Eleven RCTs were included (1,679 participants). Allocation concealment methods were adequate in one trial, inadequate in five trials, and not reported in five trials. Four trials reported adequate blinding (for patients, outcome assessors, or both), one had inadequate blinding, the methods were not reported in the other six trials. Two trials lost enough patients at follow-up for this to be judged likely to have biased the results.
First- versus second-generation: There were no differences between the groups for the incidence of further surgery (six trials; Ι²=0), the number of patients with menstrual bleeding scores less than 75 at one year (four trials; Ι²=20%), and the rate of amenorrhoea (five trials; Ι²=68%). Second-generation devices had fewer complications (RR 0.52, 95% CI 0.35 to 0.76; seven trials; Ι²=0), shortened the operating time (MD 16.6 minutes, 95% CI 12.1 to 21.2; three trials; Ι²=89%), and could more commonly be used with local anaesthesia (RR 1.87, 95% CI 1.04 to 3.37; three trials; Ι²=74%).
Second- versus second-generation: There was a significantly higher rate of amenorrhoea in patients treated with Novasure than with other second-generation devices (RR 2.60, 95% CI 1.63 to 4.14; four trials; Ι²=24%). There were no differences between these groups for the incidence of further surgery and complications.