Lin 2015b.

Methods	Parallel‐group randomised controlled trial. Single centre, university referral centre, Reproductive Medicine Centre of the Sir Run Run Shaw Hospital, Hangzhou, China. Protocol approved by IRB: yes. Study protocol registered as ISRCTN69690272 in ISRCTN Registry. Statistical power calculation done before start of trial. No conflicts of interest declared by authors. External funding: National Science Foundation of China (81270657), Zhejiang Public Welfare Technology Application Research Project (2013C33236), and Zhejiang Key Science and Technology Innovation Team Project (2011R50013‐26).
Participants	Number recruited: 207 women. Number excluded before randomisation: 6 women (3 surgical complications; 3 declined to participate). Number randomly assigned: 201 women. Number excluded after randomisation: 5 women (1 intervention group; 4 control group) with reason: protocol violation. Number lost to follow‐up: 34 women (16 intervention group; 18 control group) with reason: no second‐look hysteroscopy in time. Number analysed: 162 women (82 intervention group, 80 control group). Inclusion criteria: aged 18‐40 years; moderate‐to‐severe IUA (AFS score ≥ 5); no previous history of hysteroscopic adhesiolysis; written consent obtained; agreement to undergo second‐look hysteroscopy. Exclusion criteria: minimal adhesion (AFS score < 5); previous hysteroscopic adhesiolysis. Mean age (± SD) in intervention group: 29.7 (± 4.3) years. Mean age (± SD) in control group: 30.1 (± 5.1) years. Proportion of women with infertility in intervention group: 21/82 (26%). Proportion of women with infertility in control group: 18/80 (22%). Primary vs secondary infertility: not reported. Study duration: 20 months.
Interventions	Intrauterine balloon (intervention: n = 82) vs IUD (control: n = 80) Hysteroscopic adhesiolysis carried out by 1 of 2 experienced hysteroscopic surgeons with use of 4.5‐mm rigid hysteroscope (Storz) with 5% mannitol perfusion under 100 mmHg pressure. Procedure performed under general anaesthesia in a day surgery unit. Ultrasonographic guidance routinely used; in some cases, laparoscopy was also performed either in exceptionally difficult cases or when there was a need to inspect pelvic organs to rule out pathology such as endometriosis or to verify tubal patency. Once the extent and severity of uterine adhesion had been assessed, adhesions were divided with use of hysteroscopic scissors until normal uterine anatomy was achieved. Intervention group: immediately following operative hysteroscopy specially designed intrauterine balloon (Cook Medical) inflated with 3‐5 mL normal saline fitted into uterine cavity. Control group: immediately following operative hysteroscopy heart‐shaped copper IUD (Yantai Contraceptive Instrument) with thread knitted tail fitted into uterine cavity. Both devices were removed after 1 week in outpatient department. All participants were treated with oral cefuroxime combined with metronidazole for 7 days. In all cases, hormone therapy was also begun from the day of operation, consisting of oestradiol valerate 6 mg/day for 21 days, with medroxyprogesterone acetate 6 mg/day for the last 7 days of the oestrogen therapy. After withdrawal bleed, hormone therapy repeated for another cycle. Second‐look hysteroscopy carried out in early proliferating phase, 1‐2 months after initial operation. After assessment of extent and severity of any reformed adhesions, hysteroscopic adhesiolysis was also carried out at time of second‐look procedure, if adhesions had recurred.
Outcomes	Primary outcomes: adhesion reformation, measured by second‐look hysteroscopy 1‐2 months after surgery. Power calculation done before start of trial; reduction in adhesion scores, measured by second‐look hysteroscopy 1‐2 months after surgery. Severity and extent of IUAs scored according to AFS 1988 classification.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "After the completion of hysteroscopic adhesiolysis, recruited patients were randomized to one of the two treatment groups by computer‐generated numbers…" Comment: probably done; low risk of bias.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not reported. Comment: unclear risk of bias; query not answered.
Blinding of participants and personnel (performance bias) Live birth, pregnancy or miscarriage	Low risk	Blinding of surgeons impossible since balloon and IUD were easily recognised as being different. Blinding of participants not reported but device removed after 1 week at the outpatient department. Comment: probably no blinding of participants and personnel.
Blinding of participants and personnel (performance bias) Adhesions	High risk	Blinding of surgeons impossible since balloon and IUD were easily recognised as being different. Blinding of participants not reported but device removed after 1 week at the outpatient department. Comment: probably no blinding of participants and personnel.
Blinding of outcome assessment (detection bias) Live birth, pregnancy or miscarriage	Low risk	Quote: "The surgeon who later performed the second‐look hysteroscopy was blinded to the randomisation." Comment: probably done; low risk of bias.
Blinding of outcome assessment (detection bias) Adhesions	Low risk	Quote: "The surgeon who later performed the second‐look hysteroscopy was blinded to the randomisation." Comment: probably done; low risk of bias.
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "There were 39 women who were subsequently excluded from the study for the following reasons. In the balloon group, 1 woman was excluded because of protocol violation, in addition to 16 withdrawals because they did not proceed to second‐look hysteroscopy within the specified time frame. In the IUD group, 4 women were excluded because of protocol violation, in addition to 18 withdrawals because they did not proceed to second‐look hysteroscopy within the specified time frame." Comment: proportion of women lost follow‐up 34/201 (17%) women; high risk of attrition bias.
Selective reporting (reporting bias)	High risk	According to study protocol registered as ISRCTN69690272, secondary outcome was pregnancy rate after surgery. Not reported in study report. Comment: at high risk of selective outcome reporting since duration of study was 20 months and no data reported in final review of secondary outcome predefined in registered study protocol.
Other bias	Low risk	Quote: "No difference in baseline characteristics. Co‐treatment with antibiotics and hormone therapy in both comparison groups." Comment: no evidence for other potential sources of bias.