Heisterberg 1986

Methods	‐ Single centre, Denmark (Copenhagen) ‐ Study period not reported ‐ Follow‐up visit after 2 weeks
Participants	‐ 102 women randomised, 81 analysed ‐ Women presenting for surgical abortion in first trimester ‐ Exclusion criteria: allergy to treatment; treatment with antibiotics at the time of abortion; active haematological or neurological disease; alcohol abuse or treatment with disulfiram (Antabuse); positive test for N. gonorrhoeae ‐ Preoperative infections: 1) History of PID: all women, 43/43 in arm 1 and 38/38 in arm 2 2) Chlamydia: not reported 3) Gonorrhoea: all tested, none positive 4) Bacterial vaginosis: not reported
Interventions	Antibiotic prophylaxis compared with alternative regimens of the same antibiotic ‐ Intervention, arm 1: metronidazole 400 mg, oral (peri‐operative 3 doses, 1h before, 4h after and 8h after abortion) ‐ Control, arm 2: pivampicillin 350 mg, oral (peri‐operative 3 doses, 1h before, 4h after and 8h after abortion)
Outcomes	1) Post‐abortal PID defined as: a) Patient seen at follow‐up visit after 2 weeks, at least 3 of the following: temperature > 38 °C; continued pelvic pain; malaise with tender adnexal mass; pathologic discharge or bleeding b) Patient admitted before scheduled follow‐up, at least 2 of the following: temperature > 38 °C; moderate tenderness of the uterus; tender adnexal mass; pathologic discharge or bleeding 2) Re‐admission to hospital
Notes	Not universal antibiotic prophylaxis according to protocol definition; study population included only women without gonorrhoea at baseline. All had a history of PID
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk
Incomplete outcome data (attrition bias)	Low risk