Article Text

Download PDFPDF

Systematic review on antibiotic therapy for pneumonia in children between 2 and 59 months of age
  1. Zohra S Lassi1,
  2. Jai K Das1,
  3. Syed Waqas Haider1,
  4. Rehana A Salam1,
  5. Shamim A Qazi2,
  6. Zulfiqar A Bhutta1
  1. 1Division of Women and Child Health, The Aga Khan University, Karachi, Pakistan
  2. 2Department of Maternal, Newborn, Child and Adolescent Health, World Health Organization
  1. Correspondence to Dr Zulfiqar A Bhutta, Division of Women and Child Health, The Aga Khan University, Karachi 74500, Pakistan; Program for Global Pediatric Research, Hospital For Sick Children, Toronto; zulfiqar.bhutta{at}

Statistics from


Community acquired pneumonia (CAP) remains the leading killer globally, and accounts for 1.1 million of all deaths in children less than 5 years of age with the majority of these deaths occurring in the first 2 years of life.1 ,2 It was estimated that in 2010, there were 120 million episodes of pneumonia (14 million of which progressed to severe episodes) in children younger than 5 years.1 This pneumonia morbidity and mortality burden is disproportionally distributed with high prevalence in the low and middle income countries (LMIC) of Southeast Asia and Africa.1

The treatment of CAP requires effective antibiotics in optimum dosage and duration, with the choice of antibiotics based on the bacterial aetiology. The commonest causative bacterial pathogens responsible for CAP vary according to the age of the patient. In paediatric patients, including the 2–59-month group, Streptococcus pneumoniae and Heamophilus influenzae are the leading and most commonly targeted organisms. However, empirical antibiotics after clinical diagnosis of CAP and not aetiology itself, is currently the most important basis of pneumonia treatment guidelines even in resource-rich Western countries. Microbiological diagnosis is still warranted in special circumstances such as in those children who have severe symptoms of pneumonia, those who become hospitalised, and in children with complicated clinical course.3 Strategies do exist to prevent pneumonia and are an essential component to reduce child mortality; these include immunisation against haemophilus influenzae type B vaccine, pneumococcus, measles and pertussis, exclusive breastfeeding for the first 6 months of life and complementary feeding thereon with supplementation of zinc, and addressing environmental hygiene.4

The high pneumonia burden and mortality, especially in children less than 59 months in LMICs,1 ,2 necessitates an updated review on the most suitable antibiotic therapy based on trials involving this age group and in LMIC setting. We found previous reviews5–8 and reports9–14 on this subject published in recent years (see online table S1). Evidence from Lodha et al5 which also dealt purely with antibiotic therapy showed amoxicillin to be an effective alternative to co-trimoxazole for treating ambulatory patients with CAP. Lodha et al further recommend oral amoxicillin for children with severe pneumonia without hypoxaemia, and penicillin/ampicillin plus gentamicin for children hospitalised with severe and very severe CAP. The recommendations in the review by Lodha et al,5 However, are for children under the age of 16 years, whereas, the current review focuses on children between 2 and 59 months of age. Also, while previous reviews,6 ,8 including that of Lodha et al focused purely on a single modality, such as the choice or duration of antibiotics for treating CAP, the present review updates evidence on the overall spectrum of antibiotic recommendation for treating CAP in children, namely, the choice, optimum dose, duration, route and combination of antibiotics. This crucial information is presented with the aim of providing a targeted cure for the LMICs setting.


We included randomised controlled trials (RCT) that assessed the route, dose, combination and duration of antibiotics in the management of WHO defined non-severe/severe/very severe CAP. Only those studies that used standard WHO definition and/or classification for CAP9 were included, irrespective of language or publication status. Study participants included children of 2–59 months of age with CAP. The review included studies that compared any intervention that either examined a single drug or a combination; or a different dose, duration and route of the same drug. We included studies that analysed outcomes including clinical cure rate (ie, symptomatic and clinical recovery by end of treatment), treatment failure (ie, developing worsening clinical signs at any point in time; a respiratory rate (RR) exceeding age-specific cut-off and/or <90% oxygen saturation on pulse oximetry after completion of treatment), relapse rate (developing disease recurrence after being declared ‘cured’), change in antibiotic required and mortality rate. We grouped and analysed the outcomes according to very severe, severe and non-severe pneumonia.

Standard extraction forms were used to extract data from included studies by two independent reviewers. Errors were corrected by comparison of the extracted data and any differences in interpretation of data were resolved by contacting the trial author, where possible. All data was entered and analysed using Review Manager 5 software.15 We used Cochrane methods for risk of bias assessment to assess the quality of included studies.16 For each study, the level of attrition was noted and its impact on overall assessment of treatment effect was explored via sensitivity analysis. Analysis for all outcomes was done, as far as possible, on an intention-to-treat basis; that is, all participants randomised to each group in the analysis were included irrespective of their adherence. Fixed-effect meta-analysis model was used for analysis. Where heterogeneity was high (I2>30%, τ2>0, χ2 p<0.1), data was analysed in a random-effect model.


The review identified 8122 studies on initial search, while 22 trials with 20 593 enrolled children were included in meta-analyses (figure 1). Three studies were included for very severe pneumonia, six studies for severe pneumonia and 13 studies for non-severe pneumonia. Out of 22 included studies, 20 were free of selective reporting, 12 were double-blinded and 14 had adequate allocation concealment. Five studies included for very severe and severe pneumonia were free of other bias, while four studies in non-severe pneumonia mentioned their source of funding. The nuances of the included studies in the current review are discussed in the table of characteristics (see online tables S2–9).

Figure 1

Search flow diagram.

Very severe pneumonia (danger signs)

We found four studies17–20 for very severe pneumonia in children in LMICs, of which three17–19 met the review's criteria and were subject to meta-analysis. Two studies17 ,19 involved children 2–59 months, and one study18 involved children of 1 month to 5 years of age; antibiotics in all these studies were administered parenterally (see online supplementary tables S2, S3, S6 and S7).

Evidence from Asghar et al17 showed that a combination of ampicillin and gentamicin led to a significant decrease in failure rates on day 5 by 30% (RR 0.70; 95% CI 0.51 to 0.97), day 10 by 27% (RR 0.73; 95% CI 0.55 to 0.97), and day 21 by 25% (RR 0.75; 95% CI 0.57 to 0.98) compared to chloramphenicol group. Duke et al,18 however, reported no impact in overall adverse effects and death rates with a combination of penicillin and gentamicin compared with chloramphenicol alone. The pooled estimate of the two studies17 ,18 showed a non-significant decrease in mortality by 29% (RR 0.71; 95% CI 0.51 to 1.00), and a significant lower failure rate by 21% (RR 0.79; 95% CI 0.66 to 0.94) in the penicillin and gentamicin group compared to the chloramphenicol group (figure 2). Bansal et al19 found no differences in failure rates on comparing the combination of penicillin and gentamicin with amoxicillin-clavulinic acid (table 1).

Table 1

Summary estimates: very severe pneumonia

Figure 2

Failure rate with ampicillin and gentamicin in very severe pneumonia.

Severe pneumonia (chest indrawing pneumonia)

There were seven eligible studies21–27 for severe pneumonia, however, six20–25 met our criteria and were subject to meta-analysis. Four other studies28–31 which made a diagnosis of pneumonia radiologically were not included because the review's criteria was to include only those studies which diagnosed pneumonia according to WHO's definition of pneumonia.7 All 7 studies were conducted in LMIC settings, and children included were between 2 and 59 months of age (see online tables S4, S5, S6 and S7). Different routes were also employed for comparing antibiotics, and these routes are mentioned with their respective findings in this section (table 2).

Table 2

Summary estimates: severe pneumonia

Evidence from Cetinkaya et al21 showed a combination of parenteral penicillin and chloramphenicol versus parenteral ceftriaxone to be equally efficacious in cure rates by day 10 (RR 1.05; 95% CI 0.88 to 1.27). Straus et al22 compared oral antibiotics for severe pneumonia and reported a significant increase by 79% (RR 1.79; 95% CI 1.13 to 2.84) in treatment failure with co-trimoxazole compared to amoxicillin at 48 h.

For severe pneumonia, availability of parenteral treatment is sometimes a challenge in LMICs where a safe and equally effective alternative is warranted. This review found oral amoxicillin to have similar failure rates when compared to: combination of intravenous penicillin and oral amoxicillin,23 combination of intravenous ampicillin plus oral amoxicillin by 66%,24 or intravenous benzyl penicillin (P 0.031).25 Campbell et al26 showed similar rates of treatment failure with oral co-trimoxazole when compared to a combination of intramuscular procaine penicillin and oral ampicillin: −0.01% (95% CI −0.11 to 0.09%). Due to paucity of studies, however, the current review's objective of comparing various dosages or duration of antibiotics for the treatment of severe pneumonia in children of age 2–59 months could not be met.

Non-severe pneumonia (fast breathing pneumonia)

There were 26 eligible studies32–57 on children with non-severe pneumonia, however 13 were subject to meta-analysis (table 3).22 ,32–38 ,40 ,44–47 Most of these studies met the review's target age group of 2–59 months, and were conducted in LMIC setting (see online tables S8 and S9).

Table 3

Summary estimates: non-severe pneumonia

Pooled estimates of evidence from Straus et al, 22 CATCHUP,32 and Awasthi et al33 which compared oral co-trimoxazole versus oral amoxicillin showed failure rate to be similar between the two groups (RR 1.09; 95% CI 0.93 to 1.27) (figure 3). Also, cure rates in two of these studies 32 ,33 were similar (RR 0.99; 95% CI 0.96 to 1.01).

Figure 3

Failure rate with amoxicillin in non-severe pneumonia.

Levofloxacin is highly effective for adult pneumonia, but its efficacy or safety profile has not been widely explored in children less than 5 years of age. Bradley et al34 compared oral levofloxacin versus oral co-amoxiclavulinic acid, and showed a similar effect in cure rate. Pooled estimates from two other trials 35 ,36 which assessed efficacy of oral azithromycin with co-amoxiclavulanic acid for children with non-severe pneumonia showed no difference in rates of treatment failure (RR 1.20; 95% CI 0.45 to 3.21).

Deivanayagam et al38 examined the use of parenteral ampicillin versus combination of benzyl penicillin and chloramphenicol for non-severe pneumonia, and reported no difference in cure rate and treatment failure rate. Other trials comparing parenteral versus parenteral antibiotics were found in the current review, including evidence from Camargos et al, 39 which showed relapse of pneumonia within 24 h to significantly increase with single shot group of intravenous benzathine penicillin (75.3%) versus 7 days of intramuscular procaine penicillin (66.3%); In the same study, relapse rate was significantly lower after 24–48 hours of treatment with benzathine penicillin (8.6%) compared to treatment with IM procaine penicillin (16.9%).

To examine the ideal dose of antibiotic to manage non-severe pneumonia, evidence from Rasmussen et al40 showed no difference in clinical cure rate with a double dose of co-trimozaxole (trimethoprim plus sulfamethoxazole) compared to standard dose. Two other trials 41 ,42 reported amoxicillin twice a day to be a reasonable alternative to dosing three times a day. A similar finding was reported by Cook RC et al43 in that clinical cure rate with amoxicillin/clavulanate was not different in the thrice a day versus twice a day group (difference 3.2%; 95% CI 4.36 to 10.80).

For the duration of antibiotic treatment for non-severe pneumonia, evidence from four studies 44–47 was analysed; a 3-day therapy with either amoxicillin or co-trimoxazole was equivalent to a 5-day therapy for relapse rate (RR 0.99; 95% CI 0.98 to 1.01). A similar finding was reported by Ficnar et al48 which reported no difference in cure rate in the 5-day versus 3-day group (p=0.82) of azithromycin. To determine the ideal route of antibiotic therapy for non-severe pneumonia, the current review identified two studies,49 ,50 but which could not be analysed as the study did not specifically include children who were between 2–59 months of age.


With the last meta-analysis on the subject by Lodha et al5 and Lassi et al7 published in 2013 and 2011, respectively, the current review reanalysed pre-existing studies to look for new evidence on antibiotic therapy in children less than the age of five (table 4). The current review, however, analysed trials addressing additional components in antibiotic therapy (ie, dose, duration, route) compared to previous reviews 5 ,6 ,8 that dealt only with single modality, such as the choice or route. The evidence derived from the current review further focuses on children between 2 and 59 months of age in LMIC settings, which is an age group and setting where the highest pneumonia mortality burden is witnessed.

Table 4

Summary table

Based on trials addressing the important issue of choosing the most effective antibiotic for managing WHO-defined pneumonia with danger signs (old classification very severe pneumonia) in the 2–59-month age group, the current review favours a combination of penicillin/ampicillin and gentamicin.17–19 These trials were conducted in facilities in LMICs. An important issue in LMICs is that referral is often difficult and injection is not available, which makes it crucial to seek an equally effective oral drug as an alternative. We found oral amoxicillin to be equally efficacious as parenteral antibiotics for WHO-defined chest indrawing pneumonia (old WHO-defined severe pneumonia), and thus recommend it for treatment in LMICs.23 ,24 ,26 The 2005 WHO guidelines recommended IV/IM benzyl penicillin/ampicillin for chest indrawing pneumonia, and a switch over to oral amoxicillin once improvement was witnessed.58 There weren't sufficient studies to address the issue of the optimum dose and duration of parenteral antibiotic for treating severe and very severe pneumonia in children of age 2–59 months; this is an area where more trials are warranted. WHO's Integrated Management of Childhood Infections (IMCI) 2006 guidelines however recommend that a prereferral dose of 7.5 mg/kg intramuscular injection of gentamicin, and 50 mg/kg injection of ampicillin be used for very severe pneumonia.10 An important consideration in drafting guidelines for high-mortality diseases like pneumonia is the prevalence of HIV in a region, as it can alter pneumonia-related management of children in HIV-endemic regions. Without sufficient evidence from HIV patients in this review, we recommend WHO treatment guidelines that were developed for HIV-infected children with pneumonia in LMICs.10 ,59

For the management of fast breathing pneumonia, WHO recommended treatment with co-trimoxazole as a first-line empirical agent in countries with an infant mortality greater than 40 per 1000 live births.10 Evidence derived from our review did not favour use of co-trimoxazole over amoxicillin in managing non-severe pneumonia in children between 2 and 59 months of age, which is consistent with findings from the review by Lodha et al5 Thus, amoxicillin is a better alternative to co-trimoxazole in managing non-severe pneumonia in this age group as it is also effective against chest indrawing. The review also concludes that a 3 day course of antibiotics is as effective as 5 day therapy for managing non-severe pneumonia, which is beneficial for low-income settings where cost is an issue. It further has the added benefit of decreased antibiotic resistance and improved compliance.35–38 But this switch over from 5-day to 3-day therapy is not recommended for HIV-endemic countries. Where the optimum dose of antibiotic for non-severe pneumonia is concerned, the current review prefers a twice a day to three times a day dosing which is also favoured by the American Academy of Paediatrics.60 We further found oral antibiotic to be a safe and cost-effective alternative to injectable antibiotic for fast breathing pneumonia in children aged between 2 and 59 months.49 ,50

In this review, we used ‘clinical failure’ to demonstrate difference between antibiotics, as this outcome was consistently mentioned in included trials. But Dagan et al61 have previously argued that bacterial eradication is a more accurate determinant of antimicrobial efficacy than clinical outcome. While research gaps do exist in treatment of CAP in children, as noted in the present review, some useful interventions, besides antibiotics, to reduce childhood pneumonia mortality have also been examined. An important intervention is the early identification of pneumonia and referral of the sick child which is recommended by the IMCI. A study also demonstrated that provision of assured supply of oxygen via oxygen concentrators for children with severe pneumonia resulted in 35% mortality reduction.62 Other important interventions related to treatment of CAP warrant further exploration, such as the management of fast breathing pneumonia with potential supportive therapy sans antibiotics, role of hypoxaemia detection and management, identifying comorbidities in children with pneumonia and managing them, and community management of chest indrawing and pneumonia with danger signs in resource-poor settings. Further studies are also warranted to explore second-line antibiotics. It is important to note here that while cephalosporins are used widely in high-resource countries, there is paucity of literature about their success in LMICs. Ceftriaxone is one cephalosporin that has the additional benefit of being a once-daily drug. Thus, if its success is established in trials conducted for LMICs, its usage could potentially simplify management of CAP in such settings as well.

Finally, the need for future research should not take away from the expediency of implementing what we know. As the recent global action plan for pneumonia and diarrhoea indicates,63 we know enough in terms of key interventions for us to implement these at scale. This review supports that we know enough about the management of pneumonia among children in greatest need to implement strategies among those who need them most. This imperative to save 1.2 million pneumonia-associated deaths annually must not take away from the imperative of working concurrently on refining and improving existing interventions.

Search strategy

The electronic search was last performed on 15 March 2013 on Medline, Cochrane Library, Pubmed, Google Scholar with the following search terms: ((‘Pneumonia’ OR ‘very severe pneumonia’ OR ‘severe pneumonia’ OR ‘non-severe pneumonia’ OR ‘acute respiratory illness’ OR ‘Community acquired pneumonia’) AND (‘child*’ OR ‘infant’ OR ‘preschool*’ OR ‘schoolchild’ OR ‘school age’ OR ‘preschool’ OR ‘kid*’ OR ‘toddler*’) AND (‘treatment’ OR ‘anti-infective agent’ OR ‘anti-bacterial agents’ OR ‘antibiotic’ OR ‘management’)). The bibliographies of all relevant RCTs and reviews were cross-checked, and clinical trial registries were browsed for any on-going trials. All available titles and abstracts were screened for inclusion by two review authors independently. Where trial eligibility could not be assessed by screening of the title and abstract, full texts was retrieved to judge relevance. Any differences were resolved by discussion, or conferring with a third review author.

WHO definition of pneumonia7

According to WHO classification, non-severe pneumonia is defined by a RR per minute of ≥50 in infants 2–11 months of age and an RR of ≥40 in children 12–59 months without lower chest in-drawing (LCI); severe pneumonia is defined by LCI with or without rapid breathing, and very severe pneumonia is defined as inability to drink, convulsions, abnormally sleepy or difficulty arousing, stridor in a calm child or clinically severe malnutrition.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.