Table 1. Characteristics of included studies.

Study	Setting	Participants	Intervention	Outcomes measured	Treatment failure	Comments
Addo-Yobo et al [6] (non-blind, equivalence study)	Paediatric departments of tertiary care facilities at 9 international sites. Recruitment period: May 1999 to May 2002	Children aged 3–59 months with WHO-defined severe pneumonia. Oral amoxicillin (n = 857) or parenteral penicillin (n = 845)	Intervention arm: admitted for 48 hr and received oral amoxicillin (45 mg/kg/day), if improved discharged with a 5-day course of oral amoxicillin.Control arm: admitted for 48 hr and received parenteral penicillin (200000 IU/kg/day), if improved discharged with a 5-day course of oral amoxicillin, and if treatment failure then treated with a local alternative standard therapy.	Primary outcome: treatment failure up to or at first 48 hr.Secondary outcome: treatment failure at 5 or 14 day follow-up visits.	Any of the following: appearance of danger signs, low oxygen saturation, persistence of lower chest indrawing at 48 hr, life-threatening or serious adverse drug reaction, received another antibiotic, newly diagnosed co-morbid condition, parents or guardian withdrew consent, child left against medical advice, death.	Treatment failure rate was 19% in each arm at 48 hr. 3% of children <12 months, and 6% of >12 months age progressed to very severe disease on oral amoxicillin. Microbiological diagnosis was done. Serious adverse events reported (amoxicillin group –8, penicillin group –22) were: death (12), rash (5), diarrhoea (5), penicillin allergy (2), anaemia and malaria (one), severe malaria (3), and unspecified events (2).
Hazir et al [7](non-blind, equivalence study)	Paediatric departments of tertiary care facilities at 7 sites in Pakistan. Recruitment period: February 2005 to August 2006	Children aged 3–59 months with WHO-defined severe pneumonia. Oral amoxicillin (n = 1025) or parenteral ampicillin (n = 1012)	Hospitalized group: admitted for 48 hr and received parenteral ampicillin (100 mg/kg/day) followed by 3 days oral amoxicillin (80–90 mg/kg/day),Ambulatory group: treatment with 5 days oral amoxicillin (80–90 mg/kg/day).	Primary outcome: treatment failure up to or on day 6.Secondary outcome: treatment failure between day 6 or and day 14; relapse.	Any of the following: clinical deterioration, inability to take oral medication, co-morbid condition requiring antibiotic, persistence of fever >38°C with LCI from day 3 to 6, either fever or LCI alone at day 6, hospitalization related to pneumonia, serious adverse event, LAMA or lost to follow-up, voluntary withdrawal of consent, death.	Treatment failure in hospital and ambulatory arms was 5.8% vs 3.5% at 48 hr and 8.6% vs 7.5% by day 6. Microbiological diagnosis was not done. Five children died (ambulatory group –1, hospitalized group –4). None of the deaths were considered to be associated with study treatment, and there were no serious adverse events reported in the trial.
Bari et al [8] (cluster randomized)	Haripur district in Pakistan, community based involving lady health workers. Recruitment period: April 2008 to December 2009	Children aged 2–59 months with WHO-defined severe pneumonia. Oral amoxicillin (n = 1857) or Co-trimoxazole (n = 1354) followed by referral.	Intervention clusters: oral amoxicillin for 5 days (80–90 mg/kg/day or 375 mg BD for 2–11 months and 625 mg BD for those 12–59 months).Control clusters: first dose of oral co-trimoxazole (2–11 months, SMZ 200 mg plus TMP 40 mg; 12 months - 5 yrs, SMZ 300 mg plus TMP 60 mg) and then referred for standard care.	Primary outcome: development of clinical treatment failure by day 6.Secondary outcome: clinical relapse on days 7–14.	Appearance of a danger sign (unable to drink or breastfeed, convulsions, vomiting after ingestion of food or drink, and abnormally sleepy or difficult to wake), temperature at least 100°F and LCI on day 3, fever or LCI alone on day 6, and change of antibiotic (through self-referral or by carers).	Treatment failure in intervention and control cluster was, 9% and 18% by day 6. Microbiological diagnosis was not done. Three deaths occurred, one of which was in the intervention group. Reported adverse events were diarrhoea (n = 4) and skin rash (n = 1) in the intervention clusters and diarrhoea (n = 3) in the control clusters.
Patel et al [9] (non-blind, equivalence study)	Paediatric departments of tertiary care facilities of 6 centres in India. Recruitment period: October 2008 to March 2011	Children aged 3–59 months with WHO-defined severe pneumonia. Oral amoxicillin at home (n = 505) or oral amoxicillin at hospital (n = 499).	Hospital group: oral amoxicillin (50 mg/kg/day) administered for first 48 hr in the hospital followed by 5 days at home. Home group: 7 days oral treatment with amoxicillin (50 mg/kg/day) at home.	Primary outcome: treatment failure up to or on day 6. Secondary outcome: treatment failure between day 6 or and day 14; relapse.	Any of the following: clinical deterioration after enrolment, change of antibiotic, hospitalization, serious adverse event related to amoxicillin, LAMA, voluntary withdrawal of consent from study from enrolment by day 7, loss to follow up on day 8.	The treatment failures due to clinical deterioration in the home group –5.4% and in the hospital group –7.4%. Microbiological diagnosis was done. Two children died within the first 72 hrs, one in each group. There were no other serious adverse events. Neither of the deaths was considered to be related to oral amoxicillin.
Soofi et al [10] (cluster randomized)	Matiari district in Pakistan, community based involving lady health workers. Recruitment period: February 2008 to March 2010	Children aged 2–59 months with WHO-defined severe pneumonia. Oral amoxicillin (n = 2341) or co-trimoxazole (n = 2069) followed by referral.	Intervention clusters: oral amoxicillin for 5 days (90 mg/kg/day).Control clusters: first dose of oral co-trimoxazole (2–11 months, SMZ 200 mg plus TMP 40 mg; 12 months - 5 yrs, SMZ 300 mg plus TMP 60 mg) and then referred.	Primary outcome: treatment failure up to or on day 6.Secondary outcome: treatment failure between days 7–14; relapse.	Any of the following: appearance of any signs of very severe pneumonia, change of antibiotic treatment without objective criteria of treatment failure, persistence of fever >38°C with LCI on day 3 (after 48 h of initiation of treatment), either fever >1000F or LCI alone at day 6, death.	Treatment failure in intervention and control cluster was, 8% and 13% by day 6. Microbiological diagnosis was not done. Recorded 3 deaths, by day 6 (2 in the intervention group), and by day 9 (1in the control group).

Note: LCI = lower chest indrawing; LAMA = left against medical advice; SMZ = sulfa-methoxazole; TMP = trimethoprim; BD = twice daily.