Antibiotic therapy versus no antibiotic therapy for children aged 2 to 59 months with WHO‐defined non‐severe pneumonia and wheeze
Abstract
Background
Worldwide, pneumonia is the leading cause of death amongst children under five years of age, and accounts for approximately two million deaths annually.
Pneumonia can be classified according to the World Health Organization (WHO) guidelines. Classification includes assessment of certain clinical signs and symptoms, and the severity of the disease. Treatment is then tailored according to the classification. For non‐severe pneumonia, the WHO recommends treatment with oral antibiotics. We used the 2014 WHO definition of non‐severe pneumonia for this review: an acute episode of cough, or difficulty in breathing, combined with fast breathing and chest indrawing.
The WHO recommends treating non‐severe pneumonia with oral antibiotics. Pneumonia is more commonly caused by viruses that do not require antibiotic treatment, but pneumonia caused by bacteria needs management with antibiotics to avoid complications. There is no clear way to quickly distinguish between viral and bacterial pneumonia. It is considered safe to give antibiotics, however, this may lead to the development of antibiotic resistance, and thus, limit their use in future infections. Therefore, it is essential to explore the efficacy of antibiotics for children with WHO‐defined non‐severe pneumonia and wheeze.
Objectives
To evaluate the efficacy of antibiotic therapy versus no antibiotic therapy for children aged 2 to 59 months with WHO‐defined non‐severe pneumonia and wheeze.
Search methods
We searched CENTRAL, MEDLINE, Embase, four other databases, and two trial registers (December 2020).
Selection criteria
We included randomised controlled trials (RCTs) evaluating the efficacy of antibiotic therapy versus no antibiotic therapy for children, aged 2 to 59 months, with non‐severe pneumonia and wheeze. We defined non‐severe pneumonia as 'a cough or difficulty in breathing, with rapid breathing (a respiratory rate of 50 breaths per minute or more for children aged 2 to 12 months, or a respiratory rate of 40 breaths per minute or more for children aged 12 to 59 months), chest indrawing and wheeze'. We excluded trials involving children with severe or very severe pneumonia, and non‐RCTs.
Data collection and analysis
Our primary outcomes were clinical cure and treatment failure; secondary outcomes were relapse, mortality, and treatment harms. We used standard methodological procedures expected by Cochrane. We used GRADE to assess the certainty of the evidence. Two review authors independently assessed the search results, extracted data, assessed risk of bias and the certainty of the evidence. We contacted the authors of two included trials and the author of the trial awaiting classification to obtain missing numerical outcome data.
Main results
We included three trials involving 3256 children aged between 2 to 59 months, who exhibited features of non‐severe pneumonia with wheeze. The included trials were multi‐centre, double‐blind, randomised, placebo‐controlled trials carried out in Malawi, Pakistan, and India. The children were treated with a three‐day course of amoxicillin or placebo, and were followed up for a total of two weeks. We assessed the included trials at overall low risk of bias for random sequence generation, allocation concealment, blinding, attrition bias, and selective reporting. Only one trial was assessed to be at high risk for blinding of outcome assessors. One trial is awaiting classification
Antibiotic therapy may result in a reduction of treatment failure by 20% (risk ratio (RR) 0.80, 95% confidence interval (CI) 0.68 to 0.94; three trials; 3222 participants; low‐certainty evidence).
Antibiotic therapy probably results in little or no difference to clinical cure (RR 1.02, 95% CI 0.96 to 1.08; one trial; 456 participants; moderate‐certainty evidence), and in little or no difference to relapse (RR 1.00, 95% CI 0.74 to 1.34; three trials; 2795 participants; low‐certainty evidence), and treatment harms (RR 0.81, 95% CI 0.60 to 1.09; three trials, 3253 participants; low‐certainty evidence). Two trials (2112 participants ) reported on mortality; no deaths occurred in either group.
One trial reported cases of hospitalisation, diarrhoea (with and without dehydration), rash (without itch), tremors, mild nausea and vomiting.
Authors' conclusions
We do not currently have enough evidence to support or challenge the continued use of antibiotics for the treatment of non‐severe pneumonia. There is a clear need for RCTs to address this question in children aged 2 to 59 months with 2014 WHO‐defined non‐severe pneumonia and wheeze.
PICOs
Plain language summary
Comparing treatment of non‐severe pneumonia, in children aged 2 to 59 months, with and without antibiotics
Review question
We tried to identify whether there was a difference in the outcomes for children aged 2 to 59 months with non‐severe pneumonia and wheeze, treated with or without antibiotics.
Background
Pneumonia is an infection of the lungs. In children, it is one of the leading causes of deaths globally. Pneumonia can be classified according to the World Health Organization (WHO) guidelines. Classification includes assessment of certain clinical signs and symptoms, severity of the disease and its treatment according to the severity. For non‐severe pneumonia, the WHO recommends treatment with oral antibiotics. We used the 2014 WHO definition of non‐severe pneumonia for this review: an acute episode of cough, or difficulty in breathing, combined with fast breathing and chest indrawing.
More commonly, pneumonia is caused by viruses that require supportive care rather than antibiotic treatment; however, pneumonia caused by bacteria should be treated with antibiotics to avoid complications. Since there is no clear way to quickly distinguish which organism actually caused the pneumonia, it is considered safe to give antibiotics. However, this may lead to the development of antibiotic resistance, and limit their use in future infections. The question is whether the use of antibiotics is justified in non‐severe pneumonia.
Search date
Our evidence is current to 23 December 2020.
Study characteristics
We included three trials (3256 children). They were conducted in four hospitals in three cities in Pakistan (Islamabad, Lahore and Rawalpindi), and in hospital outpatient departments in India and Malawi. The children were treated with a three‐day course of amoxicillin (antibiotic) or placebo, and followed up for two weeks. One trial is awaiting classification.
Study funding source
The included trials were supported by the USAID through INCLEN and IndiaClen; ARI Research Cell, Children Hospital, Pakistan Institute of Medical Sciences, Islamabad, Pakistan, and by a grant from the Bill & Melinda Gates Foundation.
Key results
Limited data showed a 20% reduction in treatment failure, however, no impact was observed on clinical cure, relapse, and treatment harms. No deaths were reported in either group. Our review did not have enough evidence to support or challenge the continued use of antibiotics for the treatment of non‐severe pneumonia.
Certainty of the evidence
The certainty of evidence for clinical cure was moderate. The certainty of evidence for treatment failure, relapse, and treatments harms was low, due to downgrading for imprecision and risk of bias.
Authors' conclusions
Summary of findings
| Antibiotic therapy versus placebo compared to placebo for children aged 2 to 59 months with WHO‐defined non‐severe pneumonia and wheeze | ||||||
|---|---|---|---|---|---|---|
| Patient or population: children, aged 2 to 59 months, with WHO‐defined non‐severe pneumonia and wheeze | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
| Risk with placebo | Risk with antibiotic therapy versus placebo | |||||
| Clinical cure (symptomatic and clinical recovery by the end of treatment) | Study population | RR 1.02 | 456 | ⊕⊕⊕⊝ | ||
| 896 per 1000 | 914 per 1000 | |||||
| Treatment failure (development of more severe signs and symptoms by the end of treatment) | Study population | RR 0.80 | 3222 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 164 per 1000 | 131 per 1000 | |||||
| Relapse (children who were 'cured' but developed symptoms within 14‐day follow‐up) | Study population | RR 1.00 | 2795 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 60 per 1000 | 60 per 1000 | |||||
| Treatment harms (adverse events or side effects associated with antibiotic therapy within 14‐day follow‐up) | Study population | RR 0.81 (0.60 to 1.09) | 3253 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 55 per 1000 | 45 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aData not precise: the confidence interval included the potential for important benefit or harm | ||||||
Background
Description of the condition
Pneumonia is an acute respiratory infection (ARI) that affects the lungs. In the early 1980s, an increase in global burden of pneumonia cases resulted in the development of guidelines for pneumonia by the World Health Organization (WHO). These guidelines were developed on the basis of evidence generated in the 1970s and 1980s, which were incorporated into the original version of the Integrated Management of Childhood Illness (IMCI) in 1999 (WHO/UNICEF 1999). Revisions were made in 2010, and 2014 based on the scientific evidence. These revisions included changes in the recommendation for the first‐line antibiotic, and re‐defining the severity classification of community acquired pneumonia (CAP). The 2010 WHO guideline had four treatment categories for CAP in which children with fast breathing were classified as having pneumonia and were treated with oral cotrimoxazole (WHO 2010). Children who had chest indrawing with or without fast breathing were classified as having severe pneumonia, and were referred to a healthcare facility and treated with injectable ampicillin/penicillin. Based on the 2014 WHO revised classification of pneumonia severity, there are now two categories of pneumonia (WHO 2014). Children with lower chest wall indrawing are now classified as having non‐severe pneumonia, and are treated with oral amoxicillin with home care advice. Pneumonia with any danger sign is now classified as severe pneumonia and these cases are referred to a healthcare facility and treated with injectable ampicillin/penicillin. For this 2020 update, we have followed the 2014 WHO classification for non‐severe pneumonia (WHO 2014).
Pneumonia is one of the single largest infectious cause of death amongst children (WHO 2019). Worldwide, approximately 2.6 million deaths were attributed to lower respiratory tract infections in 2015 (Troeger 2017). According to the Global Burden of Disease (GBD) Study 2015, pneumonia is the fifth most common cause of death amongst children younger than five years of age, and is the leading cause of death due to infectious diseases (Troeger 2017). Globally, the disease burden has decreased by 37% over the last decade amongst children younger than five years of age (Troeger 2017; Whitney 2017). According to GBD estimates, the highest lower respiratory tract infection mortalities in children younger than five years of age were in sub‐Saharan Africa, in Somalia (546.8 deaths per 100,000), and Chad (511.3 deaths per 100,000); the lowest were in Finland, in western Europe (0.65 deaths per 100,000; (Troeger 2017)).
The increased focus on the reduction of child mortality, from Sustainable Development Goal (SDG) 3, has generated renewed interest in developing more accurate assessments of the causes and number of deaths in children younger than five years of age. UNICEF and WHO have shown interest in reducing the number of deaths due to pneumonia amongst children by developing new strategies, integrated as the Global Action Plan for Pneumonia and Diarrhoea (GAPPD; (WHO 2013)). The recent data show a 47% decline in deaths amongst children due to pneumonia between 2000 and 2015. However, data on the pathogen‐specific causes of pneumonia are limited (WHO 2015).
The most common causes of severe pneumonia amongst children in low‐income countries are the bacterial pathogens Streptococcus pneumoniae (S pneumoniae) and Haemophilus influenzae type b (Hib) (WHO 2019). Respiratory syncytial virus are the most common cause of non‐severe pneumonia; although to a lesser extent, bacteria can also be a cause (WHO 2019). Some less common bacteria and fungi can also cause pneumonia in children. Recent data from the global burden of disease study show that S pneumoniae is the leading cause of deaths from lower respiratory tract infections in children aged one to five years, followed by H influenzae. Mortality associated with viral infections is much lower in this age group (Lozano 2012).
Description of the intervention
Chest X‐rays and laboratory tests are accurate measures when confirming pneumonia, including the extent and location of the infection, and its cause. However, in resource‐poor settings, where qualified personnel and diagnostic facilities are often unavailable, suspected cases of pneumonia are diagnosed by their clinical symptoms. The WHO has developed integrated community‐based case management guidelines that include deployment of community health workers (CHWs) to underserved areas to increase coverage and improve access to treatment (WHO/UNICEF 2012). It includes training of CHWs to assess the cause of illness and its severity, to refer severe cases to healthcare facilities, and to provide treatment where uncomplicated pneumonia is diagnosed (WHO/UNICEF 2012). Children and infants who exhibit fast breathing (50 breaths per minute or more in infants 2 months to 12 months of age, and 40 or more in children 12 months to 5 years of age), and cough are presumed to have non‐severe pneumonia, and the WHO/UNICEF tools for CHWs recommend oral amoxicillin in two daily doses, since CHWs are not expected to treat chest indrawing (WHO 2019).
How the intervention might work
Based on findings from a meta‐analysis, implementation of integrated case‐management approach proposed by 2012 WHO guidelines (WHO/UNICEF 2012) to identify and treat pneumonia at the community level has shown to reduce acute respiratory infection‐related mortality by 32% (Das 2013). The review also reported an increase in seeking care by 13% and a reduction in treatment failure by 40%. Other reviews on community case management with antibiotics based on 2012 WHO guidelines, showed a 36% (95% confidence interval 20% to 49%; Sazawal 2003)) and a 21% reduction in acute respiratory infection‐related mortality in children up to five years of age (Theodoratou 2010). According to Hazir 2006 children who were classified as having non‐severe pneumonia (cough and fast breathing) were found not to have a clinical pneumonia when assessed by physicians, or by examining chest X‐rays. Therefore, prescribing antibiotics for fast breathing alone may lead to the spread of antibiotic resistance, as fast breathing may be caused by other conditions, and affected by other factors, such as fever.
Another concern is the addition of wheeze to the algorithm. According to the trial by Hazir 2004, 62% of children presenting with non‐severe pneumonia and wheeze responded to bronchodilators. Amongst these responders, only 15% showed clinical deterioration on follow‐up. Hazir 2011 also suggests use of bronchodilators for a pragmatic management algorithm for children with wheeze. A prospective observational study by Lochindarat 2008 found that 85% of children with fast breathing and/or lower chest indrawing and wheeze responded to bronchodilators; only 4% deteriorated on day three and 3% on days five to seven.
Another study from Bangladesh prescribed antibiotics (either ampicillin or erythromycin) to two‐thirds of children presenting with runny nose, cough, breathing difficulties, chest in‐drawing, and rhonchi, and a placebo to one‐third of the children. There was no significant difference in clinical improvements between the groups (Kabir 2009). Previously conducted studies from all over the world have shown that wheeze is more common in viral infections, particularly respiratory syncytial virus infection (Blanken 2013; Cherian 1990; Eiland 2009; Forgie 1991; Nunez 1988; O’Brien 2015; Piedimonte 2014; Scheltema 2018; Selwyn 1990; Sobĕslavský 1997; Videla 1998; Weber 1998; Wennergren 2001; Yoshihara 2013). Similarly, a recent Cochrane Review of infants with bronchiolitis found no difference in clinical outcomes when comparing antibiotics with a placebo (Farley 2014).
Why it is important to do this review
Two previously published Cochrane Reviews focused on short‐course versus long‐course antibiotic therapy for non‐severe pneumonia and severe pneumonia in children (Haider 2011; Lassi 2017). There are no reviews exploring whether supportive treatment without antibiotics in non‐severe pneumonia may or may not be beneficial. As WHO guidelines do not make a distinction between viral and bacterial pneumonia, children diagnosed with non‐severe pneumonia continue to receive antibiotics because of the concern that it may not be safe to do otherwise. Therefore, it is essential to explore the role of antibiotics in children with WHO‐defined non‐severe pneumonia and wheeze, and to develop effective guidelines for initial antibiotic treatment.
Objectives
To evaluate the efficacy of antibiotic therapy versus no antibiotic therapy for children aged 2 to 59 months with World Heatlh Organization (WHO)‐defined non‐severe pneumonia and wheeze.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs) evaluating the efficacy of antibiotic therapy versus no antibiotic therapy for children aged 2 to 59 months with non‐severe pneumonia and wheeze. We considered trials that defined non‐severe pneumonia as cough,difficulty in breathing, or fast breathing with a respiratory rate above the 2014 WHO‐defined age‐specific values (respiratory rate of 50 breaths per minute or more for children aged 2 to 12 months, or a respiratory rate of 40 breaths per minute or more for children aged 12 to 59 months), or chest indrawing and wheeze (WHO 2019). We excluded non‐RCTs.
Types of participants
We included trials involving children aged 2 to 59 months, with a cough, or difficulty in breathing, or rapid breathing or chest indrawing (as per 2014 WHO‐classified non‐severe pneumonia), and wheeze, i.e. breathing with a hoarse whistling or rattling sound in the chest, as a result of obstruction in the air passages.
We excluded trials involving children with severe or very severe pneumonia (defined on the 2014 basis of inability to drink, convulsions, and difficulty waking). We excluded trials involving children with any chronic illness or conditions requiring antibiotics; children who had been hospitalised in the past two weeks, or who had received antibiotics in the past 48 hours; or children who had experienced measles within the last month.
Types of interventions
Any antibiotic therapy compared with no other medical treatment, or placebo.
Types of outcome measures
Primary outcomes
-
Clinical cure: symptomatic and clinical recovery by the end of treatment (if the child had a fever, their temperature returned to normal, and the respiratory rate dropped to normal)
-
Treatment failure: presence of any of the following: development of chest in‐drawing, convulsions, drowsiness, or inability to drink at any time, respiratory rate above the age‐specific cut‐off point on completion of treatment
Secondary outcomes
-
Relapse: defined as children who were declared 'cured' but who developed recurrence of disease at follow‐up within a defined period
-
Mortality: death within one month
-
Treatment harms: any adverse events or side effects associated with antibiotic therapy
Search methods for identification of studies
Electronic searches
We searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2020 Issue 5) in the Cochrane Library (accessed 23 December 2020), which contains the Cochrane Acute Respiratory Infections (ARI) Group's Specialised Register, MEDLINE (1946 to 23 December 2020), Embase (January 2010 to 23 December 2020), CINAHL (1981 to 23 December 2020), LILACS (Latin American and Caribbean Health Science Information database; 1982 to 23 December 2020), Networked Digital Library of Theses and Dissertations (23 December 2020), and Web of Science (1985 to 23 December 2020).
We used the same search strategy to search CENTRAL and MEDLINE (Appendix 1). We combined the MEDLINE search with the Cochrane highly sensitive search strategy for identifying randomised trials in MEDLINE: sensitivity‐ and precision‐maximising version (2008 revision); Ovid format (Lefebvre 2011). We adapted the search strategy to search Embase (Appendix 2), CINAHL (Appendix 3), LILACS (Appendix 4), Web of Science (Appendix 5), and the Networked Digital Library of Thesis and Dissertations (NDLTD; Appendix 6). To identify child trials in each of the electronic databases, we combined the topic terms with a filter for identifying child trials, based on the work of Boluyt (Boluyt 2008). We also searched the WHO International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp/en/ (searched 23 December 2020), and the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov/ (23 December 2020) for completed and ongoing trials.We did not impose any language or publication restrictions.
Searching other resources
We searched the Database of Abstracts of Reviews of Effects (DARE), in the Cochrane Library (www.thecochranelibrary.com; accessed 23July 2013) in order to scan through the reference lists of relevant reviews. In addition, we searched related conference proceedings for relevant abstracts. We also tried to contact organisations and researchers in the field, and pharmaceutical companies for information on unpublished and ongoing trials. We checked the reference lists of all the included (Awasthi 2008a; Ginsburg 2019; Hazir 2011), excluded (Agarwal 2004; Fontoura 2010; Ginsburg 2020; Hazir 2007; MASCTM Pakistan 2002; McCollum 2019) and awaiting classification (Jehan 2020) studies identified by the above methods.
Data collection and analysis
Selection of studies
Two review authors (ZSL, ZAP) independently assessed the eligibility of the trials. We selected potentially relevant trials by screening the titles and abstracts. If we were not able to ascertain the relevance of trials by screening the title and abstract, we retrieved and reviewed the full text of the article. We resolved disagreements by discussion.
Data extraction and management
We carried out data extraction using a data extraction form designed and piloted by the review authors. Two review authors (ZSL, ZAP) independently extracted the data. We resolved discrepancies through discussion. We extracted the following information:
-
study setting (for example country, type of facility, and type of population);
-
description of antibiotic used (including type of drug, dose, duration, and frequency);
-
ethics approval for trial protocol and informed consent from trial participants (parents or guardian in this case);
-
sample size;
-
length of follow‐up;
-
randomisation procedure and blinding information;
-
outcomes as listed above; and
-
funding or sponsorship for the trial.
Assessment of risk of bias in included studies
Two review authors (ZSL, ZAP) independently assessed the risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We classified each risk of bias as low, high, or unclear, with information directly from the study. We assessed risk of bias according to the following domains (Appendix 7).
-
Random sequence generation
-
Allocation concealment
-
Blinding of participants and personnel
-
Blinding of outcome assessment
-
Incomplete outcome data
-
Selective outcome reporting
-
Other bias
Measures of treatment effect
We entered outcome data for the study into data tables in Review Manager 5 to calculate the treatment effects (Review Manager 2014). Since all the outcomes were dichotomous, we used risk ratio (RR) with 95% confidence intervals (CIs).
Unit of analysis issues
We planned to deal with cluster‐RCTs and cross‐over trials as specified in the Cochrane Handbook for Systematic Reviews of Interventions, but the included trials were RCTs (Higgins 2011).
Dealing with missing data
We contacted the authors of two included trials (Awasthi 2008a; Hazir 2011), and the author of the trial awaiting classification (Jehan 2020) to obtain missing numerical outcome data. We contacted Hazir 2011 to obtain the data of 456 children with non‐severe pneumonia and wheeze from the 900 children included in the trial. We contacted Awasthi 2008a to obtain data on adverse effects of antibiotic therapy on the children with pneumonia, and we contacted Jehan 2020 for the data of included children with non‐severe pneumonia stratified by wheeze. When this was not possible, and the missing data were thought to introduce serious bias, we planned to explore the impact of including such trials in the overall assessment of results by a sensitivity analysis. However, we were unable to conduct sensitivity analysis because all trials were at low risk for allocation concealment.
Assessment of heterogeneity
We applied tests for heterogeneity between trials, using the I² statistic or P value of the Chi² test. We planned to explore high levels of heterogeneity amongst the trials (I² > 50% and Chi² test P < 0.1) by subgroup analysis, as specified below. We performed meta‐analysis using the random‐effects model due to diverse study setting and difference in time of outcome assessment.
Assessment of reporting biases
We had planned to use funnel plots to assess reporting biases.
Data synthesis
We carried out statistical analysis using the Review Manager 5 software (Review Manager 2014). We used a Mantel–Haenszel fixed‐effect meta‐analysis for combining data when it was reasonable to assume that trials estimated the same underlying treatment effect, i.e. when trials were examining the same intervention, and we judged the trials' populations and methods sufficiently similar. In case of clinical heterogeneity, we planned to use a Mantel–Haenszel random‐effects meta‐analysis to produce an overall summary, if we considered an average treatment effect across trials clinically meaningful.
Subgroup analysis and investigation of heterogeneity
We had planned to perform a subgroup analysis on the following:
-
type of antibiotic used;
-
dosage and frequency of antibiotics used; and
-
industry‐sponsored trials versus other types of trials.
Sensitivity analysis
We had planned to carry out a sensitivity analysis to explore the effect of trial quality (assessed by concealment of allocation), by excluding trials with clearly inadequate allocation concealment. However, we were unable to conduct sensitivity analysis because all trials were at low risk for allocation concealment.
Summary of findings and assessment of the certainty of the evidence
We used the GRADE approach to assess the quality of evidence. We constructed summary of findings Table 1 to report the primary and secondary outcomes, using GRADE criteria (Higgins 2011), and GRADEpro GDT software (GRADEpro GDT). It considers within‐study risk of bias (methodological quality), directness of evidence, heterogeneity, precision of effect estimates, and risk of publication bias. We graded the certainty of evidence for each outcome as high, moderate, low, or very low. A high‐certainty rating of the evidence means that we are very confident that the estimated effect lies close to that of the estimate of the true effect; moderate‐certainty means that the estimated effect is probably close to the true effect; a low‐certainty rating means that the estimated effect might substantially differ from the estimate of the effect; and very low‐certainty means that we have very little confidence in the effect estimate, and the estimated effect is probably markedly different from the true effect.
Results
Description of studies
See: Characteristics of included studies, Characteristics of studies awaiting classification and Characteristics of excluded studies tables.
Results of the search
We identified 5440 records using the searches outlined in Electronic searches, and 18 through other sources. After removing duplicates, we screened a total of 3505 titles and abstracts for possible inclusion, and further assessed a total of seven full‐text articles for eligibility. Only three trials met the inclusion criteria (Awasthi 2008a; Ginsburg 2019; Hazir 2011), and one study was categorised under study awaiting classification (Jehan 2020) due to unavailability of stratified data of children with and without wheeze. Awasthi 2008a was initially excluded by the review authors based on the abnormal X‐ray results in a low‐income setting (where no X‐ray was available). However, the study was included in this update due to re‐screening based on the eligibility criteria. See Figure 1.
Study flow diagram
Included studies
Design
We included randomised controlled trials (RCT).
Setting
The included trials were conducted in four hospitals in three cities of Pakistan (Islamabad, Lahore, and Rawalpindi) (Hazir 2011); in outpatient departments of eight referral hospitals in India (Awasthi 2008a), and in Kamuzu Central Hospital (KCH) and Bwaila District Hospital in Lilongwe, Malawi, Africa (Ginsburg 2019).
Participants
The trials included a total of 3256 children aged from 2 to 59 months with non‐severe pneumonia and wheeze. Awasthi 2008a and Hazir 2011 included children with complaints of cough, rapid respiration, or difficulty in breathing classified according to 2010 WHO guideline. Ginsburg 2019 included children with non‐severe fast‐breathing pneumonia classified according to 2014 WHO guideline.
Awasthi 2008a investigated 1674 children, and Ginsburg 2019 investigated 1126 children. Hazir 2011 included 900 children with cough, difficulty in breathing, or both, and fast breathing, of which 456 children were identified with non‐severe pneumonia and wheeze (classified according to 2010 WHO guideline). We obtained the subset data of 456 children from Hazir 2011 as it was not reported in the published paper.
Intervention
The included studies assessed the effect of oral amoxicillin versus placebo. Doses of oral amoxicillin were 31 mg/kg/day to 54 mg/kg/day (Awasthi 2008a), 250 mg (500 mg/day for children aged 2 to 11 months, 1000 mg/day for children aged 12 to 35 months, 1500 mg/day for children aged 36 to 59 months) (Ginsburg 2019), and 5 mg/kg for 3 days (Hazir 2011).
Children were also given oral salbutamol and paracetamol when required.
Outcomes
Hazir 2011 reported on clinical cure. Awasthi 2008a and Hazir 2011 reported on mortality. Awasthi 2008a, Ginsburg 2019, and Hazir 2011 reported on treatment failure, relapse, and treatment harms.
Support and sponsorship
The included trials were supported by the US AID through INCLEN and IndiaClen (Awasthi 2008a), the ARI Research Cell, Children Hospital, Pakistan Institute of Medical Sciences, Islamabad, Pakistan (Hazir 2011), and by the grant from the Bill & Melinda Gates Foundation (Ginsburg 2019).
Excluded studies
We excluded 12 trials after reviewing the full text of the trials (Agarwal 2004; Awasthi 2008b; Fontoura 2010; Ginsburg 2020; Hazir 2007; MASCTM Pakistan 2002; McCollum 2019; NCT00130013; NCT01200706; NCT01375426; NCT03031210; NCT03208361). We excluded of 11 these trials as they had neither a placebo group nor a no treatment arm, and hence did not meet our inclusion criteria (Agarwal 2004; Fontoura 2010; Ginsburg 2020; Hazir 2007; MASCTM Pakistan 2002; NCT03208361). We excluded McCollum 2019 because it included children without wheeze.
Studies awaiting classification
One study was identified as awaiting classification (Jehan 2020). The study was conducted in primary healthcare centres in low‐income communities in Karachi, Pakistan. It included 4002 children with complaints of cough, rapid respiration, or difficulty in breathing classified according to the 2010 WHO guideline. It included children both with and without wheeze. The study provided oral amoxicillin for 3 days based on WHO‐IMCI weight bands (500 mg every 12 hours for children who weighed 4 to < 10 kg, 1000 mg every 12 hours for children who weighed 10 to < 14 kg, and 1500 mg every 12 hours for children who weighed 14 to < 20 kg). Outcomes reported were mortality, treatment failure, relapse, and treatment harms. The study was supported by grants from the Joint Global Health Trials Scheme of the Department for International Development, Medical Research Council, Bill and Melinda Gates Foundation, and from the Fogarty International Center of the National Institutes of Health.
Risk of bias in included studies
See Figure 2 for summary of risk of bias of included study.
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
Allocation
Sequence generation
Randomisation was adequately conducted in all the trials, making it low risk of bias. Sequence generation was done through a computer programme in uneven block sizes.
Allocation concealment
The included trials were at low risk of bias for allocation concealment, because the assignment of which drug was concealed from parents, participants, and study personnel.
Blinding
Blinding of participants and personnel
The trials were at low risk of bias due to adequate blinding of participants and personnel. The randomisation list consisted of unique identification numbers and codes (Hazir 2011). The placebo had a similar colour, consistency, taste, and smell as the intervention drug (Awasthi 2008a; Ginsburg 2019; Hazir 2011).
Blinding of outcome assessors
Two trials were at low risk of bias due to adequate blinding of outcome assessors (Awasthi 2008a; Hazir 2011). One trial was at high risk of blinding of outcome assessor due to unblinding of biostatistician (Ginsburg 2019).
Incomplete outcome data
The trials were at low risk of attrition bias. Loss to follow‐up was reported as 1.07% (Awasthi 2008a), 2.7% (Ginsburg 2019), and 3% (Hazir 2011).
Selective reporting
The trials had low risk of reporting bias as all outcomes mentioned in the protocol were reported in the study.
Other potential sources of bias
The trials were free from other sources of potential bias.
Effects of interventions
Primary outcomes
1. Clinical cure
Antibiotic therapy probably results in a little improvement or no difference in pneumonia at day five (risk ratio (RR) 1.02, 95% confidence interval (CI) 0.96 to 1.08; one trial; 456 participants; moderate‐certainty evidence; Analysis 1.1; summary of findings Table 1).
2. Treatment failure
Antibiotic therapy may reduce treatment failure amongst the participants with pneumonia by 20% by day five (RR 0.80, 95% CI 0.68 to 0.94; three trials; 3222 participants; low‐certainty evidence; Analysis 1.2; summary of findings Table 1).
Secondary outcomes
1. Relapse
Antibiotic therapy may result in little or no difference in relapse of pneumonia within 14 days of follow‐up (RR 1.00, 95% CI 0.74 to 1.34; three trials; 2795 participants; low‐certainty evidence; Analysis 1.3; summary of findings Table 1).
2. Mortality
Mortality was not reported in either of the groups (two trials; 2112 participants; Analysis 1.4).
3. Treatment harms
Antibiotic therapy may result in a little reduction to no difference in treatment harms within 14 days of follow‐up (RR 0.81, 95% CI 0.60 to 1.09; three trials, 3253 participants; low‐certainty evidence; Analysis 1.5; summary of findings Table 1).
Awasthi 2008a reported cases of hospitalisation, diarrhoea (with some dehydration and without dehydration), rash (without itch), tremors, and mild nausea and vomiting. Hazir 2011 did not provide any details on adverse effects caused by antibiotic therapy.
Discussion
Summary of main results
Antibiotic therapy as an intervention for children aged 2 to 59 months with non‐severe pneumonia and wheeze has not been studied comprehensively. Our review included only three multi‐centre, double‐blind, randomised, placebo‐controlled trials (3256 children), conducted in Pakistan, Malawi and India. Two trials followed the 2010 WHO guidelines (WHO 2010) and one trial followed the 2014 WHO guidelines for diagnosing non‐severe pneumonia in children (WHO 2014). The children were treated with a three‐day course of amoxicillin or placebo, and were followed up for a total of two weeks. One study is awaiting classification (Hazir 2011).
Antibiotic therapy may reduce treatment failure amongst participants with pneumonia by 20% by day five. Antibiotic therapy had no impact on clinical cure, relapse, and treatment harms. Mortality was not reported in either group.
Awasthi 2008a reported cases of hospitalisation, diarrhoea (with some dehydration and without dehydration), rash (without itch), tremors, and mild nausea and vomiting.
Overall completeness and applicability of evidence
We conducted a comprehensive search, but identified only three trials for inclusion. Additional information was acquired by contacting the trial authors of two of the included studies. The paucity of eligible trials for inclusion demonstrates a need for robust evidence on the effect of antibiotic therapy in children aged 2 to 59 months with WHO‐defined non‐severe pneumonia and wheeze. We were unable to provide a precise conclusion due to a lack of published RCTs on the subject.
Quality of the evidence
We judged the included trials to be at low risk of bias for random sequence generation, allocation concealment, blinding of participants and personnel, incomplete data outcome, and selective reporting. Only one study was at high risk of bias for blinding of outcome assessors.
We assessed the overall certainty of evidence for clinical cure as moderate. However, we assessed the certainty of evidence for treatment failure, relapse, and treatment harms as low, due to downgrading for risk of bias and imprecision.
Potential biases in the review process
We were aware of the possibility of introducing bias at every stage of the reviewing process. Therefore, we tried to minimise bias in a number of ways: two review authors independently assessed eligibility for inclusion, carried out data extraction, and assessed risk of bias, following Cochrane methodology. Nevertheless, the process of assessing risk of bias, for example, is not an exact science, and includes many personal judgements.
While we attempted to be as inclusive as possible in the search strategy, the literature identified was predominantly written in English, and published in North American and European journals. Although we did attempt to assess reporting bias, time constraints meant that this assessment largely relied on information available in the published trial reports, and thus, reporting bias was not usually apparent. We could not generate funnel plots due to small number of included studies.
Agreements and disagreements with other studies or reviews
We only identified one review investigating the use of antibiotics for the treatment of community acquired pneumonia in neonates and children based on the 2014 WHO guidelines; however, this review only studied the use of antibiotics and did not pool the outcome data to study the effect of antibiotics on clinical cure, relapse, mortality and treatment failure (Mathur 2018).
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
Comparison 1: Antibiotic therapy versus placebo, Outcome 1: Clinical cure
Comparison 1: Antibiotic therapy versus placebo, Outcome 2: Treatment failure
Comparison 1: Antibiotic therapy versus placebo, Outcome 3: Relapse
Comparison 1: Antibiotic therapy versus placebo, Outcome 4: Mortality
Comparison 1: Antibiotic therapy versus placebo, Outcome 5: Treatment harms
| Antibiotic therapy versus placebo compared to placebo for children aged 2 to 59 months with WHO‐defined non‐severe pneumonia and wheeze | ||||||
|---|---|---|---|---|---|---|
| Patient or population: children, aged 2 to 59 months, with WHO‐defined non‐severe pneumonia and wheeze | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
| Risk with placebo | Risk with antibiotic therapy versus placebo | |||||
| Clinical cure (symptomatic and clinical recovery by the end of treatment) | Study population | RR 1.02 | 456 | ⊕⊕⊕⊝ | ||
| 896 per 1000 | 914 per 1000 | |||||
| Treatment failure (development of more severe signs and symptoms by the end of treatment) | Study population | RR 0.80 | 3222 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 164 per 1000 | 131 per 1000 | |||||
| Relapse (children who were 'cured' but developed symptoms within 14‐day follow‐up) | Study population | RR 1.00 | 2795 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 60 per 1000 | 60 per 1000 | |||||
| Treatment harms (adverse events or side effects associated with antibiotic therapy within 14‐day follow‐up) | Study population | RR 0.81 (0.60 to 1.09) | 3253 (3 RCTs) | ⊕⊕⊝⊝ | ||
| 55 per 1000 | 45 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aData not precise: the confidence interval included the potential for important benefit or harm | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Clinical cure Show forest plot | 1 | 456 | Risk Ratio (M‐H, Random, 95% CI) | 1.02 [0.96, 1.08] |
| 1.2 Treatment failure Show forest plot | 3 | 3222 | Risk Ratio (M‐H, Random, 95% CI) | 0.80 [0.68, 0.94] |
| 1.3 Relapse Show forest plot | 3 | 2795 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.74, 1.34] |
| 1.4 Mortality Show forest plot | 2 | 2112 | Risk Ratio (M‐H, Random, 95% CI) | Not estimable |
| 1.5 Treatment harms Show forest plot | 3 | 3253 | Risk Ratio (M‐H, Random, 95% CI) | 0.81 [0.60, 1.09] |
