength, improving cardiovascular health, and reducing the risk of autoimmune disorders and cancer (Cheng, 2018; Xu et al., 2022). Vitamin D sufficiency and healthy lifestyle choices, like regular physical exercise, help prevent the risk of childhood diabetes and other metabolic disorders (Xu et al., 2022). 

It has been found that it is important for children’s quality of life to have proper amounts of physical activity (Wu et al., 2017). Whereas, sedentary behaviors were linked to lower quality of life in children and adolescents (Wu et al., 2017). In addition to overall quality of life, children with higher levels of physical activity have been shown to have better physical health, mental well-being, and psychosocial health compared to children with lower levels of physical activity (Wu et al., 2017). Physical activity is negatively correlated with an array of diseases, including obesity, heart disease, and cognitive impairment (Wu et al., 2017). Although chronic diseases associated with physical inactivity do not manifest until adolescence, proper physical activity habits in children are important to lower the risk of developing these chronic conditions later in life (Pate, 2002). The median physical activity level was higher for boys than girls, with 90% of children meeting physical activity Objective 22.6, set by Healthy People 2010, defined as more than 30 minutes of total physical activity at least 5 days a week (Pate, 2002). Children aged 10-12 were less likely to meet the physical activity Objective 22.6, compared to other age groups (Pate, 2002). However, very few students met the requirement of Objective 22.7, which is defined as more than 20 continuous minutes of physical activity for at least 3 days a week (Pate, 2002). Overall, children are more likely to meet the standards for moderate physical activity compared to vigorous physical activity (Pate, 2002).

Literature Review 

Vitamin D levels are strongly correlated with obesity rates in children (Cheng, 2018; Turer et al., 2013). Obesity can be prevented through a change in lifestyle habits, including increasing physical activity and consistent healthy eating (Cheng, 2018). Since there is a strong negative relationship between physical activity levels and obesity, the impact of obesity on vitamin D3 levels is important to explore (Cheng, 2018). Obese children have been found to have higher rates of vitamin D insufficiency and lower levels of serum 25(OH)D, the most abundant metabolite of vitamin D, than non-obese children (Cheng, 2018; Xu et al., 2022). Specifically, in 2013, the prevalence of vitamin D deficiency in healthy-weight children, obese, and severely obese children between the ages of 6 to 18 was 21%, 34%, and 49%, respectively (Turer et al., 2013). Obese children were twice as likely to be vitamin D deficient compared to healthy-weight children (Turer et al., 2013). This same study also found a higher prevalence of vitamin D deficiency among minority groups and female children (Turer et al., 2013). Recent literature has shown that there is a negative correlation between vitamin D levels and obesity, which can be attributed to the fact that vitamin D is necessary to make leptin, which is a hormone that regulates food intake (Xu et al., 2022). Vitamin D levels have not been clearly defined in the literature based on socio-economic background for adolescents there have been studies elsewhere that reported in Saudi Arabia, children of a lower socio-economic background had higher levels of vitamin D than their counterparts, but in the UK it was found that adults of lower socio-economic had lower levels of serum vitamin D compared to their counterparts (Al-Agda et al., 2016) (Grimes, 2011). There has not been found to be a significant difference in vitamin D levels when comparing genders, and when comparing four age cohorts <21, 21-40, 41-60, and >60 (Lippi et al., 2011). Although the negative relationship between vitamin D levels and rates of obesity is evident, further studies are needed to analyze the interconnectedness between physical activity levels, BMI, and vitamin D levels. Another significant gap in the literature is a lack of recent data on the prevalence of vitamin D deficiency amongst children aged 6 to 17 in the United States. 

Vitamin D deficiency has been linked to malnutrition and low levels of physical activity (Allam et al., 2021). High-intensity exercise was shown to significantly increase vitamin D levels in children aged 10 to 12 (Allam et al., 2021). This study found that aerobic exercises stimulated bone formation, increasing the demand for minerals, like serum 1,25-dihydroxyvitamin D3 (Allam et al., 2021). High-intensity exercise raises the heart rate and activates vitamin D metabolites, which then boost metabolism in muscles (Allam et al., 2021). Many recent studies have shown that physical activity increases vitamin D levels in children, adolescents, and adults, indicating the importance of routine physical exercise throughout one’s lifetime (Ouyang et al., 2024). Indoor physical activity has been shown to increase levels of vitamin D, indicating that vitamin D concentration is attributed to factors other than sun exposure, like physical activity (Fernandes & Barreto Junior, 2017). A research study analyzing the differences in the relationship between physical activity and vitamin D levels among different age groups and sexes found a positive relationship between physical activity levels and serum vitamin D levels (Ouyang et al., 2024). Specifically, for both males and females aged 6-11 years old, there was an increase in serum vitamin D levels associated with a higher amount and intensity of physical activity with outdoor time as a mediator, from a 28% to a 46% increase in vitamin D (Ouyang et al., 2024). However, there was no association between physical activity volume and serum vitamin D levels in children aged 3 to 5 years old, which could be due to the removal of external factors from the study analysis (Ouyang et al., 2024). These included race, ethnicity, BMI, total vitamin D intake, and PIR (Ouyang et al., 2024). Current literature on the relationship between vitamin D and physical activity is inconclusive, and thus, more research should be conducted to further investigate this relationship for different ages and sexes, with demographic factors in mind.

Objectives

Although much of the current literature has provided many links to vitamin D deficiency in children, like having a dark complexion, living at a higher latitude, and being lactose intolerant, it has not been well established if there is a significant link between low levels of exercise, indoor or outdoor, and low levels of vitamins D (Misra et al., 2008). Using NHANES, which pulls information from across the nation, a comparison can be made between the average amount of exercise per week that children ages 6-17 complete and their levels of Vitamin D to better understand the relationship between the two variables on a large scale. Having proper vitamin D levels is critical for a child's health as they grow, and nutritional deficiencies can lead to serious health problems. Thus, researching more links to the characteristics and habits of children with proper vitamin D levels is critical. The objective of this study is to understand how physical activity/exercise is associated with the level of vitamin D in children aged 6-17.

Methods

Study design and setting

For this research study, data were obtained from the National Health and Nutrition Examination Survey (NHANES). NHANES interviews approximately 5,000 individuals of all ages in their homes annually to assess factors of health and nutrition in the United States (CDC, n.d.-b). The health examination component of the survey was completed in mobile examination centers, abbreviated MECs (CDC, n.d.-b). The NHANES design has repeated cross-sectional studies since 1999, but our study is only analyzing data collected during the 2017-2018 study period. NHANES collects data from civilian, noninstitutionalized individuals residing in all 50 states and D.C. (CDC, n.d.-b). For this research study, further inclusion and exclusion criteria were applied, which will be described below. 

Sample size and sampling

NHANES collected data from 16,211 individuals from 30 different locations in the United States during the 2017-2018 survey (CDC, n.d.-b). Of these 16,211 selected individuals, 9,254 were interviewed at home and 8,704 completed the health examination in the mobile examination centers (CDC, n.d.-b). NHANES completes a multistage probability sampling design, which requires the following 4 steps. First, counties or small groups of contiguous counties are selected as primary sampling units, or PSUs. Then the PSUs were broken down into segments that constitute a block or group of blocks containing a cluster of households. Third, specific households from the segments were selected. Finally, individuals within a household were selected randomly to eliminate bias within the study design. 

During the 2017-2018 sampling, certain demographics were oversampled. This includes Hispanic persons, non-Hispanic black persons, non-Hispanic Asian persons, non-Hispanic white persons at or below 185% of the Department of Health and Human Services poverty line, and non-Hispanic white and other persons over the age of 80 years (CDC, n.d.-b). Data from the home interviews and physical health examinations completed in the MECs were used for this study. This particular research study excluded individuals below the age of 6 and over 17. Any missing, unknown, or refused to answer participants’ data were also excluded from the study. After these exclusion criteria, the number of participants in our study is 1397, which is lower than the NHANES entire sample population from 2017-2018, which was 16,211. 

Data collection methods

NHANES collected data on the health and nutrition status of civilians in the United States through at-home interviews and physical health examinations inside the MECs (CDC, n.d.-b). Demographic information, including relationship status, family-level information, and sample person data, was collected during the individual household interviews (CDC, n.d.-c). To gather laboratory data, health examinations were completed in mobile health examination centers, which serve as an ideal standardized environment for collecting high-quality data (CDC, n.d.-b). Data was collected from all 50 states and D.C., so all NHANES staff participated in cultural competency training to understand and respect cultural differences (CDC, n.d.-b). Additionally, local and medical professional interpreters were available during the data collection (CDC, n.d.-b). NHANES provides extensive data, but this study will focus on personal survey information on physical activity and laboratory data on vitamin D levels (CDC, n.d.-b). 

Variables

Dependent variable: The dependent variable for this research study is the level of vitamin D3, specifically 25-hydroxyvitamin D3 (25OHD3) measured in nmol/L. NHANES describes vitamin D levels as laboratory data that is collected, processed, and stored in MECs(CDC, n.d.-a).  MECs are a controlled laboratory environment with standardized conditions that allow for sterile and precise collection of biospecimens (CDC, n.d.-a). Participants 12 years of age or older completing a physical health examination appointment in the morning session were instructed to fast for 9 hours before their appointment (CDC, n.d.-a). During the data collection for the 2017-2018 study, 80% of children, aged 1-17 years old, provided blood specimens through phlebotomy in the MECs (CDC, n.d.-a). These blood specimens then underwent high-performance liquid chromatography-tandem mass spectrometry for the quantitative detection of 25OHD3 (Pirkle, 2020). All serum specimens were treated as potentially infectious agents, and proper laboratory precautions were followed for the protection of the laboratory team (Pirkle, 2020). Our study has divided the measurements of Vitamin D into two categories: insufficient >50.0 nmol/L, and sufficient as any value of 25OHD3  above 50.0 nmol/L (Misra et al., 2008). 

Independent variable: The independent variable for this research study was the number of days spent physically exercising for 60 minutes or more. NHANES measured physical activity in youth from the “Sample Person Questionnaire”, which was gathered during participant interviews conducted at individual households (CDC, n.d.-c). If individuals under the age of 16 were unable to answer the sample person survey questions, an adult proxy provided the information (CDC, n.d.-c). All household interview data were electronically recorded using computerized questionnaire forms and uploaded to a database system (CDC, n.d.-c). If unusual, inconsistent, or unrealistic responses were provided by the participants, the interviewer was alerted to verify or edit the response to provide the most accurate and useful questionnaire data (CDC, n.d.-c). 

Cofounders: In this study, seven confounding variables were identified: age, sex, race/ethnicity, education level, body mass index (BMI), household income ratio to poverty, and season when the NHANES data was collected. Our age confounder, measured in the number of years old, was not coded into new categories in SAS because we are limiting our study to ages 6-17 (DHHS, 2020). Coding age into smaller categories would not be helpful for our study because there was already a small range of ages, so we would better see differences between each age group. In our study, age may not play a significant role as a confounding variable, but we felt it was requisite to include it because of age’s long-documented effects on various health indicators (Giefman, 2013). Our next confounding variable is the sex of the respondent, which is divided into two categories: female and male. Sex and gender have been thoroughly documented as having an effect on health outcomes for a population, so accounting for sex as a confounder for our study is necessary (Wunsch, 2007). Race and ethnicity as a confounder were divided into 6 categories by NHANES: Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian, and others (including multiracial) (DHHS, 2020). This information was obtained via a question on the demographic survey, and the categories align with other studies' categorizations of race (Ver Stralen et al., 2010). Race has been found to affect the way the body absorbs Vitamin D, so it is important to include it here as a confounder (Freedman and Register, 2012). The confounder, education level, is divided into 14 categories by NHANES: never attended school/kindergarten,1st, 2nd, 3rd, 4th, 5th, 6th, 7th, 8th, 9th, 10th, 11th, 12th, and high school graduate (CDC, n.d.-c). The education level has been coded into three categories: elementary level (grades kindergarten-4th), middle school (5th-8th), and high school (9th-high school graduate). We believe education level, like age, for people below the age of 18 will not have a large effect on the results as a confounder but felt the need to include education level due to the same reasons as age (AlQuaiz et al., 2018). The next confounder, BMI, has been kept in its numerical form to reflect the common use of BMI in relation to children, where they are sorted into percentiles and not groups. Having a high BMI has been known to affect the body’s levels of Vitamin D, though it is unknown whether low Vitamin D contributes to obesity or if the inverse is true, so it is important to account for it in this study (Vranić et al., 2018). Our next confounder, the ratio of family income to poverty, is a binary variable divided between families with a ratio below 1.00 and families with a ratio to poverty above 1.00. Families under the poverty line, or below a value of 1.00, are more likely to experience food insecurity, which can lead to lower levels of Vitamin D (Grimes, 2011). Our final confounder is the season in which the data was collected for NHANES, from November 1 through April 30 and May 1 through October 31 (DHHS, 2020). It has been found in studies that Vitamin D levels can drop in the winter months and rise in the summer, so it is important to account for when the data was collected for NHANES, making this confounder important to the interpretation of our data (Thiedan et al., 2009). For each confounding variable, responses recorded as missing, refused, or unknown were deleted from the final dataset (DHHS, 2020). 

Statistical analyses

Accounting for each type of variable in this study, one independent, one dependent, and seven confounders, there are a total of nine variables in this study. There are six categorical variables, among them four are binary: vitamin D3 level, sex, family income ratio to poverty, and season of data collection. Race/ethnicity is the only nominal categorical variable. The only ordinal categorical variable is education level. These categories were calculated as summary statistics with frequencies and proportions to show the portion of each variable. For our three numerical variables, the number of days per week children exercise for 60 minutes, age, and BMI, univariate summary statistics were acquired by generating histograms and Q-Q plots so the normality of the distribution could be ascertained. Since all three numerical variables were not normally distributed, we calculated the median and IQR. If they had been normally distributed, we would have calculated the mean and standard deviation. 

When comparing the dependent variable to each confounder and independent variable, a t-test was used to create the bivariate test statistics. The t-test was chosen because the bivariate statistics were comparing a categorical variable and a numerical one. Additionally, t-tests were used to discern whether or not the number of days physically active, age, and BMI affected Vitamin D levels. A chi-squared test was employed to determine whether or not sex, race/ethnicity, education level, family income ratio to poverty, and the season of collection affected Vitamin D levels. Due to the scope of this study, we only used tests that assume a normal distribution. 

Finally, since our dependent variable was binary, we performed a multinomial logistic regression to assess the association of Vitamin D levels compared to the number of days with 60 minutes of exercise or more. In the regression, we accounted for each confounding variable: age, sex, race/ethnicity, education level, BMI, household income ratio to poverty, and season of NHANES data collection. To interpret the results, the odds ratio and confidence intervals were calculated from the regression, which shows the strength of association between the two variables. 

Results

Final sample size:

NHANES originally surveyed 16,211 individuals. After excluding individuals who did not participate in the at-home interviews or data collection in the mobile examination centers, those with missing data for our variables included in the study, and participants who did not meet 

our inclusion criteria, our final study population was 1397.

General characteristics of the study population

Table 1 displays the socio-demographic and health characteristics of the study population. There was an equal number of female to male participants (50.0%), with a median age of 11 out of the age range of 6 to 17 for participants (Table 1). The majority of participants were of non-Hispanic white origin (30.9%), an elementary education status (44.6%), had a family income above the poverty line (72.4%), provided data to NHANES during the sunnier season (54.3%), and had a median BMI of 20.20 (Table 1). Additionally, the majority of the study participants had sufficient levels of vitamin D3 (67.5%) and had a median number of 5 days of physical activity exceeding 60 minutes within a week (Table 1). 

Bivariate analysis between vitamin D3 levels and independent variables

Table 2 compares the characteristics of the study participants with the percentage of insufficiency and sufficiency for vitamin D3 levels. Of the study participants, 32.50% are insufficient in vitamin D3 compared to 67.5% of participants who had sufficient levels (Table 2). Through bivariate analysis, higher frequencies of sufficient levels of vitamin D3 were found in participants that were male (54.0%), of non-hispanic white origin (40.0%), had elementary level education (52.8%), had a family income above the poverty line (75.1%), and provided data to NHANES during the sunnier season (58.5%) (Table 2). Additionally, participants with sufficient vitamin D3 levels were physically active for at least 60 minutes at an average of 4.8 days out of the week, had a mean BMI of 20.4, and the average age of 10.9 (Table 2). Whereas for participants with insufficient levels, the average days of physical activity lasting longer than 60 minutes was 3.9 per week, the mean BMI was 23.6, and the average age was 12.6 (Table 2). Bivariate analysis also showed higher frequencies of insufficient levels of vitamin D3 in female participants (58.2%), individuals of non-Hispanic black origin (42.7%), with a middle-school level education (40.8%), with a family income above the poverty level (66.7%), and provided data to NHANES during the darker seasons (54.6%) (Table 2). Chi-square tests revealed that all socio-demographic characteristics of the study participants in relation to their level of vitamin D3 were statistically significant, with p-values less than 0.001 (Table 2). 

Binary logistic regression analysis

Table 3 displays the unadjusted and adjusted odds ratios and the corresponding 95% confidence intervals, demonstrating the relationship between vitamin D3 levels, possible cofounders, and number of days within a week of physical activity exceeding 60 minutes as determined by binary logistic regression analysis. Within the sample population, compared to males, females (OR: 1.65, 95% CI: 1.26-2.16, p< 0.001) had a 65% increase in the odds of having insufficient levels of vitamin D3 (Table 3). Compared to those that identified as non-Hispanic white, Mexican American (OR: 2.72, 95% CI: 1.79-4.12, p< 0.001), other Hispanic (OR: 2.58, 95% CI: 1.43-4.65, p< 0.001), non-Hispanic Asian (OR: 6.80, 95% CI: 4.17-11.09, p< 0.001), and those of another race (OR: 2.21, 95% CI: 1.33-3.67, p< 0.001) were 2.72, 2.58, 6.80, and 2.21 times more likely to have insufficient levels of vitamin D3 than non-hispanic whites (Table 3). In the study population, middle school students (OR: 2.68, 95% CI: 0.99-2.88, p< 0.001) and high school students (OR: 1.23, 95% CI: 0.52-2.88, p< 0.001) were 2.68 and 1.23 times more likely to have insufficient levels of vitamin D3 compared to elementary school student (Table 3). However, based on the 95% confidence intervals, the odds ratio for level of education is not statistically significant, which suggests that the observed increase in odds may be due to random chance. Individuals with a family income above the poverty line (OR: 0.67, 95% CI: 0.50-0.90, p<0.001) had a 33% decrease in the odds of having insufficient levels of vitamin D compared to those with a family income that is below the poverty line (Table 3). For participants whose data was collected during the sunnier months of May to October (OR: 0.55, 95% CI: 0.42-0.71, p<0.001), their vitamin D3 levels had a 45% decrease in the odds of being insufficient compared to those individuals whose data was collected during the darker months (November-April) (Table 3). For a unit increase in the number of days of physical activity lasting longer than 60 minutes (OR: 0.90, 95% CI: 0.85-0.95, p<0.001), the odds of having insufficient levels of vitamin D3 decrease by 10%, which is statistically significant (Table 3). Whereas for a unit increase in BMI (OR: 1.07, 95% CI: 1.05-1.10, p<0.001), the odds of having insufficient vitamin D3 levels increase by 7%, which is statistically significant (Table 3). Finally, for a unit increase in age (OR: 1.08, 95% CI: 0.97-1.20, p<0.001), the odds of having insufficient vitamin D3 levels increase by 8%, yet this is not statistically significant (Table 3). Therefore, the observed increase in odds associated with age may be due to random chance. 

Tables

Discussion

Synopsis of key findings compared to previous studies

The interpretation of these results shows that there is a statistically significant relationship for children ages 6-17 between the number of days they are physically active over 60 minutes per week and whether or not they have sufficient levels of Vitamin D. Notably, those who had insufficient levels of Vitamin D had on average 0.9 days less per week of one hour of physical activity. 

These findings are consistent with current research in the subject that suggests that physical activity aids in increasing serum levels of Vitamin D in children (Kim et al., 2022). This study was conducted over a period of six years and analyzed the serum vitamin D levels of 2,811 adolescents. This study identified a correlation between the amount of physical activity an adolescent does and their level of vitamin D (Kim et al., 2022).

Possible mechanisms or explanations for observed results

Indoor and outdoor physical activities have been known to increase vitamin D serum levels since exercise has been shown to increase vitamin D levels and bone mass (Kim et al., 2022). A possible explanation for this is that vitamin D is stored in fat and muscle cells and is released during the process of exercise (Kim et al., 2022). The studies in which this was studied were controlled for sunlight exposure in children to obtain these results, however, other studies saw a negative correlation between vitamin D levels and the amount of physical activity (Kim et al., 2022). More studies on the relationship between vitamin D and physical activity need to be conducted to understand this relationship. 

Strengths and limitations of the study

Assessing the strengths and weaknesses of the study, one significant strength of this cross-sectional study was the large sample size provided by the National Health and Nutrition Examination Survey (NHANES). This research analyzed the relationship between vitamin D and physical activity for 1397 children between 6 and 17 years old from across the nation, providing a large and representative sample of the United States population. NHANES, where the data was obtained from, has consistent standards for data collection, ensuring the data obtained is reliable. For the primary variable of vitamin D level, we anticipate little measurement bias; however, for the other primary variable measuring the number of days per week a child was physically active for 60 or more minutes, there is the potential for some error due to the retrospective nature of the question. Furthermore, the exclusion criteria were minimal, only looking to exclude respondents outside of the target population of children ages 6-17. This ensures the findings are generalizable to the greater population. 

A limitation of the study is that certain populations were oversampled by NHANES, including black people and people of Mexican origin (CDC, n.d.). This introduces selection bias and may skew the final results, since both black people and people of Mexican origin were more likely to have insufficient levels of vitamin D based on the results. This issue could have been avoided if the data had been adjusted before analysis, however, this was beyond the scope of this course. Additionally, since this is a cross-sectional study, it is not possible to establish a causal relationship between vitamin D and physical activity. Finally, confounding factors such as taking a vitamin D supplement, longitude of participants, and diet were not accounted for in this study, despite this information being available on NHANES. Not controlling for these confounding factors could have led to skewed final results. 

Implications for future research

Given that the data is from before the COVID-19 pandemic, it would be important to conduct this study using post-COVID-19 data to see if trends have changed due to the effects of the pandemic. Future research conducted longitudinally would also be a meaningful contribution to gaining a deeper understanding of these trends in populations as children grow. Additionally, other studies could introduce other variables more prominently, like measuring vitamin D levels across races, longitude, and season, when compared to the amount of physical activity. Based on the results of this study, children need to prioritize an hour of physical activity per day to increase their vitamin D levels if they are insufficient, in addition to taking supplements and improving their diet.