Description of the condition
Acute respiratory tract infections (ARTIs) and vitamin A deficiency are important public health problems in many low-income countries. Vitamin A deficiency is associated with impaired humoral and cellular immune function, keratinisation of the respiratory epithelium and decreased mucus secretion, which weaken barriers to infection (Ross 1996). In low-income countries, ARTIs, mostly in the form of pneumonia, are the leading cause of death in children under five years of age. The incidence of clinical pneumonia in low-income countries is estimated at 0.29 episodes per child per year. This equates to an annual incidence of 150.7 million new cases, 11 to 20 million (7% to 13%) of which are severe enough to require hospital admission. No comparable data are available for high-income countries. However, from large population-based studies, the incidence of community-acquired pneumonia among children less than five years of age is approximately 0.026 episodes per child per year (Rundan 2004). Pneumonia is associated with and causes about 3.8 million childhood deaths annually; 30.3% of these are in children under the age of five (Kirkwood 1995).
Description of the intervention
Vitamin A functions in a very different way in human metabolism. It is needed by the retina of the eye's specific metabolite, the light-absorbing molecule retinal; and as an irreversibly oxidized form retinoic acid, which is an important hormone-like growth factor for epithelial and other cells. Lack of vitamin A will cause some severe problems such as night blindness, xerophthalmia and complete blindness. Vitamin A deficiency also contributes to maternal mortality and other poor outcomes in pregnancy and lactation (WHO 2010).
How the intervention might work
Community-based clinical trials have been conducted in order to determine whether periodic high-dose vitamin A supplementation reduces the incidence, severity or both of ARTIs in children. The association between vitamin A deficiency and child mortality was first observed in the 1930s when vitamin A supplementation significantly reduced mortality among measles patients (Ellison 1932). Two meta-analyses (Fawzi 1993; Glasziou 1993) examined the relationship between vitamin A supplementation and infectious diseases. Glasziou (Glasziou 1993) reported that vitamin A reduced all-cause mortality by one third in children in low-income countries. However, the reduction in deaths from respiratory disease was seen only in the measles studies. Fawzi (Fawzi 1993) also reported that supplementation was protective against overall mortality in community-based studies (odds ratio (OR) 0.70) and highly protective against mortality in hospitalised patients with measles (OR 0.39).
From the current evidence it appears that vitamin A supplementation reduces the severity of respiratory infection and other systemic complications of measles. However, the association between vitamin A and non-measles ARTIs is unclear. The World Health Organization (WHO) Programme for the Control of Acute Respiratory Infections published a meta-analysis to assess the impact of supplementation on pneumonia morbidity and mortality (VAPWG 1995). They reported no consistent overall protective or detrimental effect on pneumonia-specific mortality and no effect on the incidence or the prevalence of pneumonia.
Why it is important to do this review
Hospital-based clinical trials examining the effectiveness of high-dose vitamin A, administered during an acute episode of non-measles ARTI, in reducing morbidity or mortality, have been performed. However, no meta-analyses have been carried out. Given the apparent effectiveness of vitamin A in reducing mortality in hospitalised patients with measles, and the inexpensiveness of the intervention, clarification of the association between vitamin A and non-measles ARTIs is of some importance.
The objective of this review was to assess the effectiveness and safety of vitamin A in children with diagnosed non-measles pneumonia.
Criteria for considering studies for this review
Types of studies
We considered only parallel-arm, randomized controlled trials (RCTs) and quasi-RCTs in which children with diagnosed non-measles pneumonia were treated with vitamin A (at the time of confirmed disease) for inclusion in this review. We did not consider studies that examined the effectiveness of vitamin A in preventing episodes of non-measles pneumonia or studies including patients with measles.
Types of participants
We included children of either gender and under 15 years of age with non-specific pneumonia that was uncomplicated by measles. The definition of pneumonia was a clinical case definition, radiological confirmation or both.
Types of interventions
Vitamin A plus standard medical treatment versus standard medical treatment with or without placebo.
Types of outcome measures
- Signs of pneumonia (for example, fever, tachypnoea, dyspnoea, chest X-ray findings).
- Clinical severity (for example, oxygen saturation, requirement for mechanical ventilation or need for supplemental oxygen, crepitations, bronchial breathing, duration of hospitalization, failure of first line treatment or change of antibiotic required).
- Adverse events associated with the intake of vitamin A such as diarrhea and signs of toxicity, including vomiting and desquamation.
Search methods for identification of studies
We searched The Cochrane Library, Cochrane Central Register of Controlled Trials (CENTRAL 2010, issue 3) which contains the Acute Respiratory Infections Group's Specialised Regsiter, MEDLINE (1996 to July week 3, 2010), EMBASE (1990 to August 2010), LILACS (1985 to August 2010), CINAHL (1990 to August 2010), Biological Abstracts (1990 to August 2010), Current Contents (1990 to August 2010) and the Chinese Biomedicine Database (CBM) (1994 to June 2010).
We used the following search strategy to search CENTRAL and MEDLINE. We combined the MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE:sensitivity- and precision- maximizing version (2008 revision); Ovid format (Lefebvre 2009). The search strategy was modified to search Embase (see Appendix 1), LILACS (see Appendix 2), CINAHL (see Appendix 3), Biological Abstracts (see Appendix 4) and Current Contents (see Appendix 5), The search strategy for the Chinese databases is presented in Figure 1. See Appendix 6 for details of original search.
|Figure 1. Search strategy in the Chinese databases.|
1 exp Vitamin A/ )
2 (vitamin a or retinol).tw,nm.
3 Dietary Supplements/
5 exp Pneumonia/
7 Respiratory Tract Infections/
8 (respiratory adj2 (infection* or disease* or acute)).tw.
In addition to electronic databases, we searched reference lists. There were no language or publication restrictions.
Searching other resources
We searched the WHO ICTRP for ongoing studies. This site contains records provided by nine primary registries (http://apps.who.int/trialsearch/).
Data collection and analysis
Selection of studies
One review author (JN) collated study citations into a single library, excluded duplicates and articles unlikely to be relevant by scanning the abstracts. We retrieved citations without abstracts in full text. Two review authors (JN, TW) independently reviewed a sample of the selection process. Two review authors (JN, WT) independently retrieved relevant articles in full text and applied the inclusion criteria.
Data extraction and management
Two review authors (JN, WJ) independently extracted data from included studies using standardized forms. They obtained a copy of the original article from Dr. Quinlan and telephoned the original authors of Chinese articles. All data were extracted from published papers and differences were resolved by discussion among the review authors.
Assessment of risk of bias in included studies
We assessed study quality using an adaptation of the method specified in the Cochrane Handbook for Systematic Reviews of Interventions (Julian 2009). The following characteristics were assessed.
Adequacy of the randomisation process
Adequate sequence generation was reported using one of following approaches: random number tables, computer-generated random numbers, coin tossing, or shuffling.
Yes: with low risk of selection bias.
Unclear: did not specify one of the adequate methods outlined above but only mentioned 'random' in which situation moderate risk of selection bias existed.
No: other methods of allocation that appeared to be biased with high risk.
Adequacy of the allocation concealment process
Adequate measures to conceal allocations such as central randomisation; serially numbered, opaque, sealed envelopes; or another description that contained convincing elements of concealment.
Yes: with low risk of selection bias.
Unclear: did not report an allocation concealment approach at all in which situation moderate risk of selection bias existed.
No: inadequately concealed allocation that reported an approach that did not fall into one of the categories described above, and did not conceal allocation. In which situation a high risk of selection bias existed.
Level of masking
Low risk of both performance and detection bias: the healthcare providers and outcome assessors were masked to know what interventions were received in each participants.
Moderate risk of both performance and detection bias: single blinding used only.
High risk of both performance and detection bias: blinding not been used.
Free of other bias
Yes: it was specified that without any potential bias, for example, conflict of interest.
No: there was conflict of interest or other risk of bias.
Measures of treatment effect
Dichotomous data were expressed as an odds ratio (OR) with 95% confidence intervals (CI). We converted continuous data to mean differences (MD). All meta-analyses were based on a fixed-effect model.
Unit of analysis issues
Only individual participants data were included and assessed.
Dealing with missing data
Original investigators of Chinese studies were contacted by telephone to request missing data include randomisation method and allocation concealment.
Incomplete outcome data bias were assessed depending on the percentage of missing data:
Low risk: only few withdraw or losses to follow up occurred.
Moderate risk: exclusions were about 10%.
High risk: exclusions of at least 10%, or wide differences in exclusions between groups.
Assessment of heterogeneity
We were able to perform only limited pooled analyses. We presented a statistical summary of treatment effects only in the absence of significant clinical or statistical heterogeneity. We tested heterogeneity using the Cochran Q statistic (Cochran 1954) with significance at P = < 0.10 (Boissel 1989; Fleiss 1986).
Assessment of reporting biases
We planed to assess publication bias by using funnel plots. However, due to the small number of included studies we did not perform this assessment. Free of selective reporting was assessed depends on whether all the outcomes were reported or not:
Yes: all outcomes were addressed and reported.
No: some of outcomes appeared in protocol but not reported in the article.
We statistically combined data when it appropriate. Some data were analyzed separately using Review Manager 5 software (RevMan 2008). We expressed the effects as OR with 95% confidence intervals (CI) for dichotomous data. However, as mortality was an unlikely event, we used Peto OR with 95% CI. For continuous outcomes mean differences (MD) with 95% CI were calculated.
Subgroup analysis and investigation of heterogeneity
We intend to explore the following potential sources of heterogeneity using subgroup analyses or meta-regression. However there was insufficient data.
1. Different doses (low, medium, high).
2. Duration of treatment.
We tested the robust of evidence by sensitivity analysis when pooling analysis takes place.
1. Repeating the analysis excluding unpublished studies (if any).
2. Repeating the analysis excluding poor quality studies, as specified above.
3. Comparing the results of fixed-effect models to random-effects models. Robust evidence should not be reversed by changing these models.
Description of studies
Results of the search
After full-text examination of the search results, we identified 72 papers. No new trials were identified for inclusion or exclusion in our updated searches.
Although we failed to contact the authors of two Chinese studies, we identified their trials as a randomized controlled trial (RCT) (Liu 1997) and a quasi-RCT (Zhang 1999) based on the information provided in the studies. One study (Fawzi 1998) conducted in Africa, and three studies conducted in South America (Nacul 1997; Rodriguez 2005; Stephensen 1998) were identified as RCTs. Finally, a total of six studies were included in the review (Fawzi 1998; Liu 1997; Nacul 1997; Rodriguez 2005; Stephensen 1998; Zhang 1999).
A total of 1740 infants and children participated in the six studies. One study (Fawzi 1998) included 687 participants; three studies (Nacul 1997; Rodriguez 2005; Stephensen 1998) included 854 participants; and two studies (Liu 1997; Zhang 1999) conducted in China included 119 participants, of which 80 children had recurrent bronchopneumonia. All participants were hospitalised children with ages ranging from one month to 14 years.
Two studies (Fawzi 1998; Nacul 1997) administered 200,000 IU of vitamin A to children aged one year and older, and 100,000 IU to infants under the age of one year, daily for two days. One study (Rodriguez 2005) administered 100,000 IU of vitamin A to children aged one year and older, and 50,000 IU to infants under the age of one year. One study (Liu 1997) administered 150,000 IU of vitamin A. One study (Stephensen 1998) used a different dosing scheme: children younger than one year old received 100,000 IU of vitamin A on admission to the study and 50,000 IU on the second day of hospitalization; children one year of age or more received 200,000 IU on the first day and 100,000 IU on the second day. The final study (Zhang 1999) administered 10,000 IU twice daily for the first six days; following this, 1500 IU/day was administered for the next 20 days. All of the studies used a placebo control.
Three studies (Fawzi 1998; Nacul 1997; Rodriguez 2005) reported mortality, two studies (Fawzi 1998; Stephensen 1998) reported the duration of hospitalization, and one study (Nacul 1997) reported a change in antibiotic use. One study (Stephensen 1998) reported clinical severity scores with eight individual components that included, for example, oxygen saturation, requirement of supplemental oxygen and signs of pneumonia from chest X-ray results. Three studies (Fawzi 1998; Nacul 1997; Stephensen 1998) reported the frequency of children requiring supplemental oxygen. We contacted the trial authors for further information but as yet no response has been forthcoming. One study (Nacul 1997) reported on adverse outcomes including vomiting, diarrhea, irritability and bulging fontanelle. Two studies (Liu 1997; Rodriguez 2005) reported variable outcomes and we only extracted data for outcomes which matched the inclusion criteria of this review: time to resolution of fever and cough, time until positive auscultation and positive chest X-ray results. The study by Zhang (Zhang 1999) reported on a lack of improvement of bronchopneumonia at short-term follow up and the recurrence rate of bronchopneumonia at long-term follow up.
One trial author (Quinlan 1996) kindly sent us her paper. After careful consideration it was excluded. Ten Chinese studies seemed to meet the inclusion criteria and we contacted eight authors by telephone. We discovered that the allocation methods they had used were not actually randomized, thus these studies were excluded.
Risk of bias in included studies
One study (Rodriguez 2005) was identified as low risk of bias and four studies (Fawzi 1998; Liu 1997; Nacul 1997; Stephensen 1998) as moderate risk. One study (Zhang 1999) was identified as high risk of bias. An overview of the study quality can be found in the Characteristics of included studies table and Figure 2 and Figure 3.
|Figure 2. Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.|
|Figure 3. Methodological quality summary: review authors' judgements about each methodological quality item for each included study.|
Four studies (Fawzi 1998; Nacul 1997; Rodriguez 2005; Stephensen 1998) used block randomisation design and two studies (Liu 1997; Zhang 1999) use individual randomisation. All studies except for Rodriguez 2005 did not mention the randomisation method and allocation concealment in detail. Four studies (Fawzi 1998; Liu 1997; Nacul 1997; Stephensen 1998) were judged to have a moderate risk of selection bias and one study (Zhang 1999) had a high risk of selection bias.
Rodriguez 2005 used block randomisation to allocate the participants and the sequence was generated by using tables of random numbers and was converted into a sequence of envelopes that contained the regimen assignments for all children in the trial. The Ethical Committee of the Corporacion Ecuatoriana de Biotecnología held the blinded randomisation codes in a secure place. The study code was not broken until all the data were entered and the initial analyses performed. It was judged to have a low risk of selection bias.
Five studies (Fawzi 1998; Liu 1997; Nacul 1997; Rodriguez 2005; Stephensen 1998) performed double-blinding by masking the containers, they were judged to have a low risk of both performance and detection bias. One study (Zhang 1999) did not perform blinding and was thus judged to have a high risk of performance and detection bias.
Incomplete outcome data
Four studies (Fawzi 1998; Nacul 1997; Rodriguez 2005; Stephensen 1998) addressed the number of withdraws or loss of follow-up. In the study by Rodriguez 2005, 48 children (16.7%) were lost to follow-up during the course of the study. There was a moderate risk of incomplete outcome data bias. Two studies (Liu 1997; Zhang 1999) did not address the incomplete outcome data.
All studies were judged as unclear for selective reporting.
Other potential sources of bias
All studies were judged as unclear for other potential sources of bias.
Effects of interventions
The outcome measures included in the meta-analysis were 1. mortality; 2. time period with signs of pneumonia, indicators of disease severity including duration of hospitalization, time with fever, time with rapid respiratory rate, time with hypoxia, time with positive findings on auscultation, time with cough, time with positive at X-ray findings, requirement for mechanical ventilation, failure of first line antibiotic treatment, slight improvement at short-term follow up, and recurrent rate of bronchopneumonia (no improvement) at long-term follow up; 3. adverse events including vomiting, diarrhea and bulging fontanelles.
We did not perform a sensitivity analysis as there were no low quality studies included in a pooled analysis.
Pooled analysis across three studies (Fawzi 1998; Nacul 1997; Rodriguez 2005) showed no statistically significant difference in deaths during hospitalization (OR 1.29; 95% CI 0.63 to 2.66) between the vitamin A and placebo groups. Two studies used the same high dose of vitamin A (Fawzi 1998; Nacul 1997); the other study (Rodriguez 2005) used a moderate dose of vitamin A. There was no statistically significant heterogeneity (P = 0.64, I
|Figure 4. Forest plot of comparison: 1 Vitamin A versus control, outcome: 1.1 Mortality during hospitalisation.|
Time with fever
Pooled analysis across three studies (Fawzi 1998; Liu 1997; Rodriguez 2005) showed no statistical significance (MD -1.11, 95% CI -5.66 to 3.44) and no heterogeneity was detected (P = 0.58, I
|Figure 5. Forest plot of comparison: 1 Vitamin A versus control, outcome: 1.2 Time with signs of pneumonia and treatment issues (continuous).|
Time with respiratory rate greater than 40 breaths/minute/day
Two studies (Fawzi 1998; Rodriguez 2005) reported this outcome; no differences were found (MD 0.03; 95% CI -0.32 to 0.38) and no heterogeneity was detected (P = 0.19, I
Time with positive finding on auscultation
Two studies (Liu 1997; Rodriguez 2005) reported the time with positive findings on auscultation; there was no difference between the two groups (MD -0.54; 95% CI -1.14 to 0.05) and no heterogeneity was detected (P = 0.10, I
Time with positive finding on chest X-ray
Two studies (Liu 1997; Stephensen 1998) reported the signs of pneumonia in chest X-ray examinations after giving high doses of vitamin A. There was no statistically significant difference between the vitamin A and placebo groups (MD 0.90; 95% CI -1.10 to 2.90) (Liu 1997). There were also no positive findings on hyperinflation (in two of 41 children receiving vitamin A versus one of 31 receiving placebo), perihilar infiltrate (three of 41 versus six of 31), interstitial infiltrate (three of 41 versus none of 31), consolidation (seven of 41 versus three of 31), atelectasis (three of 41 versus none of 31), or effusions (three of 41 versus one of 31) (Stephensen 1998) (Figure 5).
Time with cough
No improvement at short-term follow up and recurrence rate at long-term follow up
One study (Zhang 1999) reported lower rates of no improvement at both short-term follow up and of recurrence rate at long-term follow up in the low-dose vitamin A supplement group than in the control group. There were statistically significant differences (OR 0.06, 95% CI 0.01 to 0.30 and OR 0.12, 95% CI 0.03 to 0.46, respectively). These results suggest that vitamin A can promote the recovery rate at short-term follow up and decrease the recurrence rate of recurrent bronchopneumonia at long-term follow up (Figure 6).
|Figure 6. Forest plot of comparison: 1 Vitamin A versus control, outcome: 1.3 Time with signs of pneumonia and treatment issues (dichotomous).|
Time with fever, tachypnoea and hypoxaemia in children with basal serum retinol concentration > 200 ug/L
One study (Rodriguez 2005) reported the time with fever, tachypnoea and hypoxaemia in children with basal serum retinol concentration > 200 ug/L. There was a significant difference between the two groups (MD -61.40; 95% CI -119.10 to -3.7). This result suggests that vitamin A can shorten the time to remission of signs in children whose basal serum retinol was > 200 ug/L (Figure 7).
|Figure 7. Forest plot of comparison: 1 Vitamin A versus control, outcome: 1.5 Time with fever, tachypnoea and hypoxaemia of pneumonia (serum retinol > 200 ug/L).|
Indicators of disease severity
Duration of hospitalization
Duration of hospitalization was reported in two studies (Fawzi 1998; Stephensen 1998). There was no statistically significant heterogeneity (P = 0.83 and I
Time with hypoxia, less than 95% oxygen saturation per day
Requirement for mechanical ventilation
None of the studies reported on requirements for mechanical ventilation.
Failure of first line antibiotic treatment
Children that required a change of antibiotic due to failure of first line treatment were reported in only one paper (Nacul 1997). This study reported no statistically significant reduction in the odds of first line antibiotic failure (OR 0.65; 95% CI 0.42 to 1.01) in children receiving vitamin A compared to those receiving placebo (Figure 6).
One study (Stephensen 1998) assessed disease severity during hospitalization by using a clinical severity score that consisted of 12 variables. Commencing measurement at 24 hours post-treatment, the mean blood oxygen saturation levels of the placebo group increased to a higher level than the mean levels for the vitamin A group, and this difference persisted throughout hospitalization (P = 0.003 for treatment period). Significantly more children in the vitamin A group than in the placebo group eventually required supplementary oxygen (P = < 0.0005 for the treatment period). The mean respiratory rate (P = < 0.0005 for the treatment period) and heart rate (P = 0.001 for the treatment period) were significantly higher in the vitamin A group than in the placebo group when compared over time. The prevalence of auscultatory evidence of consolidation was initially the same in both groups but the prevalence decreased steadily in the placebo group while staying nearly constant in the vitamin A group, resulting in a significant difference between the groups (P = 0.0014 for the treatment period).
Three studies reported adverse effects of vitamin A intake. Four children (four out of 99) were observed with bulging fontanelle in Nacul 1997; none in the placebo group. The difference was not statistically significant (OR 8.25; 95% CI 0.44 to 155.37). The incidence of vomiting was 27 of 171 in the vitamin A group and 34 of 167 in the placebo group of one study (Nacul 1997); and one of 47 in the vitamin A group and none of 45 in the placebo group in a second study (Stephensen 1998). There was no statistical significance in the pooled analysis across these two studies (OR 0.77; 95% CI 0.45 to 1.33). The incidence of diarrhea was 19 of 206 in the vitamin A group and 30 of 199 in the placebo group (Nacul 1997); there was no statistical significance (OR 0.57; 95% CI 0.31 to 1.05). The incidence of irritability (Nacul 1997) was 38 of 157 in the vitamin A group and 38 of 149 in the placebo group. No statistical significance was apparent (OR 0.93; 95% CI 0.56 to 1.57). Nausea was observed in one child from each group (Stephensen 1998) (Figure 8).
|Figure 8. Forest plot of comparison: 1 Vitamin A versus control, outcome: 1.4 Adverse events.|
Vitamin A supplementation appears to have had little effect on the clinical course of non-measles pneumonia in children. The lack of a clear beneficial effect after vitamin A treatment is a little surprising given the protective effects of vitamin A in pneumonia associated with measles.
Summary of main results
Headaches, loss of appetite, vomiting and bulging fontanelles in infants are some of the adverse effects known to occasionally occur with the administration of high doses of vitamin A. These symptoms are minor and transitory, with no known long-term effects, and require no special treatment (Yang 2009). From three included studies in this review (Fawzi 1998; Nacul 1997; Stephensen 1998), the incidence of toxicity as an adverse event was not statistically different between the vitamin A and placebo-treated groups.
There was a borderline relative increase in the placebo group of children who failed to respond satisfactorily to first line antibiotic treatment. However, only one study reported on this and whether it represents a consistent effect of vitamin A, or is the result of chance, needs further investigation.
Rodriguez (Rodriguez 2005) conducted a subgroup analysis according to the serum retinol concentration, and found vitamin A can shorten the duration of signs in children with normal serum retinol (> 200 ug/L) but no benefit in children with vitamin A deficiency. However, this trial did not report on the nutritional status and severity of illness, which may be confounding factors. This study also found the time to remission of respiratory signs did not differ significantly between the groups of normal weight children and underweight children. However, data regarding this issue were unavailable to review.
Supplemental vitamin A in a lower dose seems to result in a beneficial effect on recurrent bronchopneumonia. This finding was from one study of poor methodological quality and lacking nutritional information about the vitamin A status of included children (Zhang 1999). More well-designed, large RCTs are needed to support this positive finding.
Fawzi (Fawzi 1998) observed a trend towards a higher incidence of death in the vitamin A group (13 of 346 participants versus eight of 341 in the placebo group) although this was not statistically significant (OR 1.63; 95% CI 0.66 to 3.97). Stephensen (Stephensen 1998) reported more severe clinical scores in children receiving vitamin A than those who received a placebo. We expect that studies in the future will provide more conclusive evidence.
The possible explanation for the lack of benefit of vitamin A in non-measles pneumonia is that the effects of vitamin A may be disease-specific, with vitamin A only being effective when pneumonia is complicated with measles.
Overall completeness and applicability of evidence
The benefit of vitamin A as an adjunct to the treatment of non-measles pneumonia was not clarified, but vitamin A might be beneficial to children with high basal serum retinol. Further RCTs, possibly with measured vitamin A levels and varying vitamin A doses, may provide sufficient evidence to clarify the role of vitamin A in non-measles pneumonia.
Quality of the evidence
The quality of the evidence of outcomes, which include mortality and adverse events (rates of diarrhea, bulging fontanelles and irritability), have a higher quality (moderate); the quality of other evidence ranks from very low to low (see Figure 9; Figure 10; Figure 11).
|Figure 9. Primary outcomes|
|Figure 10. Secondary outcomes|
|Figure 11. Adverse events and secondary outcome|
Potential biases in the review process
We included English language and Chinese language articles only as we were unable to search articles in other languages. This might result in selection bias.
Agreements and disagreements with other studies or reviews
Similar findings were reported in a meta analysis (Brown 2004) which reviewed vitamin A for children aged from one month to six years with acute respiratory infections in low-income countries. Five studies including a total of 2177 children (1067 intervention, 1110 controls) were included in this meta analysis. There were no significant differences in any of the recovery measures or mortality between the intervention and control groups. Pooled results showed no statistical significant differences between the vitamin A group and placebo group: fever: 0.03 (-0.10 to 0.17); oxygen requirement: -0.08 (-0.31 to 0.16); raised respiratory rate: -0.09 (-0.38 to 0.19); and hospital stay: -0.06 (-0.52 to 0.40). Mortality was below 2% in both groups, with a non-significant higher risk in the intervention group (odds ratio 1.16, 95% CI: 0.61 to 2.21).
Implications for practice
From the evidence available at this time and given the lack of clinical benefit associated with vitamin A treatment in children with non-measles pneumonia, it is difficult to recommend vitamin A as an adjunctive therapy in this patient group, particularly if the risk of vitamin A deficiency is low.
Implications for research
Even though six studies met the inclusion criteria, the variability in the outcomes reported and measured meant that only results from a few studies were eligible for inclusion in the meta-analyses for each of the outcomes. This limited the power of the meta-analyses to detect statistically significant differences. Large RCTs reporting clinically important outcomes are needed to increase the likelihood of obtaining a true and precise estimate of the effect. However, dosage of vitamin A could also be important and RCTs with a risk-stratified population, that is children at high risk and low risk of vitamin A deficiency, and varying administered dosage may provide evidence of the effectiveness of vitamin A treatment for non-measles pneumonia in children.
We thank Drs. George Swingler and Nelcy Rodriguez and the ARI Group editorial team for advice in writing this review. We wish to acknowledge Janet Grant, Amy Zelmer, David Ross, Bhavneet Bharti, Heather Zar, and Chanpen Choprapawon for commenting on drafts of this review. We also thank Alexandra Raulli, Elmer Villanueva and Renae Johnston for their prior involvement with this review. Finally, we thank Dr. Kyran P. Quinlan, from the Department of Pediatrics at the University of Chicago, for kindly providing her published paper for assessment.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. Embase search strategy
14. #10 AND #13
13. #11 OR #12
12. random*:ab,ti OR placebo*:ab,ti OR factorial*:ab,ti OR crossover*:ab,ti OR 'cross over':ab,ti OR 'cross-over':ab,ti OR volunteer*:ab,ti OR assign*:ab,ti OR allocat*:ab,ti OR ((singl* OR doubl*) NEAR/2 (blind* OR mask*)):ab,ti
11. 'randomized controlled trial'/exp OR 'single blind procedure'/exp OR 'double blind procedure'/exp OR 'crossover procedure'/exp
10. #4 AND #9
9. #5 OR #6 OR #7 OR #8
8. (respiratory NEAR/2 (acute OR infection OR disease)):ab,ti
7. 'respiratory tract infection'/de OR 'lower respiratory tract infection'/exp
4. #1 OR #2 OR #3
3. 'diet supplementation'/exp
2. 'vitamin a':ab,ti OR retinol:ab,ti
Appendix 2. LILACS search strategy
Mh vitamin a OR Tw vitamin a OR Tw vitamina a OR Tw retinol OR Mh dietary supplements OR Tw dietary supplement$ OR Tw suplementos dieteticos [Words] and Mh pneumonia OR Mh aspiration pneumonia or Mh bronchiolitis obliterans organizing pneumonia or Mh cryptogenic organizing pneumonia or Mh eosinophilic pneumonia or Mh lobar pneumonia or Mh mycoplasma pneumonia or Mh pneumocystis pneumonia or Mh staphylococcal pneumonia or Mh ventilator-associated pneumonia or Mh pneumonia aspiration or Mh pneumonia, bacterial or Mh pneumonia, eosinophilic or Mh pneumonia, interstitial or Mh pneumonia, interstitial plasma cell or Mh pneumonia, lipid or Mh pneumonia, lobar or Mh pneumonia, mycoplasma or Mh pneumonia, pneumococcal or Mh pneumonia, pneumocystis or Mh pneumonia, primary atypical or Mh pneumonia, radiation or Mh pneumonia, staphylococcal or Mh pneumonia, ventilator-associated or Mh pneumonia, viral or Mh chlamydia pneumoniae or Mh chlamydophila pneumoniae or Mh diplococcus pneumoniae or Mh klebsiella pneumoniae or Mh meningitis, streptococcus pneumoniae or Mh mycoplasma pneumoniae or Mh streptococcus pneumonia or Mh streptococcus pneumoniae infections or Mh idiopathic interstitial pneumonias OR Tw pneumon$ OR Tw neumonS [Words] and ((Pt randomized controlled trial OR Pt controlled clinical trial OR Mh randomized controlled trials OR Mh random allocation OR Mh double-blind method OR Mh single-blind method) AND NOT (Ct animal AND NOT (Ct human and Ct animal)) OR (Pt clinical trial OR Ex E05.318.760.535$ OR (Tw clin$ AND (Tw trial$ OR Tw ensa$ OR Tw estud$ OR Tw experim$ OR Tw investiga$)) OR ((Tw singl$ OR Tw simple$ OR Tw doubl$ OR Tw doble$ OR Tw duplo$ OR Tw trebl$ OR Tw trip$) AND (Tw blind$ OR Tw cego$ OR Tw ciego$ OR Tw mask$ OR Tw mascar$)) OR Mh placebos OR Tw placebo$ OR (Tw random$ OR Tw randon$ OR Tw casual$ OR Tw acaso$ OR Tw azar OR w aleator$) OR Mh research design) AND NOT (Ct animal AND NOT (Ct human and Ct animal)) OR (Ct comparative study OR Ex E05.337$ OR Mh follow-up studies OR Mh prospective studies OR Tw control$ OR Tw prospectiv$ OR Tw volunt$ OR Tw volunteer$) AND NOT (Ct animal AND NOT (Ct human and Ct animal))) [Words]
Appendix 3. CINAHL search strategy
S22 S11 and S21
S21 S12 or S13 or S14 or S15 or S16 or S17 or S18 or S19 or S20
S20 TI ( random* or placebo* ) or AB ( random* or placebo* )
S19 (MH "Quantitative Studies")
S18 (MH "Placebos")
S17 (MH "Random Assignment")
S16 AB ( singl* or doubl* or trebl* or tripl* ) and AB ( blind* or mask* )
S15 TI ( singl* or doubl* or trebl* or tripl* ) and TI ( blind* or mask* )
S14 TI clinic* W1 trial* or AB clinic* W1 trial*
S13 PT clinical trial
S12 (MH "Clinical Trials+")
S11 S5 and S10
S10 S6 or S7 or S8 or S9
S9 TI ( respiratory N2 infection OR respiratory N2 acute OR respiratory N2
disease ) or AB ( respiratory N2 infection OR respiratory N2 acute OR
respiratory N2 disease )
S8 (MH "Respiratory Tract Infections")
S7 TI pneumon* or AB pneumon*
S6 (MH "Pneumonia+")
S5 S1 or S2 or S3 or S4
S4 TI ( vitamin a or retinol ) or AB ( vitamin a or retinol )
S3 (MH "Dietary Supplements")
S2 (MH "Vitamin A Deficiency")
S1 (MH "Vitamin A+")
Appendix 4. Biosis Previews (Thomson ISI) search strategy
Topic=("vitamin a" or retinol or dietary supplement*) AND Topic=(pneumon*)
Refined by: Topic=(random* or placebo* or clinical trial* or singl* blind* or doubl* blind*)
Appendix 5. Current Contents (Thomson ISI) search strategy
Topic=("vitamin a" or retinol or dietary supplement*) AND Topic=(pneumon*)
Refined by: Topic=(random* or placebo* or clinical trial* or singl* blind* or doubl* blind*)
Appendix 6. Original search strategy
We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2007, Issue 3); MEDLINE (1996 to August Week 2, 2007); EMBASE (1990 to January 2007); LILACS (6 May 2007); CINAHL (1990 to August 2007); Biological Abstracts (1990 to July 2007); Current Contents (1990 to May 2007); and the Chinese Biomedicine Database (CBM) (1994 to June 2007).
We ran the following search strategy in CENTRAL and MEDLINE in combination with the highly sensitive search strategy developed by the Cochrane Collaboration for identifying randomised controlled trials (Dickersin 1994). The search strategy was modified to search the other electronic databases.
1. exp vitamin A/
2. vitamin A.mp
4. exp dietary supplements/
6. exp pneumonia/
8. exp pneumonia, bacterial/
9. exp pneumonia, lipid/
10. exp pneumonia, mycoplasma/
11. exp pneumonia, pneumococcal/
12. exp pneumonia, rickettsial/
13. exp pneumonia, staphylococcal/
14. exp pneumonia, viral/
15. exp respiratory tract infections/
16. acute adj respiratory.mp
17. respiratory adj infection.mp
18. respiratory adj disease.mp
20. 5 and 19
In addition to electronic databases, we searched reference lists. There were no language or publication restrictions.
Last assessed as up-to-date: 3 August 2010.
Protocol first published: Issue 2, 2002
Review first published: Issue 3, 2005
Contributions of authors
Taixiang Wu (TW) was responsible for drafting, editing, commenting, making amendments and updating this review.
Juan Ni (JN) was responsible for searching for studies, data extraction and analysis and drafting the review.
Jiafu Wei (JW) was responsible for searching for studies and data extraction.
Declarations of interest
Sources of support
- Chinese Cochrane Center, Chinese Centre of Evidence-Based Medicine, West China Hospital of Sichuan University, China.
- China Medical Board of New York, USA.
1. We used new standards of quality assessment described in new Cochrane Handbook of Systematic Reviews of Interventions.
2. We included a summary findings table result using GRADEprofiler to assess the quality of evidence.
Medical Subject Headings (MeSH)
MeSH check words
Child; Humans; Infant