The Prevalence and Correlates of Vitamin D Deficiency and Overweight/Obesity of School-Age Children in Colombia–Findings on the Double Burden of Malnutrition from Nationally-Representative Data

Abstract

The double burden of malnutrition (DBM)—the coexistence of overweight/obesity and micronutrient deficiency—is an emerging public health concern among school-aged children. Using data from 6063 participants in Colombia’s 2015 National Survey of Nutritional Status (ENSIN), this study estimated DBM prevalence and identified factors associated with its occurrence among children aged 5–12 years. DBM was defined as concurrent overweight/obesity (BMI-for-age z-score > 1) and vitamin D deficiency, applying thresholds of <30, <37.5, and <50 nmol/L. The prevalence of DBM ranged from 0.7% to 6.9%. Firth’s penalized logistic regression models were conducted separately for (1) overweight/obese combined, (2) overweight-only, and (3) obesity-only groups. For DBM1, insufficient physical activity was linked to higher odds across all three models. For DBM2, smaller household size and higher maternal education were associated with greater odds in the combined model. Living in large urban areas was related to lower odds compared with major metropolitan areas, a pattern also observed in the overweight-only model. For DBM3, children from the second wealth quartile (Q2) showed higher odds than those from the poorest (Q1), with a similar pattern in the overweight-only analysis. Stricter DBM definitions tended to capture behavioral and household characteristics, whereas broader thresholds reflected structural and contextual conditions. Despite its relatively low prevalence, DBM remains a relevant public health issue among Colombian schoolchildren.

1. Introduction

The World Health Organization (WHO) [1] defines the Double Burden of Malnutrition (DBM) as the coexistence of deficiency of one or more nutrients with overweight and obesity or diet-related non-communicable diseases within individuals, households, and populations and across the life course. DBM can manifest in various forms depending on its location and context and may occur at the national, community, household, and individual levels [2,3].

Micronutrient deficiencies such as those of vitamin D, zinc, iron, and vitamin B6 have also been reported among overweight and obese children and adolescents [4]. Recognizing the coexistence of multiple nutrient deficiencies alongside excess adiposity is critical to understanding the full scope of DBM and for designing comprehensive health promotion and prevention strategies. A specific example of DBM is the coexistence of overweight or obesity with low levels of vitamin D. This lower vitamin D level is highly prevalent in overweight and obese children and adolescents [5]. A meta-analysis [6] found that a significant number of obese children experienced vitamin D deficiency, indicating the presence of DBM at the individual level.

Vitamin D plays a crucial role in biological functions such as calcium homeostasis and bone mineralization [7]. Consequently, low vitamin D concentration is associated with numerous adverse outcomes, including low bone mineral density [8], poor skeletal health [9], functional disabilities [10], multiple sclerosis [11,12], poor immune system function, and increased mortality [13]. Furthermore, vitamin D is essential in regulating glucose homeostasis, insulin secretion, and inflammation associated with obesity [14,15,16]. Lower vitamin D level has been linked to metabolic disorders such as obesity, type 2 diabetes mellitus (T2DM), insulin resistance, and cardiovascular diseases [17,18,19]. Recent studies highlight the clinical importance of monitoring 25-(OH)D3 levels in school-aged children due to its impact on bone growth and metabolic health [20].

Populations with limited sun exposure are at increased risk for low vitamin D levels [21,22]. Several socioecological factors operating at multiple levels may elucidate the mechanisms behind this phenomenon. For example, at the individual level, children who are less physically active or more sedentary often experience insufficient exposure to sunlight due to prolonged use of electronic devices such as phones, televisions, or computers [23]. Environmental and social factors related to housing, neighborhood characteristics, and broader structural inequities, including socioeconomic status and urbanization, also significantly contribute to reduced sunlight exposure and obesity risk [24,25].

Increasing evidence highlights the association between low vitamin D levels and metabolic disorders among obese children [6,26,27,28,29,30,31,32,33]. A 2015 meta-analysis [31] revealed a substantial connection between vitamin D deficiency and obesity, with an odds ratio (OR) of 3.43 (95% CI: 2.33–5.06). Similarly, another meta-analysis by Fiamenghi and Mello [6] reported a prevalence ratio (PR) of 1.41 (95% CI: 1.26–1.59) within the pediatric population.

The interaction between vitamin D deficiency and excess adiposity induces adverse effects, influencing metabolic pathways by promoting the accumulation of inactive forms and reducing the availability of vitamin D [34], in addition to decreasing tissue secretion and insulin sensitivity [35]. Despite these findings, consensus on the underlying cause of decreased vitamin D levels in obese children remains elusive. The main hypothesis postulates that the absorption of vitamin D by adipose tissue, given its fat-soluble nature, may play a contributing role [35].

The global definition of Vitamin D status is complicated by persistent disagreement over clinical thresholds and variability in 25-hydroxyvitamin D [25(OH)D] assays [36]. Although international initiatives such as the Vitamin D Standardization Program (VDSP) have substantially improved the analytical accuracy of automated immunoassays—including platforms such as the Siemens ADVIA [37]—significant clinical debate persists over the appropriate cut-off points [36,38]. Generally, the scientific community recognizes two categories: deficiency and insufficiency. The most conservative threshold, used by the Institute of Medicine (IOM), defines deficiency as serum 25(OH)D ≤ 30 nmol/L (≤12 ng/mL), sufficient only to prevent skeletal diseases [39]. In contrast, the Endocrine Society defines deficiency more stringently as <50 nmol/L (<20 ng/mL) and insufficiency as 50–75 nmol/L (20–30 ng/mL), aiming for broader endocrine benefits [38].

In Colombia, some studies have examined the relationship between vitamin D deficiency and overweight/obesity [30,40]. According to Beer et al. [40], there is a significant association between vitamin D levels and the prevalence of overweight among Colombian children aged 1 to 18 years with a BMI-for-age Z score greater than 1. Specifically, they found that the prevalence is 3.3% when vitamin D levels are below 30 nmol/L, escalating to 30.1% when vitamin D levels fall below 50 nmol/L. However, these studies did not investigate broader socioecological factors influencing DBM development in Colombian schoolchildren.

To address these gaps and guided by the social-ecological model, this study aimed to assess the prevalence and factors associated with DBM, defined as the co-occurrence of vitamin D deficiency and overweight/obesity, in Colombian schoolchildren (aged 5–12 years) using nationally representative data from the Colombian National Nutrition Survey (ENSIN 2015) [41].

2. Materials and Methods

2.1. Study Population and Data

We used data from the Colombian National Nutrition Survey 2015–2016 [Encuesta Nacional de Situación Nutricional (ENSIN)] [41], a nationally representative survey designed to cover 99% of the Colombian population through a multistage, stratified sampling design. The survey targeted non-institutionalized civilian residents and included 44,202 households organized into 177 strata. Sampling followed the official ENSIN hierarchical design, comprising 238 primary sampling units (PSUs), which grouped municipalities across all 32 departments and Bogota. Within these, secondary sampling units (SSUs) consisted of clusters of contiguous city blocks within the same sector and census section, each including at least 96 households. Finally, tertiary sampling units (TSUs) included 5000 segments averaging 12 contiguous households (range: 6–17), of which 4962 were selected and 4813 contained at least one completed household interview.

The ENSIN-2015 data are anonymized, publicly available, and can be obtained upon reasonable request from the Colombian Ministry of Health. Survey participants completed detailed questionnaires on demographic information, wealth indicators, and household characteristics.

Anthropometric measurements were conducted by trained personnel following standardized procedures. Weight was measured to the nearest 100 g using a Seca 874 balance (Seca Medical Measurement Systems, Hamburg, Germany), and height was measured using a stadiometer. BMI was calculated as weight in kilograms divided by height in meters squared.

BMI Z-scores were computed according to WHO standards [42]. Children were categorized as underweight if their Z-score fell below −2, normal if their Z-score ranged from −2 to +1, and overweight if their Z-score was greater than +1 to +2. Children with Z-scores exceeding +2 were classified as obese [42].

The 25-hydroxyvitamin-D test was used to measure vitamin D status [43]. The bioanalysts drew blood from the participants’ veins while adhering to all biosafety measures. The blood samples were then centrifuged for ten minutes using portable centrifuges (EBA 20, Hettich GmbH & Co. KG, Tuttlingen, Germany). Laboratory technicians transported the blood samples in liquid nitrogen to the INS (National Institute of Health in Bogota, Colombia) for processing and analysis. Total serum 25(OH)D concentration was quantified using a chemiluminescent antibody immunoassay on an ADVIA Centaur XP analyzer (Siemens Health Care Diagnostics Inc., Tarrytown, NY, USA). This competitive immunoassay involves a mouse monoclonal anti-fluorescent antibody covalently bound to paramagnetic particles, acridinium ester-labeled mouse monoclonal anti-25(OH)D antibody, and a fluorescein-labeled vitamin D analog.

The Siemens ADVIA Centaur^® Vitamin D Total assay is distinguished by its rigorous adherence to standardized protocols, ensuring its traceability to the Ghent Reference Method (RMP), a procedure endorsed by the National Institute of Standards and Technology (NIST). This assay’s consistent participation in the Centers for Disease Control and Prevention (CDC) Vitamin D Standardization Certification Program (VDSCP) over several years highlights its commitment to ongoing validation and quality control [44]. Performance benchmarks, including a coefficient of variation (CV) of 15% and a bias of less than 10% as outlined in laboratory data models, alongside the expert-recommended DEQAS model specifications of a CV of 22% and a bias of 10%, have been successfully met by the ADVIA Centaur assay in clinical trials [45,46]. These achievements affirm the assay’s compliance with both sets of criteria, underscoring its reliability and accuracy for routine measurement procedures [46].

Due to the lack of consensus on the cutoff point for defining vitamin D deficiency [36,47], we have proposed three thresholds: <30 nmol/L [6,40], <37.5 nmol/L [48,49], and <50 nmol/L [6,40]. Among these, the threshold of <50 nmol/L is notably the most widely accepted and consistently supported across numerous studies.

DBM status was determined by the simultaneous presence of overweight/obesity and low vitamin D levels. This study considered three specific scenarios: 1. DBM1: Overweight/obese subjects with vitamin D levels below 30 nmol/L. 2. DBM2: Overweight/obese subjects with vitamin D levels below 37.5 nmol/L. 3. DBM3: Overweight/obese subjects with vitamin D levels below 50 nmol/L.

The socio-ecological model of correlates of the double burden of malnutrition in developing countries by Mahmudiono et al. [50] was the basis for selecting variables in our model (

Table 1. Socio-Ecological Model of Factors Associated with the Double Burden of Malnutrition (BMI-for-age z score > 1 and Vitamin D deficiency) among Colombian Schoolchildren.

Level	Variables
Individual	Age Biological Sex Ethnicity Meets the Minimum of Physical Activity Excessive Exposure to Screens
Interpersonal	Wealth Quartiles Household Size Type of Family Biological Sex of Head of Household Education Level of Head of Household Maternal Education Level
Community	Region Urbanicity Degree of Urbanization Living in Neighborhood with Playgrounds

).

We examined a range of individual-level sociodemographic variables, including age, biological sex, ethnicity, minimum physical activity, and excessive screen exposure. In the bivariate analysis, age was considered a discrete variable measured in completed years (y), whereas in the multivariate analysis, it was treated as a continuous variable. Biological sex was categorized into two groups: male and female. Ethnicity was self-reported, based on the participant’s appearance, skin color, or identification with specific racial or ethnic groups [43].

To assess physical activity and screen exposure, the survey investigated whether schoolchildren met the recommended 60 min of daily physical activity and whether they experienced excessive screen time, defined as more than two hours per day in front of screens [43]. These aspects were captured using two instruments tailored to different age groups, developed by the ENSIN team [43]. For children aged 3 to 5 years, the Measurement of Physical Activity and Sedentary Behavior (C-MAFYCS) questionnaire was used [43]. This parent-reported tool assessed active play outside educational settings. For children and adolescents aged 6 to 17 years, the Youth Risk Behavior Surveillance System (YRBSS) questionnaire was administered [43]. This instrument evaluated compliance with physical activity guidelines and measured excessive screen time unrelated to schoolwork, including television viewing, computer use, video gaming, tablet use, and smartphone engagement.

At the interpersonal level, we delve into both household characteristics and the attributes of the household head. Regarding the household, we scrutinize both its size and wealth. Household size was dichotomized into two categories: those with more than four members and those with fewer. We also included the variable type of family, which refers to the distribution by family structure typology and includes the following two categories: nuclear family and extended family. A nuclear household typically consists of two parents and their children living together as a single-family unit. This structure is often seen as the traditional family model. In contrast, an extended household includes not only the nuclear family but also other relatives such as grandparents, aunts, uncles, or cousins who live under the same roof, reflecting a broader support network and multigenerational living arrangements.

Wealth level was assessed using the quartile system proposed by the ENSIN team [43], ranging from one to four. Notably, the lowest quartile denotes the most financially vulnerable households. This classification evaluates household economic standing based on three critical dimensions: asset ownership, access to public services, and housing quality [43]. When examining the characteristics of the household head, we considered their biological sex and educational background. Specifically, variables pertaining to the mother’s educational attainment were incorporated into the analysis.

Finally, at the community level, we examined the region of residence of schoolchildren, including Atlantico, Oriental, Orinoquia, Amazonia, Bogota, Central, and Pacifico. We also considered urbanicity (rural vs. urban) and degree of urbanization (population size). Additionally, we included whether children lived in neighborhoods with playgrounds or other recreational spaces. Access to safe and well-equipped recreational spaces influences the amount of time children spend outdoors, which directly affects their exposure to ultraviolet B (UVB) radiation—the main source for vitamin D synthesis. Proximity to parks or playgrounds can facilitate sustained sun exposure necessary for adequate 25(OH)D production, and differences in the availability or quality of these environments may help explain variations in vitamin D status among children [51].

2.2. Statistical Analysis

We estimated the weighted prevalence of the double burden of malnutrition (DBM) among Colombian schoolchildren, defined as a BMI-for-age z-score > 1 (overweight/obesity) combined with vitamin D deficiency, using three thresholds: <30 nmol/L (DBM1), <37.5 nmol/L (DBM2), and <50 nmol/L (DBM3). Weighted prevalence was calculated as the proportion of children meeting both criteria among those with complete anthropometric and vitamin D data, relative to the total study population (n = 6063). For DBM3, we additionally calculated prevalence separately for children classified as overweight only and for those classified as obese only; for DBM1 and DBM2, the number of cases was too small to allow this stratification. All analyses accounted for the complex survey design—including primary sampling units (PSUs), strata, and sampling weights—to produce nationally representative estimates.

Bivariate associations between DBM outcomes and individual, household, and community-level variables were assessed using survey-weighted chi-square tests for categorical variables and survey-weighted logistic regression to estimate unadjusted odds ratios (crude ORs) with 95% confidence intervals. A significance level of 0.05 was applied. To address the risk of false-positive findings due to multiple comparisons, p-values were adjusted using the False Discovery Rate (FDR) method [52]. Both the raw and FDR-adjusted p-values are reported for each variable.

Covariate selection for multivariable logistic regression followed a two-step process. First, each candidate variable was assessed through survey-weighted logistic regression, accounting for clustering, stratification, and sampling weights. For categorical variables with multiple levels, the minimum p-value across levels (excluding the intercept) was used. Variables with a bivariate association of p < 0.20, based on survey-weighted Wald tests, were considered for inclusion. Additionally, age and sex were retained in all models regardless of their statistical significance, as they are potential confounders and key demographic factors commonly associated with child nutritional status. This approach combines empirical selection criteria with theoretical rationale, enhancing model stability and ensuring epidemiological relevance [53].

To assess potential multicollinearity among independent variables, we first recoded categorical variables as factors and created dummy variables for all factor levels. We then calculated the Variance Inflation Factor (VIF) using a weighted linear model, accounting for survey design weights, to identify variables that might be highly correlated. Variables with VIF values above 5 were considered indicative of problematic multicollinearity, following common epidemiological guidelines [54,55].

Survey-weighted logistic regression models were then fitted to estimate both crude and adjusted associations between DBM outcomes and explanatory variables. Variables showing a bivariate association with p < 0.20 and not exhibiting problematic multicollinearity (VIF > 5) were considered eligible for inclusion in multivariable models. Multivariable survey-weighted logistic regression models were fitted using penalized regression with Firth’s correction [56] to improve estimation stability across all DBM outcomes. This approach was applied to DBM1, DBM2, and DBM3, with stratified analyses conducted for each outcome by considering overweight and obesity both separately and in combination. Effect estimates were evaluated in terms of their magnitude and precision, with particular attention to confidence intervals rather than relying solely on p-values.

All statistical analyses were performed using R software (version 4.4.3), employing the survey package to account for the complex sampling design.

3. Results

3.1. Prevalence of Anthropometric Indicators, Vitamin D Deficiency, and DBM in Colombian Schoolchildren (5–12 Y), ENSIN 2015

Among Colombian schoolchildren aged 5–12 years (ENSIN 2015; n = 6063), most had a normal weight (73.9%), while 17.0% were classified as overweight, 7.4% as obese, and only 1.6% as underweight. The prevalence of vitamin D deficiency varied substantially depending on the cutoff applied—2.5% for concentrations below 30 nmol/L, 10.6% for below 37.5 nmol/L, and 22.6% for below 50 nmol/L. Consistent with these thresholds, the prevalence of the double burden of malnutrition (DBM)—defined as the coexistence of overweight/obesity and vitamin D deficiency—also increased as broader definitions were used: 0.7% for DBM1 (<30 nmol/L), 3.6% for DBM2 (<37.5 nmol/L), and 6.9% for DBM3 (<50 nmol/L). These results indicate that although severe DBM is relatively uncommon, a considerable proportion of schoolchildren are affected when more inclusive vitamin D thresholds are applied (

Table 2. Weighted Prevalence of Anthropometric Indicators, Vitamin D Deficiency, and Double Burden of Malnutrition (DBM: BMI-for-Age Z-Score > 1 and Vitamin D Deficiency) Among Colombian Schoolchildren Aged 5–12 Years, ENSIN 2015.

Characteristics	n	Weighted n	Prev ± SE (95% CI)
Underweight	79	90,066	1.6 ± 0.4 (0.8–2.4)
Normal	4565	4,111,243	73.9 ± 1.2 (71.5–76.4)
Overweight	1012	947,533	17.0 ± 1.1 (14.9–19.2)
Obesity	407	410,912	7.4 ± 0.8 (5.9–8.9)
Vitamin D Level < 30 nmol/L	126	138,754	2.5 ± 0.5 (1.5–3.5)
Vitamin D Level < 37.5 nmol/L	595	587,918	10.6 ± 1.5 (7.6–13.6)
Vitamin D Level < 50 nmol/L	1290	1,254,375	22.6 ± 3.0 (16.7–28.4)
* DBM1	40	41,088	0.7 ± 0.2 (0.4–1.1)
** DBM2	188	201,464	3.6 ± 0.7 (2.2–5.0)
*** DBM3	381	384,778	6.9 ± 1.2 (4.6–9.3)

* DBM1: BMI-for age z score > 1 (overweight/obese) and Vitamin D < 30 nmol/L. ** DBM2: BMI-for age z score > 1 (overweight/obese) and Vitamin D < 37.5 nmol/L. *** DBM3: BMI-for age z score > 1 (overweight or obese) and Vitamin D < 50 nmol/L. n = number of cases; Weighted n = weighted number of cases. Prev ± SE (95% CI) = prevalence (%) ± standard error (%) with 95% confidence interval.

). All prevalence estimates were weighted and accounted for the complex survey design, including primary sampling units (PSUs), strata, and survey weights, to ensure national representativeness.

3.2. Prevalence of Double Burden of Malnutrition (DBM1) by Individual, Interpersonal, and Community-Level Factors Among Colombian Schoolchildren Aged 5–12 Years

The overall weighted prevalence of DBM1—defined as the coexistence of overweight/obesity (BMI-for-Age Z Score > 1) and vitamin D deficiency (<30 nmol/L)—was extremely low among Colombian schoolchildren aged 5–12 years, affecting less than 1% of the population. Detailed prevalence estimates by individual, interpersonal, and community-level factors are presented in Supplementary Table S1.

Prevalence was uniformly low across all examined factors. At the individual level, estimates ranged from 0% to 2.1% across age groups, with minimal variation by sex, ethnicity, physical activity, or screen exposure. At the interpersonal level, small numerical differences were observed across wealth quartiles, household size, family type, and parental education, but none reached statistical significance based on FDR-adjusted p-values. At the community level, prevalence remained below 1% in most regions and urbanicity categories, with slightly higher numerical estimates in major metropolitan cities (1.3%), yet these differences were not statistically significant. Overall, DBM1 prevalence was extremely low and consistent across individual, interpersonal, and community-level factors. The limited number of cases and wide confidence intervals highlight the uncertainty of these estimates. Separate analyses for overweight-only or obesity-only subgroups were not feasible due to the small number of cases.

3.3. Prevalence of Double Burden of Malnutrition (DBM2) by Individual, Interpersonal, and Community-Level Factors Among Colombian Schoolchildren Aged 5–12 Years

The weighted prevalence of DBM2—defined as the coexistence of overweight/obesity (BMI-for-Age Z Score > 1) and vitamin D deficiency (<37 nmol/L)—was 3.6% among Colombian schoolchildren aged 5–12 years. Differences across age, sex, ethnicity, physical activity, screen time, and wealth quartiles were observed descriptively, but these did not reach statistical significance after adjusting for multiple comparisons (Supplementary Table S2). At the household level, prevalence was higher in smaller households (<4 members, 4.9%) compared with larger households (>4 members, 2.4%). Maternal education also showed variation, with the highest prevalence among children whose mothers completed secondary education (11–15 years, 5.2%) (Supplementary Table S2).

Community-level differences were observed by region and urbanization. Prevalence was highest in Bogota (7.0%) and major metropolitan cities (6.3%), with lower prevalence in the Atlantico region (1.0%) and smaller or large urban areas (2.5–2.7%). These patterns suggest variability in DBM2 prevalence across household and geographic factors, while other individual-level and behavioral characteristics showed minimal differences (Supplementary Table S2). Analyses restricted to overweight only or obesity only were not conducted, as the number of cases was too small to provide reliable estimates.

3.4. Prevalence of Double Burden of Malnutrition (DBM3) by Individual, Interpersonal, and Community-Level Factors Among Colombian Schoolchildren Aged 5–12 Years

The prevalence of DBM3, defined as concurrent overweight/obesity (BMI-for-Age Z Score > 1) and vitamin D deficiency (<50 nmol/L), was higher than that observed for DBM1 and DBM2, affecting 6.9% of Colombian schoolchildren aged 5–12 years.

3.4.1. Individual-Level Factors

These findings are descriptive and reflect observed variations in DBM3 prevalence across demographic and behavioral characteristics (

Table 3. Prevalence of Double Burden of Malnutrition (DBM3): Concurrent Overweight/Obesity (BMI-for-Age Z Score > 1) and Vitamin D Deficiency (<50 nmol/L) by Individual-Level Factors Among Colombian Schoolchildren Aged 5–12 Years.

Characteristics	n	Weighted n	Prev ± SE (95% CI)	p-Value	p-Value FDR
Individual Level
Age (Years)
5	69	38,020	6.3 ± 2 (2.4, 10.3)	0.502	0.543
6	40	56,073	7.9 ± 1.7 (4.7, 11.1)
7	37	45,272	5.5 ± 1.9 (1.8, 9.2)
8	49	54,861	7.9 ± 1.9 (4.1, 11.6)
9	45	48,388	7 ± 2 (2.9, 11)
10	48	50,756	7.6 ± 1.6 (4.5, 10.6)
11	52	56,692	8.4 ± 1.8 (5, 11.9)
12	41	34,717	5 ± 1 (3.1, 7)
Biological Sex
Male	192	185,085	6.5 ± 1.1 (4.4, 8.5)	0.250	0.332
Female	189	199,693	7.4 ± 1.5 (4.5, 10.4)
Ethnicity
Black	33	23,610	4.3 ± 2 (0.4, 8.2)	0.174	0.249
Indigenous	37	11,874	3.6 ± 1.6 (0.6, 6.7)
Without Ethnicity	311	349,294	7.5 ± 1.3 (4.9, 10)
Meets the Minimum of Physical Activity
No	283	290,250	7.5 ± 1.2 (5.1–10)	0.020	0.040
Yes	96	88,361	5.2 ± 1.1 (3.1–7.4)
Missing	2	6168	-
Excessive Exposure to Screens
Yes	256	281,328	7.8 ± 1.3 (5.2–10.5)	<0.001	<0.001
No	124	97,378	5 ± 0.8 (3.3–6.6)
Missing	1	6072	-

p-value = χ² score test result. p-value FDR: p-values adjusted for multiple testing using the False Discovery Rate (FDR) method. Prev ± SE (95% CI) = Prevalence (%) ± Standard error (%) (95% Confidence Interval). “-” indicates missing data for which prevalence could not be calculated.

). At the individual level, prevalence was higher among children who did not meet recommended physical activity levels (7.5% vs. 5.2%) and among those with excessive screen exposure (7.8% vs. 5.2%), whereas differences by sex and ethnicity were not statistically significant after FDR adjustment.

3.4.2. Interpersonal-Level Factors

At the interpersonal level, DBM3 prevalence varied across socioeconomic and household characteristics (

Table 4. Prevalence of Double Burden of Malnutrition (DBM3): Concurrent Overweight/Obesity (BMI-for-Age Z Score > 1) and Vitamin D Deficiency (<50 nmol/L) by Interpersonal-Level Factors Among Colombian Schoolchildren Aged 5–12 Years.

Characteristics	n	Weighted n	Prev ± SE (95% CI)	p-Value	p-Value FDR
Wealth Index, Quartiles
Q1	157	95,236	4.2 ± 0.6 (2.9, 5.5)	<0.0001	<0.0001
Q2	119	112,902	7.7 ± 1.3 (5.1, 10.2)
Q3	64	82,925	7.9 ± 1.8 (4.4, 11.5)
Q4	41	93,714	12.1 ± 2.8 (6.6, 17.6)
Household Size
<4 People	234	255,390	9.3 ± 1.4 (6.5, 12.2)	<0.0001	<0.0001
>4 People	147	129,388	4.6 ± 1 (2.6, 6.6)
Type of Family
Nuclear	224	221,949	7.2 ± 1.2 (4.7, 9.6)	0.576	0.599
Extended	157	162,829	6.6 ± 1.4 (4, 9.3)
Biological Sex of the Head of Household
Male	236	238,686	6.9 ± 1.2 (4.5, 9.2)	0.879	0.879
Female	145	146,093	7 ± 1.4 (4.3, 9.7)
Education Level of Head of Household (Years)
0–4	77	77,239	5.2 ± 1.1 (3, 7.4)	0.037	0.059
5–10	135	131,464	6.8 ± 1.1 (4.7, 9)
11–15	146	159,323	8.8 ± 1.9 (5, 12.6)
16–24	22	16,570	5.2 ± 1.7 (1.9, 8.5)
Missing	1	183	-
Maternal Education Level (Years)
0–4	45	39,997	5.2 ± 1.6 (2.1–8.3)	<0.0001	<0.0001
5–10	115	96,234	5 ± 0.9 (3.3–6.7)
11–15	171	206,201	9.5 ± 1.6 (6.4–12.7)
16–24	34	33,645	9.6 ± 2.7 (4.3–15)
Missing	16	8701	-

p-value = χ² score test result. p-value FDR: p-values adjusted for multiple testing using the False Discovery Rate (FDR) method. Prev ± SE (95% CI) = Prevalence (%) ± Standard error (%) (95% Confidence Interval). “-” indicates missing data for which prevalence could not be calculated.

). A clear positive gradient was observed across wealth quartiles, ranging from 4.2% in the lowest quartile (Q1) to 12.1% in the highest (Q4). Prevalence was higher among children living in smaller households (9.3%) compared with larger households (4.6%). In contrast, prevalence was similar across family type and sex of the household head, with no significant differences. DBM3 prevalence also varied by maternal education. Prevalence tended to be higher among children whose mothers had completed secondary (9.5%) or higher education (9.6%) compared with lower maternal education levels.

3.4.3. Community-Level Factors

At the community level, the prevalence of DBM3 varied across regions and cities. Bogota had the highest prevalence (13.4%), followed by the Oriental (8.5%) and Pacific (8.0%) regions, whereas the Atlantic region showed the lowest burden (1.9%) (

Table 5. Prevalence of Double Burden of Malnutrition (DBM3): Concurrent Overweight/Obesity (BMI-for-Age Z Score > 1) and Vitamin D Deficiency (<50 nmol/L) by Community-Level Factors Among Colombian Schoolchildren Aged 5–12 Years.

Characteristics	n	Weighted n	Prev ± SE (95% CI)	p-Value	p-Value FDR
Region
Atlantico	31	26,483	1.9 ± 0.7 (0.6, 3.3)	<0.0001	<0.0001
Oriental	74	83,110	8.5 ± 1.7 (5.1–11.9)
Orinoquia and Amazonia	99	8926	4.8 ± 0.7 (3.4–6.2)
Bogota	44	106,137	13.4 ± 2.7 (8.1–18.8)
Central	71	83,121	6.5 ± 1.2 (4–8.9)
Pacifico	62	77,002	8 ± 2.8 (2.5–13.6)
Urbanicity
Urban	310	312,857	7.4 ± 1.5 (4.5, 10.3)	0.257	0.332
Rural	71	71,922	5.5 ± 1 (3.5, 7.4)
Degree of Urbanization (Population Size)
Major Metropolitan Cities	70	160,755	10.8 ± 2.6 (5.7, 15.9)	0.032	0.059
Large Urban Areas	78	86,496	5.7 ± 1.4 (3, 8.4)
Small Urban Areas	233	137,527	5.4 ± 0.8 (3.9, 6.9)
Living In Neighborhood with Playgrounds
Yes	229	247,329	7.5 ± 1.3 (4.9–10)	<0.0001	<0.0001
No	151	131,377	5.9 ± 1 (4–7.8)
Missing	1	6072	-

p-value = χ² score test result. p-value FDR: p-values adjusted for multiple testing using the False Discovery Rate (FDR) method. Prev ± SE (95% CI) = Prevalence (%) ± Standard error (%) (95% Confidence Interval). Major metropolitan cities (Barranquilla, Cali, Medellin, Bogota). Large urban areas (100,001–1,000,000 inhabitants). Small urban areas (≤100,000 inhabitants). “-” indicates missing data for which prevalence could not be calculated.

). Additionally, children living in neighborhoods with playgrounds exhibited a higher prevalence of DBM3 compared with those without playgrounds (7.5% vs. 5.9%).

Across analyses of DBM3, prevalence patterns differed depending on the anthropometric threshold used to define the double burden of malnutrition (Supplementary Tables S3 and S4). In the overweight-only subgroup (BMI-for-age Z score > 1 and ≤2; Table S3), prevalence was higher among children not meeting physical activity recommendations (5.5% vs. 3.6%) and those with excessive screen exposure (5.8% vs. 3.4%). At the household and community levels, higher prevalence was observed in smaller households (<4 members: 6.8% vs. ≥4 members: 3.3%), in Bogota (11.7%) and other major metropolitan areas, and among children living in neighborhoods with playgrounds (5.5% vs. 4.1%). Degree of urbanization also showed significant differences, with higher prevalence in major metropolitan cities (8.8%) compared with large (3.2%) and small urban areas (3.9%).

In the obesity-only subgroup (BMI-for-age Z score >2; Table S4), only maternal education showed statistically significant differences, with higher prevalence among children of mothers with 11–15 or 16–24 years of schooling (2.9–3.4%) compared with lower education levels (0.8–1.3%). No other individual-, household-, or community-level factors reached statistical significance in this subgroup.

When overweight and obesity were combined (BMI-for-age Z score > 1; Table 3, Table 4 and Table 5), prevalence patterns largely reflected those of the overweight-only group, with significant associations for physical activity, screen exposure, household size, maternal education, region, wealth, and living in neighborhoods with playgrounds. However, some patterns specific to the obesity-only subgroup were less apparent in the combined analysis. For instance, the strong association of higher maternal education with obesity prevalence became less distinct, as the overweight group contributed more to the overall estimates. Similarly, regional differences in obesity prevalence were partially masked, although Bogota remained the area with the highest prevalence. These observations indicate that combining overweight and obesity can obscure certain subgroup-specific patterns, emphasizing the value of examining anthropometric thresholds separately to capture nuanced variations in DBM3 prevalence.

3.5. Unadjusted and Adjusted Associations of Individual, Interpersonal, and Community-Level Factors with the Double Burden of Malnutrition (DBM1) Among Colombian Schoolchildren

We examined associations between individual-, interpersonal-, and community-level factors and DBM1 (concurrent overweight/obesity and vitamin D deficiency <30 nmol/L) (Supplementary Table S5). In unadjusted analyses, children not meeting the minimum physical activity recommendations had higher odds of DBM1 compared with those meeting the recommendations (OR = 10.64; 95% CI: 1.98–57.25), and excessive screen exposure was also associated with higher odds (OR = 2.44; 95% CI: 1.08–5.53). At the household and community levels, children from the wealthiest households (Q4 vs. Q1) had higher odds (OR = 4.04; 95% CI: 1.05–15.55), while regional differences were observed, with lower odds in the Atlántico (OR = 0.26; 95% CI: 0.08–0.86) and Orinoquia/Amazonia (OR = 0.29; 95% CI: 0.10–0.80) compared with Bogota.

For the adjusted model, we included variables with a bivariate Wald test p-value < 0.20 that did not show evidence of multicollinearity, along with age and sex (included a priori). After adjustment, not meeting the minimum physical activity recommendations remained significantly associated with DBM1 (adjusted OR = 2.37; 95% CI: 1.02–6.67). Associations for other factors—including excessive screen exposure, household wealth, and regional differences—were attenuated and confidence intervals crossed 1, indicating no statistically significant associations in the fully adjusted model (Supplementary Table S5).

These results suggest that physical activity shows the most robust association with DBM1, whereas other observed differences at the household and community levels are less pronounced when considered jointly in the model. Additional sensitivity analyses stratified DBM1 by weight category (overweight-only vs. obesity-only) (Supplementary Table S6).

In the overweight-only subgroup, head of household education was significantly associated with DBM1 (5–10 years vs. 11–15 years: OR = 3.19; 95% CI: 1.05–10.49). In the obesity-only subgroup, physical inactivity showed a very high OR (OR = 11.73; 95% CI: 1.52–1507.37), but the extremely wide confidence interval indicates high uncertainty, likely due to a small number of cases. No other factors reached statistical significance in either subgroup.

Comparing adjusted ORs across weight definitions (combined overweight + obesity vs. overweight-only and obesity-only) suggests weight-specific patterns: some factors appear strongly associated in subgroup analyses but are attenuated in the combined group. For example, physical inactivity had a much higher OR in the obesity-only group than in the combined group, while household education remained significant only in the overweight-only group. These patterns highlight the importance of considering subgroup-specific effects when interpreting risk factors for DBM1.

3.6. Unadjusted and Adjusted Associations of Individual, Interpersonal, and Community-Level Factors with the Double Burden of Malnutrition (DBM2) Among Colombian Schoolchildren

Results for DBM2 (BMI-for-age z score > 1 and vitamin D < 37.5 nmol/L) are summarized in Supplementary Table S7. In unadjusted analyses, several factors showed statistically significant associations. At the individual level, children identifying as Black (OR = 5.86; 95% CI: 1.73–19.81) or without ethnicity information (OR = 9.01; 95% CI: 3.65–22.21) had higher odds of DBM2 compared with Indigenous children. Excessive screen exposure was also associated with higher odds (OR = 1.62; 95% CI: 1.10–2.38). Interpersonal-level factors, including household wealth, showed increasing odds with higher quartiles (Q2: OR = 2.26, 95% CI: 1.19–4.28; Q3: OR = 3.02, 95% CI: 1.60–5.72; Q4: OR = 2.75, 95% CI: 1.17–6.48), and smaller households (<4 members) had higher odds (OR = 2.05; 95% CI: 1.41–2.98). Female heads of household were associated with higher odds compared with male heads (OR = 1.43; 95% CI: 1.06–1.94). At the community level, residing in the Atlantico (OR = 0.14; 95% CI: 0.07–0.27), Orinoquia and Amazonia (OR = 0.24; 95% CI: 0.13–0.44), or Central regions (OR = 0.49; 95% CI: 0.28–0.89) was associated with lower odds compared with Bogota. Children living in large (OR = 0.38; 95% CI: 0.18–0.83) or small urban areas (OR = 0.42; 95% CI: 0.22–0.79) had lower odds compared with those in major metropolitan cities.

After adjustment, considering variables selected based on a global Wald test p-value < 0.20 and without evidence of collinearity, only a subset of associations remained statistically significant. Children from smaller households continued to show higher odds of DBM2 (OR = 1.50; 95% CI: 1.09–2.07), and those residing in large urban areas had lower odds compared with major metropolitan areas (OR = 0.47; 95% CI: 0.25–0.91). All other characteristics, including ethnicity, excessive screen exposure, household wealth, female head of household, regional differences, and small urban areas, were no longer statistically significant after adjustment.

Sensitivity analyses for DBM2 (BMI-for-age z score > 1 and vitamin D < 37.5 nmol/L) were conducted separately for overweight-only and obesity-only definitions (Supplementary Table S8). In the overweight-only group, residing in large urban areas was associated with lower odds of DBM2 compared with major metropolitan areas (OR = 0.35; 95% CI: 0.16–0.78). In the obesity-only group, children living in smaller households (<4 members) had higher odds compared with larger households (>4 members; OR = 1.96; 95% CI: 1.14–3.47). In the combined overweight + obesity definition, both associations remained detectable—residing in large urban areas (OR = 0.47; 95% CI: 0.25–0.91) and smaller household size (OR = 1.50; 95% CI: 1.09–2.07)—though the subgroup-specific magnitude is less pronounced. These findings highlight how some associations are more apparent when weight categories are analyzed separately, while others are still observable in the combined analysis. All associations are measures of statistical association and should not be interpreted as causal.

3.7. Unadjusted and Adjusted Associations of Individual, Interpersonal, and Community-Level Factors with the Double Burden of Malnutrition (DBM3) Among Colombian Schoolchildren

Before fitting the multivariable models, multicollinearity was assessed, and all predictors had VIFs below 5, indicating no problematic correlation. Variables were selected based on theoretical considerations and bivariate analyses (global Wald test p < 0.20), with age and sex retained A Priori.

3.7.1. Individual-Level Factors

At the individual level (

Table 6. Unadjusted and Adjusted Models of Individual-level Factors Associated with the Double Burden of Malnutrition (DBM3: BMI-for age z score > 1 (overweight/obese) and Vitamin D < 50 nmol/L) in Colombian Schoolchildren Aged 5 to 12 y.

Characteristics	Category	Unadjusted Model	p-Value	Adjusted Model
Individual-Level		OR (95% CI)		OR (95% CI)
Age	Per year increase	1.00 (0.92, 1.09)	0.952	1.04 (0.99, 1.08)
Biological Sex	Male (Ref)	1	0.250	1
	Female	1.16 (0.90, 1.51)		1.08 (0.87, 1.34)
Ethnicity	Indigenous (Ref)	1	0.181	1
	Black	1.18 (0.32, 4.33)		0.82 (0.47, 1.43)
	Without Ethnicity	2.13 (0.80, 5.70)		0.89 (0.60, 1.36)
Meets the Minimum of Physical Activity	Yes (Ref)	1		1
	No	1.47 (1.04, 2.08)	0.058	1.19 (0.93, 1.54)
Excessive Exposure to Screens	No (Ref)	1		1
	Yes	1.63 (1.22, 2.19)	0.001	1.10 (0.87, 1.40)

Adjusted Model include variables with bivariate Wald test p < 0.20, age and sex (A Priori); collinearity assessed. Final models fitted using Firth’s penalized logistic regression.

), no statistically significant associations were observed for age or sex in the adjusted model. Physical inactivity and excessive screen exposure showed positive associations with DBM3 in the unadjusted analyses (OR = 1.47 and OR = 1.63, respectively), but these relationships were attenuated after adjustment for other covariates. Similarly, no significant differences were detected across ethnic groups once adjustment was applied. These findings suggest that individual-level behaviors and biological characteristics were not strongly associated with DBM3 when accounting for sociodemographic and contextual factors.

3.7.2. Interpersonal-Level Factors

At the interpersonal level (

Table 7. Unadjusted and Adjusted Models of Interpersonal-level Factors Associated with the Double Burden of Malnutrition (DBM3: BMI-for age z score > 1 (overweight/obese) and Vitamin D < 50 nmol/L) in Colombian Schoolchildren Aged 5 to 12 y.

Characteristics	Category	Unadjusted Model	p-Value	Adjusted Model
Interpersonal Level		OR (95% CI)		OR (95% CI)
Wealth Index, Quartiles	Q1 (Ref)	1	<0.001	1
	Q2	1.90 (1.20, 3.00)		1.47 (1.10, 1.96)
	Q3	1.97 (1.23, 3.17)		1.23 (0.85, 1.78)
	Q4	3.15 (1.85, 5.39)		1.36 (0.86, 2.13)
Household Size	>4 people (Ref)	1		1
	<4 people	2.13 (1.59, 2.86)	<0.001	1.61 (1.28, 2.03)
Type of Family	Extended (Ref)	1	0.577	-
	Nuclear	1.09 (0.80, 1.47)		-
Biological Sex of Head of Household	Male (Ref)	1	0.879	-
	Female	1.02 (0.78–1.33)		-
Education of Head of Household (Years)	11–15 (Ref)	1	0.016	1
	0–4	0.57 (0.35, 0.92)		0.96 (0.67, 1.37)
	5–10	0.76 (0.51, 1.14)		1.13 (0.84, 1.52)
	16–24	0.57 (0.29, 1.10)		0.73 (0.42, 1.23)
Maternal Education (Years)	0–4 (Ref)	1		1
	5–10	0.96 (0.50, 1.84)	<0.001	0.91 (0.62, 1.34)
	11–15	1.92 (1.08, 3.41)		1.21 (0.81, 1.82)
	16–24	1.94 (0.80, 4.69)		2.29 (1.26, 4.12)

Adjusted Model include variables with bivariate Wald test p < 0.20, age and sex (A Priori); collinearity assessed. Final models fitted using Firth’s penalized logistic regression. “-” indicates that the variable was not included in the final model.

), household characteristics displayed notable patterns. Children from smaller households (<4 members) had higher adjusted odds of DBM3 compared with those from larger households (adjusted OR = 1.61; 95% CI: 1.28–2.03). Maternal education showed a graded pattern, with higher adjusted odds of DBM3 among children whose mothers had 16–24 years of education (adjusted OR = 2.29; 95% CI: 1.26–4.12). The sex and educational level of the head of household, as well as family structure, were not significantly associated with DBM3 after adjustment. These results indicate that differences in household composition and maternal education were statistically related to variations in DBM3, though no directional or causal inferences can be drawn.

3.7.3. Community-Level Factors

At the community level (

Table 8. Unadjusted and Adjusted Models of Community-level Factors Associated with the Double Burden of Malnutrition (DBM3: BMI-for age z score > 1 (overweight/obese) and Vitamin D < 50 nmol/L) in Colombian Schoolchildren Aged 5 to 12 y.

Characteristics	Category	Unadjusted Model	p-Value	Adjusted Model
Community level		OR (95% CI)		OR (95% CI)
Region	Bogota (Ref)	1	<0.001	1
	Atlantico	0.13 (0.06, 0.27)		0.52 (0.28, 0.93)
	Oriental	0.60 (0.38, 0.94)		1.25 (0.68, 2.26)
	Orinoquia and Amazonia	0.33 (0.24, 0.44)		0.90 (0.49, 1.63)
	Central	0.45 (0.30, 0.67)		0.76 (0.43, 1.34)
	Pacifico	0.56 (0.26, 1.21)		1.34 (0.76, 2.31)
Urbanicity	Urban (Ref)	1	0.259	-
	Rural	0.72 (0.41, 1.28)
Degree of Urbanization	Major Metropolitan Cities	1	0.057	1
	Large Urban Areas	0.50 (0.23, 1.10)		0.64 (0.40, 1.06)
	Small Urban Areas	0.47 (0.25, 0.87)		0.70 (0.43, 1.15)
Living in Neighborhood with Playgrounds	No (Ref)	1	0.004	1
	Yes	1.29 (0.99–1.69)		1.00 (0.79, 1.26)

Adjusted Model include variables with bivariate Wald test p < 0.20, age and sex (A Priori); collinearity assessed. Final models fitted using Firth’s penalized logistic regression. “-” indicates that the variable was not included in the final model.

), regional variations were evident in the unadjusted models but diminished after adjustment. Children residing in the Atlántico region had lower adjusted odds of DBM3 compared with those in Bogota (adjusted OR = 0.52; 95% CI: 0.28–0.93). No statistically significant associations were observed for other regions, degrees of urbanization, or the presence of playgrounds after adjustment. These findings suggest that regional differences in DBM3 were limited once sociodemographic and household factors were considered.

Further analyses compared adjusted odds ratios (ORs) across three alternative definitions of the double burden of malnutrition (DBM3)—combined overweight and obesity, overweight only, and obesity only—to explore convergent and definition-specific patterns across individual, interpersonal, and community-level factors (Supplementary Table S9). Variables were selected based on a bivariate Wald test (p < 0.20) and assessed for collinearity before inclusion in the final models.

In the overweight-only model, significant associations were observed for female sex (OR 1.33, 95% CI 1.03–1.73), smaller household size (OR 1.48, 95% CI 1.13–1.95), and moderate wealth (Q2) (OR 1.53, 95% CI 1.07–2.17). In the obesity-only model, significant associations included lack of physical activity (OR 2.16, 95% CI 1.34–3.67), smaller household size (OR 1.84, 95% CI 1.22–2.82), and higher maternal education (16–24 years) (OR 4.20, 95% CI 1.55–11.54). Notably, some of these ORs are large, and confidence intervals are wide, indicating considerable uncertainty, likely due to smaller sample sizes in these subgroups.

In the combined overweight + obesity model, several factors remained relevant, including smaller household size (OR 1.61, 95% CI 1.28–2.03), maternal education, moderate wealth (Q2), and residence in the Atlántico region (OR 0.52, 95% CI 0.28–0.93), highlighting persistent socioeconomic and regional differences.

Comparisons across models suggest that some characteristics, such as household size, were consistently associated across definitions, whereas others—including sex, maternal education, and physical activity—showed definition-specific patterns. These findings indicate that overweight and obesity may be linked to partially distinct sociodemographic and behavioral profiles, underscoring the value of examining them separately within DBM frameworks, while interpreting large ORs with caution due to potential imprecision.

4. Discussion

Using nationally representative data, we assessed the prevalence and associated factors of the double burden of malnutrition (DBM) among Colombian schoolchildren aged 5–12 years, defined as the co-occurrence of overweight/obesity (BMI-for-age z score > 1) and vitamin D deficiency (<30 nmol/L, <37.5 nmol/L, or <50 nmol/L). In this population, 17.0% of children were overweight and 7.4% were obese, while vitamin D deficiency affected 2.5%, 10.6%, and 22.6% of children, depending on the cutoff applied. The simultaneous presence of overweight/obesity and vitamin D deficiency (DBM) was relatively low, with prevalences of 0.7% (DBM1), 3.6% (DBM2), and 6.9% (DBM3). These estimates are lower than those reported in other countries. For instance, in Mexico, 18.2% of children had vitamin D deficiency (<50 nmol/L) alongside overweight, increasing to 24.7% when obesity was considered [57]. In the United States, 17% of overweight and 31% of obese children had vitamin D levels below 50 nmol/L [58]. Overall, these results suggest that, although a substantial proportion of Colombian schoolchildren are individually overweight/obese or vitamin D deficient, the combined occurrence of both conditions is relatively uncommon.

Although DBM prevalence is low, the co-occurrence of overweight/obesity and vitamin D deficiency remains clinically and publicly health relevant. Low vitamin D levels in school-aged children have been linked to obesity and impaired bone growth, as well as deficiencies in other micronutrients, including zinc, iron, and vitamin B6 [4]. While the cross-sectional design precludes causal inference, our findings suggest that even moderate prevalence of DBM warrants continued monitoring and interventions, particularly within the context of Colombia’s ongoing nutritional transition [59,60,61]. DBM3 (<50 nmol/L) represents the most widely accepted public health threshold, providing the most robust basis for policy and programmatic decisions, whereas DBM1 and DBM2 help identify children with severe deficiency and structural vulnerabilities.

4.1. Individual-Level Factors

Lifestyle behaviors showed distinct patterns of association with the double burden of malnutrition (DBM). In this study, school-aged children not meeting the recommended physical activity levels had higher odds of concurrent overweight/obesity and vitamin D deficiency, particularly at serum 25(OH)D concentrations below 30 nmol/L. This association was observed, though slightly attenuated, when considering a less severe threshold (<50 nmol/L) and obesity. While these findings align with prior evidence indicating that low physical activity and vitamin D deficiency often coexist in children with excess adiposity, it is important to note that the cross-sectional design precludes causal inference and that the observed ORs may be imprecise, particularly in subgroups with few cases.

Several plausible pathways may explain this association. First, low physical activity is often accompanied by limited outdoor time and, consequently, reduced exposure to ultraviolet B (UVB) radiation—currently recognized as the primary determinant of cutaneous vitamin D synthesis [62,63]. Second, children with overweight or obesity may spend less time outdoors or engaging in vigorous activities, further decreasing sunlight exposure and increasing the risk of hypovitaminosis D [64]. Additionally, excess adipose tissue can sequester vitamin D in fat depots, reducing its bioavailability and contributing to lower circulating 25(OH)D levels [65]. Recent systematic reviews suggest that physical activity per se has limited direct effects on serum vitamin D unless it increases outdoor exposure, emphasizing the behavioral mediation of this relationship [62]. Therefore, the association observed in this study likely reflects a combination of behavioral factors (e.g., reduced outdoor time and lower sun exposure) and biological mechanisms (e.g., adipose sequestration and altered metabolism), rather than a direct causal effect of physical inactivity on vitamin D deficiency. Consistent with these patterns, the national survey shows that only about 35% of children aged 6–17 meet recommended physical activity guidelines [66].

Excessive screen time was associated with DBM1, DBM2, and DBM3 in unadjusted analyses, but these associations did not persist after multivariable adjustment, suggesting that shared individual, interpersonal, or community factors may underlie the crude relationships. Excessive screen use remains a plausible behavioral risk marker given its potential role in displacing outdoor time (and thus reducing UVB-mediated cutaneous vitamin D synthesis [67]), promoting sedentarism and unhealthful snacking behaviors (increasing adiposity) [68], and correlating with demographic and socioeconomic confounders (e.g., parental education, neighborhood safety, access to green spaces). Moreover, higher adiposity itself may reduce the bioavailability of 25(OH)D via sequestration in fat tissue, thereby strengthening the convergence of obesity and vitamin D deficiency in DBM1–DBM3 configurations [69]. The attenuation of associations after adjustment implies that screen time may act more as a contextual or mediating variable rather than a direct causal factor of DBM in this population. Future longitudinal cohort studies and interventions that simultaneously measure screen time, outdoor UV exposure, diet, and body composition are needed to disentangle mediating pathways and evaluate temporal directionality.

4.2. Interpersonal-Level Factors

Household composition appeared to be associated with DBM. Children living in smaller households (<4 members) had higher odds of DBM2 and DBM3, a pattern consistent with evidence that children in smaller families tend to have higher BMI values, potentially reflecting more individualized feeding, increased access to calorie-dense foods, and fewer opportunities for active play within the home environment [70].

Socioeconomic status showed nuanced associations with DBM. In unadjusted analyses, the wealth index was associated with DBM2, but this relationship was not retained after adjustment. For DBM3, children in the second wealth quartile (Q2) had higher odds compared with those in the lowest quartile (Q1, reference), with an OR = 1.47 (95% CI: 1.10–1.96) when overweight and obesity were combined. A similar association was observed when only overweight children were considered (OR = 1.53 [95% CI: 1.07–2.17]), whereas no association was found when analyses were limited to obese children. These findings may reflect that children from middle-income households present behavioral or dietary patterns differing from those in lower or higher socioeconomic groups. This pattern aligns with previous evidence suggesting that both low- and middle-income families may experience contextual and behavioral conditions related to nutrient status and body composition [71,72,73]. Such results underscore the complexity of socioeconomic gradients in DBM, where contextual factors coexist with behavioral characteristics.

Parental education showed context-specific associations with DBM in our analyses. Household heads with incomplete secondary education (compared with those completing secondary education) were associated with higher odds of DBM1 in analyses restricted to children with overweight (excluding obesity) (OR = 3.19; 95% CI: 1.05–10.49). Similarly, mothers with high educational attainment (≥16 years) had higher odds of children presenting DBM2 (combined overweight and obesity with vitamin D deficiency) compared with mothers with ≤4 years of education (OR = 3.19; 95% CI: 1.05–10.49). For analyses focusing on obese children, these highly educated mothers also showed elevated odds of DBM3 (OR = 4.20; 95% CI: 1.55–11.54), as well as increased odds for the combined overweight and obesity category with vitamin D deficiency (OR = 2.29; 95% CI: 1.26–4.12). Some of these ORs are relatively large, and the corresponding confidence intervals are wide, particularly in subgroup analyses, reflecting limited precision and the smaller number of cases. Therefore, these estimates should be interpreted cautiously, as indicative of potential associations rather than definitive causal effects.

These patterns are consistent with the heterogeneous findings reported in the literature: the association between parental education and child overweight/obesity varies across contexts and country income levels, and in some middle-income settings higher parental education has been associated with higher child BMI or different dietary/behavioral profiles compared with lower education groups [74]. Moreover, broader reviews indicate that parental education can be linked to diverse pathways (e.g., food purchasing patterns, time allocation for child care, access to recreational spaces) that coexist with socioeconomic and environmental determinants of both adiposity and micronutrient status, which may help explain why associations differ by DBM subtype and by whether analyses include overweight alone versus obesity [74,75]. Given these context-dependent patterns and the cross-sectional nature of our data, the associations observed here should be considered exploratory and warrant further investigation through longitudinal studies or intervention research in similar settings.

4.3. Community-Level Factors

The residential environment appeared to influence DBM, although these associations should be interpreted cautiously given the cross-sectional design and potential confounding. Children living in large urban areas had lower odds of DBM2 for overweight + obesity (OR: 0.47; 95% CI: 0.25–0.91) and overweight-only (OR: 0.35; 95% CI: 0.16–0.78 (Supplementary Table S8)). Similarly, residing in the Atlántico region was associated with lower odds of overweight + obesity DBM (OR: 0.52; 95% CI: 0.28–0.93) compared with Bogota (Supplementary Table S9). These patterns may reflect contextual differences, such as warmer temperatures and cultural practices favoring outdoor activity, which provide children with greater opportunities for active play and sunlight exposure.

Data from ENSIN 2015 indicate that 63.8% of Colombian children live in neighborhoods with parks, green spaces, recreational centers, or sports facilities conducive to play, and 74.2% perceive these spaces as safe [66]. Participation in community initiatives such as Ciclovías (7.7% of children) and other physical activity programs (20.4% of adolescents) reflects proactive efforts to promote active lifestyles. However, disparities remain, particularly in suburban areas of large cities, where infrastructure for safe and accessible play is limited [76,77]. These inequalities may restrict opportunities for physical activity and contribute to nutritional imbalances [78,79].

These findings align with recent evidence from systematic reviews and longitudinal studies. A 2022 systematic review reported that greater access to green spaces and recreational areas was associated with lower BMI and higher physical activity levels among children, highlighting the potential relevance of urban planning for child health [80]. Similarly, a 2025 longitudinal study found that cumulative exposure to residential green spaces was associated with a reduced risk of childhood overweight and obesity [81]. Taken together, these results suggest that residential and neighborhood characteristics may be related to patterns of overweight, obesity, and micronutrient deficiencies, but further longitudinal research is needed to clarify the strength, direction, and causal nature of these relationships.

4.4. Strengths and Limitations

This study provides the first nationally representative estimates of DBM among Colombian schoolchildren, using standardized serum 25(OH)D measurements and multiple clinical thresholds. This approach allowed exploration of individual, household, and community factors simultaneously.

However, several limitations must be acknowledged:

• The cross-sectional design limits causal inference; associations should be interpreted as correlational.
• Some DBM categories (DBM1 and DBM2) are rare, leading to unstable logistic regression estimates; sparse data may bias odds ratios. Firth’s correction or penalized approaches may improve reliability.
• Potential residual confounding may persist due to unmeasured determinants of serum 25(OH)D, including dietary vitamin D intake, supplement use, seasonal or monthly variation in sampling, latitude- and altitude-related differences in ultraviolet exposure, air pollution, and skin pigmentation. These factors could contribute to variability in vitamin D status and partially explain the observed patterns of DBM.
• Vitamin D measurement using the ADVIA Centaur assay may underestimate deficiency compared to LC-MS/MS, the gold standard [46,47].

4.5. Implications for Policy and Practice

Our findings underscore the need for multifaceted interventions addressing both behavioral and structural determinants of DBM. Promoting physical activity and outdoor play, improving access to safe recreational spaces, and supporting parents in balancing child-rearing with other responsibilities are critical. Health promotion efforts should also consider monitoring micronutrient status, particularly vitamin D, in children with overweight or obesity. Interventions should be tailored to the social and environmental context of Colombian families and should address both behavioral risks and structural inequities.