Interventions for heavy menstrual bleeding; overview of Cochrane reviews and network meta-analysis

Background: Heavy menstrual bleeding (HMB) is excessive menstrual blood loss that interferes with women's quality of life, regardless of the absolute amount of bleeding. It is a very common condition in women of reproductive age, affecting 2 to 5 of every 10 women. Diverse treatments, either medical (hormonal or non-hormonal) or surgical, are currently available for HMB, with different effectiveness, acceptability, costs and side effects. The best treatment will depend on the woman's age, her intention to become pregnant, the presence of other symptoms, and her personal views and preferences.

Objectives: To identify, systematically assess and summarise all evidence from studies included in Cochrane Reviews on treatment for heavy menstrual bleeding (HMB), using reviews with comparable participants and outcomes; and to present a ranking of the first- and second-line treatments for HMB.

Methods: We searched for published Cochrane Reviews of HMB interventions in the Cochrane Database of Systematic Reviews. The primary outcomes were menstrual bleeding and satisfaction. Secondary outcomes included quality of life, adverse events and the requirement of further treatment. Two review authors independently selected the systematic reviews, extracted data and assessed quality, resolving disagreements by discussion. We assessed review quality using the Assessing the Methodological Quality of Systematic Reviews (AMSTAR) 2 tool and evaluated the certainty of the evidence for each outcome using GRADE methods. We grouped the interventions into first- and second-line treatments, considering participant characteristics (desire for future pregnancy, failure of previous treatment, candidacy for surgery). First-line treatments included medical interventions, and second-line treatments included both the levonorgestrel-releasing intrauterine system (LNG-IUS) and surgical treatments; thus the LNG-IUS is included in both groups. We developed different networks for first- and second-line treatments. We performed network meta-analyses of all outcomes, except for quality of life, where we performed pairwise meta-analyses. We reported the mean rank, the network estimates for mean difference (MD) or odds ratio (OR), with 95% confidence intervals (CIs), and the certainty of evidence (moderate, low or very low certainty). We also analysed different endometrial ablation and resection techniques separately from the main network: transcervical endometrial resection (TCRE) with or without rollerball, other resectoscopic endometrial ablation (REA), microwave non-resectoscopic endometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA.

Main results: We included nine systematic reviews published in the Cochrane Library up to July 2021. We updated the reviews that were over two years old. In July 2020, we started the overview with no new reviews about the topic. The included medical interventions were: non-steroidal anti-inflammatory drugs (NSAIDs), antifibrinolytics (tranexamic acid), combined oral contraceptives (COC), combined vaginal ring (CVR), long-cycle and luteal oral progestogens, LNG-IUS, ethamsylate and danazol (included to provide indirect evidence), which were compared to placebo. Surgical interventions were: open (abdominal), minimally invasive (vaginal or laparoscopic) and unspecified (or surgeon's choice of route of) hysterectomy, REA, NREA, unspecified endometrial ablation (EA) and LNG-IUS. We grouped the interventions as follows. First-line treatments Evidence from 26 studies with 1770 participants suggests that LNG-IUS results in a large reduction of menstrual blood loss (MBL; mean rank 2.4, MD -105.71 mL/cycle, 95% CI -201.10 to -10.33; low certainty evidence); antifibrinolytics probably reduce MBL (mean rank 3.7, MD -80.32 mL/cycle, 95% CI -127.67 to -32.98; moderate certainty evidence); long-cycle progestogen reduces MBL (mean rank 4.1, MD -76.93 mL/cycle, 95% CI -153.82 to -0.05; low certainty evidence), and NSAIDs slightly reduce MBL (mean rank 6.4, MD -40.67 mL/cycle, -84.61 to 3.27; low certainty evidence; reference comparator mean rank 8.9). We are uncertain of the true effect of the remaining interventions and the sensitivity analysis for reduction of MBL, as the evidence was rated as very low certainty. We are uncertain of the true effect of any intervention (very low certainty evidence) on the perception of improvement and satisfaction. Second-line treatments Bleeding reduction is related to the type of hysterectomy (total or supracervical/subtotal), not the route, so we combined all routes of hysterectomy for bleeding outcomes. We assessed the reduction of MBL without imputed data (11 trials, 1790 participants) and with imputed data (15 trials, 2241 participants). Evidence without imputed data suggests that hysterectomy (mean rank 1.2, OR 25.71, 95% CI 1.50 to 439.96; low certainty evidence) and REA (mean rank 2.8, OR 2.70, 95% CI 1.29 to 5.66; low certainty evidence) result in a large reduction of MBL, and NREA probably results in a large reduction of MBL (mean rank 2.0, OR 3.32, 95% CI 1.53 to 7.23; moderate certainty evidence). Evidence with imputed data suggests hysterectomy results in a large reduction of MBL (mean rank 1.0, OR 14.31, 95% CI 2.99 to 68.56; low certainty evidence), and NREA probably results in a large reduction of MBL (mean rank 2.2, OR 2.87, 95% CI 1.29 to 6.05; moderate certainty evidence). We are uncertain of the true effect for REA (very low certainty evidence). We are uncertain of the effect on amenorrhoea (very low certainty evidence). Evidence from 27 trials with 4284 participants suggests that minimally invasive hysterectomy results in a large increase in satisfaction (mean rank 1.3, OR 7.96, 95% CI 3.33 to 19.03; low certainty evidence), and NREA also increases satisfaction (mean rank 3.6, OR 1.59, 95% CI 1.09 to 2.33; low certainty evidence), but we are uncertain of the true effect of the remaining interventions (very low certainty evidence).

Authors' conclusions: Evidence suggests LNG-IUS is the best first-line treatment for reducing menstrual blood loss (MBL); antifibrinolytics are probably the second best, and long-cycle progestogens are likely the third best. We cannot make conclusions about the effect of first-line treatments on perception of improvement and satisfaction, as evidence was rated as very low certainty. For second-line treatments, evidence suggests hysterectomy is the best treatment for reducing bleeding, followed by REA and NREA. We are uncertain of the effect on amenorrhoea, as evidence was rated as very low certainty. Minimally invasive hysterectomy may result in a large increase in satisfaction, and NREA also increases satisfaction, but we are uncertain of the true effect of the remaining second-line interventions, as evidence was rated as very low certainty.