WikiJournal of Medicine/Vitamin D as an adjunct for acute community-acquired pneumonia among infants and children: systematic review and meta-analysis

Introduction
Worldwide, community acquire pneumonia (CAP) is the leading cause of mortality among under-five children, accounting for approximately 1.3 million deaths annually. Most of these deaths occur in South Asia and Africa where the prevalence of malnutrition among children is high. The consequences of pneumonia are even more severe on account of lower immunity among children with malnutrition. Low-cost interventions for better management of pneumonia, particularly those which can boost the immune response, might hence contribute to decreased mortality rates from pneumonia in children.

The role of vitamin D in immunity and consequently in the pathogenesis, treatment and prevention of human infectious diseases, particularly of the respiratory tract, has come into limelight in the last few decades, although the role of vitamin D in tuberculosis has been postulated for more than a century. The vitamin D receptor (VDR) and CYP27B1 (the enzyme which converts vitamin D into its active form) has been found in macrophages, monocytes, dendritic cells and respiratory epithelial cells, which play a pivotal role in innate as well as adaptive immune responses.

Vitamin D deficiency has been linked to increased susceptibility to various infectious diseases, pneumonia and tuberculosis being the most prominent among them. The role of vitamin D in preventing infections has been studied in multiple randomised controlled trials (RCTs). Previous systematic reviews which have investigated the role of vitamin D supplementation for prevention of respiratory infections indicate a protective effect of vitamin D.  Vitamin D levels have been correlated to the severity of pneumonia in children in various case-control studies. In rickets, which is a very severe form of vitamin D deficiency (leading to even skeletal malformations and calcium deficiency), the risk of respiratory infections in children has been found to be increased.

Given that previous systematic reviews have focused on prevention of pneumonia through vitamin D supplements, and given evidence from observational studies on the link between vitamin D and pneumonia, we set out to systematically review the evidence available from RCTs on the use of vitamin D as an adjunct for treatment of CAP in infants and children.

Methods
We searched for published and unpublished RCTs on infants and children from 1 month to 5 years of age given vitamin D (in any dose, regimen or route) as an adjunct to standard therapy (as defined by trial authors) for acute CAP and compared to placebo (or nothing). All aetiologies of pneumonia (except aspiration pneumonia) and all degrees of severity were included. The protocol for this systematic review and meta-analysis has been registered with PROSPERO 2014:CRD42014010259. The systematic review followed the PRISMA guidelines.

We only included trials which had used either of the two diagnostic criteria for acute CAP as given below:
 * 1) 	Diagnosis of acute CAP by a physician on the basis of clinical examination and/or radiological features; or,
 * 2) 	History of cough or respiratory distress with a respiratory rate above the World Health Organization (WHO) defined age-specific respiratory rate of > 50 breaths per minute for children aged two to 11 months, or respiratory rate > 40 breaths per minute for children aged 12 to 59 months); and either documented fever of > 38.33 °C (101 °F) or chest in-drawing.

Trials on patients with congenital abnormalities, rickets, and protein-energy malnutrition or proven fungal pneumonia were excluded. We also excluded trials in which vitamin D was administered along with other vitamins or supplements. The review included all papers irrespective of language or publication status.

The primary outcomes we considered were:
 * 1) 	Clinical cure rate, which was defined as clinical recovery (that is no fever, no tachypnoea and no chest in-drawing) by the end of treatment;
 * 2) 	Time to clinical recovery of pneumonia;
 * 3) 	Time to radiological resolution; and
 * 4) 	All-cause mortality at end of follow-up (as defined by trial authors).

The secondary outcomes we considered were:
 * 1) 	Total duration (in hours) of hospital stay (time from randomisation to discharge);
 * 2) 	Requirement of any additional interventions like mechanical ventilation, corticosteroids, vasopressor agents, or anything else;
 * 3) 	Rates of any of the following complications of pneumonia: parapneumonic effusions and empyema, necrotising pneumonia, septicaemia and metastatic infection.

We searched for trials in four electronic databases adopting the following PubMed search strategy “(((vitamin D OR Cholecalciferol OR Ergocalciferol OR calcitriol))) AND (pneumon* OR bronchopneumon*)” without restricting for language or date:
 * 1) 	PubMed (search updated to 24 January 2017)
 * 2) 	Cochrane Central Register of Controlled Trials (CENTRAL) (search updated to 24 January 2017)
 * 3) 	CINAHL (searched updated to 24 January 2017)
 * 4) 	Global Health (searched 20 June 2016)

We searched two clinical trial registries (ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP)) using the following search strategy “Pneumonia AND (vitamin D OR cholecalciferol OR ergocalciferol OR calcitriol)” in February 2016. We also searched for grey literature by contacting researchers and searching abstracts of scientific meetings (Supplementary Data File 1) and references of included trials found by other methods.

Two review authors independently screened all the search results initially for consideration of inclusion as per eligibility criteria based on their titles and abstracts. After the initial screening full texts were obtained and assessment for inclusion and extraction was done by two authors independently. The risk of bias assessment on each study was independently conducted by two authors according to the methodology laid down in the Cochrane Handbook for Systematic Reviews of Interventions. Any discrepancies were resolved by consensus.

We used risk ratios (RR) for dichotomous data and mean difference (MD) for continuous data with 95% confidence interval (CI) using Review Manager 5 software. Where data was sufficient, we combined data using intention-to-treat (ITT) analysis and calculated a summary statistic for each outcome by a meta-analysis. The statistical heterogeneity was determined by a combination of visual inspection of graphs of RRs as well as using the I² statistic, and the Chi² test.

Where appropriate we conducted meta-analyses. We used a used a random-effects model if studies were statistically heterogeneous, otherwise we used a fixed-effect model. When random-effects model was used, we additionally conducted a sensitivity check using a fixed-effects model to understand differences in results.

We had planned to conduct a sub-group analysis if sufficient number of trials were found for several parameters (severity of pneumonia, dosage, frequency, type and route of antibiotics used and dosage, frequency and route of vitamin D supplementation) but this was not possible. We summarised the main findings of the review in the summary of findings table using the GRADE approach.

Characteristics of the included RCTs
The literature search identified a total of 365 papers and 25 clinical registry records (PRISMA Diagram Figure 1), of which six trials met the eligibility criteria, of which three have been completed  and three are yet to be published (NCT02054182, NCT02185196, NCT02936895).

The main features of the completed trials are summarised in Table 1. The ongoing trials are being conducted in South Africa, Bangladesh and Iran (the latter completed as per record but not yet published). The characteristics of these trials as evident from the clinical trial registries are presented in Table 2. Reasons for excluding other trials are documented in Supplementary Data File 1. The PRISMA compliance checklist is presented in Supplementary Data File 2.

The completed trials were from India and Afghanistan. These trials enrolled children of different age groups, severity of pneumonia, and varied in terms of the dosage and duration of vitamin D supplementation. Both studies from India included only children with severe pneumonia but while the earlier study included children of 2 months to 5 years, the recent trial included children in the age group 6 months to 5 years of age. The definition of severe pneumonia was different in the two Indian trials. In one trial severe pneumonia was defined as those with pneumonia with either chest in-drawing or at least one of the three danger signs (inability to feed, lethargy, and cyanosis). The 2016 Indian trial defined severe pneumonia as the presence of lower chest in-drawing with cough and difficult breathing. Both the Indian trials excluded children with rickets, severe acute malnutrition, asthma, hypertension, complicated pneumonia or severe illnesses (Table 1). The trial from Afghanistan included children of 1 week to 3 years of age presenting with non-severe, severe and very severe pneumonia. The trial from Afghanistan excluded children with rickets whereas five of the included children in the Indian trial had rickets (2 in vitamin D and 3 in placebo group).

The earlier Indian trial used vitamin D [more specifically vitamin D3 or cholecalciferol, dr. P. Gupta, personal communication] (1000 IU to children less than a year and 2000 IU to children between 1-5 years of age) for five days in the intervention arm, while the recent Indian trial used a much higher but single dose of 100,000 IU of oral cholecalciferol. In the trial from Afghanistan a single dose of 10,000 IU of Vitamin D was given orally in the intervention arm. Baseline and follow-up vitamin D status was not determined in the earlier two RCTs but the recent Indian trial has measured it.

The completed trials were judged to be low risk of bias for most parameters except some. Figure 2 provides a graphical summary of the risk of bias and the basis for these assessments is provided in Table 3.

Clinical cure rates
Only one trial reported this outcome and found no effect of Vitamin D as an adjunct on clinical cure rates (RR 1.01; 95% CI 0.91, 1.13) (Figure 3).

Time-to-clinical recovery of pneumonia
All trials reported this outcome and individually none of the trials found any significant difference in the time to clinical recovery of pneumonia in between those given vitamin D and those given placebos. In the 2012 trial from India, the median time of recovery in the vitamin D group was 72 hours (standard error (SE) 4.5, 95% CI 48, 96 hours) and 64 hours (SE, 95% CI 48, 88 hours) in the placebo group (P 0.33). The trial from Afghanistan found that mean time-to-clinical recovery in vitamin D group was 4.74 ± 2.22 days and in placebo group was 4.98 ± 2.89 days (P 0.17). The 2016 trial from India found no statistically significant difference for time to complete recovery from pneumonia between the vitamin D and placebo arm (Log Rank Chi² P 0.382). We could not pool the results from these studies as it is not appropriate to analyse time-to-event data by using methods for pooling continuous outcomes (using mean or median) since the relevant time is known only for those participants who had the event. Individual patient data meta-analyses should be conducted in such a scenario but this was not possible since patient-level data from all three trials was not available.

All-cause mortality
All three trials reported all-cause mortality, but these trials were heavily underpowered for this outcome with only three events in the vitamin D and three events in the placebo group in all the three trials together. Overall, children with vitamin D supplementation had a non-significant 1.01 risk of death ratio compared to children with no supplementation (RR 1.01; 95% CI 0.23, 4.41; three trials, 977 participants, Figure 4 and Table 5). A fixed effects model was used since no heterogeneity was detected (I²= 0 % and Chi² non-significant).

Total duration of hospital stay
Both the Indian trials reported this outcome. While the median duration of hospitalization was longer, but statistically non-significant in the vitamin D group compared to the placebo group in the 2012 Indian trial (112 hours (interquartile range (IQR) 96-136) versus 104 hours (IQR 88-128) (P 0.29)), the mean duration of hospital stay in the 2016 Indian trial (in hours) was shorter, but statistically non-significant between the vitamin D group (104.7 ± 37.9) and the placebo group (109.4 ± 46.0). We could not pool the results from the studies as it is not appropriate to analyse time-to-event data by using methods for pooling continuous outcomes (using mean or median) since the relevant time is known only for those participants who had the event. Individual patient data meta-analyses was not possible since patient-level data from all three trials was not available.

Time to radiological resolution
None of the trials reported any data on the outcomes of time to radiological resolution, requirement of any additional interventions and rates of complications.

Safety outcomes
The earlier trials did not report on any safety outcomes e.g. incidence of hypercalcemia but the recent trial from India reported that serious adverse events were similar across vitamin D (12.2%) and placebo (12.6%) groups.

Summary of findings
The GRADE approach was used to summarise the findings and rate the quality of evidence. The quality of evidence was low for the result that vitamin D has no effect on both clinical cure rates and all-cause mortality. A GRADE rating of low implies future research is very likely to change the estimate. Pooling of results was not possible for the other outcomes of time to clinical recovery of pneumonia and time to total duration of hospital stay because individual patient data was not available. However, the results for these outcomes individually for the study also show no statistically significant effect of vitamin D for these outcomes.

Summary of main results
The results of the three completed trials, all conducted in an in-hospital setting in South Asia, are insufficient to permit any judgement on the clinical effect of vitamin D as an adjunctive treatment for CAP. We found low quality evidence that vitamin D had no effect on clinical cure rates and all-cause mortality in children with CAP. No statistically significant effect was found in any of the studies which reported total duration of hospital stay or time to clinical recovery.

Strengths and limitations of the systematic review
This systematic review has used explicit, pre-specified methods in the process and we have searched for RCTs in multiple electronic databases and also by other methods. We believe that no relevant trials have been missed.

Assessment of publication bias, through funnel plots could not be done because of the small number of studies found. The earlier trial from India which we had included in the systematic review had included 5 participants with rickets, which is contrary to the eligibility criteria as per our protocol. However, considering the paucity of evidence and that those with rickets were divided almost equally in the two randomisation arms (2 in vitamin D and 3 in placebo) in the trial, we decided to include this trial in the full systematic review.

Research in context
A recent Cochrane Review on preventive use of vitamin D for infectious diseases in under-five children found no evidence to support the use of vitamin D for prevention of pneumonia. Contrary to this, an individual patient data meta-analysis published in February 2017 reported that vitamin D prevented acute respiratory infections and “reduced the risk of acute respiratory tract infection among all participants”. However, post-publication, the methodology of the study has been debated in the article's rapid responses. The use of vitamin D as an adjunct for treatment during pneumonia for children, we believe is a different research question from the preventive aspects of vitamin D. The use of vitamin D for infectious disease prevention has been an active area of research and scores of clinical trials and systematic reviews are already available addressing this topic. We found no evidence of benefit for the use of vitamin D as an adjunct during treatment of pneumonia but the quality of evidence for this is low.

Implications for practice and research
Current available evidence is insufficient to justify the use of vitamin D as an adjunct therapy to antibiotics and standard respiratory support in CAP among infants and children. Data from observational studies have indicated that vitamin D insufficiency is related to disease severity and prognosis but the results of our systematic review indicate that current evidence from RCTs does not support any change in practice. There is a need to conduct more placebo-controlled double blinded RCTs with adequate sample size, and appropriate dosage (discussed further below) for evaluating the safety and efficacy of vitamin D as an adjunct to antibiotics for childhood pneumonia of varying severity. Three more RCTs are already underway. However small trials like the trial from Iran (NCT02936895), which is yet to be published and uses composite severity scores as a primary outcome, is unlikely to contribute to the understanding of vitamin D in CAP. Though identification of aetiological agents is not imperative for the clinical management of pneumonia, trials should try to identify causative respiratory pathogens to evaluate if there is a differential effect of vitamin D on pneumonia due to different causative microbial organisms.

There are large differences between the included and ongoing trials (Tables 1 and 2) in the dose and regimen of vitamin D being used, ranging from 1000 IU daily to a single dose of 100,000 IU, and some trials have used age-dependent dosing too. There is however a trend in more recent trial designs to use a higher dosage of vitamin D. There is however a need to understand which dose and regimen of vitamin D might have immunomodulatory functions with respect to CAP in infants and children. Future studies should also consider evaluating the underlying vitamin D deficiency status of the populations and quantify the amount of improvement in 25-hydroxyvitamin D level as a result of the intervention. The 2016 Indian trial shows that although the proportion of vitamin D deficient children declined, there was no significant effect on serum levels of immune system mediators like cathelicidin, IgG and IgA, probably demonstrating that the high dose used did not clearly show any immunomodulatory effects. Use of even higher doses might be limited by safety concerns particularly for a disease like CAP where measurement of serum levels of vitamin D for monitoring in a clinical setting would have resource implications in low- and middle-income countries.

VDR polymorphisms have been shown to affect outcomes in a trial evaluating high dose vitamin D as an adjunct to tuberculosis treatment. Hence future trials on vitamin D as an adjunct to pneumonia treatment should also investigate the effect of host gene polymorphisms related to the vitamin D mediated immune response particularly VDR, vitamin D binding protein and CYP27B1 polymorphisms. Future trials should also standardise their definition and measurement of outcomes as well as their dosage and duration of intervention to ensure effective comparisons and adaptability across countries and settings. This is particularly important in terms of the eligibility criteria wherein the definitions of pneumonia and severe pneumonia are used in a uniform manner across studies to allow more meaningful comparison. We found that different trials have used different definitions for this purpose. As a standard the WHO recommended definitions should be used.

Conclusion
Vitamin D is cheap and has the theoretical potential to improve childhood mortality and morbidity due to CAP via its immunomodulatory effects. The role of vitamin D is also being investigated for other infectious diseases like tuberculosis, diarrhoea and malaria. Currently however there is no evidence to support the use of vitamin D as an adjunctive treatment for CAP in under-five children in low and middle-income countries, where mortality due to CAP is highest. More research is needed to understand whether vitamin D might have any effect on outcomes for specific aetiologies of CAP, genetic groups, or nutritional groups or at other doses and regimens or outcomes related to radiological resolution of pneumonia on which no data is currently available.

Conflict of interest
The authors report that they have no conflicts of interest.

Funding: none.