Primary Outcomes
Both treatments reduced postmenstrual spotting statistically significantly compared to baseline (P<0.001, see table 2). Median spotting reduced from 7 days at baseline to 2 days after hysteroscopic niche resection at 3th, 6th, 9th and 12thmonth, while after LNG-IUD it reduced to 2 days at 3thmonth follow-up to 0 day from 6th month onwards (see table 3). Comparing postmenstrual spotting during the first year, using linear mixed model on ITT analyses base, LNG-IUD was more effective in reducing postmenstrual spotting than the hysteroscopic niche resection (p=0.009, see table 2).
Total number of women with more than 2 days reduction in spotting at one-year follow-up without removal of the IUD because of symptoms and without additional applied therapies (the effective rate) was 33 out of 36 (91.7%) in the LNG-IUD group and 28 out of 38 (73.7%) in the hysteroscopic niche resection group (p=0.042). The effective rate increased over time within 1 year after the insertion of LNG-IUD (66.7%, 86.1%, 91.7%, 91.7%, p for trend=0.003), while no trend change was observed in hysteroscopy group (76.3%, 73.7%, 71.1%, 73.7%, p for trend=0.743) (see figure S4). Also the reduction in total bleeding days per month including postmenstrual spotting was statistically significantly stronger in the LNG-IUD group from 6 months onwards (see table 3).
In the LNG-IUD group, the rate of amenorrhea increased during the first year from 28% at 3 months to 39% at 12 months follow-up. And irregular bleeding reduced from 23% at 3 months to 3% at 12 months follow-up. Despite the good results, 11 out of 36 (31%) patients in the LNG-IUD group were concerned about their ovarian function because of the reduction in menstrual bleeding and requested for removal of the LNG-IUD. However, after repeated one-by-one health education about the function of the LNG-IUD all of these patients came back on their request and kept the LNG-IUD.
In the hysteroscopic niche resection group, 2 patients who got pregnant at the 7th and 8th month were amenorrhoic, all others had regular menstrual cycles at 12 months follow-up.
We also conducted the per protocol analyses, 2 patients removed their IUD in the LNG-IUD group and 2 patients got pregnant in the hysteroscopy group and 2 received an additional IUD in the hysteroscopy group and were excluded at the corresponding time points respectively; 34 patients finally finished the one-year follow-up based on the per protocol analyses in both groups (see table S1). The results at each follow up points were comparable with the results of ITT analyses.