Dashow 1986.

Methods	RCT. Individual women. 5‐arm study.
Participants	Inclusion criteria Women undergoing CS. N = 204 in this review (though 360 in study which included a placebo group). Exclusion criteria Women with a history of penicillin or cephalosporin allergy, those taking antibiotics, those with known infectious process at the time of operation (e.g. chorioamnionitis or UTI).
Interventions	Intervention 1: cephalosporin (B1). Cephapirin. 2 g; lavage. N = 70 ‐ reported in Table 4 Post‐operative morbidity (but 70 reported in Table 2 Risk factors). Intervention 2: cephalosporin (B2). Cefamandole. 2 g; lavage. N = 64 ‐ reported in Table 4 Post‐operative morbidity (but 70 reported in Table 2 Risk factors). Comparison: penicillin (A4). Ampicillin. 2g; lavage. N = 70. 4th group ‐ moxalactam disodium (1‐oxa‐beta‐lactam antibiotic: N = 64 in Table 4 (but 79 in Table 2). 5th group ‐ placebo (N = 77). Vitamin added to each solution for disguise.
Outcomes	Endometritis (temperature > 37.8 ºC, uterine tenderness, pelvic irritation without other localizing signs); wound infection (breakdown, positive culture and/or cellulitis): febrile morbidity (temperature > 100.4 x 2. 6 hour apart, excluded first 24 hours); mean duration of hospital stay.
Notes	Dates: 1 December 1982 to 31 May 1984. Setting: Madigan Army Medical Center, US. Subgroups Type of CS not defined. Not defined with respect to timing of cord clamping. Lavage during operation. Not relevant for irrigation. Comparisons 1(subgroup 1); 2; 3; 4. The numbers of women reported in each group differed in the Tables. We will write to the authors to check the figures and in the meantime we have used the data from Table 4 'Post‐operative morbidity outcomes: All patients'. No information about funding source of study.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“A computer‐generated table of pseudo‐random numbers using the mixed congruential method was used by the pharmacy to assign each patient to one of five groups.”
Allocation concealment (selection bias)	Low risk	“A computer‐generated table of pseudo‐random numbers using the mixed congruential method was used by the pharmacy to assign each patient to one of five groups.”
Blinding of participants and personnel (performance bias) All outcomes	Low risk	“Patients and physicians were unaware of the group assignment until after completion of the study...”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	“Patients and physicians were unaware of the group assignment until after completion of the study...”
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	No explanation for differences between the numbers of the initially randomised groups and the groups included in the morbidity analysis (cephapirin 79 vs 70, cefamandole 70 vs 64, moxalactam 64 vs 79. The total of women included is the same (360); therefore we can assume that women originally assigned to one group received other treatment and they were not analysed by intention to treat. The uneven numbers may be due to lack of block‐randomisation.
Selective reporting (reporting bias)	Unclear risk	We did not assess trial protocol.
Other bias	Unclear risk	Baseline characteristics similar on: age; parity; gestation. Small difference on gravidity but not considered clinically important. However, other aspects of bias unclear. No information about funding source of study.