Salam 1998.

Methods	Randomized controlled trial Generation of allocation sequence: computer generated list of random numbers Allocation concealment: allocated by Bayer AG Pharma and not available to researchers, double dummy technique Blinding: participants, providers and outcome assessor blinded Inclusion of all randomized participants: inadequate, 84% Duration: 1 year and 8 months, from August 1995 to March 1997
Participants	Number of participants enrolled: 143 Number of participants analysed: 120 Loss to follow up: 10 Inclusion criteria: children aged 2 years to 15 years; dysentery (passage of grossly bloody‐mucoid stools for 72 hours or less); who had not received any antimicrobial treatment (agent known to be effective in vivo against shigellosis and active in vitro against the Shigella strain isolated from the patient); gave informed consent Exclusion criteria: co‐existing disorders that required antimicrobial therapy
Interventions	Ciprofloxacin suspension (10 mg/kg, every 12 hours, maximum of 500 mg, for 5 days, total 10 doses with placebo of pivmecillinam) Pivmecillinam (15 to 20 mg/kg, maximum of 300 mg, every 8 hours for 5 days, total 15 doses with placebo of ciprofloxacin)
Outcomes	(1) Clinical failure (if patient did not have persistent dysentery on day 3, and if on day 5 a patient had 6 stools or less, no bloody‐mucoid stools, no more than 1 watery stool and no fever) (2) Bacteriological failure (bacteriological success: if the initial Shigella species could not be identified in culture on day 3 or later) (3) Fever less than 24 hours (4) Number of patients with bloody‐mucoid stools more than 3 days (5) Relapse (6) Adverse event ‐ limp (one of the adverse reactions to the antibiotic therapy could be a LIMP on walking due to joint pain caused by the antibiotics) (7) All adverse events
Notes	Location: Bangladesh Setting: participants were hospitalized for 6 days after the first dose and then discharged for follow up Follow‐up period: 180 days Antibiotic sensitivity pattern of Shigella isolates: all in both groups were sensitive to ciprofloxacin. 58/60, in the ciprofloxacin group and 57/60 in the pivmecillinam group were sensitive to pivmecillinam. Funding source(s): Bayer AG, Wuppertal, Germany ICDDR, Bangladesh
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"Drug allocation used a computer‐generated list of random numbers".
Allocation concealment?	Low risk	"...list of random numbers, which was not available to the researchers".
Blinding? All outcomes	Low risk	"...double dummy technique". Participant, provider and outcome assessor blinded.
Incomplete outcome data addressed? All outcomes	High risk	13/143 (6 in the ciprofloxacin group and 7 in the pivmecillinam group) were excluded from analysis because they were found not eligible (12 did not grow Shigella in their stool culture and 1 had taken nalidixic acid before study entry). Further 10 (5 in each group) withdrew before study completion. 84% follow up.
Free of selective reporting?	Low risk	The study's prespecified outcomes, which were of interest in this review, have been reported
Free of other bias?	Low risk	The study appears to be free of other sources of bias