Pneumonia is a major cause of death in children under five years of age, causing approximately one million deaths worldwide annually . In 2015, 22 million (16%) of 138 million pneumonia cases internationally were classified as severe by WHO criteria . The WHO Integrated Management of Childhood Illness (IMCI) criteria defines children with severe pneumonia as those with signs of pneumonia and at least one of the IMCI danger signs; not being fed well, convulsions, reduced consciousness level, reduced movement, fever (>38°C), or hypothermia (<35.5°C), and central cyanosis [3,4]. The case fatality rate for severe pneumonia is up to six times higher than for non-severe pneumonia (4.2% compared to 0.65%) . Therefore, low cost, accessible adjunctive therapies to improve clinical outcomes are needed, particularly in younger children.
The current WHO IMCI recommendation for the first-line treatment of severe pneumonia for children aged 2 to 59 months is parenteral ampicillin (50mg/kg every six hours for at least five days) or penicillin, and gentamicin (7.5mg/kg once a day for at least five days) . Some proposed additional therapies include vitamin supplementation such as zinc [6–8], vitamin A , vitamin D [10,11], and other micronutrients such as folic acid  due to the high rate of malnutrition in low- and middle-income countries (LMICs). Also, macrolide antibiotics are recommended for severe pneumonia in high-income countries to cover atypical bacteria – a recommendation that is not currently included in WHO guidelines . Although diagnostic tests for influenza are not widely available in LMICs, influenza has been shown to frequently cause viral pneumonia in children. The role of neuraminidase inhibitors such as oseltamivir (Tamiflu) and zanamivir in addition to standard care for children with severe pneumonia in this setting has not been established. Finally, oral corticosteroids, along with azithromycin, have been associated with a reduction in the duration of hypoxemia and dyspnoea in children with pneumonia due to Mycoplasma pneumoniae in China .
Therefore, this review aims to explore the role of additional treatments with vitamins, neuraminidase inhibitors, macrolides, and oral steroids on clinical outcomes, specifically for children aged one month to nine years with severe pneumonia in LMICs.
MEDLINE, Embase, PubMed, and the Cochrane library were searched in September 2020. Authors of identified studies that were still recruiting or had an unclear recruitment status were contacted for information on study status and availability of results. References of relevant reviews and included studies were searched for additional studies.
The search strategy included terms related to pneumonia (the condition of interest) and treatment outcomes. It also included terms related to specific additional therapies, including vitamin supplementation, oseltamivir, oral steroids, and macrolides. This included synonyms for these treatments, such as other medications within the same class and relevant specific names of vitamins. The search strategy also included an extensive list of terms related to the setting of interest (LMICs). See Appendix S1 in the Online Supplementary Document for the full MEDLINE search strategy.
Included studies involved children aged one month to nine years with severe pneumonia (by any definition) in LMICs according to World Bank income level. Studies that looked at children with severe pneumonia caused only by one specific pathogen such as measles or M. pneumoniae were excluded. Those studies reporting additional treatment with either vitamins, neuraminidase inhibitors (oseltamivir or zanamivir), macrolides or oral steroids, and that compared clinical outcomes to standard care (as per the study setting) were included. Systematic reviews (SRs), cohort studies, and clinical trials published in English between 2000 to the search date were included. Original studies were included and presented separately even when the results had been reported in the identified SRs. Literature reviews, case series, case reports, and conference, meeting, and poster abstracts were excluded.
Titles and/or abstracts of retrieved studies were screened by a single investigator (MHN) to identify studies meeting the inclusion criteria. Any queries about the studies’ inclusion eligibility were discussed with a second reviewer (AG). Separate standardised tables were used for data extraction. Extracted data included: study year, WHO region, study aim, population data (including age and standard treatment), definition of severe pneumonia, details of the additional treatment, baseline deficiency (for studies of nutritional supplements only), pathogens identified, and key clinical outcomes. For SRs, the extracted data included: Author, year, whether it was a Cochrane review, inclusion and exclusion criteria, outcomes, search results and analysis, study interventions, key clinical outcomes, and adverse events.
Risk of bias in individual studies
One investigator (MHN) assessed the risk of bias of included SRs using AMSTAR 2 , while individual studies were assessed using the Effective Public Health Practice Project (EPHPP) quality assessment tool for quantitative studies .
A narrative synthesis of findings based on the results of included studies was planned.
2054 of 2147 identified studies were excluded after titles/abstracts were reviewed, leaving 93 full-text articles to be assessed for eligibility. We were unable to locate one of these full-text articles, and a further 60 were excluded based on eligibility criteria, leaving 32 studies for final inclusion. The most common exclusion reasons were the population (19 studies), outcomes (19 studies), and intervention (9 studies). Additionally, one study was identified after contacting the lead investigator for a trial identified from clinicaltrials.gov, and five were identified after a manual searching of all references (Figure 1).
Figure 1. PRISMA Flow diagram for review.
Table 1 summarises the included SRs for effectiveness of adjunctive therapies in children with severe pneumonia in LMICs. A total of 32 studies were identified, including seven SRs and 25 RCTs. All SRs studied nutritional supplements as an adjunct treatment of pneumonia, with two evaluating zinc [17,18] and one each evaluating vitamin A , vitamin D , and vitamin C . Two of these SRs examined the effectiveness of multiple adjunct therapies (zinc and vitamin A) [19,20].
Table 1. Included systematic reviews for effectiveness of adjunctive therapies in children with severe pneumonia in LMICs
AE – adverse events, Zn – zinc, M – month, HR – hazard ratio, OR – odds ratio, CI – confidence interval, Mod – moderate, d – day, SAE – serious adverse events, Vit – vitamin, y – year, SR – systematic review, cRCT – cluster randomized controlled trial, WMD – weighted mean difference, RR – risk ratio, WHO – World Health Organization, ICTRP – International Clinical Trials Registry Platform, LMIC – low and middle-income country, IU – international units, OD – once daily, MD – mean difference, BD – twice daily, hr – hour, Sa02 – oxygen saturation, Resp Rate – respiratory rate, CAP – community acquired pneumonia, sx – symptoms
*Outcomes for children with severe pneumonia given unless otherwise stated.
†Tachypnoea, fever & hypoxemia.
Table 2 shows the included individual studies for effectiveness of adjunctive therapies in children with severe pneumonia in LMICs. 25 studies identified were all RCTs; 17 examined the effect of zinc as an additional therapy, four evaluated vitamin D, and one studied vitamin A and vitamin C. Two studies evaluated combinations of different micronutrients, including a multi-nutrient containing vitamin A, C, E, folic acid, and zinc (one study) as well as a combination of vitamin C and E (one study). Most studies were from the Southeast Asia region (14 studies), with the remaining studies being from the Eastern Mediterranean (five studies), Americas (three studies), African (two studies), and Western Pacific region (one study). In 12 studies, the WHO definition of severe pneumonia was specifically used, in four studies the definition was not stated, and the remaining studies used alternative definitions. Most studies included young children aged 2-59 months, and in 18 of the 25 included studies, children with malnutrition were excluded.
Table 2. Included individual studies for effectiveness of adjunctive therapies in children with severe pneumonia in LMICs
SE Asia – South-East Asia region, AB – antibiotics, BD – twice daily, IV – intravenous, OD – once daily, PO – per os (taken orally), STAT – immediately, Mod – moderate, BC – blood culture, CAP – community acquired pneumonia, d – day, grps – group, h – hour, m – months, NP – nasopharyngeal, NS – not stated, Pt – patient, RCT – randomised control trial, RA – room air, Resp Rate – respiratory rate, Vit – vitamin, w – week, WHO, World Health Organisation, y – year, CXR – chest x-ray, RA – room air, CI – confidence interval, Diff – difference, HR – hazard ratio, IQR – Interquartile range, RR – risk ratio, SD – standard deviation, SE – standard error, OD – odds ratio, L – low
When results were below lower limits of normal zinc reference range. Normal reference range based on using highest value for males aged 3-9 years, morning nonfasting (65, standard error 0.7 μg/dL; 9.94 ± 0.11 μmol/L) .
*Used to indicate strong evidence for outcome based on statistical testing where P value <0.05 or CI did not include the null value.
†Danger signs: Cyanosis, inability to feed/drink, lethargy and convulsions.
‡Using Definition 2 (Chest indrawing, O2 saturation <93%, RR>50) from article, as this definition closest to the WHO’s definition of severe pneumonia (tachypnoea +/− chest indrawing) + danger sign.
§No chest indrawing, respiratory rate ≤50 bpm, SaO2 ≥ 95% on RA.
¶Modified definition: Including the presence of cough and/or chest wall indrawing, tachypnoea (>50 bpm in 2-<12, >40 bpm in children 12-59 months), hypoxemia (SpO2 < 90%), and ≥ one of the following by auscultation: rales, diminished breath sounds, bronchial breath sounds, or pleural rub.
║Participants with clinical rickets were given a mega-dose of Vitamin D (6 000 000 IU).
**For all participants (pneumonia or severe pneumonia), and also recorded those for which date of recovery was not recorded.
††Micronutrient group: 200 children given all micronutrients (Vitamin A, C, E, folic acid, zinc), remaining 200 divided into 5 groups of 40 and given a specific treatment.
‡‡Summary average statistic used not stated.
The four SRs examining the effect of zinc supplementation as an adjunct to antibiotic treatment for severe pneumonia found no evidence of clinical benefit. Two of these studies were of moderate to high quality, and two were of critically low quality. The most recent SR published in 2020 included 11 RCTs involving 6497 children aged 2-60 months . In the pooled analysis of 10 studies, the addition of zinc provided no protective effect against treatment failure (pooled OR = 0.93, 95% CI = 0.75, 1.14) nor a reduction in time to clinical improvement (7 RCTs, Pooled HR = 1.01, 95% CI = 0.89, 1.14). This SR  included all the original studies identified in the three previous SRs except for two studies [8,40]. One of these studies, an RCT including 106 children, showed no evidence for the benefit of zinc on clinical outcomes . The data from this trial were included in the meta-analysis  that found zinc had no significant effect on time to clinical improvement (Pooled HR = 1.12; 95% CI = 0.89, 1.41) or length of stay (Pooled HR = 1.04; 95% CI = 0.89, 1.22). One study focused only on children with measles-related pneumonia, and therefore this study did not fulfil inclusion criteria for this review . None of the included SRs did a sub-analysis for those children with baseline malnutrition. Notably, no SR reported serious adverse events associated with zinc treatment.
Of the 17 original studies, 6/15 RCTs showed a significant improvement in time to resolution of symptoms [7,24,25,29,32,35] and 5/13 RCTs showed a reduced length of hospital stay [7,25,31,32,35]. One of three RCTs reporting on death  showed that zinc reduced mortality by 67% (RR = 0.33; 95% CI = 0.15, 0.76)  while the other two found no difference (RR = 1.0; 95% CI = 0.3, 4.0)  and no deaths in either group . Notably, this study  did not exclude those with malnutrition, and 26% of the zinc group and 25% of the placebo group were stunted at baseline (height-for-age z-score less than 2 standard deviations). This was also one of five studies where the enrolled children had baseline zinc deficiency [27,28,34,36,37] but was the only one of these studies that showed strong evidence for the clinical benefit of adjunctive zinc therapy . Another study found no association between zinc supplementation and risk of death (RR = 1.0, 95% CI = 0.3, 4.0) despite severely underweight children comprising 21% and 25% of the zinc and placebo groups, respectively . The remaining three studies [27,36,37] included children with baseline zinc deficiency and similarly found no evidence that zinc improved clinical outcomes, with one study showing an increased length of hospital stay in those treated with zinc compared to the placebo group (zinc = 9.0 ± 6.0 days vs placebo = 7.0 ± 4.0 days) . These differing outcomes are not explained by variations in zinc dosing, given that 16 studies used doses of 10 to 20 mg per day.
Three SRs in total studied the effect of vitamin A in children with pneumonia [19–21], with two looking at multiple outcomes of both vitamin A and zinc [19,20]. Both studies of vitamin A and zinc were of critically low quality [19,20], and the third SR was of moderate to high quality . One SR identified 9 RCTs and performed a meta-analysis for the duration of hypoxia (4 RCTs), duration of tachypnoea (5 RCTs), length of stay (3 RCTs), and mortality (6 RCTs) . These analyses found no difference between the vitamin A and placebo groups in any of these domains . This SR primarily included children with non-measles pneumonia, as four of the six studies excluded measles pneumonia. Also, only one of the included studies  included outcomes specifically for children with severe pneumonia. Similarly, a Cochrane review of vitamin A on outcomes of non-measles pneumonia  included the same RCT reporting on children with severe pneumonia . Another SR  reported no significant differences between Vitamin A groups and placebo for outcomes of mortality, duration of hospitalization, illness, and complications and side effects. There were no further original studies identified in this review since the publication of these systematic reviews.
Adverse effects were reported in three of the six included studies in the Cochrane review. A pooled analysis from two studies showed no association between vitamin A and vomiting (two studies, OR = 0.77; 95% CI = 0.45, 1.33), diarrhoea (one study, OR = 0.57; 95% CI = 0.31, 1.05), bulging fontanelles (one study, OR = 8.25; 95% CI = 0.44, 155.37), or irritability (one study, OR = 0.93; 95% CI = 0.56, 1.57) .
A moderate to high quality Cochrane review of vitamin D in addition to antibiotic treatment for children with pneumonia identified 7 RCTs involving a total of 1529 children, with 749 (49%) having severe or very severe pneumonia . A meta-analysis showed no clinical benefit of adjunct vitamin D on the duration of illness or hospitalization as well as mortality. Vitamin D doses varied between 1000 IU to 100 000 IU, and a subset analysis was not done based on dose. No serious adverse events were reported .
A double-blind placebo-controlled RCT provided strong evidence that a single dose of vitamin D 100 000 IU orally reduced the time to resolution of severe pneumonia (Adjusted HR = 1.39; 95% CI = 1.11, 1.76) . However, three other RCTs showed no clinical benefit [10,11,39]. Among these, two studies [10,39] used low-dose vitamin D 1000 to 2000 IU/d for five days in children with severe pneumonia, and in the other study  that used a single dose of vitamin D3 100 000 IU, only 74 of 453 (16%) included children had severe pneumonia, which may explain the differing results. None of these studies reported any serious adverse events; however, one study reported one episode of vomiting and one of diarrhoea in the vitamin D group (1%) . Only one study  reported baseline serum vitamin D deficiency, where deficiency (defined as serum 25(OH)D<12 ng/mL) was observed in 40% of the control group and 38% of the vitamin D group.
A recent Cochrane review  included five RCTs of children under five years of age mostly from LMICs. However, the results of the studies could not be pooled due to different effect measures and data provided. This Cochrane review was assessed as being of moderate to high quality. Two of the studies included were published before the year 2000 [43,44], and one did not enrol children with severe pneumonia . Both were excluded from this current review. One study from the Eastern Mediterranean region showed that a greater proportion of children treated with vitamin C had resolution of tachypnoea in less than four days and improvement in hypoxia in less than one day compared with placebo (tachypnoea = 76% vs 62%, P = 0.03; hypoxia = 97% vs 86%, P = <0.01) . However, this study was assessed as being of poor quality due to a lack of description of randomization, and key differences between the groups at baseline not being adjusted for in the analysis. Similar results were found in an RCT of children in Southeast Asia that reported a significant reduction in the time to resolution of tachypnoea and chest indrawing by 0.5 days, and a 12.9% reduction in length of stay when children were treated with five micronutrients including vitamin A, C, and E, folic acid, and zinc . This study was also of weak quality due to a lack of reporting on group baseline characteristics and giving a poor explanation for children not included in the analysis. There was only one study  that reported baseline serum vitamin C levels, with the average baseline serum concentration from all participants (n = 800) being just below the lower limit of normal (mean baseline serum vitamin C concentration = 32.50 ± 0.15 (standard deviation (SD)) μmol/L, lower limit of normal = 34.00 ± 113.00 μmol/L). Neither of these studies reported any adverse events [12,41].
Combined nutritional supplements
There was no evidence from one high-quality RCT  for the addition of vitamin C and E to antibiotic treatment in symptom resolution (feeding difficulty: RR = 1.01, 95% CI = 0.72, 1.41; tachypnoea: RR = 1.12, 95% CI = 0.77, 1.64). This study recorded the mean weight for age z-score as being -1.78 (SD = ±1.18) in the control group and -1.92 (SD = ±1.18) in the intervention group, respectively, and excluded children with obvious marasmus or oedema .
This review identified studies evaluating the clinical effectiveness of adding zinc, vitamins A, C, D, E, and micronutrients to antibiotic treatment in children with severe pneumonia in LMICs. We have found that adjunctive zinc and vitamin A do not improve clinical outcomes such as time to resolution of symptoms and length of hospital stay. For vitamin D, a meta-analysis showed no clinical benefit of adjunctive vitamin D. However, the results for low dose regimens (1000-2000 IU/d) and a single high dose (100 000 IU) were pooled. Of the two high-quality RCTs of high dose vitamin D identified in this review, the larger study  included 324 children with severe pneumonia and found an improvement in the time to resolution of symptoms in the context of a high proportion of deficiency at baseline, while the other study  which included only 74 children with severe pneumonia and unknown vitamin D status at baseline, found no benefit. It is possible that this beneficial effect for vitamin D on time to resolution of symptoms may be greater for those with vitamin D deficiency; however, it is difficult to assess this based on the limited evidence presented in this study , as it did not record vitamin D at baseline, but rather referenced previous surveys showing high proportions of vitamin D deficiency in Kabul, where the study was set . Vitamin D has an important role in immune function through effects on the innate immune system such as engagement of toll-like receptors , and therefore high doses may be beneficial in boosting the immune response. A beneficial effect has been reported in a study of children with tuberculosis that showed faster symptom resolution in participants who received adjunctive vitamin D . However, this is not routinely recommended in the management of tuberculosis. For adjunctive vitamin C, there was low-quality evidence showing that it may reduce the time to symptom resolution and average length of stay. However, the included studies had major limitations and further high-quality studies are needed prior to the routine use of vitamin C for severe pneumonia. Finally, for combined micronutrients, only one study provided weak evidence suggesting a small reduction in resolution of symptoms and length of hospital stay.
A limitation of this review is that 18 of the 25 individual studies excluded malnourished children or those with chronic disease for which micronutrient supplementation would likely be more beneficial. Furthermore, studies that did include these children did not perform subgroup analyses to explore the effect of micronutrient supplementation in this group. Malnourished children have immune dysfunction that increases their susceptibility to common infections , and conversely, children with HIV are more likely to be malnourished . However, micronutrient supplementation (with zinc and a multivitamin syrup included) is already recommended as standard management for children with severe acute malnutrition by WHO, and these children should already be receiving these adjunctive treatments . However, the dose of vitamin D that showed benefit in the studies in this review is much larger than the amount in multivitamins and in ready-to-use-therapeutic foods (RUTF) . Similarly, the amount of vitamin C in RUTF is also less than the amount used in one study  (50 mg per 100 g of RUTF, compared to 200 mg OD) . Future research should, among else, examine the effect of increased doses of nutritional supplements in malnourished children and evaluate children with severe nutritional deficiency at baseline or other forms of malnutrition such as stunting, where children may not receive vitamin or micronutrient supplements.
Another limitation of this review is that, despite the high mortality rates reported in children with severe pneumonia, only three studies reported mortality as a clinical outcome. Also, few of the original studies described the bacterial pathogen and it was not possible to study the relationship between the pneumonia aetiology and response to adjunct therapies.
Our review did not identify any studies from LMICs on the addition of macrolides, oral steroids, or neuraminidase inhibitors to standard antibiotic treatment in children presenting with severe pneumonia. Most of the studies of neuraminidase inhibitors excluded during full-text review studied the role of oseltamivir in preventing pneumonia in children with influenza [51,52] or did not study children with severe pneumonia . A Cochrane review found that neuraminidase inhibitors such as oseltamivir (Tamiflu) and zanamivir did not significantly reduce the risk of pneumonia in children with confirmed or suspected exposure to influenza (Pooled RR = 1.06, 95% CI = 0.62, 1.83) and the benefit of treatment in those with severe pneumonia is unknown . Additionally, the cost of neuraminidase inhibitors may still be prohibitive for some LMICs, with zanamivir costing US$44, and oseltamivir US$44 per dose [55,56]. A Cochrane review on the role of oral corticosteroids on pneumonia outcomes included four RCTs with children, two from high-income countries and two from China; however, the studies from China only included children with M. pneumoniae pneumonia and were excluded from this review . Three of the four RCTs found that oral steroids reduced the time to clinical cure compared to standard antibiotic treatment (Mean difference (MD) = -1.57 days, 95% CI = -2.55, -0.60). Only one study looked at length of hospital stay in bacterial pneumonia showed a significant reduction in the oral steroid group (MD = -4.70 days with steroids, 95% CI = -7.50, -1.90) . Also, the benefit of additional macrolide treatment reported in studies from the US showed that children given combination β-lactam and macrolide therapy compared to β-lactam monotherapy had a reduced length of stay (adjusted RR = 0.80, 95% CI = 0.75-0.86)  and reduced rates of treatment failure in those aged older than 5 years (4.0% vs 12.9%, adjusted OR = 0.51, 95% CI = 0.28-0.95). However, there was no evidence of benefit in children aged five years or less .
This review found that adjunctive therapy with zinc or vitamin A in addition to antibiotic treatment does not improve clinical outcomes in children with severe pneumonia in LMICs. High-dose vitamin D at 100 000 IU may be beneficial, although further studies are needed to determine which populations would benefit the most. However, it would be reasonable to consider a single high dose of vitamin D in children with known vitamin D deficiency given that no adverse effects were reported. There is weak evidence of clinical benefit for vitamin C and combined nutritional supplements and further high-quality studies are needed.
Future research should focus on the evaluation of adjunctive nutritional supplementation in malnourished and HIV positive children who are most likely to benefit from treatment. Also, further studies on high dose vitamin D 100 000 IU, vitamin C, and combined nutritional supplements in children with severe pneumonia are needed. Similar to some practices in high-income countries, studies of the addition of macrolide antibiotics, corticosteroids and neuraminidase inhibitors in children presenting with severe pneumonia are required, including a cost-effectiveness analysis.
Disclaimer: The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the views, decisions or policies of the World Health Organization.
Acknowledgements: Thanks to Poh Chua, research librarian, for substantial technical support in setting up and running the database searches, and Helen Thomson and Haset Samuel for administrative support.