Kayihura 2003.

Methods	RCT. Individual women. 2‐arm study.
Participants	Inclusion criteria Women having an emergency CS. N = 288 but data on 241. Exclusion criteria Women with allergies to antibiotics; use of antibiotics within previous 24 hours; pathology that should be treated with antibiotics; chorioamnionitis; fever on admission; need of transfusion before or during the CS; ruptured membranes > 24 hrs; body weight > 132 kg; elective CS.
Interventions	Intervention: aminoglycoside (G) + nitroimidazole (I) combination. 160 mg gentamicin (G) and metronidazole (I) IV. IV before operation starts, so before cord clamping. No more antibiotics given post‐operatively. N = 143 randomised but 116 in analysis. Comparison: penicillins (A1) + nitroimidazole (I) + macrolide (E) combination. Penicillin 4,000,000 UI IV for 6 hrs and metronidazole 500 mg IV 8 hours during 1st 24 hrs. Then erythromycin 500 mg 6‐hourly orally and metronidazole 500 mg 8‐hourly orally during 6 days. IV for 1st 24 hours then orally for 6 days. No other antibiotics given. N = 145 randomised but 125 in analysis.
Outcomes	Maternal endometritis; would infection; UTI; peritonitis; evisceration; post‐operative infection; stillbirth.
Notes	Dates: January to June 2000. Setting: Maternity Unit, Hospital Central de Maputo, Mozambique. Quaternary level care. Subgroups Non‐elective CS. Mixed: before cord clamping for intervention vs after cord clamping for comparison. IV administration vs combination of IV and oral administration. Single dose of intervention vs multiple dose of comparison. Comparison 15. No information about funding source of study.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	“...randomly constituted...”.
Allocation concealment (selection bias)	Unclear risk	“The anaesthetist administered the antibiotics according to the code written in the exercise book in the sequential order of admission to the theatre...the code was then written on the patient’s file so that the surgeon could prescribe...the doctor on duty was not aware of the group to which the patient was allocated. The principle investigator was not allowed either to select cases to be enrolled in the study or to follow the patients in the ward.”
Blinding of participants and personnel (performance bias) All outcomes	High risk	Non‐blinded comparative study. Although the doctor on duty was reported as not aware of the group to which the woman was allocated, the anaesthetist and surgeon were aware and overall the study was described as not blinded.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Non‐blinded comparative study. “Medical doctors allowing the patients to leave the maternity ward knew the regimen followed by the patients in order to not modify the antibiotics given…”.
Incomplete outcome data (attrition bias) All outcomes	Low risk	288 women randomised and analysis on 241 so 47/288 (16.3%) loss.
Selective reporting (reporting bias)	Unclear risk	We did not assess trial protocol.
Other bias	Unclear risk	Baseline characteristics similar for:age; parity; gestation. However, other possible biases are unclear. No information about funding source of study.