Importance of vitamin D in critically ill children with subgroup analyses of sepsis and respiratory tract infections: a systematic review and meta-analysis

Background Critical care and sepsis remain high priority concerns in children. Observational studies report high prevalence of vitamin D deficiency and present mixed results regarding the correlation between vitamin D status and adverse outcomes. Associations between deficiency and mortality, particularly in children with sepsis, remain unclear. We performed a systematic review and meta-analysis to address this uncertainty. Methods PubMed, OVID and Google Scholar were searched for observational studies in critically ill children. We obtained pooled prevalence estimates for vitamin D deficiency and odds ratios for the association of mortality in critically ill children treated in intensive care units, with subgroup analysis for those with sepsis and those with respiratory tract infections. Meta-regression and sensitivity analyses were used to investigate heterogeneity. Findings Forty-eight studies were included. Total sample size was 7,199, with 1,679 (23%) children acting as controls in case-control studies. Of 5,520 critically ill children, 2,664 (48%) were vitamin D deficient (< 50 nmol/L). Results of the random effects model demonstrated a pooled prevalence of vitamin D deficiency of 54·9% (95% CI 48·0-61·6, I2=95·0%, 95% CI 94·0-95·8, p < 0·0001). In subgroup analysis of children with sepsis (16 studies, 788 total individuals) we observed higher prevalence of deficiency (63·8%, 95% CI 49·9-75·7, I2=90·5%, 95% CI 86·2-93·5%, p < 0·0001). In patients admitted for respiratory tract infections (24 studies, 1,683 total individuals), prevalence was 49·9% (95% CI 37·6-62·2; I2= 93·9%, 95% CI 92·1-95·3, p < 0·0001). Only one identified study assessed vitamin D levels in sepsis and mortality. A meta-regression model with all available variables (year of publication, total study sample size, quality score, study design, country group and clinical setting) explained 37·52% of I2(F = 5·1119, p = 0·0005) with clinical setting and country groups being significant predictors for prevalence. Meta-analysis of mortality (18 studies, 2,463 total individuals) showed an increased risk of death in vitamin D deficient critically ill children both with random (OR 1·81, 95% CI 1·24-2·64, p-value = 0·002) and fixed effects (OR 1·72, 95% CI 1·27-2·33, p= 0·0005) models with low heterogeneity (I2= 25·7%, 95% CI 0·0-58·0, p = 0·153) and low evidence of publication bias (p = 0·084, Egger’s test). There were insufficient studies to perform meta-analyses for sepsis and respiratory tract infection related mortality. Interpretation Circulating vitamin D deficiency is common amongst critically ill children, particularly in those with sepsis. Our results suggest that vitamin D deficiency in critically ill children is associated with increased mortality. Clinical trials, studies with larger sample sizes and standardized approaches are needed to further assess associations between circulating levels of vitamin D and mortality and other outcomes in the paediatric population. Funding Medical Research Council UK Registration PROSPERO (CRD42016050638) Copyright Open access article under terms of CC BY Research in context Evidence before this study Vitamin D deficiency is common worldwide and has been associated to numerous diseases in observational studies. The extent of deficiency and relevance to mortality in children receiving acute and intensive care is unclear and only recently has gained more attention. We searched PubMed, OVID, Google Scholar and the Cochrane Library from inception up until 5thNovember 2017 without language restrictions. Search terms used across these databases included: “critical care”, “acute care”, “vitamin D”, “pediatric”, “child”, “neonate”, “toddler”, “intensivecareunit”, “sepsis” and “septic shock” (full search terms are listed in the appendix). Most (81·3%) included studies were published between 2014 and 2017. We did not identify randomised clinical trials assessing the value of vitamin D supplementation in this population. We did not find sufficient studies to perform meta-analyses for mortality from sepsis or respiratory tract infection. Added value of this study Our systematic review and meta-analysis provides an in-depth assessment of the magnitude and relevance of vitamin D circulating levels in paediatric acute and critically ill patients with pre-specified sub-group analyses. We found that studies were highly heterogeneous across a number of important study variables including clinical setting, patient age groups, sample size, geographic location, case definitions, study quality, study design, biomarker thresholds and assay measurements. Pooled estimates of prevalence of vitamin D deficiency were overall extremely high, showing that around half of patients in general and acute paediatric care are deficient. Estimates were similar for respiratory tract infections but higher in sepsis, with overlapping confidence intervals across all outcomes. Individual study estimates were highly variable however. We analysed this heterogeneity using meta-regression and identified clinical setting and country of study as important contributors, likely indicating that patient age and environmental exposure to vitamin D, amongst other likely important factors, are key determinants and should be adequately assessed and reported. Pooled estimates for mortality outcomes showed a clear increased risk with lower vitamin D levels, despite the variation in study characteristics. We were unable to assess the importance of vitamin D levels in sepsis and respiratory tract infections due to the small number of studies investigating these outcomes. Implications of the available evidence Vitamin D deficiency in acute and critical care settings is common and associated with increased mortality in paediatric patients. Our review highlights the heterogeneous nature of the study population however and emphasizes the need for adequate power and control of confounding factors in future work. Few studies have investigated specific diseases such as sepsis and respiratory tract infections in relation to vitamin D despite their high prevalence, social and economic costs. Understanding the causal nature and therapeutic value of vitamin D in paediatric critical care remain key areas for investigation.


Summary Background
Critical care and sepsis remain high priority concerns in children. Observational studies report high prevalence of vitamin D deficiency and present mixed results regarding the correlation between vitamin D status and adverse outcomes. Associations between deficiency and mortality, particularly in children with sepsis, remain unclear. We performed a systematic review and metaanalysis to address this uncertainty.

Methods
PubMed, OVID and Google Scholar were searched for observational studies in critically ill children. We obtained pooled prevalence estimates for vitamin D deficiency and odds ratios for the association of mortality in critically ill children treated in intensive care units, with subgroup analysis for those with sepsis and those with respiratory tract infections. Meta-regression and sensitivity analyses were used to investigate heterogeneity.
In patients admitted for respiratory tract infections (24 studies, 1,683 total individuals), prevalence high, showing that around half of patients in general and acute paediatric care are deficient.
Estimates were similar for respiratory tract infections but higher in sepsis, with overlapping confidence intervals across all outcomes. Individual study estimates were highly variable however.
We analysed this heterogeneity using meta-regression and identified clinical setting and country of study as important contributors, likely indicating that patient age and environmental exposure to vitamin D, amongst other likely important factors, are key determinants and should be adequately assessed and reported. Pooled estimates for mortality outcomes showed a clear increased risk with lower vitamin D levels, despite the variation in study characteristics. We were unable to assess the importance of vitamin D levels in sepsis and respiratory tract infections due to the small number of studies investigating these outcomes.

Implications of the available evidence
Vitamin D deficiency in acute and critical care settings is common and associated with increased mortality in paediatric patients. Our review highlights the heterogeneous nature of the study population however and emphasizes the need for adequate power and control of confounding factors in future work. Few studies have investigated specific diseases such as sepsis and respiratory tract infections in relation to vitamin D despite their high prevalence, social and economic costs. Understanding the causal nature and therapeutic value of vitamin D in paediatric critical care remain key areas for investigation.

Introduction
Vitamin D is an essential nutrient 1, 2 representing a group of fat soluble secosteroids with key endocrine functions. 3 It is synthesized in the skin upon sunlight exposure 4 while dietary sources, such as oily fish, egg yolk, certain fungi and supplements, are usually secondary sources. Vitamin D is critical in bone metabolism 5 and calcium homeostasis, 6 as well as acting as an important regulator in extra-skeletal metabolic processes, 7 cardiovascular and immune systems. 8 Many observational and laboratory studies have observed the anti-inflammatory properties of vitamin D, 9 including direct regulation of endogenous anti-microbial peptide production. 10 It is therefore crucial for humans to have sufficient vitamin D levels to maintain bone health and possibly improve response to infection. 6,11,12 Infants and children are especially dependent on vitamin D to achieve healthy bone development and growth. 13,14 Well-known functional outcomes of adequate vitamin D levels in children include rickets prevention, higher bone mineral content and reduced bone fracture rates. 5,14 In otherwise healthy children in the United States, the reported prevalence of vitamin D deficiency (25OHD levels of < 25 nmol/L) ranges from 9 to 18%. 15 The Endocrine Society Clinical Practice Guidelines and the Institute of Medicine (IOM) suggest that vitamin D levels less than 50 nmol/L (20 ng/mL) reflect a deficient state. 4,16 Studies in adults reflect a high prevalence of vitamin D deficiency both in general intensive care unit (ICU) and sepsis patients and strongly suggest an association between low vitamin D and poor clinical outcomes, including increased mortality, particularly in those suffering from sepsis. 2,17 Recent clinical trials of vitamin D supplementation in adults appear promising in both general critical care 18,19 and sepsis. 20 Sepsis remains a challenging clinical entity with high social and economic costs. 21 Each year there are approximately 123,000 sepsis cases and around 37,000 deaths in England alone. 22 Recent reports show an increased prevalence of paediatric sepsis, 23 likely a reflection of an increased population with chronic comorbidities, higher rates of opportunistic infections and multidrug resistant organisms. 24 Respiratory tract infections account for a large proportion of underlying diagnoses in acute and critical care conditions 24, 25 but remain understudied. 26 The magnitude and relevance of vitamin D deficiency in children receiving acute care is not clear.
Several recent studies have addressed these questions with mixed results. We sought to summarise the evidence regarding the implications of vitamin D deficiency and its prevalence in general ICU, respiratory tract infection and sepsis patients in the paediatric population. We carried out a systematic review and meta-analysis of circulating vitamin D levels to assess the prevalence of vitamin D deficiency (≤ 50 nmol/L) and its association with mortality in these conditions.

Methods
We planned and conducted our systematic review and meta-analysis according to the PRISMA guidelines 27 and since we did not include randomized controlled trials we reported following the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) guidelines. 28

Search strategy and selection criteria
Our population of interest consists of paediatric patients with acute conditions and/or treated in ICU or emergency units for acute conditions whose vitamin D status was assessed prior to or during admission. We included published cross sectional, case-control and cohort studies that measured circulating 25(OH) D levels and either reported prevalence, odds ratios (OR) or data to enable calculation. Studies were excluded if they were reviews, case reports, surveys, commentaries, replies, not original contributions, experimental in vitro or if they recruited patients who were not treated in emergency, neonatal intensive care units (NICUs), paediatric intensive care units (PICUs) or for acute conditions. Studies were also excluded if they only enrolled vitamin D deficient patients, investigated healthy populations only or did not measure circulating 25(OH) D levels as an indicator of vitamin D status. When we identified more than one publication utilising the same cohort, we included the publication which shared our review's objective to investigate vitamin D levels and prevalence of deficiency.
For purposes of our review, we classified vitamin D deficiency as being less than 50 nmol/L (equivalent to 20 ng/mL) as suggested by the IOM. 16 Different age categories were used to designate patients as "children" in the studies reviewed. We therefore included all "children" as defined by each treating facility and this included "neonates", "infants", "toddlers", "children" and "adolescents".
We searched PubMed, OVID, Google Scholar and the Cochrane Library from inception up until 5 th November 2017, with no language restrictions. Search terms used across these databases included: "critical care", "vitamin D", "pediatric", "child", "neonate", "toddler", "intensive care unit", "sepsis" and "septic shock". Search terms used in OVID and PubMed are listed in the Additional Tables 1A and 1B. Literature searches were performed by two investigators independently and included initial screening of titles and abstracts, followed by full text screening.
Any disagreements for study eligibility were resolved by discussion between the authors.
Reference lists of the selected papers, including reviews, were also checked for relevant titles.
Abstracts of relevant titles were then assessed for eligibility. A data extraction form was designed a priori.

Study quality assessment
The quality of each included study was assessed using the Newcastle-Ottawa Scale (NOS) for cohort, case-control and cross-sectional study designs. 29 We classified studies as low (1-3), medium (4-6) or high quality (7-9) for purposes of sensitivity analysis (Additional Table 2 A, B and C).

Prevalence and mortality outcomes
In the majority of studies (n = 36), prevalence of vitamin D deficiency was extracted as reported with a threshold of ≤ 50 nmol/L. If prevalence was not reported directly, it was calculated using data provided in each study (cases ≤ 50 nmol/L / total number of study participants, (Additional Table 3 A and B). Extracted or calculated prevalence values were then combined in a metaanalysis. For mortality, we calculated unadjusted odd ratios (OR) as: OR = (vitamin D deficient patients who died * vitamin D non-deficient patients who did not die)/ (vitamin D deficient patients who did not die * vitamin D non-deficient patients who died) We had sufficient information to calculate ORs < 50 nmol/L for 36 studies (75%). For the 12 studies with insufficient information, we used the lower cut-off values reported as a conservative approximation (Additional Table 4). We converted 25(OH) D values using: nmol/L = ng/mL * 2·496.

Data analysis
We obtained proportions of vitamin D deficiency with 95% confidence intervals (CI) using the Clopper-Pearson method 30 in R. We used a random effects model 31 to account for the variation observed within and between studies due to the different ages and acute conditions in the populations considered. For mortality, we also obtained pooled proportions and pooled ORs with fixed effect model for sensitivity analysis and to avoid false conclusions that could result from small-study effects. 32 We investigated possible sources of heterogeneity using sensitivity and subgroup analyses. The I 2 statistic was used to estimate the percentage of total variation across studies which can be attributed to heterogeneity. A Q value of < 0·05 was considered significant and an I 2 statistic greater or equal to 75% indicated a high level of variation due to heterogeneity. 33,34 We used Egger's regression test to present results for publication bias and funnel plot asymmetry 35 and generated funnel plots for visual assessment and screen for evidence of publication bias.
To further assess heterogeneity, we utilised meta-regression to identify predictor variables that could explain variation in study prevalence estimates. We used restricted maximum likelihood (REML) estimations in the model to account for residual heterogeneity 36 and the Knapp-Hartung method to adjust confidence intervals and test statistics. This method estimates between study variance using a t-distribution, rather than a z-distribution, yielding a more conservative inference. 37 We tested the following continuous predictors: year of study publication, total sample size and quality score. Categorical variables included study setting (PICU, NICU), study design (case-control, cross-sectional and cohort) and country group by geographic region and economic development (group 1, group 2, and group 3) and were dummy coded.
We used R version 3·5·0 and Microsoft Excel 2010 for analyses and data collection. The R packages "meta" 38 and "metafor" 39 were used for analyses. Only results of the random effects model are reported for prevalence due to the expected heterogeneity between populations being considered. Our protocol is registered in PROSPERO (CRD42016050638).

Role of the funding source
The study received funding from the UK Medical Research Council. The funders had no role in data collection, analysis, interpretation or writing of the report. All authors had access to the data in the study.

Screening and study characteristics
After title and abstract screening, we identified 2,890 potentially relevant studies ( Figure 1) and eighty-five full text articles were assessed for eligibility. Rationale for study exclusion included: studies including adults, study populations other than critically ill children or with acute conditions, studies of circulating vitamin D levels and deficiency in healthy children or in children with chronic conditions. Four studies [40][41][42][43] were excluded due to insufficient data reporting (Additional Table 5). We also excluded three studies 44-46 that used the same cohort of children and included a single study to represent the cohort. 47 Ultimately, 48 studies met criteria for inclusion (Additional Table 6).

Figure 1 Flow chart of study selection process
The primary objective of the majority of studies was to determine circulating vitamin D concentration ("status") in children and/or prevalence of vitamin D deficiency. Secondary objectives included investigation of associations between deficiency of circulating vitamin D and various outcomes, such as hospital mortality length of stay, requirement of ventilation and/or illness severity (Additional Table 7).
All included studies reported vitamin D measurement assay methods used (Additional Table 8) and stated that samples were collected and analysed within the first 24 hours of hospital admission.
Studies reported ethical approval and consent for participation from parents or guardians (Additional Table 9). Included studies were published between 2004 and 2017, with the majority (n = 39, 81·3%) published between 2014 and 2017 (Additional Table 6). In total, 5,520 children were hospitalized in paediatric or neonatal intensive care units or emergency units. Sample sizes of critically ill children ranged from 25 48 to 511. 49 In 16 studies the total number of cases was greater than 100.
Studies originated from 15 countries, with the majority from India 8, 50-58 (n = 10) or Turkey 48, 59-64 (n = 7) (Additional Table 6). All were of medium or high quality (NOS score median 6•5, range 4-8). The score range for cohort studies was 6 to 8 (n = 20), for case-control studies 5 to 8 (n = 24) and for cross sectional 4 to 6 (n = 4). Studies used a broad range of ages to classify patients as "children". Six studies (12·5%) 48, 60, 62-65 included only neonates. In two 60, 65 of these six studies, neonates were preterm. The largest age range was seen in the study of Ayulo et al 2014, which included individuals between 1 and 21 years of age (Additional Table 10).
All studies included both female and male participants. For mortality, four of the 18 studies (22%) carried out multivariate regression analysis with adjustment for confounders. The remaining studies presented results using a variety of methods, including Spearman's correlation analysis, chi-square or Fisher's exact tests or descriptive statistics.
When examining results by sample size (defining "large" as ≥ 100 and "small" as < 100), we found that the 16 8 Table 13).
Variation attributable to heterogeneity was still high in the three subgroups (I 2 > 90%).

Post-hoc investigation to determine sources of heterogeneity
To investigate the substantial heterogeneity observed in prevalence estimates, we incorporated study-specific characteristics (year of publication, total study sample size, quality score, study design, country group and clinical setting) as covariates in a random effects meta-regression model. We identified clinical setting and country groups as significant predictors (Figure 3). We found that the model fitted with all available covariates can explain 37·52% of I 2 with F = 5·1119, p = 0·0005 (Additional Table 14). We also conducted univariate meta-regressions for each of the six predictors (Additional Figure 4).

Figure 3 Bubble plots of univariate meta-regressions. Each study is represented by a circle. Predictor variables;
A clinical setting and B country groups are shown on the x-axis and the effect measure logit transformed proportion shown on the vertical (y-axis). NICU = Neonatal Intensive Care Unit; PICU = Pediatric Intensive Care Unit; grp = country group; country group 1 = USA, Chile, Australia, Canada, Ireland, Japan, Spain; country group 2 = South Africa, China, Egypt, Iran, Turkey, Saudi Arabia; and country group 3 = Bangladesh, Thailand, and India
There was no material change in prevalence estimates when analysed according to study design.

Figure 5 Pooled odds ratio (OR) of risk of mortality in vitamin D deficient versus vitamin D non-deficient critically ill children.
Forest plot shows result from the random effects model. Diamond represents the overall OR (with corresponding 95% Confidence Interval). Each square shows the odds ratio of each study and the horizontal line across each square represents the 95% confidence interval (CI) of the estimate.

Mortality in patients with sepsis and respiratory tract infections
We were unable to identify a sufficient number of studies assessing vitamin D and mortality for meta-analysis in individuals with sepsis. Three studies 8, 58, 60 measured vitamin D levels in paediatric patients with sepsis. One study 8 assessed mortality and did not find a significant association in children from 1 to 12 years with sepsis (n=124). None of the studies with children admitted for respiratory tract infections looked at the association of vitamin D deficient versus vitamin D not deficient children with mortality.

Discussion
Vitamin D deficiency is highly prevalent worldwide, even in countries with abundant sunshine.
Studies demonstrated high prevalence of vitamin D deficiency in otherwise healthy children from high-income countries (9 to 24%) but also from middle and low-income countries in Indian subcontinent (36 to 90%). 8 We identified 48 studies representing a total of 5,520 children treated in ICU or emergency units for acute conditions who had blood vitamin D levels measured close to or upon admission. Our analysis shows that prevalence of vitamin D deficiency is high (range 5% 66 to 95% 54 ) across ICU and emergency units in the paediatric population, particularly in individuals with sepsis.
Importantly, our analysis showed a significantly increased risk of mortality in critically ill children with vitamin D deficiency. We carried out several analyses for sensitivity including fixed effects models, by study design, country group, age and sample size and found consistent results. Although sepsis is a leading cause of paediatric mortality and morbidity worldwide, 82 we found few studies assessing the relationship between vitamin D status and mortality in this population.
We were unable to identify sufficient studies including patients with sepsis to perform a metaanalysis of vitamin D status and mortality. Sepsis remains an area of unmet need with high social and financial costs. 24 Diagnostic criteria, 83 a lack of adequate biomarkers 84 and targeted treatment remain important challenges in research on sepsis. We did not find studies that assessed the risk of mortality in relation to vitamin D deficiency in children admitted for respiratory tract infections either.
Strengths of our review include the large number of studies and large total sample size, allowing a high-powered investigation to identify meaningful associations. For our systematic review and meta-analysis, we followed pre-specified eligibility criteria and used the PRISMA 27  vitamin D deficiency and mortality could be due to reverse causation and future studies will need to control for these covariates and other confounders.
Although included studies were generally of good quality, sample sizes varied considerably and were typically small. Over half of studies included less than 100 cases and only 10 studies (19·6%) had a total sample size of more than 200 individuals. In addition, studies used a variety of definitions and age ranges to designate individuals as children. Our analysis only included mortality as a clinical outcome. A further general limitation is the difference in thresholds for vitamin D deficiency, particularly in the levels which are considered normal for infants and young children. Our assessment used the currently recommended threshold for deficiency (≤ 50 nmol/L) 16 and used a conservative estimate for studies which used different criteria.
Vitamin D remains an attractive biomarker and potential therapeutic agent in acute and critical care patients. Carefully designed and adequately powered studies are needed to determine the importance and therapeutic value of vitamin D in the general and septic paediatric critical care population.

Availability of data and materials
Data and computational code used for processing and analysis are available at https://github.com/margarc/VitaminD_children

Author contributions
AJBT conceived the study. AJBT and IT designed the study. MC collected data and performed the analysis with input from MAC, IT, ABJT and EE. MC and AJBT wrote the manuscript with contributions from all authors.

Declaration of interests
The authors declare no conflicts of interest.
Additional Table 14 Multivariate meta-regression model for prevalence Additional Table 15 Characteristics of studies included in the meta-analysis for prevalence in individuals with sepsis Additional Table 16 Sensitivity analyses for prevalence of vitamin D deficiency in critically ill children with sepsis Additional Table 17 Sensitivity analyses for mortality

Additional Figures
Additional Figure 1