Vitamin D levels and its influencing factors in children and adolescents in mainland China: a systematic review and meta-analysis

Abstract

Objectives

Childhood vitamin D deficiency is a public health issue. This study aims to systematically evaluate vitamin D nutritional status among children and adolescents in Mainland China through a quantitative analysis of literature, providing evidence-based strategies for prevention.

Design

This is a systematic review and meta-analysis, conducted following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

Data sources

A comprehensive search strategy was implemented across eight electronic databases (PubMed, Embase, Web of Science, Cochrane Library, CNKI, VIP, Wanfang and CBM) from inception to February 2024.

Eligibility criteria

We included cross-sectional studies that measured serum 25-hydroxyvitamin D levels and analysed influencing factors (eg, age, season, region) in healthy children and adolescents aged 0–18 years in Mainland China; studies reporting prevalence data of vitamin D deficiency/insufficiency based on standardised thresholds (deficiency <50 nmol/L, insufficiency 50–75 nmol/L) and using validated detection methods (eg, liquid chromatography-tandem mass spectrometry, chemiluminescence immunoassay, ELISA) were prioritised.

Data extraction and synthesis

Two independent reviewers systematically searched, screened and extracted data using predefined protocols. Study quality was assessed with the Agency for Healthcare Research and Quality (AHRQ) tool. Meta-analyses were performed using random-effects models in Review Manager V.5.3 and Stata V.16.0, with subgroup analyses by age, season and region. Sensitivity analysis and Egger’s test were applied to evaluate robustness and publication bias. Findings were synthesised through narrative summaries and quantitative pooling.

Results

The pooled prevalence of vitamin D deficiency among children and adolescents in Mainland China was 48% (95% CI: 40% to 57%), with extreme heterogeneity across studies (I² = 99.98%, p<0.001). Sensitivity analysis confirmed the stability of pooled estimates. Subgroup analyses revealed significant age-related declines (infants: 82.35 nmol/L vs adolescents: 50.98 nmol/L, p<0.05). Study quality assessed by AHRQ criteria showed 16% of included studies were high-quality (scores 8-11/11), 84% moderate (4-7). Evidence of publication bias was detected via Egger’s test (p<0.05) and funnel plot asymmetry.

Conclusions

The study highlights the widespread nature of vitamin D deficiency among children and adolescents in Mainland China, particularly in older children and during winter months. Effective interventions are necessary to address this issue. Future research should prioritise methodological standardisation to reduce heterogeneity and address potential publication bias.

PROSPERO registration number

CRD42023479183.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This systematic review and meta-analysis rigorously followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines and dual-reviewer protocols, covering over 90% of provinces in Mainland China to capture geographic diversity.

• The analysis was conducted using a prespecified random-effects model to account for anticipated heterogeneity across diverse methodologies and populations, with sensitivity analyses confirming the robustness of findings.

• Variations in vitamin D measurement methods (eg, Liquid Chromatography-Tandem Mass Spectrometry,liquid chromatography-tandem mass spectrometry vs Chemiluminescence Ie chemiluminescence immunoassay) across studies introduced potential comparability bias, despite standardised inclusion criteria.

• The exclusion of studies with sample sizes <5000, while improving statistical power, may limit insights into smaller population subgroups.

• Evidence of publication bias (Egger’s test p<0.05) suggests possible underrepresentation of small-scale studies reporting non-significant vitamin D deficiency rates, which could inflate pooled prevalence estimates.

Introduction

It is widely believed that vitamin D affects calcium-phosphate metabolism and contributes significantly to bone health in the human body.¹ For the crucial role of it in maintaining serum calcium and phosphate, its deficiency can have long-term influences on children, especially leading to rickets, which results in skeletal abnormalities, stunted growth, delayed development or poor growth.² However, more and more studies indicate that the role of it extends beyond bone health to influencing areas such as the nervous system, immune system, endocrine system, respiratory diseases, cardiovascular system and cancer.³ Recently, vitamin D deficiency remains common globally, with a particularly high prevalence among children.⁴

Serum vitamin D levels exhibit significant variations across different countries. Data from the National Health and Nutrition Examination Survey in the USA from 2011 to 2014 show that 10.98% of 3498 children aged 1–11 years had 25-hydroxyvitamin D (25(OH)D) levels <50 nmol/L, while 31.4% of 2355 individuals aged 12–19 years had that in range 25–50 nmol/L.⁵ A meta-analysis found that the overall occurrence of low vitamin D status was 18.46%, with a mean of 67.78 nmol/L.⁶ Furthermore, a study in Iran in 2018 found a prevalence of 31% (95% CI 30% to 31%) among children and adolescents.⁷ Even within the same country, there are differences in the nutritional status of vitamin D. For instance, literature reports that only 41.2% of 3630 children in Suzhou have sufficient vitamin D levels.⁸ Yang and Wu found that the vitamin D levels in children in northern China range from 40.0 nmol/L to 50.25 nmol/L, with a deficiency rate of 30–70%, while in southern China, the levels range from 52.0 nmol/L to 124.0 nmol/L, with a deficiency rate of 10–40%.⁹

Due to differences in research subjects, time, regions, diagnostic criteria and detection methods, studies published at different times and in different regions show obvious variation in the detection rates of it. Currently, there is a lack of comprehensive multicentre large-sample studies analysing the nutritional status of it in Chinese children. Therefore, this systematic review and meta-analysis aims to evaluate the vitamin D nutritional status among children and adolescents in mainland China. The population of interest comprises healthy individuals aged 0–18 years from various regions across mainland China.

Materials and methods

The present research was registered with PROSPERO (CRD42023479183), establishing a structured framework for our research. This registration aligns with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (online supplemental table S1).

Literature retrieval

A literature search was performed on PubMed, Embase, Web of Science, Cochrane Library and Chinese databases (CNKI, VIP, Wanfang and CBM) for articles published in English or Chinese, published up to 19 February 2024. The search terms included “Child,” “children,” “adolescents,” “Vitamin D,” “25-hydroxyvitamin D” and “25(OH)D,” with “China,” “People’s Republic of China,” and “Mainland China” as population limiters. This search encompassed subject terms, keywords and abstracts. The complete search strategy is provided in online supplemental table S2.

Vitamin D deficiency screening method: according to relevant Guidelines, Treatment, and Prevention of Vitamin D Deficiency released by an American organisation, serum 25(OH)D levels below 50 nmol/L indicate a vitamin D deficiency; 50–75 nmol/L denote insufficiency; levels above 75 nmol/L are regarded as adequate.¹⁰

Literature screening

Inclusion criteria: (1) the study subjects are healthy children and adolescents, aged under 18 years; (2) original articles employing either cross-sectional designs; (3) the research articles are written in either Chinese or English; (4) the primary outcome measure is serum 25(OH)D, with the secondary outcome measure being the occurrence rate of deficiency, regardless of the measurement methods used; (5) the study population consists of individuals from mainland China; (6) analysis of impacting factors; (7) the research should specify the study duration, location, the criteria used to determine vitamin D levels.

Exclusion criteria: (1) indices, commentaries, editorials; publications duplicated elsewhere, prioritising articles with the most comprehensive data related to this research; (2) absence of the full text; (3) studies with incomplete data or without relevant outcome measurement indicators; (4) for studies from the same region and year, preference is given to articles of higher quality; (5) studies with a sample size of less than 5000.

Two independent reviewers (PJ and ZL) screened titles/abstracts and assessed full texts based on predefined inclusion/exclusion criteria. Discrepancies were resolved through discussion, and unresolved cases were adjudicated by a third reviewer (BC).

Literature quality assessment

Two independent reviewers (WQ and YS) assessed study quality using the 11-item Agency for Healthcare Research and Quality (AHRQ) tool,¹¹ with responses being ‘yes,’ ‘no’ or ‘unclear.’ A ‘no’ or ‘unclear’ response scores 0 points. A total score of 0–3 points means low quality, 4–7 points and 8–11 points, respectively, means moderate and high quality. Initial discrepancies in scoring were resolved through iterative discussions between the reviewers. For persistently conflicting items, a senior methodologist (BC) reviewed the full text and made final determinations based on AHRQ guidelines. Final quality scores were cross-validated against the inclusion criteria to ensure internal consistency. All studies scored ≥4 (moderate/high quality), and none were excluded based on quality alone. The certainty of evidence for the pooled prevalence of vitamin D deficiency was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. We assessed the risk of bias, inconsistency (I²), indirectness, imprecision (95% CI width) and publication bias. Evidence certainty was categorised as high, moderate, low or very low.

Data extraction

Two scholars extracted and cross-verified information. Discrepancies were resolved through forming a third party. For literature selection, we initially read the titles, excluded those clearly irrelevant to present research and then reviewed the abstracts and full texts for inclusion. Data extracted from the eligible studies included: (1) first author, year of study, region, type of study and quality assessment score; (2) sample size, age, gender and season of blood collection; (3) average levels of vitamin D, methods of measurement and influencing factors, the prevalence of vitamin D deficiency and insufficiency, with deficiency defined as serum 25(OH)D <50 nmol/L, insufficiency as 50–75 nmol/L and adequacy as >75 nmol/L; (4) vitamin D levels were reported in various units across studies, including ng/mL, μg/L and nmol/L. To ensure consistency, all measurements were converted to nmol/L using the conversion factor 1 ng/mL=2.5 nmol/L.

Statistical analysis

Meta-analysis, subgroup analysis and sensitivity analysis were conducted using Review Manager V.5.3 and Stata V.16.0 software. Based on prior evidence of methodological and population diversity in paediatric vitamin D studies across Mainland China, we prespecified the use of a random-effects model for all analyses to conservatively account for expected heterogeneity. For the meta-analysis of average serum vitamin D levels, we pooled the mean values from the included studies using a random-effects model, weighted by the inverse of the variance. For the prevalence of vitamin D deficiency, we pooled the proportions using a random-effects model. To validate this approach, sensitivity analyses were conducted using a fixed-effect model. Additionally, a one-study-removed sensitivity analysis was performed to assess the robustness of the pooled prevalence estimates. Heterogeneity was quantified via I² statistics for descriptive purposes. Publication bias was assessed through visual inspection of funnel plots and quantified using Egger’s linear regression test. To address potential publication bias identified by funnel plot asymmetry, a trim-and-fill analysis was performed using Stata V.16.0. Subgroup analyses were predefined to explore heterogeneity across key demographic and methodological variables, including age (categorised as infants (<1 year), toddlers (1–2 years), preschoolers (3–5 years), school-age children (6–12 years) and adolescents (13–18 years)), gender, season of blood collection (spring, summer, autumn, winter), geographic region (north vs south of the Qinling-Huaihe Line) and vitamin D quantification methods (liquid chromatography-tandem mass spectrometry (LC-MS/MS), chemiluminescence immunoassay (CLIA), electrochemiluminescence immunoassay (ECLIA), enzyme-linked immunosorbent assay(ELISA), fluorescence immunoassay (FIA)). Meta-regression analyses were performed to evaluate the association between study publication year and pooled prevalence estimates. The significance level was set at 0.05.

Patient and public involvement

No patients or the public were directly involved in the design, conduct or interpretation of this systematic review and meta-analysis. As the study synthesised existing published data, public involvement was not required for data collection or analysis.

Results

Literature search results

An initial comprehensive search included 4199 studies. Following a thorough screening, 31 researches were found to meet the inclusion criteria.12,42 Of 31 studies, 26 (83.9%) were moderate quality (AHRQ 4–7), 5 (16.1%) high quality (AHRQ 8–11) and none low quality (AHRQ 0–3). The AHRQ scoring table is presented in online supplemental table S3. Figure 1 illustrates the selection process. This study covered a decade (2011–2021) and included over 90% of the provinces in Mainland China. The cohort comprised 877 419 children and adolescents, aged 0–18 years. The population’s mean vitamin D level was 71.65 nmol/L (95% CI 68.24 to 75.06). The ELISA method was the most prevalent for measuring vitamin D levels, used in 35.48% (11 out of 31) of the studies. Most studies identified age and seasonal variations as the primary factors influencing vitamin D levels. Detailed features of these studies were listed in online supplemental table S4.

Figure 1. The procedure of selecting studies on vitamin D levels in objects.

Pooled prevalence of vitamin D deficiency/insufficiency in Chinese children and adolescents

Heterogeneity testing was conducted on the 31 included studies, showing obvious heterogeneity among the researches with I²=99.98%, p<0.001. Hence, a random-effects model was used for analysis. And it was found that the pooled prevalence of vitamin D deficiency/insufficiency in children and adolescents in China is 48% (95%CI: 40% to 57%), with the forest plot as figure 2.

Figure 2. Forest plot of the pooled prevalence of vitamin D deficiency/insufficiency among children and adolescents in China, based on a random-effects meta-analysis.

Subgroup analysis

To discern the nuanced differences in Vitamin D concentrations across various subgroups of children and adolescents in Mainland China, our meta-analysis stratified the data accordingly (table 1). The summarised rates of concentration varied widely, with a marked increase noted from infancy to adolescence: infants had an average concentration of 82.35 nmol/L (95% CI: 77.86 to 86.83), while adolescents showed a significantly lower average concentration of 50.98 nmol/L (95% CI: 43.53 to 58.44), indicating a statistically significant age-related decline (p<0.05).

Table 1. Subgroup analysis of vitamin D concentration among objects in Mainland China.

Grouping method	Number of studies	Heterogeneity test		Effect model	Meta-analysis (95% CI)	P value (inter-subgroup)
		I2 (%）	P value
Gender
Male	28	99.9	＜0.05	Random	72.29 nmol/L(67.77 to 76.80)	0.76
Female	28	99.9	＜0.05	Random	71.19 nmol/L(66.59 to 75.79)
Season
Spring	18	99.9	＜0.05	Random	69.65 nmol/L(62.51 to 76.78)	0.44
Summer	18	99.9	＜0.05	Random	75.38 nmol/L(68.22 to 82.55)
Autumn	18	99.9	＜0.05	Random	76.07 nmol/L(68.47 to 83.67)
Winter	18	99.9	＜0.05	Random	66.47 nmol/L(60.23 to 72.71)
Age
Infants	21	99.8	＜0.05	Random	82.35 nmol/L(77.86 to 86.83)	＜0.05
Toddlers	21	99.9	＜0.05	Random	78.28 nmol/L(71.84 to 84.71)
Preschoolers	20	99.8	＜0.05	Random	62.57 nmol/L(58.68 to 66.46)
School-age children	10	99.8	＜0.05	Random	54.11 nmol/L(48.11 to 60.11)
Adolescents	6	99.4	＜0.05	Random	50.98 nmol/L(43.53 to 58.44)
Region
South	24	100	＜0.05	Random	70.36 nmol/L(65.74 to 74.98)	0.617
North	6	100	＜0.05	Random	76.70 nmol/L(63.50 to 89.90)
Methodology
LC-MS/MS	6	99.9	＜0.05	Random	73.02 nmol/L(69.81 to 76.24)	0.323
CLIA	5	99.9	＜0.05	Random	62.89 nmol/L(52.80 to 72.98)
ECLIA	6	100	＜0.05	Random	72.34 nmol/L(55.09 to 89.59)
ELISA	11	100	＜0.05	Random	71.38 nmol/L(64.89 to 77.88)
FIA	3	99.9	＜0.05	Random	83.1 nmol/L(72.47 to 93.72)

CLIA, chemiluminescence immunoassay; ECLIA, electrochemiluminescence immunoassay; ELISA, Enzyme-Linked Immunosorbent Assay; FIA, fluorescence immunoassay; LC-MS/MS, liquid chromatography-tandem mass spectrometry.

Differences in Vitamin D levels also appeared along gender lines, although to a lesser extent, with males averaging 72.29 nmol/L (95% CI: 67.77 to 76.80) and females averaging 71.119 nmol/L (95% CI: 66.59 to 75.79), though this disparity was not statistically significant (inter-subgroup p value=0.76).

When the data were segmented by season, minor variations were apparent, with the highest average concentration occurring in the summer at 75.38 nmol/L (95% CI: 68.22 to 82.55) compared with the winter at 66.47 nmol/L (95% CI: 60.23 to 72.71), suggesting a possible impact of seasonal changes on that. However, the absence of statistical significance (inter-subgroup p value=0.44) points to the need for more robust research to confirm any definitive seasonal patterns.

Considering geographical variations, our analysis indicated that there exists a slightly higher risk of vitamin D deficiency in the southern regions (70.36 nmol/L, 95% CI: 65.74 to 74.98) as opposed to the northern regions (76.70 nmol/L, 95% CI: 63.50 to 89.90). Nonetheless, these regional differences did not achieve statistical significance (inter-subgroup p value=0.617), suggesting that latitude-related sunlight exposure may not be the only factor influencing Vitamin D levels.

Moreover, methodological differences in assessing Vitamin D were substantial. Notably, the LC-MS/MS method reported an average concentration of 73.02 nmol/L (95% CI: 69.81 to 76.24), in contrast to the CLIA method, which indicated a considerably lower concentration of 62.89 nmol/L (95% CI: 52.80 to 72.98). These significant method-based differences (inter-subgroup p value=0.323) highlight the critical need for methodological standardisation in future research to ensure the comparability and reliability of findings.

The high degree of heterogeneity captured within our analysis (I²=100% across most subgroups) necessitated the adoption of a random-effects model to accommodate the substantial variability, implying the impact of additional, unexamined factors on Vitamin D insufficiency and deficiency beyond the subgroups analysed. Therefore, further investigation is essential, potentially including meta-regression or additional subgroup distinctions, to identify and elucidate these underlying factors. It is also vital to consider the potential biases and the implications of study quality and sample size in interpreting the results.

Meta-regression analyses of prevalence by study year

The regression result indicated that the year of study publication did not have obvious influence on the prevalence of it (p=0.525). Figure 3 displays the meta-regression of vitamin D deficiency/insufficiency prevalence (r) against publication year. Each circle represents an individual study, with circle size proportional to the study’s sample size (larger circles denote larger studies).

Figure 3. Meta-regression of vitamin D deficiency/insufficiency prevalence (r) by publication year. Circles represent individual studies, sized by sample weight (larger=larger sample size).

Sensitivity analysis results

Sensitivity analysis was conducted based on the one-study-removed approach, and the results indicated that the combined rates did not change obviously, suggesting good stability of the outcomes.

Publication bias

Significant funnel plot asymmetry was observed (figure 4), corroborated by Egger’s test (p<0.05). The trim-and-fill analysis imputed 12 hypothetical missing studies (online supplemental figure S1). After adjustment, the pooled prevalence of vitamin D deficiency decreased to 33% (95% CI: 25% to 40%), suggesting that the original estimate of 48% (95% CI: 40% to 57%) may have been influenced by publication bias favouring studies reporting higher deficiency rates.

Figure 4. Funnel plot analysis of vitamin D deficiency/insufficiency rates.The x-axis represents the prevalence rate ("r").

GRADE evidence quality assessment

The GRADE assessment for the primary outcome (pooled prevalence of vitamin D deficiency) indicated low certainty of evidence, downgraded by two levels (table 2).

Table 2. GRADE evidence profile for the pooled prevalence of vitamin D deficiency.

Domain	Assessment	Rationale
Risk of bias	Moderate	83.9% (26/31) studies had moderate quality (AHRQ 4–7); 16.1% (5/31) high quality (AHRQ 8–11).
Inconsistency	Serious	Extreme heterogeneity (I² = 100%, p<0.001).
Indirectness	Not serious	All studies focused on children/adolescents in Mainland China.
Imprecision	Not serious	Narrow 95% CI (40% to 57%) excludes clinically irrelevant ranges.
Publication bias	Likely	Significant funnel plot asymmetry (Egger’s test p<0.05).
Overall certainty	Low	Downgraded by two levels due to inconsistency and publication bias.

AHRQ, agency for healthcare research and quality; GRADE, grading of recommendations assessment, development and evaluation.

Discussion

Children have a high metabolic rate, requiring additional energy for growth. A typical diet may not suffice to meet daily vitamin D requirements.³⁷ Given the profound impact of vitamin D status on children, our primary focus is on assessing that in this demographic. Moreover, vitamin D has significant biological activity. Beyond maintaining calcium and phosphorus metabolism balance, it is related to cellular proliferation and apoptosis. It influences neuromuscular and immune system functions and is pivotal in children’s growth and development.⁴³ Vitamin D deficiency in the human body may cause not only a disruption in calcium and phosphorus levels, risking rickets in children, but it also correlates with multiple health risks. These risks encompass respiratory diseases, immune dysfunctions, gestational diabetes in pregnant women and, in adults, chronic conditions such as osteoporosis and cardiovascular diseases.⁴⁴

A systematic review in Southeast Asia reported a vast range of vitamin D insufficiency, with concentrations below 50 nmol/L occurring in 0.9–96.4% of the population and noted that half of the newborns suffer from a deficiency.⁴⁵ In our synthesis of 31 studies,12,42 which included 877 419 children, the mean serum 25(OH)D level was 71.65 nmol/L (95% CI: 68.24 to 75.06), and the combined rate of deficiency or insufficiency was 48% (95% CI: 40% to 57%). This underscores that vitamin D levels among children are suboptimal on a broad scale. The pervasive issue of that is evidently a significant challenge for the healthy development of children and adolescents in our nation.

The emergence of widespread deficiency is attributed to a complex interplay of factors. The expansive territory of China features regions where limited sunlight exposure is common, due to geographic, climatic and environmental concerns, including air pollution.⁴⁶ Rapid urbanisation and a shift in eating habits have contributed to fewer outdoor activities and a growing dependence on processed foods, leading to subpar nutritional consumption.⁴⁷ Additionally, the economic divide means that children from less affluent backgrounds often have limited access to vitamin D-fortified products or supplements, intensifying their risk of deficiency.⁴⁸ Moreover, the intense focus on scholastic achievement adds to this issue by promoting sedentary lifestyles and curtailing time spent outdoors.

The analysis reveals a significant decline in vitamin D levels with advancing age. Infants exhibit an average level of 82.35 nmol/L (95% CI: 77.86 to 86.83), which decreases to 50.98 nmol/L (95% CI: 43.53 to 58.44) during adolescence, aligning with prior research.³⁸ This decrease is likely due to the increased vitamin D requirements during growth spurts, coupled with reduced outdoor exposure, which is crucial for natural synthesis of vitamin D through sunlight.⁴⁹ The notably higher serum contents of 25(OH)D in infants are primarily due to two factors. First, it is standard practice in China for full-term infants to start prophylactic vitamin D intervention of 400 IU/day 2 weeks post birth (preterm infants or twins begin with 800 IU/day 1 week after birth), which continues until the age of two.⁵⁰ Second, there is the transplacental transfer of vitamin D from the fetus, especially in the later stages of pregnancy, enabling mothers to deposit vitamin D for their newborns.⁵¹ While this study observes no significant gender disparity in serum 25(OH)D deficiency occurrence, other research^{14 18} notes higher levels in boys compared with girls. The discrepancy may arise from differences in physical activity and outdoor time, with boys generally being more active and having more sun exposure and contemporary parental practices which often involve greater sun protection for girls, potentially resulting in higher vitamin D levels in boys within some regions.

The concentration of 25(OH)D varies in different latitudinal regions with the changing duration. In Wenling City, a notable variation was observed in the serum levels of 25(OH)D, with higher levels in summer than winter and spring.³⁰ A study in Shizuishan City, Ningxia, involving 1103 children with ages of 2–6 years, reported vitamin D deficiency rates of 41.16% during winter and spring, which plummeted to 2.52% in the warmer seasons.⁵² Moreover, serum 25(OH)D contents in 290 normal infants and toddlers aged 0–48 months from Shizuoka or Tokyo were significantly higher in summer and autumn.⁵³ Although the seasonal difference had no obvious significance (p=0.44), the highest average level observed in summer was 75.38 nmol/L (95% CI: 68.22 to 82.55), with the lowest in winter at 66.47 nmol/L (95% CI: 60.23 to 72.71). Skin synthesis through photoreaction is a critical source of vitamin D.⁵⁴ This synthesis fluctuates seasonally, due to varying daylight durations and the differential skin exposure influenced by clothing.⁵⁵ Additionally, climatic conditions such as diminished solar radiation during overcast, rainy or smoggy days, and factors like air quality, altitude and outdoor activity levels, act as potential modifiers of serum 25(OH)D concentrations.⁵⁶

While Liu et al⁵⁷ reported a high prevalence of vitamin D inadequacy in Chinese populations (46.8% in children/adolescents and 63.2% in adults), our study provides novel insights through exclusive focus on 0–18 year-olds and geographically comprehensive sampling (90% of provinces). Unlike prior work, we identified significant seasonal variations (eg, winter nadir at 66.47 nmol/L) and methodological inconsistencies (LC-MS/MS vs CLIA differences), which were not addressed in national assessments. Additionally, our stratified analysis of infants vs adolescents offers actionable evidence for age-specific supplementation guidelines.

These findings further explain why detection methods (eg, LC-MS/MS vs CLIA) showed no significant effect in our study, as methodological heterogeneity itself emerged as a key confounder. There were observed discernible differences in vitamin D levels when various detection methods were used. Despite an overall p value of 0.323, indicating a lack of statistical significance, the variations are nonetheless noteworthy. The mean level of vitamin D as determined by LC-MS/MS was 73.02 nmol/L, compared with a lower average of 62.89 nmol/L obtained via the CLIA method. ECLIA provided an average of 72.34 nmol/L, ELISA yielded 71.38 nmol/L and FIA reported the highest average level at 83.1 nmol/L. Such variations may reflect the inherent differences in sensitivity and specificity across different technologies. Although these differences have no obvious significance, they underscore the importance of considering the impact of the chosen technique on study outcomes. Previous studies found that disparate vitamin D detection methods can lead to varying results,^{58 59} which holds significant clinical implications for diagnostic standards and public health monitoring. Future work should advocate for the use of uniform, validated methods to enhance comparability of results across different studies.

This research finds no obvious difference in the occurrence of vitamin D deficiency in children between the northern and southern regions (p=0.617). However, regional differences in public health strategies may still be relevant. Previous research suggests that due to shorter sunlight exposure and lower ultraviolet radiation, the incidence of that tends to be higher in northern areas.⁶⁰ Nonetheless, these studies may not have fully accounted for other factors affecting vitamin D levels, like dietary intake, supplement use and lifestyle differences. Moreover, methodological differences between studies might contribute to inconsistencies in results. Therefore, further research should consider these potential influencing factors and use standardised measurement methods to explore differences between regions. In particular, considering that air pollution might block ultraviolet radiation, thus affecting vitamin D synthesis, it is necessary to delve deeper into the association between environmental factors and vitamin D deficiency.⁶¹

While aiming to comprehensively analyse data on the levels of deficiency in children in Mainland China and its influencing factors, it is crucial to consider inherent limitations. First, despite including a large number of studies (31 studies involving 877 419 children), there was a high degree of heterogeneity among them (I²=99.98%) due to differences in study design and measurement methods. Therefore, we used a random-effects model for analysis, but caution is still needed in interpreting the results. Sources of heterogeneity may be diverse, including differences in sunlight exposure, lifestyle, dietary habits and methods to detect vitamin D. Second, although we attempted to explore potential factors influencing vitamin D levels through subgroup analysis (age, gender, season and geographic location), we did not find significant statistical differences between different subgroups, suggesting other unconsidered influencing factors. Additionally, our study did not cover all possible factors affecting vitamin D levels, such as outdoor activity levels, use of relevant supplements and dietary patterns. Third, trim-and-fill analysis indicated that publication bias may have inflated the original prevalence estimate. This implies a potential under-representation of studies reporting lower deficiency rates, possibly due to selective publication of high-prevalence findings. Despite this adjustment, the prevalence remains clinically significant, highlighting the need for targeted interventions. While the GRADE framework highlighted low certainty of evidence, this reflects challenges inherent to synthesising real-world observational studies with methodological diversity (eg, LC-MS/MS vs ELISA). Importantly, our conclusions align with prior national surveys in China,⁵⁷ suggesting that the pooled estimates are clinically plausible despite statistical heterogeneity.

In summary, our study results provide useful information on the vitamin D status of children in mainland China, but the above limitations must be considered when interpreting and applying these findings. Future research should adopt more standardised methods, explore more possible influencing factors and conduct more targeted studies in different regions and populations.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.