Short-Course vs Long-Course Antibiotic Therapy for Children With Nonsevere Community-Acquired Pneumonia

Key Points

Question

Is a shorter course of antibiotics noninferior to a longer course for the treatment of childhood nonsevere community-acquired pneumonia?

Findings

In this systematic review and meta-analysis of 11 143 children with nonsevere community-acquired pneumonia, more than 95% of the participants were aged 2 to 59 months, and treatment failure occurred in 12.8% vs 12.6% of those randomized to a shorter vs a longer course of antibiotics. The comparison met the prespecified 5% noninferiority margin.

Meaning

Clinicians should consider prescribing a shorter course of antibiotics for nonsevere community-acquired pneumonia in children aged 2 to 59 months in accordance with antimicrobial stewardship principles.

Abstract

Importance

Short-course antibiotic therapy could enhance adherence and reduce adverse drug effects and costs. However, based on sparse evidence, most guidelines recommend a longer course of antibiotics for nonsevere childhood community-acquired pneumonia (CAP).

Objective

To determine whether a shorter course of antibiotics was noninferior to a longer course for childhood nonsevere CAP.

Data Sources

MEDLINE, Embase, Web of Science, the Cochrane Library, and 3 Chinese databases from inception to March 31, 2022, as well as clinical trial registries and Google.com.

Study Selection

Randomized clinical trials comparing a shorter- vs longer-course therapy using the same oral antibiotic for children with nonsevere CAP were included.

Data Extraction and Synthesis

Random-effects models were used to pool the data, which were analyzed from April 15, 2022, to May 15, 2022. Grading of Recommendations Assessment, Development and Evaluation (GRADE) was used to rate the quality of the evidence.

Main Outcomes and Measures

Treatment failure, defined by persistence of pneumonia or the new appearance of any general danger signs of CAP (eg, lethargy, unconsciousness, seizures, or inability to drink), elevated temperature (>38 °C) after completion of treatment, change of antibiotic, hospitalization, death, missing more than 3 study drug doses, loss to follow-up, or withdrawal of informed consent.

Results

Nine randomized clinical trials including 11 143 participants were included in this meta-analysis. A total of 98% of the participants were aged 2 to 59 months, and 58% were male. Eight studies with 10 662 patients reported treatment failure. Treatment failure occurred in 12.8% vs 12.6% of participants randomized to a shorter vs a longer course of antibiotics. High-quality evidence showed that a shorter course of oral antibiotic was noninferior to a longer course with respect to treatment failure for children with nonsevere CAP (risk ratio, 1.01; 95% CI, 0.92-1.11; risk difference, 0.00; 95% CI, –0.01 to 0.01; I² = 0%). A 3-day course of antibiotic treatment was noninferior to a 5-day course for the outcome of treatment failure (risk ratio, 1.01; 95% CI, 0.91-1.12; I² = 0%), and a 5-day course was noninferior to a 10-day course (risk ratio, 0.87; 95% CI, 0.50-1.53; I² = 0%). A shorter course of antibiotics was associated with fewer reports of gastroenteritis (risk ratio, 0.79; 95% CI, 0.66-0.95) and lower caregiver absenteeism (incident rate ratio, 0.74; 95% CI, 0.65-0.84).

Conclusions and Relevance

Results of this meta-analysis suggest that a shorter course of antibiotics was noninferior to a longer course in children aged 2 to 59 months with nonsevere CAP. Clinicians should consider prescribing a shorter course of antibiotics for the management of pediatric nonsevere CAP.

This systematic review and meta-analysis investigates whether a shorter course of antibiotics is noninferior to a longer course for the treatment of nonsevere community-acquired pneumonia in children aged 2 to 59 months.

Introduction

Community-acquired pneumonia (CAP) is one of the most common serious infections in children. Annually, an estimated 920 000 children younger than 5 years die of pneumonia.¹ Although respiratory viruses are the most common cause of childhood CAP, most deaths are attributed to bacterial infection.² Therefore, providing an effective and safe antibiotic treatment course is of great importance. Moreover, limiting antibiotic use to the shortest effective duration is critical for enhancing adherence and for reducing adverse drug effects, antimicrobial resistance, and costs.³ However, an inappropriately shortened course of therapy may lead to treatment failure, impaired lung function, or, in some cases, even death,⁴ which raises a vital and difficult question for health care workers: how long should antibiotics be given to a child with CAP?

The World Health Organization recommends a 3-day course of oral amoxicillin for treatment of fast-breathing pneumonia in immunocompetent children and a 5-day course for chest-indrawing pneumonia in children.⁵ In contrast, most national guidelines in both high- and low-income countries recommend antibiotics for 5 to 10 days (eTable 1 in the Supplement). However, these recommendations are based on sparse evidence.⁶ Therefore, a systematic review is needed so that evidence-based recommendations can be formulated to guide the proper duration of antibiotic treatment for children with CAP.

A Cochrane review published in 2008 identified 4 trials and concluded that a 3-day course of antibiotic therapy was as effective as a 5-day course for pediatric CAP.⁷ However, new trials have emerged,^8,9,10,11,12 including a trial reported with contrary findings; a 3-day regimen had a higher failure rate than a 5-day or 10-day regimen, raising concerns about the appropriateness of the short-course recommendation.¹² Therefore, an update of the Cochrane review is needed to examine these discrepancies.¹³ Furthermore, the previous review did not analyze a number of important outcomes, including antibiotic-associated adverse events, antimicrobial resistance, patients’ or their caregivers’ absenteeism, and cost. In addition, the previous review examined only children with fast-breathing pneumonia in low-income countries. It is unclear whether these results can be extrapolated to other populations and settings.

To resolve the gaps, we performed a systematic review and meta-analysis to investigate whether a shorter course of antibiotics is noninferior to a longer course for the treatment of childhood CAP in different populations and settings. The findings from our review can help clinicians in their daily practice and may inform future clinical guidelines.

Methods

Our systematic review and meta-analysis were conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions Version 6.3,¹⁴ and reporting conformed to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 2020 statement.¹⁵ The review protocol was prospectively registered with PROSPERO (CRD42022321435).

Search Strategy and Selection Criteria

We searched MEDLINE (via PubMed), Embase, Web of Science, the Cochrane Library, China Biology Medicine, China National Knowledge Infrastructure, and the WanFang Data databases for eligible studies from inception to March 31, 2022, without language restrictions. Further details of the search strategy can be found in eTable 2 in the Supplement. We also searched ClinicalTrials.gov, the World Health Organization International Clinical Trial Registry Platform, and Google.com. Searches were supplemented by hand searching the reference lists of the included publications and previous meta-analyses.

We included randomized clinical trials comparing a shorter course (eg, 3 days vs 5-10 days and 5 days vs 7-10 days) with a longer course of therapy using the same oral antibiotic for children (<18 years) with nonsevere CAP. Community-acquired pneumonia was defined as pneumonia acquired outside of the hospital.⁸ Nonsevere CAP was defined as the absence of any general danger signs of CAP (eg, lethargy, unconsciousness, seizures, or inability to drink) and not requiring referral or injection therapy.⁵ We excluded trials that included only neonates, in which treatment groups received different antibiotics even if 1 group received a shorter course (eg, azithromycin for 3 days vs cotrimoxazole for 5 days) or different doses of antibiotics (eg, standard vs double dose of amoxicillin), and publications that did not present research findings (eg, narrative reviews, protocols, opinions, editorials, and reports). Two groups of investigators (group 1: Q.L. and Q.Z.; group 2: L.S., G.Z., and X.T.) independently performed study selection; discrepancies were resolved through consultation with a third investigator (Z.L.).

Data Extraction and Quality Assessment

Two groups of investigators extracted data independently. Data from each study were tabulated and checked by a third investigator (Z.L.) before analysis. We used a predesigned spreadsheet to collect the following variables: age, sex, and country; diagnostic criteria, as well as classification and infection types of pneumonia; type, dose, frequency, and duration of antibiotics; length of follow-up; and outcomes. The primary outcome was treatment failure, which was defined by persistence of pneumonia or the new appearance of any general danger signs of CAP, elevated temperature (>38 °C) after completion of treatment, change of antibiotic, hospitalization, death, missing more than 3 study drug doses, loss to follow-up, or withdrawal of informed consent. Secondary outcomes included relapse, adverse events, antimicrobial resistance, patients’ or their caregivers’ absenteeism, and direct medical costs. Outcomes are defined in eTable 3 in the Supplement. Intention-to-treat data sets were selected if more than 1 set of results was reported. The outcome recorded at the last visit was selected if the patients were assessed more than once. Authors were contacted if data were missing or if the reporting format was not suitable for the meta-analysis.

We assessed the risk of bias with the Cochrane Risk of Bias tool.¹⁴ We classified risk of bias as low, high, or unclear for each study. Disagreements in these assessments were resolved by a third investigator (Y.C.).

Statistical Analysis

For dichotomous outcomes, data are presented as pooled risk ratios (RRs) and 95% CIs. To facilitate interpretability, we also present risk differences (RDs) according to the probability of achieving the noninferiority margin. Continuous variables are presented as mean differences with 95% CIs. The noninferiority margin has not been not definitively established; however, the generally accepted margin in the literature is 5% to 10%.^{8,9,10,11,12,16,17,18,19} We conservatively selected the smallest margin (5%) to avoid overestimating treatment effects. Heterogeneity was assessed with the I² statistic, and the values above 50% suggested substantial statistical heterogeneity.²⁰ We used the Mantel-Haenszel random-effects model because of diverse study settings and differences in the definitions of pneumonia and treatment failure.

We performed prespecified subgroup analyses on the following variables: method of diagnosis of pneumonia (clinical diagnosis or radiographically confirmed diagnosis), classification of pneumonia (fast-breathing pneumonia or chest-indrawing pneumonia), country income level (high income or low and middle income), age range (2-59 months or 5-10 years), duration of antibiotics (3 days, 5 days, 7 days, or 10 days), type of antibiotics (amoxicillin or cotrimoxazole), dose of amoxicillin (high dose or low dose), and frequency of administration of amoxicillin (twice daily or 3 times daily). Further details of the definitions of the subgroups can be found in eTable 3 in the Supplement. We performed a sensitivity analysis to study the association of different definitions of pneumonia and treatment failure with the results by excluding 1 trial for every analysis. Small study effects were assessed by the Egger test,²¹ with 2-sided P values; the threshold for significance was .05. We performed data analyses with Stata, version 15.0 (StataCorp), and RevMan, version 5.4 (Nordic Cochrane Center, Cochrane Collaboration). Data were analyzed from April 15, 2022, to May 15, 2022.

We assessed the quality of the evidence with the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for all outcomes.²² The quality of the evidence was rated as high, moderate, low, or very low.

Results

Initial literature retrieval produced 7978 articles, and 34 full-text articles were assessed for eligibility. After exclusion of 25 studies (eTable 4 in the Supplement), 9 randomized clinical trials involving 11 143 participants were included in the review^{8,9,10,11,12,16,17,18,19} (Figure 1). Among the 9 included studies, all participants were immunocompetent, not infected with HIV, and without underlying medical conditions; 58% were male and 42% were female; and 98% of the participants were aged 2 to 59 months. The characteristics of the included randomized clinical trials are shown in Table 1. We collected data on racial and ethnic categories but did not analyze them because only 2 studies reported the data. The sample size of these studies was only 9% of the total sample size, so their results were not representative of the overall racial or ethnic characteristics.

Figure 1. Study Flow Diagram.

CAP indicates community-acquired pneumonia; CBM, China Biology Medicine; CNKI, China National Knowledge Infrastructure; and RCT, randomized clinical trial.

Table 1. Characteristics of the Included Studies.

Source	Country	Sample size	Age	Method of diagnosis	Classification of pneumonia	Type of drug	Dose of drug	Frequency of drug	Duration of drug (intervention vs control), d
SCOUT-CAP,8 2022	US	385	6-71 mo	Clinical	NR	Amoxicillina	80-100 mg/kg/d	Twice daily	5 vs 10
CAP-IT,9 2021	UK and Ireland	824	6-59 mo	Clinical	Chest-indrawing or fast-breathing pneumoniab	Amoxicillin	Standard dose: 35-50 mg/kg/d; high dose: 70-90 mg/kg/d	Twice daily	3 vs 7
SAFER,10 2021	Canada	281	6 mo-10 y	Radiographic	Chest-indrawing or fast-breathing pneumoniab	Amoxicillin	75-100 mg/kg/d	3 Times daily	5 vs 10
Ginsburg et al,11 2020	Africa	3000	2-59 mo	Clinical	Chest-indrawing pneumonia	Amoxicillin	2-11 mo: 500 mg/d; 12-35 mo: 1000 mg/d; 36-59 mo: 1500 mg/d	Twice daily	3 vs 5
Greenberg et al,12 2014	Israel	140	6-59 mo	Radiographic	Not reported	Amoxicillin	80 mg/kg/d	3 Times daily	3 vs 10
									5 vs 10
ISCAP,16 2004	India	2188	2-59 mo	Clinical	Fast-breathing pneumonia	Amoxicillin	31-54 mg/kg/d	3 Times daily	3 vs 5
Kartasasmita et al,17 2003	Indonesia and Bangladesh	2022	2-59 mo	Not reported	Not reported	Cotrimoxazole	Not reported	Not reported	3 vs 5
MASCOT,18 2002	Pakistan	2000	2-59 mo	Clinical or radiographicc	Fast-breathing pneumonia	Amoxicillin	45 mg/kg/d	3 Times daily	3 vs 5
Lupisan et al,19 1999	Philippines	303	2-59 mo	Clinical	Fast-breathing pneumonia	Cotrimoxazole	≥12 mo: 80 mg/dose; 2-12 mo: 40 mg/dose	Twice daily	3 vs 5

Abbreviations: CAP-IT, Community-Acquired Pneumonia: a randomized controlled trial; ISCAP, INDIACLEN Short Course Amoxicillin Pneumonia Study Group; MASCOT, Multicentre Amoxycillin Short Course Therapy; SAFER, Short-Course Antimicrobial Therapy for Pediatric Respiratory Infections; SCOUT-CAP, Short-Course Outpatient Therapy of Community Acquired Pneumonia.

^a A total of 91% of patients received amoxicillin; 9% received amoxicillin-clavulanate or cefdinir.

^b Some of the included patients had chest-indrawing pneumonia and some had fast-breathing pneumonia.

^c All patients received a diagnosis of pneumonia by clinical manifestations. and 14% of them had further confirmation by radiographic assessment.

Eight studies had adequate randomization, allocation concealment, and complete outcome data and were free from selective outcome reporting and other biases; 7 had adequate blinding of the participants and researchers. Insufficient information in 1 study did not allow us to make any judgment.¹⁷ Details of the risk-of-bias assessments are provided in eFigure 1 in the Supplement.

Eight studies with 10 662 patients reported treatment failure.^{8,9,10,11,12,16,17,18} Treatment failure occurred in 12.8% vs 12.6% of participants randomized to a shorter vs a longer course of antibiotics. Overall, a shorter course of oral antibiotics was noninferior to a longer course with respect to treatment failure for children with CAP (RR, 1.01; 95% CI, 0.92-1.11; RD, 0.00; 95% CI, –0.01 to 0.01; I² = 0%) (Figure 2 and eFigure 2 in the Supplement). In the subgroup analysis, noninferiority was met for children aged 2 to 59 months (RR, 1.01; 95% CI, 0.91-1.11; RD, 0.00; 95% CI, –0.01 to 0.01; I² = 0%) (Table 2) but not met for children older than 5 years (RR, 2.07; 95% CI, 0.76-5.63; RD, 0.15; 95% CI, –0.05 to 0.36) (Table 2). A 3-day course of antibiotic treatment was noninferior to a 5-day course for the outcome of treatment failure (risk ratio, 1.01; 95% CI, 0.91-1.12; I² = 0%), and a 5-day course was noninferior to a 10-day course (risk ratio, 0.87; 95% CI, 0.50-1.53; I² = 0%). Noninferiority continued to be met in other subgroups, except the comparison between a 3-day and a 10-day course (Table 2).

Figure 2. Results for the Outcome of Treatment Failure.

CAP-IT indicates Community-Acquired Pneumonia: a randomized controlled trial; ISCAP, INDIACLEN Short Course Amoxicillin Pneumonia Study Group; MASCOT, Multicentre Amoxycillin Short Course Therapy; RR, risk ratio; SAFER, Short-Course Antimicrobial Therapy for Pediatric Respiratory Infections; and SCOUT-CAP, Short-Course Outpatient Therapy of Community Acquired Pneumonia.

^aThree days vs 10 days.

^bFive days vs 10 days.

Table 2. Subgroup Analyses of the Outcome of Treatment Failure.

Variable	No. of trials	Risk difference (95% CI)	Relative risk
			Risk ratio (95% CI)	I2, %	P value
Age range
2-59 mo	8	0.00 (–0.01 to 0.01)	1.01 (0.91 to 1.11)	0	.92
5-10 y	1	0.15 (–0.05 to 0.36)	2.07 (0.76 to 5.63)	NA	.15
Duration of antibiotics, d
3 vs 5	4	0.00 (–0.01 to 0.02)	1.01 (0.91 to 1.12)	0	.81
3 vs 7	1	0.00 (–0.04 to 0.05)	1.01 (0.70 to 1.46)	NA	.96
3 vs 10	1	0.40 (0.06 to 0.74)	6.55 (0.41 to 105.10)	NA	.18
5 vs 10	3	0.00 (–0.02 to 0.02)	0.87 (0.50 to 1.53)	0	.64
Method of diagnosis of pneumonia
Clinical	5	0.01 (–0.01 to 0.02)	1.11 (0.98 to 1.27)	0	.11
Radiographically confirmed	3	–0.01 (–0.04 to 0.03)	0.92 (0.64 to 1.32)	0	.65
Classification of pneumonia
Fast breathing	2	0.01 (–0.01 to 0.03)	1.04 (0.90 to 1.21)	0	.57
Chest indrawing	1	0.01 (–0.01 to 0.02)	1.14 (0.85 to 1.55)	NA	.38
Country income level
High	4	0.00 (–0.03 to 0.03)	0.99 (0.73 to 1.34)	0	.95
Low and middle	4	0.00 (–0.01 to 0.02)	1.01 (0.91 to 1.12)	0	.81
Type of antibiotics
Amoxicillin	7	0.00 (–0.01 to 0.01)	1.05 (0.93 to 1.18)	0	.43
Cotrimoxazole	1	0.01 (–0.05 to 0.02)	0.94 (0.79 to 1.11)	NA	.46
Dose of amoxicillin
High	5	0.00 (–0.01 to 0.02)	1.10 (0.86 to 1.39)	0	.45
Low	3	0.00 (–0.02 to 0.02)	1.03 (0.90 to 1.19)	0	.64
Administration frequency of amoxicillin
Twice daily	3	0.00 (–0.01 to 0.02)	1.08 (0.86 to 1.36)	0	.52
Three times daily	4	0.01 (–0.02 to 0.03)	1.04 (0.90 to 1.20)	0	.60

Abbreviation: NA, not applicable.

Six studies with 9447 patients reported rates of relapse.^{10,11,16,17,18,19} Overall, a shorter course of oral antibiotics was noninferior to a longer course (RR, 1.12; 95% CI, 0.94-1.34; RD, 0.00; 95% CI, 0.00-0.01; I² = 0%) (Figure 3; eFigure 3 in the Supplement). In the subgroup analysis, noninferiority was met for children aged 2 to 59 months (RR, 1.10; 95% CI, 0.92-1.32; RD, 0.00; 95% CI, –0.00 to 0.01; I² = 0%) but not met for children older than 5 years (RR, 6.43; 95% CI, 0.32-128.60; RD, 0.07; 95% CI, –0.04 to 0.19). The results for relapses in other subgroup analyses are shown in eFigure 4 in the Supplement.

Figure 3. Results for the Outcome of Relapse.

ISCAP indicates INDIACLEN Short Course Amoxicillin Pneumonia Study Group; MASCOT, Multicentre Amoxycillin Short Course Therapy; RR, risk ratio; and SAFER, Short-Course Antimicrobial Therapy for Pediatric Respiratory Infections.

Five studies with 4475 patients reported adverse events.^8,9,10,11,16 The risks of gastroenteritis (RR, 0.79; 95% CI, 0.66-0.95) and rash (RR, 0.79; 95% CI, 0.65-0.97) were significantly lower in the short-course groups compared with the long-course groups. We found no differences between the interventions in the risk of other nonserious events or any serious adverse events (eFigure 5 in the Supplement).

Four studies with 2876 patients reported antimicrobial resistance, but the results varied between studies.^8,9,16,17 In terms of β-lactamase resistance, the Short-Course Outpatient Therapy of Community Acquired Pneumonia trial reported that the median number of β-lactamase resistance genes per prokaryotic cell during days 19 to 25 was significantly lower during the 5-day treatment compared with the 10-day treatment (0.55 [range, 0.18-1.24] vs 0.60 [range, 0.21-2.45]).⁸ However, the CAP-IT trial (Community-Acquired Pneumonia: a randomized controlled trial) did not identify significant differences in day 28 pneumococcal penicillin nonsusceptibility (14 of 205 vs 7 of 232) or day 28 pneumococcal amoxicillin resistance or nonsusceptibility (2 of 205 vs 2 of 232) between 3-day and 7-day treatments.⁹ In terms of cotrimoxazole resistance, ISCAP (INDIACLEN Short Course Amoxicillin Pneumonia Study Group) reported that the proportion of Streptococcus pneumoniae isolates resistant to cotrimoxazole on day 14 was significantly lower with 3-day compared with 5-day treatment (66.7% vs 78.2%).¹⁶ However, Kartasasmita and Saha¹⁷ found no significant difference in day 15 cotrimoxazole-resistant S pneumoniae (61.5% vs 64.1%) between 3-day vs 5-day treatment.

One study (SAFER [Short-Course Antimicrobial Therapy for Pediatric Respiratory Infections]) reported caregiver and child absenteeism.¹⁰ Caregiver work absenteeism was significantly lower in the 5-day group than in the 10-day group (incident rate ratio, 0.74; 95% CI, 0.65-0.84). Child absenteeism was similar in the 2 groups (incident rate ratio, 0.95; 95% CI, 0.71-1.27). Another study (ISCAP) with 2188 patients reported treatment-related costs. The mean direct medical costs of treating 1000 cases of nonsevere pneumonia were lower in the 3-day treatment group ($1100) than the 5-day treatment group ($1250).¹⁶

The results remained unchanged with sensitivity analysis (eTable 5 in the Supplement). We found no evidence of small study effects for the primary outcome among 8 eligible studies (eAppendix in the Supplement). The quality of evidence for treatment failure was high overall and for children aged 2 to 59 months but low for children older than 5 years owing to imprecision. Further details of the GRADE assessment are summarized in eTable 6 in the Supplement.

Discussion

In this systematic review and meta-analysis, we summarized the evidence from 9 randomized clinical trials comparing a shorter and longer antibiotic course for the treatment of children with CAP. We found that a shorter course of oral antibiotics was as effective as a longer course in treating children aged 2 to 59 months who had nonsevere CAP. We also found that a shorter course of antibiotics was associated with fewer adverse events, including gastroenteritis and rash, and lower caregiver absenteeism, and it was also associated with lower medical costs. Furthermore, our results showed that a 3-day course of antibiotic treatment was noninferior to a 5-day course and that a 5-day course was noninferior to a 10-day course.

A previous Cochrane review reported that a shorter course was as effective as a longer course in treating CAP in children.⁷ However, the authors of that review defined pneumonia solely on the basis of fast breathing. Fast breathing is a nonspecific symptom of pneumonia and may be caused by other diseases, such as bronchiolitis and asthma. This simplified syndromic diagnosis of childhood CAP poses a challenge in the accurate diagnosis of CAP. Chest radiographs are a more accurate approach for confirming pneumonia than fast breathing is.²³ We found that noninferiority continued to be met for patients with radiologically confirmed pneumonia who were more likely to have actual CAP. Thus, our findings are robust across various definitions of CAP.

Although we included children of any age, most of the children were younger than 5 years. Thus, the number of older children was small, and the quality of evidence for older children was rated downward owing to imprecision. Therefore, our results may apply only to children aged 2 to 59 months. More studies are needed to investigate whether short courses of antimicrobial therapy are noninferior to longer courses for children older than 5 years.

Furthermore, we compared the effectiveness of different regimens of short-course amoxicillin and found that short courses were effective when given in lower doses (25-50 mg/kg/d) and less frequently (twice daily). This latter finding is consistent with recommendations from the World Health Organization and other international organizations to administer amoxicillin twice daily^5,24,25 because of the equivalence of thrice-daily and twice-daily dosing demonstrated by several randomized clinical trials and supported by pharmacokinetics studies.^26,27 Moreover, lower frequency of administration could enhance adherence. In contrast to our finding regarding drug dose, the World Health Organization and the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America recommend high doses of amoxicillin in view of resistance in pneumococci.^5,24 With a decrease in penicillin resistance after the introduction of the pneumococcal vaccine,²⁸ it might be plausible to consider a lower dose of amoxicillin. In addition, data from adults suggest that gastrointestinal absorption of amoxicillin may be saturable, limiting the expected utility of high-dose regimens.²⁹ Nevertheless, the appropriate dose of amoxicillin is still controversial, and further studies are needed in this area. Dosage recommendations should be based on local susceptibility data.

Finally, we compared the effectiveness of 3-day, 5-day, and 10-day courses for pediatric CAP. Current recommendations on treatment duration vary greatly across guidelines. The World Health Organization recommends 3 days of amoxicillin for fast-breathing pneumonia and a 5-day course to treat chest-indrawing pneumonia in children.⁵ The recommendation of a 5-day course was not based on systematic reviews but on indirect evidence comparing 5-day oral antibiotics at home with 5-day parenteral antibiotics in the hospital for chest-indrawing pneumonia.^30,31 Our systematic review suggests that a 3-day course was noninferior to a 5-day course for children with chest-indrawing pneumonia. We believe that our current findings may allow for harmonization and simplification of treatment courses to 3 days for both fast-breathing and chest-indrawing pneumonia in children. The studies in high-income countries did not provide separate data on fast-breathing and chest-indrawing pneumonia for subgroup analyses. Therefore, this conclusion may be applicable only to low- and middle-income countries. The Pediatric Infectious Diseases Society, the Infectious Diseases Society of America, and other pediatric societies recommended 7 to 10 days of treatment based on limited and low-quality evidence.^{24,32,33,34,35} High-quality evidence from our review showed that a 5-day regimen might be sufficient. We think our review helps fill an evidence gap and provides a basis for evidence-based decision-making. Guideline developers and clinicians from these countries could consider shifting to a 5-day course for the management of children with CAP. Although current evidence supports the shorter courses, the optimal duration of antibiotic therapy is not “1 size fits all” but should be individualized and consider pathogen factors and patients’ response to treatment.³⁶ However, in clinical practice, knowing the true pathogen responsible for CAP and treating CAP according to the patient’s response are challenging. Future research is needed to promote the individualization of antimicrobial therapy for pediatric CAP.

Limitations

Our review has several limitations. First, we included multiple infection types. It is possible that the optimal duration of antibiotics required differs by different types. However, there were insufficient data for subgroup analyses for each infection type because this information is infrequently reported, given that microbiologic testing is not routinely performed in outpatient and inpatient settings. Even in inpatient settings, neither chest radiographs nor inflammatory biomarkers can reliably discriminate among children with viral, atypical, and bacterial CAP.^37,38 Therefore, new technologies, such as gene expression signatures for rapidly identifying etiologic pathogens in pneumonia, are needed to address these important issues. Second, the definitions of pneumonia and treatment failure varied across studies, which may have led to heterogeneity in results. Therefore, we pooled the data with a random-effects model and found no statistical heterogeneity in the meta-analysis. Furthermore, we performed subgroup analyses and sensitivity analyses and found that the results remained unchanged and that statistical heterogeneity was also not present. Nevertheless, there is a need for greater standardization in future research. Development of a uniform definition of childhood CAP and adherence to core outcome sets of CAP are possible solutions.³⁹ Third, we did not analyze long-term outcomes because of the lack of data. These important long-term outcomes should be monitored in future research.

Conclusions

This systematic review and meta-analysis found high-quality evidence that a shorter course of antibiotics was noninferior to a longer course for the outcome of treatment failure among children aged 2 to 59 months who had nonsevere CAP. Clinicians should consider prescribing a shorter course of antibiotics for the management of pediatric nonsevere CAP in accordance with antimicrobial stewardship principles.

Supplement.

eTable 1. Summary of Recommendations From Guidelines for First-line Antibiotics for Children With Community-Acquired Pneumonia

eTable 2. Search Strategies

eTable 3. Definitions of Terms Used in Included Studies

eTable 4. Characteristics of Excluded Studies

eTable 5. Sensitivity Analysis Excluding Each Study for the Outcome of Treatment Failure

eTable 6. GRADE Assessment (Summary of Findings Table)

eFigure 1. Assessment of the Risk of Bias for Included Studies

eFigure 2. Risk Difference for Treatment Failure Among Pediatric Patients With Community-Acquired Pneumonia

eFigure 3. Risk Difference for Relapse Among Pediatric Patients With Community-Acquired Pneumonia

eFigure 4. Subgroup Analyses of Relapse for Included Studies

eFigure 5. Adverse Events Reported in the Included Studies

eAppendix. Small-Study Effects of the Included Studies

eReferences.