Abstract
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.
Plain language summary
Which is the best treatment for heavy menstrual bleeding?
Key results
Evidence suggests that the levonorgestrel‐releasing intrauterine system (LNG‐IUS) is the best first‐line option for reducing menstrual bleeding, while antifibrinolytics are probably the second best, and long‐cycle progestogens are the third best. Because of some limitations in the evidence, we are not sure what the true effect of these first‐line treatments is for the perception of improvement and satisfaction.
For second‐line treatments, evidence suggests any type of hysterectomy is the best treatment for reducing bleeding, even though this is a major surgery, and resectoscopic endometrial ablation (REA) and non‐resectoscopic endometrial ablation (NREA) are second and third best. We are uncertain of the true effect of the second‐line treatments on amenorrhoea (absence of menstrual blood loss). Evidence suggests that minimally invasive hysterectomy results in a large increase in satisfaction, and NREA increases satisfaction, but we are uncertain of the true effect of the remaining interventions.
What is heavy menstrual bleeding?
Heavy menstrual bleeding is defined as excessive menstrual blood loss that interferes with the quality of life of people who menstruate. It is very common and can affect 20% to 50% of people who menstruate during their reproductive years. There are different treatments available, each with their own pros and cons. The best treatment depends on the person's age, whether they have or want to have children, their personal preferences, and their medical history, among other things.
What did we want to find out?
We wanted to get an overview of all the published evidence on different treatments for heavy menstrual bleeding. We were most interested in finding out if the treatments were effective for reducing menstrual bleeding and for improving women's satisfaction. We also wanted to know how the treatment affected quality of life, what side effects it caused, and whether women required further treatment.
What did we do?
This study is an overview of reviews, which means we looked for published studies that synthesised the results of other studies on different treatments for heavy menstrual bleeding. Then we tried to give a broad overview of all that evidence. We analysed the certainty of the evidence based on factors like study size and methodological rigour. We categorised the treatments based on patient characteristics, including the desire (intention) for future pregnancy, failure of previous treatment or having been referred for surgery. First‐line treatment included medical interventions and second‐line treatment included the LNG‐IUS plus surgical interventions; thus, the LNG‐IUS was included in both first‐ and second‐line treatments. We used network meta‐analysis, a statistical method that compares all the interventions at the same time, to find out which treatments produced the best results for patients.
What did we find?
We found nine reviews with 104 studies, involving a total of 11,881 participants. Altogether, the data we analysed came from 85 trials and 9950 participants. The medical interventions included were: non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics (tranexamic acid), combined oral contraceptives (COC), combined vaginal ring (CVR), long‐cycle and luteal oral progestogens, the LNG‐IUS, ethamsylate and danazol (included only to provide indirect evidence). These were compared to placebo (sham treatment). The surgical interventions included were: open (abdominal), minimally invasive (vaginal or laparoscopic) and unspecified (or surgeon's choice of) route of hysterectomy, REA, NREA and unspecified endometrial ablation (EA).
What are the limitations of the evidence?
Our confidence in some evidence is moderate, but for most of it, our confidence is low to very low. The main reasons were because the studies were often not blinded, which means the participants knew which treatment they were receiving, and that could have changed their perception; the direct and indirect evidence was not similar enough to compare in the network; and the range of the results was too wide.
How up to date is this evidence?
The last search for reviews was in July 2021.
Background
Description of the condition
Heavy menstrual bleeding (HMB) is excessive menstrual blood loss that interferes with the physical, emotional, social and material quality of life of people who menstruate, and it can occur alone or in combination with other symptoms (Munro 2012; NICE 2018a), such as dysmenorrhoea, fatigue or headache. HMB can lead to other serious health issues, such as iron deficiency anaemia, which occurs in one in four people with HMB (Morrison 2008).
Abnormal uterine bleeding is often used interchangeably with HMB. However, abnormal uterine bleeding is much broader and includes many gynaecological diseases. The International Federation of Gynecology and Obstetrics (FIGO) classifies the gynaecological diseases related to abnormal uterine bleeding as having structural causes (polyp, adenomyosis, leiomyoma, malignancy, and hyperplasia); being unrelated to structural causes (coagulopathy, ovulatory dysfunction, endometrial dysfunction, and iatrogenic); or as not yet classified (Munro 2012). This review will be restricted to HMB.
HMB is objectively defined as menstrual blood loss of 80 mL or more per menstrual period (Cole 1971; Hallberg 1966), which is not related to pregnancy or any systemic condition (for example, bleeding or thyroid disorders) or gynaecological disease. HMB is also known as menorrhagia.
Historically, the gold standard for objectively assessing menstrual bleeding is the haematin alkaline method: this requires the woman to provide all the pads and tampons she uses, and then in the laboratory, a sodium hydroxide solution allows calculation of haemoglobin, and thus of the amount of menstrual bleeding (Shaw 1972). However, in a clinical setting, the objective measurement of menstrual blood loss is impractical (Edlund 2011). Currently, with the clinical diagnosis centred on the women's perception of HMB, the measured amount of blood loss has a secondary role; rather, evaluating the clinical success of treatment should be focused on the woman's perception of improvement.
The prevalence of HMB based on objective measurement ranges from 9% to 14%, but in studies that subjectively assess HMB, it is as high as 20% to 52% (Fraser 2009; NICE 2007). In England and Wales, every year around 30,000 women undergo surgical treatment for HMB (RCOG 2012). HMB is estimated to account for around 30% of total gynaecological visits in the USA and represents an important encumbrance for more than 10 million women (Liu 2007; Miller 2015). Treatment costs ascend to approximately USD 1.3 billion, and lost productivity to around USD 12 billion to USD 36 billion per year (Liu 2007; Miller 2015). In Japan, 19% of women 15 to 49 years of age reported HMB (Tanaka 2013). In low‐ and middle‐income countries, the incidence of HMB appears to be similar to that of high‐income countries, although data are limited (Haththotuwa 2011). Data from a systematic review in 2004 assessing the epidemiology of menstrual disorders in low‐ and middle‐income countries reported a prevalence of HMB of 15%, ranging from 5% in rural Gambia to 20% in China (Harlow 2004).
The prevalence of HMB varies by age. In a population‐based prevalence study with nearly 1000 healthy adolescent girls, approximately 40% had experienced HMB (Friberg 2006). The annual rate of presentation to health services with HMB in the UK is around 2% before the age of 40 and increases to 5% between 45 and 49 years of age (NICE 2007). A Swedish study conducted to assess quality of life in women with HMB reported that prevalence in women aged 40 to 45 years was 32% (Karlsson 2014).
Adequate assessment, followed by appropriate treatment, considerably improves the quality of life of women with HMB (Hurskainen 2007).
Description of the interventions
Diverse medical and surgical treatments are available for treating HMB, with different effectiveness, tolerability, acceptability and cost for the patient. The woman's age, intention to become pregnant, associated symptoms, preferences and values are important for selecting the treatment.
First‐line treatments for HMB include all medical interventions (non‐hormonal and hormonal). Second‐line treatments include the levonorgestrel‐releasing intrauterine system (LNG‐IUS) and all surgical interventions; they may be appropriate when previous medical treatment has failed, when women have completed their family or when they are surgical candidates. LNG‐IUS is considered both a first‐ and second‐line treatment.
Medical interventions
Non‐steroidal anti‐inflammatory drugs (NSAIDs)
NSAIDs, such as ibuprofen and mefenamic acid, are a non‐hormonal oral medical intervention. NSAIDs have potent anti‐inflammatory, analgesic and antipyretic activity and are among the most widely used drugs worldwide (Bacchi 2012). NSAIDs should be avoided in women with severe asthma, gastrointestinal ulceration or kidney disease.
Antifibrinolytic agents
Tranexamic acid is the most commonly used antifibrinolytic agent. Antifibrinolytics should be avoided in women with thromboembolic disease or a history of convulsions (Lecker I 2016).
Combined hormonal contraceptives
Combined hormonal contraceptive methods contain oestrogen and progestin. There are different regimens and delivery routes available, including oral, vaginal, transdermal and intramuscular. Not all delivery routes have been assessed for the management of HMB. Combined hormonal contraceptives have been linked to a higher risk of thrombotic side effects, although the oestrogen dose has been gradually reduced from 150 μg in the original preparations to 30 μg or less at the time of writing, which has reduced the risk considerably (Speroff 2010). Combined hormonal contraceptives are still associated with a small increased risk of venous thromboembolism (De Bastos 2014). This overview included two different types of combined hormonal contraceptives: combined oral contraceptive (COC) and combined vaginal ring (CVR).
Cyclical oral progestogens
During the 1990s, oral progestogens were the most commonly prescribed drug for HMB (Coulter 1995). Oral progestogens can be taken during the luteal phase (day 14 to 24 of the menstrual cycle) or as an extended regimen for 20 to 25 days per cycle. Cyclical progestogens are not contraceptives. The prolonged use of high‐dose progestogens is associated with side effects, such as weight gain, nausea, headaches and decreased libido.
Progestogen‐releasing intrauterine systems (Pg‐IUS or LNG‐IUS)
The progesterone‐releasing intrauterine systems are small T‐shaped systems, inserted into the uterine cavity through the cervix, releasing small amounts of a progestogen (levonorgestrel) locally. The LNG‐IUS is available in different formats with varying doses of levonorgestrel released over varying periods.
Danazol (†)
Danazol is an oral drug that interferes with the pituitary gland, preventing the secretion of follicle‐stimulating hormone and luteinising hormone. Danazol is contraindicated in pregnancy and should be used along with safe contraception in sexually active women (Brunskill 1992). Danazol is currently used for a myriad of haematological conditions (Audia 2016), and it used to be commonly prescribed for HMB during the 1990s. However, although its adverse events are infrequent, other safer and fairly similarly effective alternatives are available. Reported adverse events are usually androgenic and anti‐oestrogenic, and haematologically can be either anticoagulant or procoagulant (Alvarado 2001). In a cohort of 530 patients, 29% reported at least one adverse event: 6% androgenic, 6% hypo‐oestrogenic, 12% some combination and 5% other non‐androgenic and non‐hypo‐oestrogenic (Jick 1995). Danazol has also been linked to liver damage (Alvaro 1996).
(†) Despite being effective in reducing menstrual bleeding, its adverse events advise against its use for HMB. We included danazol in this overview to provide indirect comparisons in the network meta‐analysis.
Ethamsylate
Ethamsylate is a synthetic haemostatic drug used for capillary bleeding (Garay 2006). It was widely used for HMB in the 1970s and 1980s (Harrison 1976). With less promising results, its use for HMB over the past 20 years has been negligible ((D) Bonnar 1996; Bongers 2004). We included ethamsylate in this overview to provide indirect comparisons in the network meta‐analysis.
Surgical interventions
Endometrial resection and ablation
Endometrial resection and endometrial ablation (EA) are different techniques aiming to remove or destroy the endometrium (inner layer of the uterus). They are usually categorised by generation: first‐generation techniques require direct visualisation of the uterine cavity, while second‐ or third‐generation procedures do not. Although these categories are widely used in the literature, there is evidence that endometrial ablation started in the late 19th century with non‐resectoscopic techniques (Famuyide 2018). In 1894, Sneguireff applied steam to arrest profuse bleeding during the removal of a parasitic liver cyst and then started using it to control surgical bleeding on the lungs, kidneys, long bones and, eventually, uterine bleeding in women (Famuyide 2018). In the late 1890s, the equipment was improved, adding some safety features and defining indications and contraindications (Blacker 1902). Simultaneously, Pincus defined two types of procedures: atmocausis, which used steam directly on the endometrium, and zestocausis, which used metal to provide the heat (Famuyide 2018; Pincus 1899a; Pincus 1899b). Honouring the first attempts of endometrial ablation, we will use the categories based on the procedures' characteristics (Famuyide 2018; Munro 2018).
Resectoscopic endometrial ablation (REA) involves different forms of energy delivered through an operative hysteroscope to remove or destroy the endometrium, such as electrocoagulation or desiccation, transcervical endometrial resection (TCRE), or vaporisation (Munro 2018).
Non‐resectoscopic endometrial ablation (NREA) is a group of devices designed to destroy the endometrium without a resectoscope (not requiring fluid distention of the uterine cavity). Although it is a non‐hysteroscopic technique, pre‐ and postprocedure diagnostic hysteroscopy or intraprocedural ultrasound guidance may be useful for safety and intracavitary device placement (Laberge 2015).
Hysterectomy
Hysterectomy is a major surgical procedure, consisting of removing the uterus, with or without the adnexa. Some examples of hysterectomy complications are infection, thromboembolic complications, genitourinary and gastrointestinal tract injuries, bleeding and vaginal cuff dehiscence (Clarke‐Pearson 2013). The complications vary depending on the route and the surgical technique.
How the intervention might work
Medical interventions
Non‐steroidal anti‐inflammatory drugs (NSAIDs)
The endometrium of women with HMB has high prostaglandin levels compared to women with normal menstrual bleeding. NSAIDs (such as mefenamic acid and ibuprofen, among others) reduce prostaglandin levels by inhibiting the enzyme cyclooxygenase (Rees 1987; Smith 1981). NSAIDs taken regularly during menses can alleviate HMB.
Antifibrinolytic agents
Increased levels of plasminogen activators (group of enzymes that cause fibrinolysis) have been found in the endometrium of women with HMB compared to women with normal menstrual bleeding (Gleeson 1994). Tranexamic acid is a true plasminogen activator inhibitor, making it an alternative drug for reducing menstrual bleeding.
Combined hormonal contraceptives
Combined hormonal contraceptives contain different combinations of oestrogen and progestin. Oestrogen provides negative feedback on follicle‐stimulating hormone secretion and prevents the development of a dominant follicle (Bradley 2016), providing endometrial stability and growth, improving the progestational impact. At the same time, progestin impedes the rise of luteinising hormone and consequently prevents ovulation and creates an atrophic endometrial lining. This combination reduces overall menstrual blood loss (Fritz 2012). Different routes of delivery have been assessed to manage HMB (Uhm 2014), such as the CVR and oral contraception ((D) Dahiya 2016; Micks 2013).
Cyclical oral progestogens
The role of progestogens in treating HMB might work "by stabilizing endometrial fragility; inhibiting the growth of the endometrium by triggering apoptosis; inhibiting angiogenesis; and stimulating the conversion of estradiol to the less active oestrone" (Fritz 2012). Progestogens can be used in different regimens, either during the entire cycle, during the luteal phase only, or for a longer period of 20 to 25 days. When used as a long‐course treatment, oral progestogens prevent ovulation and ovarian steroidogenesis. As a consequence, they discontinue the production of oestrogen receptors and the oestrogen‐dependent stimulation of the endometrium, leading to an atrophic endometrium (Bradley 2016).
Progestogen‐releasing intrauterine systems
With this device, hormones are released locally into the uterine cavity, resulting in high progestogen levels in the endometrial tissue but low systemic circulation levels. As a consequence of the local action of the progesterone, endometrial growth is suppressed (Herman 2013), and in turn, menstrual bleeding decreases.
Danazol
Danazol is a synthetic steroid ethisterone. It prevents pituitary secretion of follicle‐stimulating hormone and luteinising hormone. Danazol has a weak androgenic influence, producing atrophy of endometrial tissue, which reduces menstrual loss and may lead to amenorrhoea in some women ((N) Chimbira 1980).
Surgical interventions
Endometrial resection and ablation
Endometrial resection and endometrial ablation destroy the endometrium (lining of the cavity of the uterus). They are safe, effective, minimally invasive procedures, performed through the cervix, with different methods for removing (resection) or destroying (ablation) the endometrium (Kumar 2016).
Hysterectomy
Although surgically removing the uterus is invasive, it represents the most definitive treatment option for HMB. It is important to assess the relative risk and the plans for future childbearing before making the decision. There are four different surgical approaches: abdominal hysterectomy, vaginal hysterectomy, laparoscopic hysterectomy and robot‐assisted laparoscopic hysterectomy (Van der Heijden 2017).
Why it is important to do this overview
HMB has a high prevalence amongst otherwise healthy women of reproductive age. Different interventions are currently available, including pharmacological treatments (hormonal and non‐hormonal) and surgical procedures. This overview aims to assess and rank the safety and efficacy of the treatments with regard to blood loss reduction, satisfaction with treatment, quality of life, adverse effects and treatment failure, and to summarise the data in an accessible way for patients and their families, physicians, healthcare providers and policymakers. Network meta‐analysis compares more than two interventions at the same time. This allows the comparison of interventions that have not been compared directly in the studies, as long as they form a connected network, providing a complete summary of evidence. There is an overview of interventions for HMB available during pandemics (Bofill Rodriguez 2020a), which was developed in 2020 and inspired by the challenges of the pandemic to the health system; it includes only medical interventions that required minimal (or no) face‐to‐face contact; thus, it did not include LNG‐IUS or any surgical procedure. It also only reported pairwise meta‐analysis.
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
Criteria for considering reviews for inclusion
We included Cochrane Reviews of interventions for heavy menstrual bleeding (HMB) in otherwise healthy women, which are published in the Cochrane Library; we did not include non‐Cochrane reviews in the overview. Not all studies included in the Cochrane Reviews contributed data to the overview, only studies reporting on the overview outcomes. The references to studies contributing with data to the overview are in the 'Additional references' section and are designated by a (D) for data prefacing the name of the study.
We have formulated our overview in the PICO (participants, interventions, control, outcome) format.
Type of participants
Women in reproductive years with HMB, assessed objectively or semi‐objectively, using the haematin alkaline method, pictorial blood assessment chart (PBAC), or patient perception of HMB.
We conducted a subgroup analysis based on whether the participants were surgical candidates or not. We considered the 'the desire (or intention) of future pregnancy', 'women completed their families', 'with indication of hysterectomy' or 'medical treatment failure' as inclusion or exclusion criteria to determine if the participants were surgical candidates. According to that, we performed the network meta‐analysis for first‐ or second‐line treatment. The progestogen‐releasing intrauterine system (LNG‐IUS) was part of both groups.
Type of interventions
We included medical and surgical treatments for HMB that were included in Cochrane Reviews.
Medical interventions
Nonsteroidal anti‐inflammatory drugs
Antifibrinolytic agents
Combined hormonal contraceptives divided according to the delivery route in combined oral contraceptives and combined vaginal ring
Cyclical oral progestogens, divided by the length of the cycle in luteal‐phase or long‐cycle oral progestogens (three to four weeks per cycle)
Progestogen releasing intrauterine system
Danazol and ethamsylate (included only to provide indirect evidence)
Progestin‐only contraceptives, such as injections and implants, usually reduce menstrual blood loss in the general population and can cause amenorrhoea (Di Carlo, 2015; Jacobsten 2014). They are commonly used off‐label for HMB; we performed a search on the topic. See Appendix 1 for details.
Surgical interventions
Resectocopic endometrial ablation (REA), categorised as transcervical endometrial resection, with or without rollerball, and other REA
Non‐resectoscopic endometrial ablation (NREA), categorised as microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA
Hysterectomy, categorised by the route: minimally invasive (vaginal or laparoscopic), open (abdominal) and unspecified (or route at surgeon's discretion)
Control
We compared interventions with placebo or no intervention and with other interventions.
Types of outcome measures
Primary outcomes
-
Menstrual bleeding reduction
Objectively assessed by the haematin alkaline method, mean blood loss during treatment, mean reduction of blood loss from baseline during treatment or mean blood loss at different periods
Semi‐objectively assessed using the PBAC, mean difference of PBAC at different periods or the proportion of women with reduction in bleeding as measured by PBAC under different cutoffs
Subjectively assessed patient perception of HMB: participants' perceived blood loss (better, same, worse)
Prevalence of amenorrhoea or hypomenorrhoea after treatment
Satisfaction with treatment at one year follow‐up, reported as a dichotomous outcome as the proportion of women satisfied with treatment. Some trials reported this using a four‐level scale questionnaire, either very satisfied, satisfied, somewhat dissatisfied, dissatisfied, or liking treatment very well, well, moderately, poorly; in both cases, we considered the two first categories satisfactory.
Secondary outcomes
Quality of life improvement: women's perceived change in quality of life, where it was recorded in a reproducible and validated format
Adverse effects of any severity recorded within the included Cochrane Reviews
Requirement of further surgery or additional medical treatment at one year follow‐up.
Search methods for identification of reviews
We searched the Cochrane Database of Systematic Reviews (CDSR) in the Cochrane Library for any reviews with heavy menstrual bleeding (HMB) or menorrhagia in the title, abstract or keyword fields. The last search was in July 2021.
Data collection and analysis
Selection of reviews
Two overview authors (MBR and JB) independently assessed all potentially eligible reviews identified by the search strategy. We resolved any disagreements by discussion. During the first review search, one of the authors was involved in one HMB review (JB, combined hormonal contraceptives for HMB), and the other (MB) was not involved in any reviews.
We created 'Characteristics of reviews' tables, including the following information: review title; author team; number of included trials and participants; participant characteristics (inclusion and exclusion criteria); interventions and comparisons; primary and secondary outcomes and date of the last search.
Data extraction and management
Two review authors (MB and VJ) independently extracted data from each included review, resolving any disagreements by discussion or adjudication by a third overview author. In order to obtain data in a valid format for the network meta‐analysis (NMA), we extracted data from individual studies contributing to each of the outcomes. We summarised key information from each study contributing to the NMA on the data extraction form, including participant's details, the interventions, comparisons and outcomes (see Table 1 and Table 2 for the characteristics of studies contributing to the NMA). Outcomes were focused on bleeding reduction, satisfaction, improvement of quality of life and adverse effects.
1. Characteristics of studies contributing data to the network meta‐analysis of first‐line treatments.
| Study | Interventions | Inclusion | Exclusion | Outcomes | |||
| A | B | C | D | ||||
| (D) Agarwal 2016 | COC (20 μg of ethinyl oestradiol and 120 μg of desogestrel) |
Combined vaginal ring | — | — |
|
|
|
| (D) Ashraf 2017 | Long‐cycle progestogen (Norethisterone) | LNG‐IUS | — | — |
|
|
|
| (D) Bonduelle 1991 | Luteal progestogen (Norethisterone 5 mg 3 times/day, day 19 to 26 of the cycle) | Danazol (200 mg daily) |
— | — |
|
|
|
| (D) Bonnar 1996 | Tranexamic acid regular dose (3‐4 g/day) |
Ethamsylate | NSAIDs (mefenamic acid) | — |
|
|
|
| (D) Buyru 1995 | Danazol | Luteal progestogen (Norethisterone) | — | — |
|
|
|
| (D) Callender 1970 | Tranexamic acid regular dose (3‐4 g/day) |
Placebo | — | — |
|
|
|
| (D) Cameron 1990 | Luteal progestogen (Norethisterone) | NSAIDs (mefenamic acid) |
— | — |
|
|
|
| (D) Chamberlain | NSAIDs (mefenamic acid) |
Ethamsilate | — | — |
|
|
|
| (D) Dahiya 2016 | COC (30 µg ethinyl estradiol and 150 µg levonorgestrel) |
Combined vaginal ring | — | — |
|
|
|
| (D) Dockeray 1989 | NSAIDs (mefenamic acid) |
Danazol | — | — |
|
Not stated |
|
| (D) Dunphy 1998 | Danazol | Luteal progestogen (medroxyprogesterone acetate) | — | — |
|
|
|
| (D) Edlund 1995 | Kabi pro Tranexamic 600 mg | Kabi Pro‐tranexamic 1200 mg | Placebo | — |
|
|
|
| (D) Endrikat 2009 | LNG‐IUS | COC (20 µg ethinyl estradiol and 1 mg NETA) |
— | — |
|
|
|
| (D) Fraser 1981 | NSAIDs (mefenamic acid) |
Placebo | — | — |
|
|
|
| (D) Fraser 1991 | COC (30 µg ethinyl estradiol and 150 µg levonorgestrel) |
NSAIDs (naproxen) | NSAIDs (mefenamic acid) | Danazol |
|
|
|
| (D) Fraser 2011 | COC (Estradiol valerate and dienogest) |
Placebo | — | — |
|
|
|
| (D) Freeman 2011 | Tranexamic acid low dose (< 2.5 g/day) |
Tranexamic acid regular dose (3‐4 g/day) |
Placebo | — |
|
|
|
| (D) Goshtasebi 2013 | Tranexamic acid low dose (<2.5 g/day) |
Long cycle progestogen (medroxi progesterone acetate) | — | — |
|
|
|
| (D) Grover 1990 | NSAIDs (mefenamic acid) |
Placebo | — | — |
|
|
|
| (D) Hashim 2012 | Combined vaginal ring | Long cycle progestogen (norethisterone) | — | — |
|
|
Primary
Secondary
|
| (D) Higham 1993 | Luteal progestogen (norethisterone) | Danazol (has 2 different danazol dosages that for the NMA were combined in one group) |
— | — |
|
|
|
| (D) Irvine 1988 | Long‐cycle progestogen (norethisterone) | LNG‐IUS | — | — |
|
|
|
| (D) Jaisamrarn 2006 | Tranexamic acid regular dose (3‐4 g/day) |
Luteal progestogen (norethisterone) |
NSAIDs (mefenamic acid) |
— |
|
|
|
| (D) Jensen 2011 | COC (estradiol valerate and dienogest) |
Placebo | — | — |
|
|
|
| (D) Kaunitz 2010 | Luteal progestogen (medroxy progesterone acetate) | LNG‐IUS | — | — |
|
|
|
| (D) Kiseli 2016 | Tranexamic acid regular dose (3‐4 g/day) |
Luteal progestogen (Norethisterone) |
LNG‐IUS | — |
|
|
|
| (D) Kriplani 2006 | Tranexamic acid low dose (<2.5 g/day) |
Long cycle progestogen (Medroxy progesterone acetate) |
— | — |
|
|
|
| (D) Lukes 2010 | Tranexamic acid regular dose (3‐4 g/day) |
Placebo | — | — |
|
|
|
| (D) Muggeridge 1983 | NSAIDs (mefenamic acid) |
Placebo | — | — | Menstrual blood loss > 75 mL/cycle | Pelvic pathology | Menstrual blood loss (alkaline haematin method) Dysmenorrhoea (numerical score) Adverse events |
| (D) Preston 1995 | Tranexamic acid regular dose (3‐4 g/day) |
Luteal progestogen (Norethisterone) |
— | — |
|
|
|
| (D) Sayed 2011 | LNG‐IUS | COC (30 µg ethinyl estradiol and 150 µg levonorgestrel) |
— | — |
|
Pregnancy History of ectopic pregnancy, puerperal sepsis, pelvic inflammatory disease Evidence of defective coagulation, Abnormalities on ultrasound (including submucous fibroids of any size distorting the cavity of the uterus or intramural or subserous fibroids > 5 cm in diameter) History of malignancy or evidence of hyperplasia in the endometrial biopsy Incidental adnexal abnormality on ultrasound Previous endometrial ablation/resection Uninvestigated postcoital bleeding, untreated abnormal cervical cytology Contraindication to COCs. |
|
| (D) Shabaan 2011 | COC (30 µg ethinyl estradiol and 150 µg levonorgestrel) |
LNG‐IUS | — | — |
|
|
|
| (D) Tsang 1987 | NSAIDs (mefenamic acid) |
Placebo | — | — | Women aged 26 to 47 years with either a history of heavy menstrual bleeding or menstrual blood loss > 80 mL/cycle (measured objectively) and regular menstrual cycles | Use of hormonal contraceptives or anti‐inflammatory drugs and use of intrauterine contraceptive device | Menstrual blood loss (measured by alkaline haematin method) |
| (D) van Eijkeren 1992 | NSAIDs (mefenamic acid) |
Placebo | — | — |
|
|
|
| (D) Ylikorkala 1986 | NSAIDs (naproxen) |
Placebo | — | — | Menstrual blood loss > 80 mL/cycle Regular cycles and normal pelvic findings | Not stated |
|
| (D) Zhang 2008 | Tranexamic acid regular dose (3‐4 g/day) |
Luteal progestogen (Norethisterone) |
— | — |
|
|
|
COC: combined oral contraceptive; D & C: dilation and curettage; FSH: follicle‐stimulating hormone; HMB: heavy menstrual bleeding; HRQoL: health‐related quality of life; IUS: intrauterine system; LNG‐IUS: levonorgestrel‐releasing intrauterine system; MIQ: Menorrhagia Impact Questionnaire; NET: norethisterone; NETA: norethisterone acetate; NMA: network meta‐analysis; NSAIDs: non‐steroidal anti‐inflammatory drugs; PBAC: pictorial blood assessment chart; QoL: quality of life; VTE: venous thromboembolism.
2. Characteristics of studies contributing data to the network meta‐analysis of second‐line treatments.
| Study | Interventions | Inclusion criteria | Exclusion criteria | Outcomes | Comments | |
| A | B | |||||
| (D) Abbott 2003 | Bipolar NREA (Novasure) | Balloon NREA (Cavaterm) |
|
|
|
— |
| (D) Athanatos 2015 | Microwave NREA | Bipolar NREA (Novasure) |
|
|
At 3 months:
At 12 months:
|
All participants received GnRH 3 months pretreatment |
| (D) Barrington 2002 | LNG‐IUS | Balloon NREA (Thermachoice) |
Women with menorrhagia refractory to medical treatment referred by GPs to gynaecology clinic in district hospital | Cavity > 12 cm; subserous fibroids; malignant or pre‐malignant pathology (from endometrial biopsy) |
|
GnRH 1 month pretreatment on the EA arm |
| (D) Bhattacharya 1997 | Laser REA | TCRE/rollerball |
|
None reported |
|
— |
| (D) Bongers 2004 | Bipolar NREA (Novasure) |
Balloon NREA (Thermachoice) |
|
|
Primary:
Secondary:
|
— |
| (D) Brun 2006 | Balloon NREA (Cavaterm) | TCRE REA |
|
|
Primary:
Secondary:
|
— |
| (D) Clark 2011 | Bipolar NREA (Novasure) | Balloon NREA (Thermachoice) |
|
|
Primary:
Secondary:
|
Novasure versus Thermachoice III |
| (D) Cooper 1999 | Microwave NREA | TCRE + Rollerball REA |
|
Histopathological abnormalities of the endometrium | Primary:
Secondary:
|
— |
| (D) Cooper 2002 | Bipolar NREA (Novasure) | TCRE + Rollerball REA |
|
|
|
— |
| (D) Cooper 2004 | Microwave NREA | Rollerball REA |
|
|
Primary:
Secondary:
|
All participants received 1 month GnRh pre‐treatment |
| (D) Cooper 2019 | NREA (either Thermal balloon or radiofrequency) |
Minimally invasive hysterectomy (Laparoscopic supracervical hysterectomy, LASH) |
|
Previous endometrial ablation If laparoscopic surgery was contraindicated, or if they were unable to give informed consent or complete trial paperwork |
Primary
Secondary outcome measures
|
— |
| (D) Corson 2000 | Electrode balloon NREA (Vesta) | TCRE + rollerball REA |
|
|
|
— |
| (D) Corson 2001 | Hydrotermalablation NREA (Hydro Thermablator) |
Rollerball REA |
|
|
|
— |
| (D) Crosignani 1997 | Rollerball REA | Minimally invasive hysterectomy (Vaginal hysterectomy) |
|
|
|
— |
| (D) Crosignani 1997a | Rollerball REA | LNG‐IUS |
|
|
Primary outcome
|
— |
| (D) De Souza 2010 | LNG‐IUS | Balloon NREA (Thermachoice) |
|
Not reported |
|
— |
| (D) Dickersin 2007 | Endometrial ablation (any method) | Hysterectomy (unspecified) |
|
|
|
Resectoscopic endometrial ablation with electrodesiccation/coagulation or vaporisation OR NREA ablation with thermal balloon. Hysterectomy vaginal, laparoscopic or abdominal abdominal |
| (D) Duleba 2003 | Cryoablation NREA | Rollerball REA |
|
|
|
— |
| (D) Dwyer 1993 | TCRE REA | Open hysterectomy |
|
|
|
Participants on the EA arm received medroxyprogesterone six weeks pre‐treatment. |
| (D) Ergun 2012 | LNG‐IUS | Rollerball REA |
|
|
|
— |
| (D) Gannon 1991 | TCRE REA |
Open hysterectomy | Women waiting for abdominal hysterectomy for menorrhagia |
|
|
— |
| (D) Ghazizadeh 2014 | Bipolar NREA (Novasure) | TCRE REA |
|
|
|
— |
| (D) Hawe 2003 | Balloon NREA (Cavaterm) | Laser REA |
|
|
|
Cavaterm YAGlaser |
| (D) Herman 2013 | Bipolar NREA | LNG‐IUS |
|
None reported |
|
— |
| (D) Hurskainen 2004 | Hysterectomy (unspecified) | LNG‐IUS |
|
|
Primary
Other outcomes
|
Hysterectomies were either abdominal, vaginal or laparoscopic) |
| (D) Istre 1998 | TCRE REA | LNG‐IUS |
|
|
|
— |
| (D) Jain 2016 | Balloon NREA (LiNAMenotreat system) |
Minimally invasive hysterectomy (Vaginal) |
|
|
|
Ht vaginal |
| (D) Kittelsen 1998 | LNG‐IUS | TCRE REA |
|
|
|
— |
| (D) Laberge 2017 | Bipolar NREA (Minerva) | Rollerball REA |
|
|
|
— |
| (D) Malak 2006 | TCRE REA | LNG‐IUS |
|
|
|
— |
| (D) McClure 1992 | Laser REA | TCRE REA |
|
|
|
|
| (D) Meyer 1998 | Balloon NREA (Thermachoice) | Rollerball REA |
|
|
|
— |
| (D) O'Connor 1997 | TCRE REA | Hysterectomy (unspecified) |
|
|
|
Hysterectomies: 28 abdominal and 28 vaginal. Outcomes not reported by type of hysterectomy |
| (D) Ozdegirmenci 2011 | LNG‐IUS | Open hysterectomy | Not specifically reported ‐ women with adenomyosis by sonogram and MRI with menorrhagia and or dysmenorrhoea (all women had menorrhagia) |
|
|
— |
| (D) Pellicano 2002 | Balloon NREA (Cavaterm) | TCRE REA |
|
|
Primary:
Secondary:
|
— |
| (D) Penninx 2010 | Bipolar NREA (Novasure®) | Hydrotermalablation NREA (Hydro ThermAblator®) | • Women with menorrhagia (defined by Higham minimum score of 150 points) • Normal uterine cavity (length 6 to 12 cm and histologically benign endometrium) • Normal Pap smear • Negative Chlamydia test • Premenopausal (FSH < 40 IU/L) • Desire for ablation after looking at other options for Rx | • Presence of coagulopathies • Use of anticoagulants • Desire to preserve fertility • Prior uterine surgery (except low‐segment CS) • Suspected or confirmed uterine malignancy | Primary: • Amenorrhoea at 12 months after surgery Secondary: • Reduction in bleeding • Patient satisfaction • Complications • Re‐intervention for hysterectomy | |
| (D) Penninx 2016 | Bipolar NREA (Novasure®) | Balloon NREA (Thermablate®) | • Women with HMB PBAC > 150 points and follicle‐stimulating hormone (FSH) level < 40 IU/L • Normal uterine cavity (cavity length 6 to 12 cm), confirmed by saline infusion sonography or diagnostic hysteroscopy • No endometrium pathology (histologically benign; confirmed within 6 months of screening by endometrium in the office (Pipelle1, CooperSurgical, Trumbull, USA) • Normal Pap smear | • Coagulopathies or use of anticoagulants • Desire to preserve fertility. • Prior uterine surgery other than low‐segment caesarean section • (Suspected) uterine malignancy. • Preferred to be treated in an outpatient setting • US with intracavitary pathology, except for women with intracavitary polyps < 1 cm | • PBAC • Amenorrhoea rate • Pain • Satisfaction • Requirement for further treatment | |
| (D) Perino 2004 | Laser thermal ablation NREA (ELITT (endometrial laser intrauterine thermal therapy) | TCRE REA |
• Dysfunctional uterine bleeding not associated with organic pathology and not responding to medical treatment | not reported | Primary: • Amenorrhoea and other menstrual status • Satisfaction rates Secondary: • Intraoperative complication rate • Operation time • Pain • Further treatment with hysterectomy | |
| (D) Pinion 1994 | Endometrial Ablation (unspecified) | Hysterectomy (*) | • Women under 50 years and weight less than 100 kg. • Clinical diagnosis of dysfunctional uterine bleeding, uterus < 10 weeks of gestational size, normal endometrial histology | not reported | • Satisfaction rate at one and four years • Change in general health and MBL Duration of surgery (minutes) and duration of hospital stay (days) • Return to work (weeks) • Complication rate • Quality of life (Hospital Anxiety and Depression Scale, Golombok Rust Inventory of Marital State) • Health service costs at one and four years • Participant costs at one year • Requirement for further surgery | Laser ablation or TCRE. EA arm received GnRh for 5 weeks pre‐treatment |
| (D) Romer 1998 | Balloon NREA (Cavaterm®) | Rollerball REA | • Recurrent menorrhagia not responsive to medical therapy • No desire for future fertility | • Intrauterine abnormality • Fibroids • Hyperplasia | • Satisfaction rate • Amenorrhoea or hypomenorrhoea rate | |
| (D) Sambrook 2009 | Microwave NREA (MEA TM ) | Balloon NREA (Tball) | • Women reporting heavy menstrual loss and requesting endometrial ablation • Premenopausal • Completed their families • Uterine size equivalent to a 12‐week pregnancy or less • No histopathological abnormalities • No fibroids obstructing the uterine cavity • Lower‐segment caesarean section if scar thickness > 10 mm on transvaginal US | not reported | Primary: • Satisfaction (6‐point scale) and menstrual scores at 1 year (PBAC) Secondary: • Operative differences • Acceptability of treatment • Health‐related quality of life | |
| (D) Sesti 2011 | Balloon NREA (Tball) | Minimally invasive hysterectomy (Laparoscopic subtotal hysterectomy) | • PBAC score ≥ 100 (average of two consecutive cycles) • Completed family • Normal smear • Pelvic ultrasound scan and endometrial biopsy | • Previous endometrial resection/ablation or levonorgestrel intrauterine system • Any uterine pathology on pelvic ultrasound scan or hysteroscopy. • Any pathology whereby hysterectomy was indicated • Uninvestigated abnormal bleeding or postmenopausal bleeding. | Primary: • Menstrual bleeding (PBAC score) at three, six, 12 and 24 months Secondary: • Quality of life (SF‐36 score) at 24 months • Improvement in bleeding patterns (frequency and duration of bleeding) at three, six, 12 and 24 months • Haemoglobin levels at three, six, 12 and 24 months • Intensity of postoperative pain • Early postoperative complications | endometrial ablation via Thermachoice® III thermal balloon ablation laparoscopic subtotal hysterectomy |
| (D) Sesti 2012 | Minimally invasive hysterectomy (Laparoscopic subtotal hysterectomy) | LNG‐IUS | • Women with HMB PBAC > or = 100 (average of 2 cycles) unresponsive to medical treatment • Age 35 to 50 years, completed family • Failed appropriate first‐line oral medical therapy • Normal PAP smear; no pelvic pathology on US; normal endometrial biopsy. | • Previous endometrial resection/ablation or insertion of LNG‐IUS • Uterine pathology on scan or hysteroscopy • Pathology where hysterectomy indicated • Postmenopausal bleeding | Primary: • Satisfaction (6‐point scale) and menstrual scores at 1 year (PBAC) Secondary: • Operative differences • Acceptability of treatment • Health‐related quality of life | |
| (D) Shaw 2007 | Balloon NREA (Menotreat) | LNG‐IUS | •Women with idiopathic menorrhagia aged 25 to 49 years in whom prior appropriate oral medical treatment had failed. •Family complete •Normal histology, normal ultrasound (fibroids up to 2.5 cm OK), normal cervical smear, PBAC score > 120 (mean over 2 cycles) | •Previous LNG‐IUS or endometrial resection/ablation •Abnormal uterine bleeding, other pathology, submucous fibroid •Uterine cavity < 7 cm or over 11 cm | •Change in PBAC score at 12 months (median and range) •Changes in Hb and ferritin at 6 months •Patient satisfaction •Hysterectomy rate at 2 years •Treatment discontinuation (e.g. for adverse events, menorrhagia, LNG‐IUS expulsion) | |
| (D) Soysal 2002 | Balloon NREA (Uterine Balloon Therapy system TM , Gynecare) |
LNG‐IUS | •Women aged over 40 with menorrhagia who refused or did not respond to medical treatment. •PBAC scores of > 150 for 2 consecutive months before randomisation •Family complete •Normal blood tests, transvaginal ultrasonography, hysteroscopy, endometrial suction biopsy or cervical smear examination •No intramural or subserous myomas > 2 cm diameter •Regular uterine cavity ≤ 8‐week pregnancy and < 190 ml on ultrasonography | •Any medical disorder other than iron deficiency anaemia •Abnormal intrauterine pathology | •PBAC reduction and haemoglobin •Health‐related quality of life: SF‐36, HAD depression scale •Treatment satisfaction •Additional treatment received •Adverse effects | Participants on the EA arm received 2 months of GnRH analogues pre‐treatment. |
| (D) Talis 2006 | Balloon NREA (Thermachoice®) | LNG‐IUS | • Women 25 to 50 years, with self described HMB who had completed their family and regular cycle. | •U.S. Submucosal fibroids, intramural fibroids > 3 cm diameter, large subserosal fibroids, endometrial polyps. •Lab: Follicle‐stimulating hormone level (FSH) > 30 IU/l, adverse endometrial histology. •Hysteroscopy: submucosal fibroids, endometrial polyps. •Adnexal abnormality on ultrasound. •Severe inter‐menstrual bleeding, severe dysmenorrhoea or severe pre‐menstrual pain, chronic pelvic pain. •Medical contraindications to either study treatment •Previous endometrial ablation or resection •Uninvestigated post‐coital bleeding or untreated abnormal cervical cytology | •PBAC •Satisfaction •Quality of life and menstrual symptoms, measured by questionnaire •Haemoglobin level •Treatment side effects •Treatment failure (for the LNG‐IUS this was confirmed expulsion, completed removal or the initiation of alternative therapy. For thermal balloon ablation this was the initiation of medication or the completion of alternative surgery, such as hysterectomy). •Direct and indirect costs of treatment, subsequent medical treatment, lost income and medical treatment for failed procedures. | |
| (D) Tam 2006 | Balloon NREA (Thermachoice®) | LNG‐IUS | •Women aged over 40 years with a documented history of excessive menstrual bleeding for at least 3 months, family complete. •Prior oral medical treatment unsuccessful, not on any hormonal treatment | •Uterus > 10 weeks' gravid size, submucosal fibroids, endometrial polyps. •Contraindications to interventions. Possible malignancy | •Menstrual pattern at 1 year (self reported) •Adverse effects •Haemoglobin and iron status •Health status: SF‐36 (using norms for Hong Kong Chinese) | |
| (D) van Zon‐Rabelink 2003 | Balloon NREA (Thermachoice®) | Rollerball REA | • Menorrhagia without sufficient relief from medical therapy by GP • Menstrual blood loss score = 185 points in 2 periods due to dysfunctional uterine bleeding according to US and diagnostic hysteroscopy | not reported | • Technical safety aspects • Reduction in menstrual bleeding • Success rate (PBAC < 185) • Satisfaction | |
| (D) Vercellini 1999 | Vaporising REA | TCRE REA |
• > 35 years • Referred for hysterectomy • Uterine volume < 12‐week pregnancy • Normal uterine cavity at hysteroscopy • No evidence of atypical hyperplasia • No adnexal tumours on clinical and ultrasonographic examination | • Women uncertain about future children • Recent use of hormonal agents or drugs that might affect menstrual blood loss • Intramural or subserous fibroids of ≥ 3 cm • "Unstable" general conditions | • Extent of absorption of distension fluid • Duration of surgery • Difficulty of surgery • Satisfaction rate • Proportion with amenorrhoea • Proportion with amenorrhoea and hypomenorrhoea • PBAC score | |
| (D) Zupi 2003 | TCRE REA |
Minimally invasive hysterectomy (Laparoscopic subtotal hysterectomy) | •Women with HMB unresponsive to medical treatment, younger than 50 years of age; weight less than 100 kg. •Not seeking contraception •Normal endometrial histology and Pap smear within the previous six months •Uterus not greater than 12 weeks of pregnancy in size; without submucosal fibroids, adnexal masses or endometriosis | not reported | •Pain (immediately after surgery and then for a week) •Duration of vaginal bleeding •Date resumed normal activities, sexual intercourse, work •Quality of life (SF‐36) •Further surgery •Operative outcomes (duration of surgery, blood loss, complications, hospital stay) | |
DVT: deep vein thrombosis; EA: endometrial ablation; FSH: follicle‐stimulating hormone; GnRH: gonadotropin releasing hormone; HMB: heavy menstrual bleeding; HT: hormone therapy; IUS: intrauterine system; LNG‐IUS: levonorgestrel releasing intrauterine system; MMAS: menorrhagia multi‐attribute scale; NREA: non‐resectoscopic endometrial ablation; OC: oral contraceptives; PBAC: pictorial blood loss assessment chart; PID: pelvic inflammatory disease; PMS: premenstrual syndrome; QoL: quality of life; REA: resectoscopic endometrial ablation; SAQ: Seattle Angina Questionnaire ; SF‐12: Short Form Health Survey; TCRE: transcervical endometrial resection; UFS‐QoL: Uterine Fibroid Symptom Health‐Related Quality of Life Questionnaire; VAS: visual analogue scale.
We extracted the following characteristics from each included Cochrane Review.
Review title and authors.
Date that the review was last assessed as up‐to‐date or declared a stable review.
Number of trials.
Number of participants.
Participant characteristics.
Inclusion criteria.
Exclusion criteria.
Outcomes.
Where available, the GRADE assessment and any relevant comments made in the review's 'Summary of findings' table regarding trial quality and risk of bias (only for pairwise meta‐analysis).
For the trials contributing to the NMA, we extracted the same information except for the GRADE assessment, as it is not available for individual studies.
We solved any disagreements by discussion between the author team. We contacted review and trial authors where necessary to clarify data included in the reviews.
Where multiple Cochrane Reviews used the same data, we used it only once to avoid duplication of evidence and consequent double‐counting of trial data. We reconciled data across the various reviews to ensure that all the information on each arm of the study was included appropriately. If there was any discrepancy between reviews, two review authors reviewed the original publication. As we anticipated in the protocol, we had various trials of treatment for HMB in more than one review.
As the Review Manager (RevMan) format for overviews allows the inclusion of reviews only, we added the individual studies providing data for the NMAs in the additional references adding a (D), as in data, for easy access and to differentiate them from additional references, and with (N) for the studies that were included but did not contribute data to the NMA.
Assessment of methodological quality of included reviews
Methodological quality of included Cochrane Reviews
Two overview authors assessed the included reviews with the AMSTAR 2 quality measurement tool (Shea 2017). SL assessed all of the reviews, MW assessed five reviews, MB assessed three reviews and VJ assessed one review (to avoid review authors assessing the quality of their own reviews). The overview authors solved any discrepancies by discussion; when necessary, a third overview author was included to adjucate (CF).
The original AMSTAR tool included 11 items (Shea 2007); the new AMSTAR 2 tool includes 16 items in total, and the response categories are simpler with a more comprehensive user guide, including reasons for the final assessment. The final assessments are rated 'yes', 'partial yes' or 'no'. The overall rating is based on weaknesses in critical domains (Shea 2017). See Appendix 2 for AMSTAR 2 details.
Quality of trial evidence included in Cochrane Reviews
We did not reassess the risk of bias for trials included in reviews, as they were assessed in the original reviews. Studies in the danazol review (which has not been updated) were also included in fully assessed reviews. There were only a few inconsistencies between reviews in the risk of bias assessment. When inconsistency was present, one overview author (MB) acted as a third party, checking the reasons for the assessment and the original trials and discussing the assessments with another overview author (CF). For example, review authors judged one study as being at uncertain risk of bias for all categories, as they only had access to a short conference publication and were not able to contact the authors, while in another review, review authors had access to an unpublished copy of the study; thus we used the risk of bias assessment of the review with access to the data. In one review, review authors rated three studies as being at high risk, and in another review, the authors rated the risk of bias as unknown for not mentioning either allocation or blinding; we considered the risk to be unknown. In another review, authors rated one trial from the 1990s as being at high risk of selective reporting because it did not have a protocol, but another review rated the risk as low because the study reported all the pre‐specified outcomes; we considered it to be at low risk.
Where possible, for pairwise comparisons, we used the Cochrane Review authors' GRADE assessments of relevant outcomes, as presented in the summary of findings tables of included Cochrane Reviews. Where GRADE summaries were not available for our outcomes, two overview authors (MBR and AL) independently assessed the quality using GRADEpro (GRADEpro GDT 2015). The GRADE summary of findings in a Cochrane Review includes an overall judgement of the risks of bias in the specific trials contributing data to the pooled effect estimate for each outcome displayed in the table. In addition to the risk of bias, for each outcome the GRADE assessment takes into consideration the following domains: imprecision of effects (due to wide confidence intervals, sparse data or both); unexplained inconsistency between trials (as measured by the I2 statistic value/heterogeneity); indirectness (differences in the population, intervention, comparison or outcome of trials); and evidence of publication bias, where the meta‐analysis includes sufficient trials. Pooled evidence for randomised controlled trials (RCTs) with serious problems in any of these domains is downgraded one level. If problems are very serious, we may downgrade evidence by two levels. The Guideline Development Tool assesses these downgrading decisions and assigns the pooled estimate for each outcome a rating of either high, moderate, low or very low certainty (GRADE handbook; GRADEpro GDT 2015; Guyatt 2008). There are standard definitions to aid the interpretation of GRADE ratings, as follows.
High: further research should not alter our confidence in evidence rated as high certainty.
Moderate: future research will likely impact our confidence in moderate certainty evidence and could change the estimates.
Low or very low: there is considerable uncertainty surrounding the effect estimates considered to be of low or very low certainty, and further research will impact our confidence in these effect estimates and change the estimates (Guyatt 2008; Guyatt 2011a).
We intended to exclude evidence of low or very low certainty, as the GRADE rating of evidence is an important finding of the overview. Unfortunately, as most trials were at high risk of performance bias, most of the evidence was low certainty, so we included it.
A GRADE assessment may be further translated into a summary statement that incorporates the clinical importance of the effect, for clarity, using the following guide.
We have interpreted the evidence following the 'GRADE guidelines 26: Informative statements to communicate the findings of systematic reviews of interventions' by Santesso 2020.
For outcomes subject to NMA, we present a summary of findings table, adapted from Yepes‐Nunez 2019. For NMA, one summary of findings table is produced for each outcome and contains effect estimates, credible intervals or confidence intervals, certainty of evidence (for each included intervention compared to one comparator) and a ranking. The NMA summary of findings table from Yepes‐Nunez 2019 reports the number of trials and patients providing direct evidence between the intervention and the common comparator and uses credible intervals. We adapted the table, including the number of trials and participants providing direct evidence for each intervention, and we reported confidence intervals.
The certainty of the evidence for each intervention takes into account both the certainty of the evidence provided by direct and indirect comparisons, inconsistency or incoherence (consistency between direct and indirect evidence), and imprecision (of the network estimate).
To rate the certainty of evidence of the NMA and interpret it, we considered the direct, indirect and network estimate for each comparison and outcome (Brignardello‐Petersen2017; Brignardello‐Petersen 2019; Puhan 2014). The rating of the direct estimate considered the risk of bias, inconsistency‐heterogeneity, indirectness and publication bias of the direct evidence. If the direct evidence was of high certainty and direct evidence contributed as much as the indirect, we intended to rate the network estimate directly (no comparison was rated of high certainty evidence). If it was moderate, low or very low certainty, we rated the indirect estimate. We rated the indirect evidence considering the most dominant first‐order loop (if available, and if more than one was available we intended to choose the one with a larger number of participants and trials), looking at its direct estimates. Indirect evidence was rated lower than the direct evidence providing evidence on the loop. A loop is a section of the network diagram, with three or more interventions interconnected with direct evidence, and each direct source of evidence can be complemented by an indirect source of evidence for the same comparison. The certainty was considered to be the same as whichever piece of direct evidence on the loop had the lowest certainty. The indirect evidence was also assessed for transitivity. The transitivity assumption requires that the distribution of effect modifiers be similar for all sources of direct evidence (Salanti 2014); this was evaluated conceptually. Finally, we rated the network estimate, choosing the estimate that contributed the most to the comparison (either directly or indirectly, or the highest if both contributed similarly and there was no incoherence) and examined for incoherence (inconsistency test) and imprecision (intervals and direction of effect).
We did not interpret the findings from the NMA as proposed by Brignardello‐Petersen 2020, as we did not have a threshold of clinical significance for the outcomes.
Data synthesis
Types of outcomes
We included Cochrane Reviews that reported one or both of the primary outcomes: menstrual blood loss reduction or treatment satisfaction. We included clinically important secondary outcome measures, such as quality of life, side effects and requirement of further treatment. We planned to impute data where missing for primary outcomes and conduct a sensitivity analysis, testing such results with and without the imputed data.
Data synthesis and presentation
We organised the review evidence for each subgroup (first and second‐line treatments).
NMA is a method that allows synthesis of information from a network of trials assessing the same question, but comparing different interventions. It allows inferences on interventions that have not been compared directly in any studies and can increase the precision for comparisons with limited data (Caldwell 2014; Salanti 2014). NMA enables comparisons using direct and indirect evidence simultaneously and can rank all interventions coherently. Direct evidence is obtained from trials that directly compare two or more interventions, and indirect evidence is also combined when two interventions have been compared to a common comparator. With NMA it is possible to compare multiple interventions at the same time, even if some of them have not been compared to each other. However, the network needs to be connected. The interventions in each NMA are represented by a diagram, where nodes represent the interventions, and the size of the nodes is proportional to the number of participants randomised to that intervention. Nodes are connected by lines; the width of the lines is proportional to the number of trials available for that comparison. An NMA requires included studies to be sufficiently similar so that the distribution of the effect modifiers is similar for all sources of direct evidence (Higgins 2021; Salanti 2014). This is referred to as consistency (also coherence or transitivity) (Dias 2019).
We conducted network meta‐analysis using the Stata 16 network package (White 2015), as RevMan does not support NMA; the NMA model estimation was based on multivariable meta‐analysis. We conducted the NMA using individual studies from the included reviews (checking and deleting duplicates), for the following outcomes.
-
First‐line treatments.
Menstrual blood loss reduction: mean blood loss (combining mean blood loss at the end of treatment and change from baseline); mean blood loss at the end of treatment (sensitivity analysis); and perception of bleeding improvement.
Satisfaction.
Adverse events: side effects and serious adverse events.
-
Second‐line treatments.
Menstrual blood loss reduction: PBAC reduction; proportion with blood loss of PBAC under 75 mL (normal menstrual blood loss) or acceptable improvement (participant perception).
Satisfaction.
Requirement of further surgery (endometrial ablation or hysterectomy) up to one year follow‐up.
Bleeding reduction, and in consequence the requirement of further surgery for treatment failure, are not related to the route of hysterectomy but rather the type of hysterectomy (total or supracervical/subtotal), so we combined all routes of hysterectomy for bleeding outcomes and requirement of further surgery due to treatment failure.
We conducted a second subgroup analysis for endometrial resection and ablation procedures.
We divided the EA subgroup by type of procedure into resectoscopic and non‐resectoscopic ablation, and each group into different types using the most common ones, leaving the remaining as a group.
REA was divided in TCRE with or without rollerball, and other REA.
NREA was divided by procedure into microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA. Outcomes were:
bleeding: amenorrhoea at one year follow‐up;
satisfaction;
adverse event: perforation;
requirement of further surgery or HMB treatment: any (EA or hysterectomy) and requirement of further hysterectomy.
We assumed a common estimate for the heterogeneity variance across the different comparisons. Fixed‐effect and random‐effects models were fitted where possible. Random‐effects models were fitted only where there were three or more studies in at least one treatment comparison. We inspected the heterogeneity estimate, Tau, and where it was close to zero, we presented results for the fixed‐effect model. Where Tau was moderate to large, we presented results for the random‐effects model. We compared Tau to the size of the estimated relative effects and considered it in relation to the scale of the outcomes. Where Tau was consistently larger than the estimated relative treatment effects when considered on the same scale, we qualitatively described it as large, since this implies that the variability between relative treatment effects can be larger than the effects themselves. Where Tau was similar or smaller than the estimated relative effects, we qualitatively described it as moderate.
For the selected model (fixed‐effect or random‐effects), we checked consistency for each network using the node‐split model and inconsistency models (Dias 2010; Dias 2018; Van Valkenhoef 2016; White 2015). If evidence of inconsistency in loops using the node‐split model or overall inconsistency model was detected (P < 0.05), we attempted to explain it by checking the data and study inclusion for errors. Where we could not identify an explanation, we expressed caution in interpreting the results. When the inconsistency was extremely high, we did not perform NMA.
We reported the network estimates for mean difference (MD) or odds ratio (OR), with 95% confidence interval (CI), the certainty of evidence (moderate, low or very low), mean rank and the surface under the cumulative ranking curve (SUCRA).
SUCRA is a numeric representation of the overall ranking. It has a single number associated with each treatment, and the values range from 0% to 100%. The larger the SUCRA, the higher the chances a treatment is ranked highest among all the available treatments. On the contrary, the lower the SUCRA, the more likely the intervention is to be ranked near the bottom. We presented a cumulative SUCRA graph for each outcome, which indicates the cumulative probability of being the best treatment, the second‐best, etc. until the worst treatment for the specific outcome. Interventions with high probabilities for low‐ranking values are likely to be better than interventions starting at low probabilities for low ranks. The ranking goes from best to worst for all outcomes, meaning for positive or good outcomes such as bleeding reduction and satisfaction, a higher rank indicates a higher bleeding reduction or a bigger proportion satisfied with treatment. On the other hand, for negative or bad outcomes, such as adverse events, treatment failure or requirement of further surgery, a higher rank indicates a lower incidence of adverse events, treatment failure and requirement of further surgery.
Where NMA was not conducted (network not connected or inappropriate to pool), we presented the results from pairwise comparisons.
For pairwise comparisons we used risk ratios (RRs) for dichotomous outcomes and MDs for continuous outcomes, with their respective 95% CIs. We used the individual studies used in the reviews. We checked for any duplication, and if more than two studies compared the same treatments, we calculated a random‐effects summary in a pairwise meta‐analysis.
We used RevMan for pairwise meta‐analyses and Stata for NMA (Review Manager 2020; White 2015).
Subgroup analysis
We conducted a subgroup analysis based on whether the participants were surgical candidates or not. We considered that 'the desire (or intention) of future pregnancy', 'women completed their families', 'with indication of hysterectomy' or 'medical treatment failure' as inclusion or exclusion criteria to determine if the participants were surgical candidates. According to that judgement, we performed the NMA for first‐ or second‐line treatment. The progestogen‐releasing intrauterine system (LNG‐IUS) was part of both groups.
A second subgroup analysis was conducted comparing the different endometrial ablation and resection techniques by type of procedure.
Methodology note
Methodologies for both overviews and NMAs have been developed since the publication of the protocol in 2018. It may be appropriate for the update to consider evolving the overview to a review of interventions, which would facilitate the process, by allowing a single search with all the NMA details, instead of updating each review.
Results
Description of included reviews
In July 2021 we identified 20 Cochrane Reviews on heavy menstrual bleeding.
We included nine reviews in this overview: Beaumont 2007; Bofill Rodriguez 2019a; Bofill Rodriguez 2019b; Bofill Rodriguez 2019c; Bofill Rodriguez 2020; Bofill Rodriguez 2021; Bryant‐Smith 2018; Lethaby 2019 and Marjoribanks 2016. The last search was performed after January 2016 in all but Beaumont 2007, a review of danazol that has been stable since 2007. See Figure 1 for details of the selection process.
1.
Review flow diagram
Table 3 presents details of included review characteristics (review title and author, when it was updated, number of studies and participants, type of interventions, inclusion and exclusion criteria and outcomes).
3. Review characteristics.
| Review | Review authors | No. studies (participants) | Participants' characteristics |
Interventions‐ Comparison |
Outcomes | LAST SEARCH | ||
| Inclusion criteria | Exclusion criteria | Primary outcomes | Secondary outcomes | |||||
| Antifibrinolytics for heavy menstrual bleeding (Bryant‐Smith 2018) |
|
7 (N = 1312) | Women of reproductive age With regular heavy periods (measured either objectively or subjectively), undertaken at least 2 months' follow‐up whilst on treatment, recruited from primary care, family planning, or a specialist clinic setting |
|
Antifibrinolytics versus:
|
|
|
November 2017 |
| Endometrial resection and ablation versus hysterectomy for heavy menstrual bleeding (Bofill Rodriguez 2021) |
|
10 (N = 1966) | Women of reproductive years with heavy menstrual bleeding (including both heavy regular periods (menorrhagia) and heavy irregular periods (metrorrhagia), measured objectively or subjectively |
|
|
|
|
September 2019 |
| Combined hormonal contraceptives for heavy menstrual bleeding (Lethaby 2019) |
|
8 (N = 805) | Women of reproductive years Regular heavy periods measured either objectively or subjectively assessed at baseline for at least one‐month follow‐up Type of settings: primary care, family planning, or specialist clinic | Not stated | COC versus:
CVR versus progestogens |
|
|
September 2018 |
| Cyclical progestogens for heavy menstrual bleeding (Bofill Rodriguez 2019b) |
|
15 (N = 1071) | Women of reproductive age Women with regular heavy periods measured either subjectively by the woman, objectively by the alkaline haematin method (more than 80 mL per cycle) or semi‐objectively by the pictorial blood assessment chart Women attending primary care, family planning or specialist clinics |
|
|
|
|
January 2019 |
| Danazol for heavy menstrual bleeding (Beaumont 2007) |
|
9 (N = 353) | Women of reproductive years Regular (21‐35 days cycle) heavy menstrual blood loss, subjectively or objectively defined (for example by alkaline haematin method) Recruitment from primary care, family planning or specialist clinic setting |
|
Danazol versus:
|
|
|
April 2007 * Stable |
| Endometrial resection and ablation techniques for heavy menstrual bleeding (Bofill Rodriguez 2019a) |
|
28 (N = 4287) | Women of reproductive years with regular heavy periods, measured objectively or subjectively |
|
REA versus REA techniques:
REA versus NREA techniques:
NREA versus NREA techniques:
|
|
|
May 2018 |
| Non‐steroidal anti‐inflammatory drugs for heavy menstrual bleeding (Bofill Rodriguez 2019c) |
|
25 (N = 759) | Women of reproductive years Regular heavy periods measured either objectively (greater than 80 mL) for one or more cycles prior to the intervention or subjectively by the patient |
|
NSAIDs versus
|
|
|
April 2019 |
| Progestogen‐releasing intrauterine systems for heavy menstrual bleeding (Bofill Rodriguez 2020) |
|
25 (N = 2511) | Women of reproductive years Regular heavy periods measured either objectively (by the alkaline haematin method), semi‐objectively (by PBAC score) or subjectively (patient perception) |
|
LNG‐IUS versus:
|
|
|
July 2019 |
| Surgery versus medical therapy for heavy menstrual bleeding (Marjoribanks 2016) |
|
15 (N = 1289) | Women of reproductive age Regular heavy menstrual periods measured either objectively (e.g. via the alkaline haematin test) or subjectively (e.g. via the pictorial blood loss assessment chart (PBAC), a menstrual blood loss diary or according to a woman's personal judgement) |
|
|
|
|
January 2016 |
COC: combined oral contraceptive; CVR: combined vaginal ring; HMB: heavy menstrual bleeding; LNG‐IUS:levonorgestrel‐releasing intrauterine system; NSAIDS: non‐steroidal anti‐inflammatory drugs; PBAC: pictorial blood assessment chart; Pg‐IUS: progestogen intrauterine system; TCRE: transcervical resection of the endometrium.
The nine reviews included a total of 138 studies. We identified 26 studies that were part of two reviews and 4 studies that were part of three. After removing the duplicate data, we were left with a final total of 104 unique studies with 11,881 participants.
Nineteen studies with 1931 participants did not contribute data for the following reasons.
Three because one of the treatment arms was not randomised (it was chosen from several options either by the doctor or the women) ((N) Cooper 1997; (N) Gupta 2013; (N) Kupperman 2004).
One because the comparison was the same type of intervention (different dosages of danazol) ((N) Chimbira 1980).
Eight because the data reported did not allow comparisons included in this overview ((N) Andersch 1988; (N) Hall 1987; (N) Khajehei 2013; (N) Lamb 1987; (N) Najam 2010; (N) Onoglu 2007; (N) Reid 2005; (N) Shravage 2011).
One compared TCRE to rollerball ((N) Boujida 2002).
Three because the second intervention was not specified in the included interventions of this overview ((N) Fathima 2012; (N) Goshtasebi 2015; (N) Thabet 2010); therefore they do not contribute with indirect evidence.
Three because the participants had other characteristics; in one, all women had a valve replacement and were receiving anticoagulants ((N) Kilic 2009); in another two, there were different causes of HMB ((N) Cameron 1987; (N) Makarainen 1986).
Altogether, then, 85 studies with 9950 participants contributed data for the overview.
Thirty‐five studies with 2702 participants contributed data for the first‐line treatments network ((D) Agarwal 2016; (D) Ashraf 2017; (D) Bonduelle 1991; (D) Bonnar 1996; (D) Buyru 1995; (D) Callender 1970; (D) Cameron 1990; (D) Chamberlain; (D) Dahiya 2016; (D) Dockeray 1989; (D) Dunphy 1998; (D) Edlund 1995; (D) Endrikat 2009; (D) Fraser 1981; (D) Fraser 1991; (D) Fraser 2011; (D) Freeman 2011; (D) Goshtasebi 2013; (D) Hashim 2012; (D) Higham 1993; (D) Irvine 1988; (D) Jaisamrarn 2006; (D) Jensen 2011; (D) Kaunitz 2010; (D) Kiseli 2016; (D) Kriplani 2006; (D) Lukes 2010; (D) Muggeridge 1983; (D) Preston 1995; (D) Sayed 2011; (D) Shabaan 2011; (D) Tsang 1987; (D) van Eijkeren 1992; (D) Ylikorkala 1986; (D) Zhang 2008). See Table 1 for characteristics of studies contributing with data to the network meta analysis (NMA).
Fifty studies with 7248 participants contributed data to the second‐line treatments network ((D) Abbott 2003; (D) Athanatos 2015; (D) Barrington 2002; (D) Bhattacharya 1997; (D) Bongers 2004; (D) Brun 2006; (D) Clark 2011; (D) Cooper 1999; (D) Cooper 2002; (D) Cooper 2004; (D) Cooper 2019; (D) Corson 2000; (D) Corson 2001; (D) Crosignani 1997; (D) Crosignani 1997a; (D) De Souza 2010; (D) Dickersin 2007; (D) Duleba 2003; (D) Dwyer 1993; (D) Ergun 2012; (D) Gannon 1991; (D) Ghazizadeh 2014; (D) Hawe 2003; (D) Herman 2013; (D) Hurskainen 2004; (D) Istre 1998; (D) Jain 2016; (D) Kittelsen 1998; (D) Laberge 2017; (D) Malak 2006; (D) McClure 1992; (D) Meyer 1998; (D) O'Connor 1997; (D) Ozdegirmenci 2011; (D) Pellicano 2002; (D) Penninx 2010; (D) Penninx 2016; (D) Perino 2004; (D) Pinion 1994; (D) Romer 1998; (D) Sambrook 2009; (D) Sesti 2011; (D) Sesti 2012; (D) Shaw 2007; (D) Soysal 2002; (D) Talis 2006; (D) Tam 2006; (D) van Zon‐Rabelink 2003; (D) Vercellini 1999; (D) Zupi 2003). See Table 2 for characteristics of studies contributing with data to the NMA.
See Figure 2 for the risk of bias assessment of studies contributing data to the NMA.
2.
Risk of bias of individual trials contributing with direct evidence for the network meta‐analysis.
Our search for trials assessing the use of progestin‐only contraceptives for HMB did not identify any available or ongoing clinical trials.
We excluded 11 HMB reviews. Ten had ineligible participants: in seven reviews women had HMB due to fibroids (Gupta 2014; Liu 2013; Murji 2017; Ping Liu 2013; Sangkomkamhang 2013; Song 2013; Tristan 2012), one related to anovulation (Hickey 2012), one to bleeding disorders (Ray 2016), and one to intrauterine devices (Grimes 2006). The last excluded review had an ineligible intervention – the use of preoperative endometrial thinning agents for women with HMB undergoing surgery (Tan 2013).
Methodological quality of included reviews
We used the AMSTAR 2 tool to rate the quality of the included reviews (Shea 2017). See Table 4 for details. A summary of the assessment is as follows.
4. AMSTAR 2 assessmenta,b,c.
| Review | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | (14) | (15) | (16) |
| Antifibrinolytics for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Endometrial resection and ablation versus hysterectomy for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Combined hormonal contraceptives for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Cyclical progestogens for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | ‐ | + | + | + | + |
| Danazol for heavy menstrual bleeding | + | P | + | P | ‐ | + | + | + | + | + | + | ‐ | ‐ | + | ‐ | + |
| Endometrial resection and ablation techniques for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Non‐steroidal anti‐inflammatory drugs for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Progestogen‐releasing intrauterine systems for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
| Surgery versus medical therapy for heavy menstrual bleeding | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + | + |
See Appendix 2 for details.
a(+) Yes. b(P) Partial yes. c(‐) No.
All reviews included the components of participant, intervention, comparison and outcomes (PICO) in the research question and inclusion criteria.
All but Beaumont 2007 contained an explicit statement that the review methods were established prior to the conduct of the review, and the report justified if there was any significant deviations from the protocol.
All reviews included randomised controlled trials (RCTs). The authors based their methods on the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
In all but Beaumont 2007, authors reported using a comprehensive literature search strategy and included authors with subject‐specific expertise.
Review authors reported performing study selection in duplicate in all reviews except Beaumont 2007.
Authors of all of the reviews reported performing data extraction in duplicate.
Authors of all of the reviews provided a list of excluded studies and justified the exclusions.
Authors of all of the reviews described the included studies in adequate detail.
Authors of all of the reviews used a satisfactory technique for assessing the risk of bias in individual studies that were included in the reviews. In Beaumont 2007 only allocation concealment was assessed, according to the Cochrane Handbook at that time.
Authors of all of the reviews reported on the sources of funding for the studies included in the reviews.
All of the reviews intended to perform a meta‐analysis if suitable data were available. The review authors used appropriate statistical methods and based their methods on the Cochrane Handbook.
All but two of the reviews performed sensitivity analysis (Beaumont 2007; Bofill Rodriguez 2019b).
All but Beaumont 2007 described methods to take the risk of bias into account when interpreting and discussing the results of the review.
Authors of all of the reviews provided a satisfactory explanation for, and discussion of, any heterogeneity observed in the results of the review.
All the reviews except Beaumont 2007 intended to investigate publication bias using methods described in the Cochrane Handbook.
Authors of all of the reviews reported any potential sources of conflict of interest.
We consider points 1 to 10 of the AMSTAR 2 to be the most important ones for this specific overview. By performing an NMA, we used data from individual trials and did not base the conclusions on the review's analysis.
The risk of bias of studies providing data for the overview is summarised in Figure 2.
Effect of interventions
First‐line treatments
There were 10 first‐line interventions compared: placebo, non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics (tranexamic acid), combined oral contraceptives (COC), combined vaginal ring (CVR), long‐cycle oral progestogens, luteal phase oral progestogens, levonorgestrel‐releasing intrauterine system (LNG‐IUS), danazol and ethamsylate.
Primary outcomes
1. Menstrual blood loss reduction
Mean blood loss (combined)
Twenty‐six studies reported mean blood loss (either mean blood loss at the end of treatment or mean change from baseline), comparing all 10 treatments including placebo. See Table 5 for the summary of findings for this outcome.
5. Summary of findings for first‐line treatments: mean blood loss (combined data).
| Estimated effects, confidence intervals, and certainty of the evidence for menstrual bleeding reduction (mean blood loss at the end of treatment and change from baseline) with first‐line treatment | |||||
|
BENEFITS Patient or population: women with heavy menstrual bleeding Interventions: non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics, combined oral contraceptive (COC), combined vaginal ring (CVR), luteal cyclical progestogen (luteal Pg), long‐cycle progestogen (long‐cycle Pg), levonorgestrel intrauterine device (LNG‐IUS), danazol and ethamsylate Comparator (reference): placebo (mean blood loss with placebo was 160 U/cycle) Outcome: menstrual blood loss (mean blood loss at the end of treatment and change from baseline) Setting: clinical Network geometry plot: Figure 3 | |||||
| Total studies: 26 trials, 1770 women with direct evidence for all comparisons. 8 trials with 710 women compared the interventions versus placebo | |||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Mean difference | Certainty of evidence |
Mean rank (1 to 10) and SUCRAb |
|
NSAIDs Other direct evidence 5 trials, 205 women Comparing NSAIDs to ethamsylate, antifibrinolytics, luteal Pg and danazolc |
5 trials, 145 women |
−40.67 (−84.61 to 3.27) |
Mean blood loss with NSAIDs was 40.67 lower (84.61 lower to 3.27 higher) than placebo | ⊕⊕⊝⊝ Lowd,e | 6.4 (SUCRA 40%) |
|
Antifibrinolytics Other direct evidence 4 trials, 358 women Comparing antifibrinolytics to luteal and long‐cycle Pg |
4 trials, 565 women |
−80.32 (−127.67 to −32.98) |
Mean blood loss with antifibrinolytics was 80.32 lower (127.67 to 32.98 lower) than placebo |
⊕⊕⊕⊝ Moderatee |
3.7 (SUCRA 70%) |
|
COC Other direct evidence 4 trials, 276 women Comparing COC to CVR, LNG‐IUS and danazolc |
No direct evidence with comparator | −56.08 (−140.88 to 28.72) |
Mean blood loss with COC was 56.08 lower (140.88 lower to 28.72 higher) than placebo | ⊕⊝⊝⊝ Very lowd,f | 5.7 (SUCRA 50%) |
|
CVR Other direct evidence 1 trial, 95 women Comparing CVR to long‐cycle Pg |
No direct evidence with comparator | −81.53 (−177.56 to 14.50) |
Mean blood loss with CVR was 81 lower (177.56 lower to 14.5 higher) than placebo | ⊕⊝⊝⊝ Very lowd,f | 3.9 (SUCRA 70%) |
|
Luteal Pg Other direct evidence 2 trials, 50 women Comparing it to danazol) |
No direct evidence with comparator |
−19.10 (−87.81 to 49.61) |
Mean blood loss with luteal Pg was 19.1 lower (87.81 lower to 49.61 higher) than placebo |
⊕⊝⊝⊝ Very lowe,g | 7.8 (SUCRA 20%) |
|
Long‐cycle Pg Other direct evidence 1 trial, 76 women Comparing it to LNG‐IUS) |
No direct evidence with comparator |
−76.93 (−153.82 to −0.05) |
Mean blood loss with long‐cycle progestogen was 76.93 lower (153.82 to 0.05 lower) than placebo |
⊕⊕⊝⊝ Lowd,e | 4.1 (SUCRA 70%) |
| LNG‐IUS | No direct evidence with comparator |
−105.71 (−201.10 to −10.33) |
Mean blood loss with LNG‐IUS was 105.7 lower (201.1 to 10.3 lower) than placebo |
⊕⊕⊝⊝ Lowd,e | 2.4 (SUCRA 80%) |
| Ethamsylate | No direct evidence with comparator |
10.20 (−73.73 to 94.12) |
Mean blood loss with ethamsylate 10.2 higher (73.73 lower to 94.12 higher) than placebo |
⊕⊝⊝⊝ Very lowe,g | 8.9 (SUCRA 10%) |
| Placebo | Reference comparator |
Not estimable | Not estimable | Reference comparator | 8.9 (SUCRA 10%) |
| CI: confidence interval; NMA: network meta‐analysis. | |||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||
aEstimates are reported as mean difference. bMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). cDanazol: despite being effective in reducing menstrual bleeding, its adverse events advise against its use for HMB. Included in the NMA to provide indirect evidence. Danazol results are not included in the summary of findings. dDowngraded one level for serious risk of bias of direct evidence. eDowngraded one level for serious imprecision. fDowngraded two levels for very serious imprecision. gDowngraded 2 levels for very serious risk of bias of direct evidence.
73.
The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
The network was connected; Figure 4 shows the NMA diagram for mean blood loss.
3.
Network diagram for mean menstrual blood loss with first‐line treatments (combined at the end of treatments and change from baseline) (26 studies, 10 treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
As there was substantial heterogeneity (Tau = 51.05), we used a random‐effects model. There was no evidence of incoherence using the node‐split or inconsistency model (P = 0.99 and increase in heterogeneity). See Appendix 3 for details.
See Figure 5 for the forest plot comparing all interventions to placebo.
4.
Forest plot for menstrual blood loss (mean blood loss at the end of treatment and change from baseline) for first‐line treatments. Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
Figure 6 shows the cumulative probabilities for each treatment of being at each possible rank on reducing menstrual blood loss.
5.
Cumulative rankogram comparing each treatment for menstrual blood loss (mean blood loss at the end of treatment and change from baseline) for first‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a Cumulative Ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
Danazol (†): despite being effective in reducing menstrual bleeding, its adverse events advise against its use for heavy menstrual bleeding. Please refer to the last paragraph of the medical interventions in Description of the interventions.
The best treatment for bleeding reduction was LNG‐IUS (SUCRA 80%, mean rank 2.4, MD −105.71 mL/cycle, 95% CI −201.10 to −10.33; low certainty evidence).
Second, danazol (†) (SUCRA 80%, mean rank 3.1, MD −92.43 mL/cycle, 95% CI −167.06 to −17.81; very low certainty evidence).
Third, antifibrinolytics (SUCRA 70%, mean rank 3.7, MD −80.32 mL/cycle, 95% CI −127.67 to −32.98; moderate certainty evidence).
Fourth, CVR (SUCRA 70%, mean rank 3.9, MD −81.53 mL/cycle, 95% CI −177.56 to 14.50; very low certainty evidence).
Fifth, long‐cycle cyclical progestogen (SUCRA 70%, mean rank 4.1, MD −76.93 mL/cycle, 95% CI −153.82 to −0.05; low certainty evidence).
Sixth, combined oral contraceptives (COC) (SUCRA 50%, mean rank 5.7, MD −56.08 mL/cycle, 95% CI −140.88 to 28.72; very low certainty evidence).
Seventh, NSAIDs (SUCRA 40%, mean rank 6.4, MD −40.67 mL/cycle, 95% CI −84.61 to 3.27; low certainty evidence).
Eighth, cyclical luteal progestogen (SUCRA 20%, mean rank 7.8, MD −19.10 mL/cycle, 95% CI −87.81 to 49.61; very low certainty evidence).
Ethamsylate and placebo were the worst treatments (SUCRA 10%, mean rank 8.9, MD 10.20 mL/cycle, 95% CI −73.73 to 94.12; very low certainty evidence, for ethamsylate; placebo (SUCRA 10%, mean rank 8.9 reference comparator).
Figure 7 summarises the risk of bias of individual trials contributing data to the NMA. Ninety‐five per cent to 100% of the trials were either at low or unknown risk of selection bias (random sequence generation and allocation concealment) and other bias (such as baseline characteristics). Half were at high risk of performance and detection bias (blinding), probably because blinding is difficult due to different routes and timing of administration (COC is usually one pill daily, antifibrinolytics is only during the days of bleeding, the intrauterine system is inserted only once, the vaginal ring lasts for three weeks, etc.); The outcome is likely to be influenced by the lack of blinding. Forty per cent were at high risk of attrition (incomplete outcome data), and 20% were at high risk of reporting bias (selective reporting).
6.
Risk of bias of trials contributing with data for menstrual blood loss (mean blood loss at the end of treatment and mean change from baseline) for first‐line treatments.
Sensitivity analysis for menstrual blood loss reduction
We performed a sensitivity analysis of the mean blood loss (excluding the change from baseline data). There were 23 studies. See Table 6 for summary of findings table for this outcome.
6. Summary of findings for first‐line treatments: mean blood loss (sensitivity analysis).
| Estimated effects, confidence intervals, and certainty of the evidence for menstrual bleeding reduction Mean blood loss at the end of treatment with first‐line treatment | |||||
|
BENEFITS Patient or population: women with heavy menstrual bleeding Interventions: non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics, combined oral contraceptive (COC), combined vaginal ring (CVR), luteal cyclical progestogen (luteal Pg), long‐cycle progestogen (long‐cycle Pg), levonorgestrel intrauterine system (LNG‐IUS ), danazol and ethamsylate Comparator (reference): placebo Outcome: menstrual bleeding reduction (PBAC reduction) Setting: Clinical Network geometry plot: Figure 8 | |||||
| Total studies: 23 trials, 1193 women with direct evidence for all comparisons. 8 trials with 710 women compared the Interventions versus placebo. | |||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Mean difference | Certainty of evidence | Mean rank (1 to 10) and SUCRAb |
|
NSAIDs Other direct evidence 5 trials, 215 women Comparing NSAIDs to ethamsylate, antifibrinolytics, luteal progestogen and danazolc |
5 trials, 145 women |
−45.81 (−78.83 to −12.80) |
Mean blood loss with NSAIDs 138.19 45.81 lower (78.83 to 12.80 lower) than placebo |
⊕⊝⊝⊝ Very lowd,e,f | 6.9 (SUCRA 30%) |
|
Antifibrinolytics Other direct evidence 4 trials, 358 women Comparing antifibrinolytics to luteal and long‐cycle progestogens |
2 trials, 84 women |
−107.93 (−155.12 to −60.73) |
Mean blood loss with antifibrinolytics 107.93 lower (155.12 to 60.73 lower) than placebo |
⊕⊝⊝⊝ Very lowd,e,f,g | 3.1 (SUCRA 80%) |
|
COC Other direct evidence 3 trials, 170 women Comparing COC to CVR, LNG‐IUS and danazolc |
No direct evidence with comparator |
−48.48 (−111.89 to 14.93) |
Mean blood loss with COC 48.48 lower (111.89 lower to 14.93 higher) than placebo |
⊕⊝⊝⊝ Very lowd,e,f | 6.9 (SUCRA 30%) |
|
CVR Other direct evidence 1 trial, 95 women Comparing CVR to long‐cycle progestogen |
No direct evidence with comparator |
−87.04 (−157.23 to −16.86) |
Mean blood loss with CVR 87.04 lower (157.23 to 16.86 lower) than placebo |
⊕⊝⊝⊝ Very lowd,e,f | 4.4 (SUCRA 60%) |
|
Luteal Pg Other direct evidence 2 trials, 50 women Comparing it to danazolc |
No direct evidence with comparator |
−39.17 (−92.76 to 14.41) |
Mean blood loss with luteal Pg 39.17 lower (92.6 lower to 14.4 higher) than placebo |
⊕⊝⊝⊝ Very lowd,e,f,g | 7.3 (SUCRA 30%) |
|
Long‐cycle Pg Other direct evidence 1 trial, 76 women Comparing to LNG‐IUS |
No direct evidence with comparator |
−110.32 (−170.75 to −49.9) |
Mean blood loss with long‐cycleprogestogen MD 110.32 lower (170.75 to 49.9 lower) than placebo | ⊕⊝⊝⊝ Very lowd,e,f | 3.0 (SUCRA 80%) |
| LNG‐IUS | No direct evidence with comparator |
−175.34 (−248.09 to −102.58) |
Mean blood loss with LNG‐IUS 175 lower (248 to 102 lower) than placebo |
⊕⊝⊝⊝ Very lowd,e,f | 1.0(SUCRA 100%) |
| Ethamsylate | No direct evidence with comparator |
−2.56 (−66.84 to 61.73) |
Mean blood loss with ethamsylate 2.56 lower (66.84 lower to 61.73 higher) than placebo |
⊕⊝⊝⊝ Very lowd,e,f | 9.1 (SUCRA 10%) |
| Placebo | Reference comparator |
Not estimable | Not estimable |
Reference comparator | 9.4 (SUCRA 10%) |
| CI: confidence interval; NMA: network meta‐analysis;PBAC: pictorial blood assessment chart. | |||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||
aEstimates are reported as mean difference. bMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). cDanazol: despite being effective in reducing menstrual bleeding, its adverse events advise against its use for HMB. Included in the NMA to provide indirect evidence. Danazol results are not included in the summary of findings. dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded one level for network incoherence.
The network was connected. The NMA diagram is represented in Figure 8.
7.
Network for menstrual blood loss mean blood loss at follow‐up (23 studies, 10 treatments, sensitivity analysis) for first‐line treatments.The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
We used a random‐effects model (Tau = 32.6). The node‐split model showed some evidence of incoherence (See Appendix 4 for details) in loops formed by CVR, COC, long‐cycle progestogens and LNG‐IUS. The inconsistency model gave a P value of 0.32, however, there was a reduction in heterogeneity (Tau = 27.66), suggesting some incoherence. In the presence of incoherence, results should be interpreted with caution.
See Figure 9 for the forest plot comparing all interventions to placebo.
8.
Forest Plot for mean blood loss (sensitivity analysis) for first‐line treatments.
All the evidence has very low certainty.
Figure 10 shows the cumulative probabilities for each treatment at each possible rank for reducing menstrual blood loss.
9.
Cumulative rankogram comparing each treatment for mean blood loss at the end of treatment (sensitivity analysis) for first‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA, the higher its rank among all available treatments.
Danazol (†): despite being effective in reducing menstrual bleeding, its adverse events advise against its use for heavy menstrual bleeding. Please refer to the last paragraph of the medical interventions in Description of the interventions.
The best treatment for reducing menstrual bleeding was LNG‐IUS (SUCRA 100%, mean rank 1.0, MD −175.34 mL/cycle, 95% CI −248.09 to −102.58; very low certainty evidence).
Second, long‐cycle progestogen (SUCRA 80%, mean rank 3.0, MD −110.32 mL/cycle, 95% CI −170.75 to −49.9; very low certainty evidence).
Third, antifibrinolytics (SUCRA 80%, mean rank 3.1, MD −107.93 mL/cycle, 95% CI −155.12 to −60.73; very low certainty evidence).
Fourth, danazol (†) (SUCRA 70%, 3.8, MD −95.64 mL/cycle, 95% CI −150.71 to −40.56; very low certainty evidence).
Fifth, CVR (SUCRA 60%, mean rank 4.4, MD −87.04 mL/cycle, 95% CI −157.23 to −16.86; very low certainty evidence).
Sixth, COC and NSAIDs (both have SUCRA 30%, mean rank 6.9; COC: MD −48.48 mL/cycle, 95% CI −111.89 to 14.93; very low certainty evidence).
Eight, cyclical luteal progestogen (SUCRA 30%, mean rank 7.3, MD −39.17 mL/cycle, 95% CI −92.76 to 14.41; very low certainty evidence).
Ninth, ethamsylate (SUCRA 10%, mean rank 9.1, MD −2.56 mL/cycle, 95% CI −66.84 to 61.73; very low certainty evidence)
Worst, placebo (SUCRA 10%, mean rank 9.4).
Figure 11 summarises the risk of bias of individual trials contributing data to the sensitivity NMA, and it is similar to the main analysis. There was a low or unknown risk of selection bias (random sequence generation and allocation concealment) in 95% of the trials, and of other bias (such as baseline characteristics) in 100%. Half were at high risk of performance and detection bias (blinding), 40% were at high risk of attrition bias (incomplete outcome data), and 20% were at high risk of reporting bias (selective reporting).
10.
Risk of bias of trials contributing with data for mean blood loss at the end of treatment, sensitivity analysis, for first‐line treatment.
The sensitivity analysis removing imputed data resulted in very low certainty evidence, and we are uncertain of the effect of the interventions.
Perception of bleeding improvement
Sixteen studies reported patients' perception of bleeding improvement, comparing all 10 treatments. See Table 7 for the summary of findings table for this outcome.
7. Summary of findings for first‐line treatments: perception of improvement.
| Estimated effects, confidence intervals, and certainty of the evidence for women's perception of heavy menstrual bleeding improvement with first‐line treatment | |||||||
|
BENEFITS Patient or population: women with heavy menstrual bleeding Interventions: Non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics, combined oral contraceptive (COC), combined vaginal ring (CVR), luteal cyclical progestogen (luteal Pg), long‐cycle progestogen (long‐cycle Pg), levonorgestrel intrauterine device (LNG‐IUS), danazol and ethamsylate Comparator (reference): placebo Outcome: women's perception of heavy menstrual bleeding improvement Setting: clinical Network geometry plot: Figure 12 | |||||||
| Total studies: 16, total participants: 1300 (direct evidence) for all comparisons. Interventions versus placebo: 4 RCTs, 487 women | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb(95% CI) | Certainty of evidence |
Mean rank (1 to 10) and SUCRAc |
||
|
Without intervention (with comparator, placebo) |
With intervention | Difference | |||||
|
NSAIDs Other direct evidence 3 RCTs, 272 women Comparing it to ethamsylate, antifibrinolytics,and luteal progestogen) |
1 RCT, 80 women |
7.24 (1.19 to 44.01) |
367 per 1000 | 808 per 1000 |
441 more per 1000 (41 to 595 more) |
⊕⊝⊝⊝ Very lowd,e,f | 5.3 (SUCRA 50%) |
|
Antifibrinolytics Other direct evidence 4 RCTs, 330 women Comparing it to cyclical progestogen (luteal ad long‐cycle) and LNG‐IUS) |
1 RCT, 68 women |
11.13 (1.79 to 69.30) |
367 per 1000 | 866 per 1000 |
499 more per 1000 (142 to 609 more) |
⊕⊝⊝⊝ Very lowd,e,f | 3.9 (SUCRA 70%) |
|
COC Other direct evidence 1 RCT, 50 women Comparing COC to CVR |
2 RCTs, 339 women |
5.43 (1.19 to 24.73) |
367 per 1000 | 759 per 1000 |
392 more per 1000 (41 to 568 more) |
⊕⊝⊝⊝ Very lowd,e,f | 6.0 (SUCRA 40%) |
|
CVR |
No direct evidence with comparator |
14.49 (0.86 to 244.30) |
367 per 1000 | 894 per 1000 | 527 more per 1000 (34 fewer to 626 more) |
⊕⊝⊝⊝ Very lowd,e,f | 3.8 (SUCRA 70%) |
|
Luteal Pg Other direct evidence 2 RCTs, 54 women Comparing it to danazolg |
No direct evidence with comparator |
3.30 (0.44 to 24.68) |
367 per 1000 | 657 per 1000 | 290 more per 1000 (164 fewer to 568 more) |
⊕⊝⊝⊝ Very lowd,e,f | 7.6 (SUCRA 30%) |
|
Long‐cycle Pg Other direct evidence: 2 RCTs, 107 women Comparing it to LNG‐IUS |
No direct evidence with comparator |
5.78 (0.43 to 77.71) |
367 per 1000 | 770 per 1000 | 403 more per 1000 (167 fewer to 611 more) |
⊕⊝⊝⊝ Very lowd,e,f | 5.9 (SUCRA 50%) |
| LNG‐IUS | No direct evidence with comparator |
20.73 (1.60 to 267.83) |
367 per 1000 | 923 per 1000 | 556 more per 1000 (114 to 627 more) |
⊕⊝⊝⊝ Very lowd,f,h | 2.6 (SUCRA 80%) |
| Ethamsylate | No direct evidence with comparator |
3.84 (0.28 to 52.54) |
367 per 1000 | 690 per 1000 | 323 more per 1000 (227 fewer to 601 more) |
⊕⊝⊝⊝ Very lowe,f,i | 6.8 (SUCRA 40%) |
| Placebo | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator |
9.5 (SUCRA 10%) |
| CI: confidence interval; NMA: network meta‐analysis. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and Confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded two levels for serious imprecision. fDowngraded one level for network incoherence. gDanazol: despite being effective in reducing menstrual bleeding, its adverse events advise against its use for HMB. Included in the NMA to provide indirect evidence. Danazol results are not included in the summary of findings. hDowngraded one level for serious imprecision. iDowngraded two levels for very serious risk of bias of the direct evidence.
The network was connected; the NMA diagram for mean blood loss is represented in Figure 12.
11.
Network for women with perception of improvement for first‐line treatments (16 studies, 10 treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
We used a random‐effects model (Tau = 1.03). The node‐split model showed some evidence of incoherence (see Appendix 5 for details) in loops formed by NSAIDs, antifibrinolytics and LNG‐IUS. The inconsistency model gave a P value of 0.003, and there was a reduction in heterogeneity (Tau = 0.39), suggesting incoherence. In the presence of incoherence, results should be interpreted with caution.
See Figure 13 for the forest plot comparing all interventions to placebo.
12.
Forest plot for perception of bleeding improvement forfirst‐line treatments.
All evidence has very low certainty.
Figure 14 shows the cumulative probabilities for each treatment at each possible rank for reducing menstrual blood loss.
13.
Cumulative rankogram comparing each treatment for perception of improvement forfirst‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc.
The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
Danazol (†): despite being effective in reducing menstrual bleeding, its adverse events advise against its use for heavy menstrual bleeding. Please refer to the last paragraph of the medical interventions in Description of the interventions.
The best HMB treatment in terms of perception of improvement was LNG‐IUS (SUCRA 80%, mean rank 2.6, OR 20.73, 95% CI 1.60 to 267.83; very low certainty evidence).
Second, danazol (†) (SUCRA 70%, mean rank 3.6, OR 20.73, 95% CI 1.60 to 267.84; very low certainty evidence).
Third, CVR (SUCRA 70%, mean rank 3.8, OR 14.49, 95% CI 0.86 to 244.30; very low certainty evidence).
Fourth, antifibrinolytics (SUCRA 70%, mean rank 3.9, OR 11.13, 95% CI 1.79 to 69.30; very low certainty evidence).
Fifth, NSAIDs (SUCRA 50%, mean rank 5.3, OR 7.24, 95% CI 1.19 to 44.01; very low certainty evidence).
Sixth, long‐cycle progestogens (SUCRA 50%, mean rank 5.9, OR 5.78, 95% CI 0.43 to 77.71; very low certainty evidence).
Seventh, COC (SUCRA 40%, mean rank 6.0, OR 5.43, 95% CI 1.19 to 24.73; very low certainty evidence).
Eighth, ethamsylate (SUCRA 40%, mean mean rank 6.8, OR 3.84, 95% CI 0.28 to 52.54; very low certainty evidence).
Nineth, luteal progestogen (SUCRA 30%, mean rank 7.6, OR 3.30, 95% CI 0.44 to 24.68; very low certainty evidence).
Worst, placebo (SUCRA 10%, mean mean rank 9.5).
Figure 15 summarises the risk of bias of individual trials contributing data to the NMA. Over 90% were at either low or unknown risk of selection bias (random sequence generation and allocation concealment) and other bias (such as baseline characteristics). Over 70% were at high risk of performance and detection bias (blinding), so the outcome is likely to be influenced by the lack of blinding. Eighty per cent were at low or unclear risk of attrition bias (incomplete outcome data), and 100% were at low or unclear risk of reporting bias (selective reporting).
14.
Risk of bias of trials contributing with data to proportion of women with perception of improvement for first‐line treatment.
2. Satisfaction
Three studies with six treatments reported satisfaction, although only two studies with four interventions (antifibrinolytics, COC, luteal progestogens and LNG‐IUS) were in a connected network. See Table 8 for the summary of findings table for this outcome.
8. Summary of findings for first‐line treatments: satisfaction.
| Estimated effects, confidence intervals, and certainty of the evidence for satisfaction with first‐line treatment | ||||||
|
BENEFITS Patient or population: women with heavy menstrual bleeding Interventions: antifibrinolytics, luteal cyclical progestogen (Luteal Pg) and levonorgestrel intrauterine system (LNG‐IUS) Comparator (reference): combined oral contraceptive (COC) Outcome: treatment satisfaction Setting: Clinical Direct evidence: 2 trials with 99 women; 2 three‐arm studies) Network geometry plot: Figure 16 | ||||||
| Intervention |
Relative effectb (95% CI) (from network) |
Anticipated absolute effectsa (95% CI) | Certainty of the evidence (GRADE) for direct evidence |
Mean rank (1 to 4) and SUCRAc |
||
|
Without intervention (with comparator, placebo) (direct evidence with comparator) |
With intervention | Difference | ||||
| Antifibrinolytics | 1.05 (0.12 to 9.12) |
623 per 1000 | 623 per 1000 | 12 more (452 fewer to 324 more) |
⊕⊝⊝⊝ Very lowd,e | 3.1 (SUCRA 30%) |
| Luteal progestogen |
1.40 (0.15 to 12.61) |
688 per 1000 | 688 per 1000 | 76 more (420 fewer to 341 more) |
⊕⊝⊝⊝ Very lowd,e | 2.6 (SUCRA 50%) |
| LNG‐IUS |
3.39 (0.72 to 16.07) |
842 per 1000 | 842 per 1000 | 231 more (8 fewer to 351 more) |
⊕⊝⊝⊝ Very lowe,f | 1.3 (SUCRA 90%) |
| COC | Reference comparator |
Not estimable | Not estimable | Not estimable | Reference comparator |
3.1 (SUCRA 30%) |
| CI: confidence interval; NMA: network meta‐analysis. | ||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | ||||||
aAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. bEstimates are reported as odds ratio (OR) and confidence interval (CI). cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded two levels for very serious risk of bias of the direct evidence. eDowngraded two levels for very serious imprecision. fDowngraded one level for serious risk of bias of the direct evidence.
The analysis was done on the connected subnetwork; see Figure 16. As there is only a single study per comparison, we used a fixed‐effect model.
15.
Network for satisfaction for first‐line treatments (2 studies, 4 treatments).The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
See Figure 17 for the forest plot. We are uncertain of the true effect of antifibrinolytics, luteal progestogens, LNG‐IUS and COC for satisfaction (all very low certainty evidence).
16.
Forest plot for satisfaction with first‐line treatments All evidence has very low certainty.
We used COC as a comparator, as there were no studies assessing satisfaction using placebo as a comparator. Figure 18 presents the cumulative probabilities for each treatment at each possible rank in terms of satisfaction compared to COC (cumulative SUCRA).
17.
Cumulative rankogram comparing each treatment for satisfaction forfirst‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
Best, LNG‐IUS (SUCRA 90%, mean rank 1.3, OR 3.39, 95% CI 0.72 to 16.07; very low certainty evidence).
Second, luteal progestogen (SUCRA 50%, mean rank 2.6, OR 1.39, 95% CI 0.15 to 12.61; very low certainty evidence).
Third, antifibrinolytics (SUCRA 30%, mean rank 3.1, OR 1.05, 95% CI 0.12 to 9.12; very low certainty evidence) and COC (SUCRA 30%, mean rank 3.1).
Figure 19 summarises the risk of bias of the trials contributing data to the NMA. All were at low or unknown risk of selection and reporting bias. Half were at high risk of attrition and other bias. All were at high risk of performance and detection bias, as blinding is very unlikely among these interventions, and knowing the intervention is likely to interfere with the satisfaction.
18.
Risk of bias of studies included on the network for satisfaction for first‐line treatments.
Secondary outcomes
1. Quality of life
Data extracted were not suitable to combine in an NMA because they were reported as mean differences (with an SD) measured on different scales, and this was not enough to calculate an SMD. It was therefore inappropriate to pool results. Using standardised means, there was extremely high inconsistency in the NMA. We present the results from pairwise comparisons. See Table 9 for the summary of findings table for this outcome.
9. Summary of findings for first‐line treatments: quality of life.
| Intervention | Comparator | Outcomes | Anticipated absolute effectsa (95% CI) | Relative effect (95% CI) | № of women (trials) | Certainty of the evidence (GRADE) | |
| Risk with comparator | Risk with intervention | ||||||
| Antifibrinolytics | Long‐cycle progestogen | Quality of life (SF‐36 general health) |
There was no clear evidence of difference when comparing long‐cycle progestogen with antifibrinolytics (MD 5.00, 95% CI −2 to 12) | — | 90 women (1 RCT) |
⊕⊕⊝⊝ Lowb |
|
| Short‐cycle progestogen | Quality of life (proportion reporting improvement) |
300 per 1000 women reported improvement in quality of life with short‐cycle progestogen |
501 per 1000 women (228 to 1000 women) reported improvement in quality of life with antifibrinolytics |
RR 1.67 (0.76 to 3.64) |
44 women (1 RCT) |
⊕⊝⊝⊝ Very lowc,d | |
| LNG‐IUS | Combined oral contraceptive | Quality of life (good or excellent) | 235 per 1,000 women reported good or excellent quality of life with COC |
282 per 1,000 women (169 to 471 women) reported good or excellent quality of life with |
RR 1.20 (0.72 to 2.00) |
170 (2 RCTs) | ⊕⊕⊝⊝ Lowc,e |
| CI: confidence interval; MD: mean difference; RCT: randomised controlled trial; RR: relative risk | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aThe risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). bDowngraded two levels for very serious risk of bias. cDowngraded one level for serious risk of bias. dDowngraded two levels for imprecision. eDowngraded one level for imprecision.
Antifibrinolytics versus long‐cycle progestogen
Bryant-Smith 2018 reported there may be little to no difference in quality of life, measured with the SF‐36 general health domain, when comparing antifibrinolytics (tranexamic acid) with long‐cycle progestogen (medroxyprogesterone acetate; MD 5.00 points, 95% CI −2.49 to 12.49; 1 RCT, 90 women; low certainty evidence).
Antifibrinolytics versus short‐cycle progestogen
Bryant-Smith 2018 reported that the evidence is uncertain regarding any differences in quality of life, measured by the proportion of women reporting an improvement, when comparing antifibrinolytics (tranexamic acid) with short‐cycle progestogen (norethisterone) (RR 1.67, 95% CI 0.76 to 3.64; 1 RCT, 44 women; very low certainty evidence).
Levonorgestrel‐releasing intrauterine‐device (LNG‐IUS) versus combined oral contraceptives (COC)
Bofill Rodriguez 2020 reported there may be little to no difference in quality of life, measured by the proportion of women reporting good or excellent quality of life, when comparing LNG‐IUS to combined oral contraceptives (RR 1.20, 95% CI 0.72 to 2.00; I2 = 0; 2 RCTs, 170 women; low certainty evidence).
2. Adverse events
Adverse events for first‐line treatments were reported mainly as individual side effects, and trials reported a wide variety, such as acne; anaemia; anxiety; arthralgia; back, chest and breast pain; breast tenderness; cervical dysplasia; depression; dizziness; dysmenorrhoea; dyspepsia; fatigue; headache; hypertension; insomnia; intermenstrual bleeding; leukorrhoea; metrorrhagia; migraine; nausea; vaginitis; voice and skin changes; and weight gain and loss. The abundance of adverse effects makes it impossible to combine due to the risk of double counting. Two trials comparing COC to placebo reported over 15 specific adverse events ((D) Fraser 2011; (D) Jensen 2011); although there was clear evidence of a difference favouring placebo for any adverse event, the only specific adverse events with clear evidence of difference was breast pain and metrorrhagia.
Fourteen studies reported 'any adverse events', comparing eight interventions: placebo, NSAIDs, antifibrinolytics (tranexamic acid), COCs, long‐cycle and luteal oral progestogens, the LNG‐IUS and danazol. See Table 10 for the summary of findings table for this outcome.
10. Summary of findings for first‐line treatments: any adverse event.
| Estimated effects, confidence intervals, and certainty of the evidence for any adverse events with medical treatments for heavy menstrual bleeding (HMB) with first‐line treatment | |||||||
|
BENEFITS Patient or population: women with heavy menstrual bleeding Interventions: non‐steroidal anti‐inflammatory drugs (NSAIDs), antifibrinolytics, combined oral contraceptive (COC), luteal cyclical progestogen (luteal Pg), long‐cycle progestogen (long‐cycle Pg), levonorgestrel intrauterine system (LNG‐IUS) and danazol Comparator (reference): placebo Outcome: any adverse event Setting: clinical Network geometry plot: Figure 20 | |||||||
| Total studies: 14 trials, 1341 women with direct evidence for all comparisons. Three trials with 643 women compared the interventions versus placebo | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 8) and SUCRAc | ||
|
Without intervention (with comparator, placebo) |
With intervention | Difference | |||||
|
NSAIDs Other direct evidence 2 trials, 143 women Comparing NSAIDs antifibrinolytics and luteal progestogen |
Only indirect evidence |
1.14 (0.24 to 5.48) |
505 per 1000 | 537 per 1000 | 32 more (308 fewer to 343 more) |
⊕⊕⊝⊝ Lowd,e | 2.8 (SUCRA 70%) |
|
Antifibrinolytics Other direct evidence 5 trials, 415 women Comparing antifibrinolytics to luteal, long‐cycle progestogens and LNG‐IUS |
1 trial, 297 women |
1.29 (0.63 to 2.63) |
505 per 1000 | 568 per 1000 | 63 more (114 fewer to 224 more) |
⊕⊕⊝⊝ Lowd,e | 2.7 (SUCRA 80%) |
|
COC Other direct evidence 1 trial, 39 women Comparing COC to LNG‐IUS |
2 trials, 346 women |
2.21 (1.43 to 3.41) |
505 per 1000 | 692 per 1000 |
188 more (88 to 272 more) |
⊕⊕⊝⊝ Lowf |
5.2 (SUCRA 40%) |
|
Luteal Pg Other direct evidence 3 trials, 101 women Comparing luteal progestogen to danazolg |
No direct evidence with comparator |
2.10 (0.82 to 5.38) |
505 per 1000 | 681 per 1000 | 177 more (49 fewer to 341 more) |
⊕⊝⊝⊝ Very lowf,h | 5.0 (SUCRA 40%) |
| Long‐cycle Pg | No direct evidence with comparator |
2.62 (0.96 to 7.17) |
505 per 1000 | 727 per 1000 | 223 more (10 fewer to 375 more) |
⊕⊝⊝⊝ Very lowd,f | 5.8 (SUCRA 30%) |
| LNG‐IUS | No direct evidence with comparator |
2.10 (0.69 to 6.38) |
505 per 1000 | 681 per 1000 | 177 more (92 fewer to 362 more) |
⊕⊕⊝⊝ Lowd,e | 4.9 (SUCRA 40%) |
| Placebo | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 1.9 (SUCRA 90%) |
| CI: confidence interval; NMA: network meta‐analysis. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded two levels for very serious imprecision. gDanazol: despite being effective in reducing menstrual bleeding, its adverse events advise against its use for HMB. Included in the NMA to provide indirect evidence. Danazol results are not included in the summary of findings. hDowngraded two levels for very serious risk of bias of the direct evidence.
The network was connected; Figure 20 presents the NMA diagram for any adverse event.
19.
Network diagram for any adverse events withfirst‐line treatments (combined at the end of treatments and change from baseline) (26 studies 26, 10 treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
We used a fixed‐effect model. There is no evidence of inconsistency using the node‐split or inconsistency models (P value for inconsistency model 0.35). See details in Appendix 6.
See Figure 21 for the forest plot comparing all interventions to placebo for any side effects. Figure 22 shows the cumulative probabilities for each treatment at each possible rank in terms of adverse events.
20.
First‐line treatments. Forest plot any adverse event.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
21.
Cumulative rankogram comparing each treatment for any adverse events withfirst‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
Danazol (†): despite being effective in reducing menstrual bleeding, its adverse events advise against its use for heavy menstrual bleeding. Please refer to the last paragraph of the medical interventions in Description of the interventions.
The best HMB treatment in terms of adverse events was placebo (SUCRA 90%, mean rank 1.9, reference comparator).
Second, antifibrinolytics (SUCRA 80%, mean rank 2.7, OR 1.29, 95% CI 0.63 to 2.63; low certainty evidence).
Third, NSAIDs (SUCRA 70%, mean rank 2.8, OR 1.14, 95% CI 0.24 to 5.48; low certainty evidence).
Fourth, LNG‐IUS (SUCRA 40%, mean rank 4.9, OR 2.10, 95% CI 0.69 to 6.38; low certainty evidence).
Fifth, luteal progestogens (SUCRA 40%, mean rank 5.0, OR 2.10, 95% CI 0.82 to 5.38; very low certainty evidence).
Sixth, COC (SUCRA 40% mean rank 5.2, OR 2.21, 95% CI 1.43 to 3.41; very low certainty evidence).
Seventh, long‐cycle progestogens (SUCRA 30%, mean rank 5.8, OR 2.62, 95% CI 0.96 to 7.17; very low certainty evidence).
Worst, danazol (SUCRA 0, mean rank 7.8, OR 7.58, 95% CI 2 to 28.75; very low certainty evidence).
Figure 23 summarises the risk of bias of individual trials contributing data to the NMA. Ninety per cent of trials were at either low or unknown risk of selection bias, 100% were at low or unknown risk of allocation concealment. Over 50% were at high risk of performance and detection bias due to lack of blinding, and over 80% were at low or unknown risk of selective reporting and other bias.
22.
Risk of bias of studies included on the network for any adverse event for first‐line treatments.
Severe adverse events
Severe adverse events were reported in only one trial comparing antifibrinolytics versus placebo (Table 11).
11. Summary of findings for first‐line treatments: serious adverse events.
| Interventions compared with placebo or no treatment | |||||||
| Intervention | Comparator | Outcomes | Anticipated absolute effectsa (95% CI) | Relative effect (95% CI) | N women (trials) |
Certainty of the evidence (GRADE) |
|
| Risk with placebo | Risk with antifibrinolytics | ||||||
| Antifibrinolytics | Placebo | Serious adverse events | 7 per 1000 women experienced a serious adverse event with placebo | 1 per 1000 women (range 1 to 35 women) experienced a serious adverse event with antifibrinolytics | RR 0.10 (0.00 to 2.46 | 468 (2 RCTs) |
⊕⊕⊕⊝ Moderateb |
| CI: confidence interval; RCT: randomised controlled trial; RR: relative risk. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aThe risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). bDowngraded one level for serious risk of bias.
Bryant‐Smith 2018 reported that there is probably little to no difference in the presence of serious adverse events (thrombosis) in women receiving antifibrinolytics versus placebo (RR 0.10, 95% CI 0.00 to 2.46, 2 RCTs, 468 women; moderate certainty evidence).
3. Requirement of additional medical treatment
The intrauterine device system review reported on the outcome of requirement of additional medical treatment in combination with treatment failure (Bofill Rodriguez 2020). Four studies reported failure related to a PBAC cutoff. Two studies, comparing COC and LNG‐IUS, defined treatment failure as requirement of further treatment. See Table 12 for the summary of findings table for this outcome.
12. Summary of findings for first‐line treatments: requirement for further treatment.
| Interventions compared with placebo or no treatment | ||||||
| Intervention | Comparator | Anticipated absolute effectsa(95% CI) | Relative effect (95% CI) | N women (trials) | Certainty of the evidence (GRADE) | |
| Risk with comparator | Risk with intervention | |||||
| LNG‐IUS | Combined oral contraceptives | 254 per 1000 women with combined oral contraceptives experienced treatment failure or required further treatment | 109 per 1000 (61 to 201) women with LNG‐IUS experienced treatment failure or required further treatment |
RR 0.43 (0.24 to 0.79) |
208 (2 RCTs) |
⊕⊕⊝⊝ Lowb,c |
| CI: confidence interval; RCT: randomised controlled trial; RR: relative risk. | ||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | ||||||
aThe risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). bDowngraded one level for serious risk of bias. cDowngraded one level for imprecision.
Levonorgestrel‐releasing intrauterine‐device (LNG‐IUS) versus combined oral contraceptives (COC)
Bofill Rodriguez 2020 reported that evidence suggests that women using LNG‐IUS are less likely to require further treatment due to treatment failure than women using COC (results from pairwise meta‐analysis: RR 0.43, 95% CI 0.24 to 0.79; I2 = 0%; 2 RCTs, 208 women; low certainty evidence).
Second‐line treatments
Primary outcomes
1. Menstrual blood loss reduction
For this outcome, we grouped all hysterectomy routes into a single intervention (hysterectomy).
Bleeding improvement
Eleven studies reported the proportion of women with bleeding improvement (PBAC under 75 points or acceptable bleeding improvement) at one year follow‐up, comparing four interventions: LNG‐IUS, REA, NREA and hysterectomy. See Table 13 for the summary of findings table for this outcome.
13. Summary of findings for second‐line treatments: bleeding (pictorial blood assessment chart) improvement (no imputed data).
| Estimated effects, confidence intervals, and certainty of the evidence for bleeding improvement (PBAC) with second‐line treatment for heavy menstrual bleeding (no imputed data) | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: resectoscopic endometrial ablation (REA), non‐resectoscopic endometrial ablation (NREA) and hysterectomya (all routes of hysterectomy) Comparator (reference): LNG‐IUS Outcome: bleeding improvement (without inputted data) Setting: clinical Network geometry plot: Figure 24 | |||||||
| Total studies: 11 trials, 1790 women with direct evidence for all comparisons. Six trials with 508 women comparing interventions versus LNG‐IUS (comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effectb (95% CI) (from network) |
Anticipated absolute effectc (95% CI) | Certainty of evidence | Mean rank (1 to 4) and SUCRAd | ||
|
Without intervention (with comparator, LNG‐IUS) |
With intervention | Difference | |||||
|
NREA Other direct evidence 5 trials, 1282 women Comparing NREA to REA |
2 trials, 104 women |
3.32 (1.53 to 7.23) |
822 per 1000 | 939 per 1000 |
117 more (54 to 149 more) |
⊕⊕⊕⊝ Moderatee | 2.0 (SUCRA 70%) |
| REA | 3 trials, 179 women |
2.70 (1.29 to 5.66) |
822 per 1000 | 926 per 1000 |
104 more (34 to 141 more) |
⊕⊕⊝⊝ Lowf | 2.8 (SUCRA 40%) |
| Hysterectomya | 1 trial, 225 women. | 25.71 (1.50 to 439.96) |
822 per 1000 | 992 per 1000 |
169 more (52 to 177 more) |
⊕⊕⊝⊝ Lowg | 1.2 (SUCRA 90%) |
| LNG‐IUS | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator |
4.0 (SUCRA 0%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aAs bleeding control and requirement of further surgery for HMB are related to total or subtotal hysterectomy and to to the route of the hysterectomy, in this case all hysterectomies are grouped as one intervention. bEstimates are reported as odds ratios (OR) and confidence interval (CI). cAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. dMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). eDowngraded one level for serious risk of bias of the direct evidence. fDowngraded two levels for very serious risk of bias of the direct evidence. gDowngraded one level for serious imprecision.
74.
The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.
The network is connected: Figure 25 presents the NMA diagram for PBAC improvement.
23.
Network for bleeding (PBAC) improvement (no imputed data) for second‐line treatments (11 studies, 4 treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
We used the fixed‐effect model as the estimated Tau value was practically zero. There is no evidence of inconsistency using the node‐split or inconsistency models (P value for inconsistency model 0.15). See consistency details in Appendix 7.
See Figure 26 for the forest plot. Figure 27 shows the cumulative probabilities for each treatment at each possible rank for improving the PBAC. Treatment hierarchies are presented with the surface under the cumulative curve (SUCRA) and mean rank (one to four).
24.
Forest plot bleeding (PBAC) reduction forsecond‐line treatments. With and without imputed data.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
25.
Cumulative rankogram comparing each treatment for bleeding (PBAC) improvement for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best second‐line treatment without imputed data regarding bleeding improvement was hysterectomy (SUCRA 90%, mean rank 1.2, OR 25.71, 95% CI 1.50 to 439.96; low certainty evidence).
Second, NREA (SUCRA 70%, mean rank 2.0, OR 3.32, 95% CI 1.53 to 7.23; moderate certainty evidence).
Third, REA (SUCRA 40%, mean rank 2.8, OR 2.70, 95% CI 1.29 to 5.66; low certainty evidence).
Worst, LNG‐IUS (SUCRA 0%, mean rank 4, reference comparator).
Figure 28 summarises the risk of bias of each trial contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), selective reporting and other bias (such as baseline characteristics), but all trials were at high risk of performance and detection bias (blinding), as blinding is impossible between some of the interventions. We are uncertain whether the outcome is likely to be influenced by the lack of blinding. The risk of attrition bias (incomplete outcome data) was low or unclear in 80% of trials.
26.
Risk of bias of studies included on the network for for bleeding (PBAC) improvement for second‐line treatments.
Bleeding improvement with imputed data
We imputed data from the mean PBAC value at one year, which a very limited number of studies reported (see Appendix 8 for imputation details) and combined it with the proportion with PBAC under 75 points or acceptable bleeding improvement. We obtained data from 15 studies comparing four interventions: LNG‐IUS, REA, NREA and hysterectomy. See Table 14 for the summary of findings table for this outcome.
14. Summary of findings for second‐line treatments: bleeding improvement (with imputed data).
| Estimated effects, confidence intervals, and certainty of the evidence for bleeding improvement (PBAC) with second‐line treatment for heavy menstrual bleeding (with imputed data) | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: resectoscopic endometrial ablation (REA), non‐resectoscopicendometrial ablation (NREA) and hysterectomy* (all routes of hysterectomy) Comparator (reference): LNG‐IUS Outcome: bleeding improvement (with inputted data) Setting: clinical Network geometry plot: Figure 29 | |||||||
| Total studies: 15 trials, 2241 women with direct evidence for all comparisons. 7 trials with 582 women compared one intervention versus LNG‐IUS | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 4) and SUCRAc | ||
|
Without intervention (with comparator, LNG‐IUS) |
With intervention | Difference | |||||
|
NREA Other direct evidence 7 trials, 1591 women Comparing NREA to REA |
2 trials, 104 women | 2.87 (1.29 to 6.05) |
838 per 1000 | 937 per 1000 |
99 more (33 to 131 more) |
⊕⊕⊕⊝ Moderated | 2.2 (SUCRA 60%) |
| REA | 3 trials, 179 women |
2.65 (1.29 to 5.45) |
838 per 1000 | 932 per 1000 |
94 more (32 to 128 more) |
⊕⊝⊝⊝ Very lowe,f | 2.7 (SUCRA 40%) |
|
Hysterectomyg Other direct evidence 1 trial, 68 women Comparing hysterectomy and NREA |
2 trials, 299 women |
14.31 (2.99 to 68.56) |
838 per 1000 | 987 per 1000 |
149 more (101 to 159 more) |
⊕⊕⊝⊝ Lowd,f | 1.0 (SUCRA 100%) |
| LNG‐IUS | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.0 (SUCRA 0%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and Confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded two levels for very serious risk of bias of the direct evidence. fDowngraded one level for serious imprecision. gAs bleeding control and requirement of further surgery for HMB are related to total or subtotal hysterectomy and to to the route of the hysterectomy, in this case all hysterectomies are grouped as one intervention.
The network is connected: Figure 29 presents the NMA diagram for PBAC improvement with imputed data.
27.
Network for bleeding (PBAC) improvement for second‐line treatmentswith imputed data (15 studies, five treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
There is no evidence of inconsistency using the node‐split or inconsistency models in comparisons (P value for inconsistency model is 0.26). We used the fixed‐effect model. See consistency details in Appendix 9.
See Figure 26 for the forest plot (combined with and without imputed data).
Figure 30 shows the cumulative probabilities for each treatment at each possible rank for improving the PBAC.
28.
Cumulative rankogram comparing each treatment for bleeding (PBAC) improvement with imputed data for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best second‐line treatment for the bleeding improvement (with imputed data) was hysterectomy (SUCRA 100%, mean rank 1.0, OR 14.31, 95% CI 2.99 to 68.56; low certainty evidence).
Second, NREA (SUCRA 60%, mean rank 2.2, OR 2.87, 95% CI 1.29 to 6.05; moderate certainty evidence).
Third, REA (SUCRA 40%, mean rank 2.7, OR 2.65, 95% CI 1.29 to 5.45; very low certainty evidence).
Worst, LNG‐IUS (SUCRA 0%, mean rank 4, reference comparator).
Figure 31 shows the risk of bias of each trial contributing with data to the NMA. The risk of selection bias (random sequence generation and allocation concealment), selective reporting and other bias (such as baseline characteristics) was either low or unknown in all trials, while performance and detection bias (blinding) were at high risk in 90%, because although blinding may be feasible between different endometrial ablation techniques, it is almost impossible between the other interventions. We are uncertain whether the outcome is likely to be influenced by the lack of blinding. The risk of attrition bias (incomplete outcome data) was low or unclear in 80% of trials.
29.
Risk of bias of studies included on the network for for bleeding (PBAC) improvement for second‐line treatmentswith imputed data.
Amenorrhoea
Eighteen studies reported the rate of amenorrhoea up to one year, comparing four treatments: REA, NREA, LNG‐IUS and hysterectomy. See Table 15 for the summary of findings table for this outcome.
15. Summary of findings for second‐line treatments: amenorrhoea.
| Estimated effects, confidence intervals, and certainty of the evidence for amenorrhoea with second‐line treatment for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: resectoscopic endometrial ablation (REA), non‐resectoscopicendometrial ablation (NREA) and hysterectomya (all routes of hysterectomy) Comparator (reference): LNG‐IUS Outcome: amenorrhoea Setting: clinical Network geometry plot: Figure 32 | |||||||
| Total studies: 18 trials, 2484 women with direct evidence for all comparisons. Six trials with 329 women compared the interventions versus LNG‐IUS (comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effectb (95% CI) (from network) |
Anticipated absolute effectc (95% CI) | Certainty of evidence | Mean rank (1 to 4) and SUCRAd | ||
|
Without intervention (with comparator, LNG‐IUS) |
With intervention | Difference | |||||
|
NREA Other direct evidence 12 trials, 2155 women Comparing NREA to REA |
2 trials, 83 women |
1.14 (0.57 to 2.29) |
595 per 1000 | 626 per 1000 | 31 more (139 fewer to 176 more) |
⊕⊝⊝⊝ Very lowe,f,g |
2.9 (SUCRA 40%) |
| REA | 3 trials, 169 women |
1.16 (0.58 to 2.31) |
595 per 1000 | 630 per 1000 | 35 more (135 fewer to 178 more) |
⊕⊝⊝⊝ Very lowf,h |
2.8 (SUCRA 40%) |
| Hysterectomyh | 1 trial, 77 women |
62.11 (3.57 to 1079.05) |
595 per 1000 | 989 per 1000 |
394 more (251 to 404 more) |
⊕⊝⊝⊝ Very lowh,j |
1.0 (SUCRA 100%) |
| LNG‐IUS | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator |
3.3 (SUCRA 20%) |
| CI: confidence interval; HMB: heavy menstrual bleeding. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aAs bleeding control and requirement of further surgery for HMB are related to total or subtotal hysterectomy and to to the route of the hysterectomy, in this case all hysterectomies are grouped as one intervention. bEstimates are reported as odds ratio (OR) and confidence interval (CI). cAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. d Mean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). eDowngraded one level for serious risk of bias of the direct evidence. fDowngraded two levels for very serious risk of bias of the direct evidence. iDowngraded one level for serious imprecision (wide intervals). gDowngraded one level for high heterogeneity on the direct evidence. hAs bleeding control and requirement of further surgery for HMB are related to total or subtotal hysterectomy and to to the route of the hysterectomy, in this case all hysterectomies are grouped as one intervention. jDowngraded two levels for very serious imprecision.
The network is connected: Figure 32 presents the NMA diagram for amenorrhoea.
30.
Network for amenorrhoea at one year for second‐line treatments (18 studies, 4 treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison.Multiarm trials contribute to more than one comparison.
There is no evidence of inconsistency using the node‐split or inconsistency models (P value for inconsistency model 0.24). We used the fixed‐effect model. See consistency details in Appendix 10.
See Figure 33 for the forest plot. Figure 34 shows the cumulative probabilities for each treatment at each possible rank for achieving amenorrhoea.
31.
Forest plot amenorrhoea for second‐line treatments. All evidence has very low certainty. All evidence has very low certainty.
32.
Cumulative rankogram comparing each treatment for amenorrhoea at one year follow‐up for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best second‐line treatment for the bleeding improvement was hysterectomy (SUCRA 100%, mean rank 1.0, OR 62.11, 95% CI 3.57 to 1079.05; very low certainty evidence).
Second, REA (SUCRA 40%, mean rank 2.8, OR 1.16, 95% CI 0.58 to 2.31; very low certainty evidence).
Third, NREA (SUCRA 40%, mean rank 2.9 respectively, OR 1.14, 95% CI 0.57 to 2.29; very low certainty evidence).
Worst, LNG‐IUS (SUCRA 20%, mean rank 3.3, reference comparator).
Figure 35 summarises the risk of bias of each trial contributing with data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment) and selective reporting, while 90% were at high risk of performance and detection bias (blinding), because even though blinding is feasible between different endometrial ablation and resection procedures, it is very unlikely with LNG‐IUS insertion or open surgery. The outcome is likely to be influenced by the lack of blinding. The risk of attrition bias (incomplete outcome data) was low or unclear in 80% of trials, and over 90% of trials were at unclear or low risk of other bias (such as baseline characteristics).
33.
Risk of bias of studies contributing with direct evidence for the network for amenorrhoea for second‐line treatments.
2. Satisfaction
Twenty‐seven studies reported the rate of satisfaction with treatment at up to one year follow‐up, comparing seven treatments: REA, NREA, EA (unspecified), LNG‐IUS, hysterectomy (any/unspecified), minimally invasive hysterectomy and open hysterectomy. See Table 16 for the summary of findings table for this outcome.
16. Summary of findings for second‐line treatments: satisfaction.
| Estimated effects, confidence intervals, and certainty of the evidence for satisfaction with second‐line treatment for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: resectoscopic endometrial ablation (REA), non‐resectoscopicendometrial ablation (NREA), non specified endometrial ablation (EA), levonorgestrel releasing intrauterine device (LNG‐IUS), minimally invasive hysterectomy and open hysterectomy Comparator (reference): LNG‐IUS Outcome: satisfaction Setting: clinical Network geometry plot: Figure 36 | |||||||
| Total studies: 27 trials, 4284 women with direct evidence for all comparisons; 9 trials with 981 women compared the interventions versus hysterectomy (any/unspecified route), the comparator | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectc (95% CI) | Certainty of evidence | Mean rank (1 to 7) and SUCRA d | ||
|
Without intervention (with the comparator) |
With intervention | Difference | |||||
|
NREA Other direct evidence 14 trials, 2591 women Comparing NREA to REA and minimally invasive hysterectomy) |
6 trials, 622 women |
1.59 (1.09 to 2.33) |
792 per 1000 | 859 per 1000 |
67 more (15 to 107 more) |
⊕⊕⊝⊝ Lowd | 3.6 SUCRA 60% |
|
REA Other direct evidence 2 trials, 313 women Comparing REA to hysterectomy (any/ unspecified) and open hysterectomy |
2 trials, 127 women |
1.25 (0.80 to 1.96) |
792 per 1000 | 827 per 1000 | 34 more (39 fewer to 9 more) |
⊕⊝⊝⊝ Very lowd,e | 5.0 SUCRA 30% |
|
EA (unspecified) Other direct evidence 2 trials, 399 women Comparing EA (unspecified) to hysterectomy (any/unspecified) |
No direct evidence with comparator | 1.00 (0.36 to 2.78) |
792 per 1000 | 792 per 1000 | 0 fewer (214 fewer to 122 more) |
⊕⊝⊝⊝ Very lowf,g | 5.8 SUCRA 20% |
| Hysterectomy (any/unspecified route) | 1 trial, 232 women |
1.34 (0.56 to 3.21) |
792 per 1000 | 836 per 1000 | 44 more (111 fewer to132 more) |
⊕⊝⊝⊝ Very lowf,g | 4.4 SUCRA 40% |
| Minimally invasive hysterectomy | No direct evidence with comparator |
7.96 (3.33 to 19.03) |
792 per 1000 | 968 per 1000 |
176 more (135 more to 194 more) |
⊕⊕⊝⊝ Lowd | 1.3 SUCRA 100% |
| Open hysterectomy | No direct evidence with comparator | 5.13 (1.32 to 19.92) |
792 per 1000 | 951 per 1000 |
158 more (42 to 195 more) |
⊕⊝⊝⊝ Very lowd,e | 1.8 SUCRA 90% |
| LNG‐IUS | Reference comparator | Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 6.1 SUCRA 20% |
| CI: confidence interval; HMB: heavy menstrual bleeding. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded two levels for very serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded one level for serious risk of bias of the direct evidence. gDowngraded two levels for very serious imprecision
The network is connected: Figure 36 presents the NMA diagram for satisfaction.
34.
Network for satisfaction for second‐line treatments (27 studies, seven treatments). The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
There are two loops in the network. There is no evidence of inconsistency using the node‐split or inconsistency models (P value for inconsistency model 0.25). We used the fixed‐effect model. See consistency details in Appendix 11.
See Figure 37 for the forest plot. Figure 38 shows the cumulative probabilities for each treatment at each possible rank for achieving satisfaction.
35.
Forest plot for satisfaction. Second‐line treatments.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
36.
Cumulative rankogram comparing each treatment for satisfaction at one year follow‐up for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in terms of satisfaction was minimally invasive hysterectomy (SUCRA 100%, mean rank 1.3, OR 7.96, 95% CI 3.33 to 19.03; low certainty evidence).
Second, open hysterectomy (SUCRA 90%, mean rank 1.8, OR 5.13, 95% CI 1.32 to 19.92; very low certainty evidence).
Third, NREA (SUCRA 60%, mean rank 3.6, OR 1.59, 95% CI 1.09 to 2.33; low certainty evidence).
Fourth, hysterectomy (unspecified route) (SUCRA 40%, mean rank 4.4, OR 1.34, 95% CI 0.56 to 3.21; very low certainty evidence).
Fifth, REA (SUCRA 30%, mean rank 5.0, OR 1.25, 95% CI 0.80 to 1.96; very low certainty evidence).
Sixth, unspecified EA (SUCRA 20%, mean rank 5.8, OR 1.00, 95% CI 0.36 to 2.78; very low certainty evidence).
Worst, LNG‐IUS (SUCRA 20%, mean rank 6.1, reference comparator).
Figure 39 summarises the risk of bias of each trial contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), but 90% were at high risk of performance and detection bias (blinding), probably because blinding is feasible between different endometrial ablation and resection procedures, but it is difficult with LNG‐IUS insertions and hysterectomies. The lack of blinding is likely to influence the outcome. Less than 20% of trials were at high risk for attrition (incomplete outcome data), reporting and other bias.
37.
Risk of bias of studies included on the network for satisfaction for second‐line treatments.
Secondary outcomes
1. Quality of life
There was substantial disagreement between the direct and indirect evidence comparing NREA to minimally invasive hysterectomy ((D) Sesti 2011). This caused substantial inconsistency in the network (node‐split model P < 0.001 and inconsistency model P < 0.001). In addition, inspection of the study estimates showed that the two studies comparing NREA and LNG‐IUS provide substantially different estimates of this relative treatment effect ((D) Talis 2006; (D) Tam 2006). We therefore decided not to formally combine these results in an NMA and do not present results. Instead, we present the results from pairwise comparisons. Seven trials reported quality of life up to two years using the SF‐36 and one, the Menorrhagia Multiatribute Scale (MMAS). The MMAS was reported dichotomously as the proportion with MMAS = 1, which is the best possible result. See Table 17 for the summary of findings for this outcome.
17. Summary of findings for second‐line treatments: quality of life.
| Intervention | Comparator | Outcome | Anticipated absolute effectsa (95% CI) | Relative effect (95% CI) |
N trials (participants) |
Certainty of the evidence (GRADE) | |
| Risk with comparator | Risk with intervention | ||||||
| NREA | LNG‐IUS | SF‐36 general health | There may be little to no difference between NREA and LNG‐IUS (MD 2.9, 95% CI −3.10 to 9.02) |
— | 2 trials 98 women |
⊕⊕⊝⊝ Lowb,c |
|
| Minimally invasive hysterectomy |
SF‐36 general health | Women with minimally invasive hysterectomy may be more likely to have a better quality of life up to two years follow‐up than women with NREA (MD −10.90, 95% CI −15.81 to −5.99) |
— | 1 trial 68 participants |
⊕⊕⊝⊝ Lowd |
||
| Proportion with MMAS = 100 | 583 per 1000 women in the minimally invasive group | 478 per 1000 women in the NREA group | RR 0.82 (0.70 to 0.95) |
1 trial 616 participants |
⊕⊕⊕⊝ Moderateb |
||
| REA | Minimally invasive hysterectomy | SF‐36 general health | There may be little to no difference between REA and minimally invasive hysterectomy (MD −9.90, 95% CI −19.89 to 0.09) |
— | 1 trial 67 participants |
⊕⊕⊝⊝ Very lowb,d |
|
| Open hysterectomy | SF‐36 general health | There may be little to no difference between REA and open hysterectomy (MD −5.30, 95% CI −11.90 to 1.30) |
— | 1 trial 155 women |
⊕⊕⊝⊝ Very lowb,d |
||
| EA unspecified | Hysterectomy (unspecified) |
SF‐36 general health | There may be little to no difference between EA (unspecified) and hysterectomy (unspecified route) (MD −1.90, 95% CI −8.67 to 4.87) |
— | 1 trial 204 women |
⊕⊕⊝⊝ Very lowb,d |
|
| LNG‐IUS | Minimally invasive hysterectomy | SF‐36 general health | There may be little to no difference between LNG‐IUS and minimally invasive hysterectomy (MD −1.50 points, 95% CI −4.28 to 1.28) |
— | 1 trial 72 women |
⊕⊕⊝⊝ Very lowb,d |
|
| Hysterectomy (unspecified) |
SF‐36 general health | There may be little to no difference between LNG‐IUS and hysterectomy (unspecified route) (MD 2.20, 95% CI −2.93 to 7.33) |
— | 1 trial 221 women |
⊕⊕⊝⊝ Very lowb,c |
||
| CI: confidence interval; EA: endometrial ablation; LNG‐IUS: levonorgestrel releasing intrauterine device; MD: mean difference; MMAS: Menorrhagia Multiatribute Scale (100 = best possible result); NREA: non‐resectoscopic endometrial ablation; RCT: randomised controlled trial; REA: resectoscopic endometrial ablation; RR: relative risk; SF‐36: Short Form Health Survey. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aThe risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). bDowngraded one level for serious risk of bias. cDowngraded one level for imprecision. dDowngraded two levels for imprecision.
Data showed that there may be little to no difference on the following comparisons using the SF‐36.
EA (unspecified) versus hysterectomy (unspecified route): MD −1.90 points, 95% CI −8.67 to 4.87; 1 study, 204 women; low certainty evidence).
NREA versus LNG‐IUS: MD 2.9 points, 95% CI −3.10 to 9.02; I2 =81%; 2 studies, 98 women; low certainty evidence).
The comparison reported evidence of difference using the SF‐36.
Evidence favours minimally invasive hysterectomy compared to NREA (MD −10.90 points, 95% CI −15.81 to −5.99; 1 study, 68 women; low certainty evidence).
The comparisons reported difference using the MMAS.
Women having minimally invasive hysterectomy are probably more likely to have a better quality of life measured with MMAS compared to the NREA (RR 0.82, 95% CI 0.70 to 0.95; 1 study, 616 participants; moderate certainty evidence).
The evidence is very uncertain about the following comparisons.
REA versus minimally invasive hysterectomy (MD −9.90 points, 95% CI −19.89 to 0.09; 1 study, 67 women; very low certainty evidence).
REA versus open hysterectomy (MD −5.30 points, 95% CI −11.90 to 1.30; 1 study, 155 women; very low certainty evidence).
Minimally invasive hysterectomy versus LNG‐IUS (MD −1.50 points, 95% CI −4.28 to 1.28; 1 study, 72 women; very low certainty evidence)
LNG‐IUS versus hysterectomy (unspecified, or at surgeon's discretion) (MD 2.20 points, 95% CI −2.93 to 7.33; 1 study, 221 women; very low certainty evidence).
2. Adverse events
Trials mostly reported individual adverse events, and we were unable to group them due to the risk of double counting.
Twelve studies comparing six treatments (LNG‐IUS, REA, NREA, minimally invasive hysterectomy, open hysterectomy and hysterectomy (any/unspecified)) reported 'any adverse event'. The severity of the outcome varied among studies, which reported infection, bladder injury, haematoma, uterine perforation, blood transfusion, pelvic pain, vaginal discharge, abdominal pain, breast tenderness, headache, acne, mood changes, genital ulceration, decreased libido, hair loss, acne, anxiety‐depression, hypertension and leg pain. See Table 18 for the summary of findings table for this outcome.
18. Summary of findings for second‐line treatments: any adverse event.
| Estimated effects, confidence intervals, and certainty of the evidence for any adverse event with second‐line treatment for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: non‐resectoscopicendometrial ablation (NREA), resectoscopic endometrial ablation (REA), EA (unspecified), hysterectomy (any/unspecified), minimally invasive hysterectomy and open hysterectomy. Comparator (reference): LNG‐IUS Outcome: any adverse event Setting: clinical Network geometry plot: Figure 40 | |||||||
| Total studies: 12 trials, 1878 women with direct evidence for all comparisons; 3 trials with 201 women comparing interventions versus LNG‐IUS (comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 6) and SUCRAc | ||
|
Without intervention (with comparator, LNG‐IUS) |
With intervention | Difference | |||||
| NREA Other direct evidence 4 trials, 967 women Comparing NREA to REA and minimally invasive hysterectomy |
1 trial, 72 women |
0.28 (0.11 to 0.69) |
570 per 1000 | 271 per 1000 |
299 fewer (443 fewer to 92 fewer) |
⊕⊝⊝⊝ Very lowd,e | 2.9 (SUCRA 70%) |
| REA Other direct evidence 4 trials, 508 women Comparing REA to minimally invasive, open and any/ unspecified hysterectomy |
2 trials, 129 women |
0.29 (0.15 to 0.59) |
570 per 1000 | 285 per 1000 |
285 fewer (404 fewer to 131 fewer) |
⊕⊝⊝⊝ Very lowe,f | 3.2 (SUCRA 60%) |
| EA (unspecified) Other direct evidence one trial, 202 women Comparing EA unspecified to hysterectomy any/unspecified |
No direct evidence with comparator | 0.23 (0.01 to 4.00) |
570 per 1000 | 234 per 1000 | 336 fewer (557 fewer to 271 more) |
⊕⊝⊝⊝ Very lowe,f | 3.0 (SUCRA 70%) |
| Hysterectomy (any/unspecified) | No direct evidence with comparator | 1.30 (0.20 to 8.37) |
570 per 1000 | 633 per 1000 | 63 more 36 fewer to 347 more) |
⊕⊝⊝⊝ Very lowd,e | 6.2 (SUCRA 10%) |
| Minimally invasive hysterectomy |
No direct evidence with comparator | 0.36 (0.11 to 1.16) |
570 per 1000 | 323 per 1000 | 247 fewer (443 fewer to 36 more) |
⊕⊝⊝⊝ Very lowd,e | 3.9 (SUCRA 50%) |
| Open hysterectomy |
No direct evidence with comparator | 0.21 (0.30 to 1.41) |
570 per 1000 | 218 per 1000 | 352 fewer (285 fewer to 81 more) |
⊕⊝⊝⊝ Very lowd,e | 2.6 (SUCRA 70%) |
| LNG‐IUS | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator |
6.1 (SUCRA 10%) |
| CI: confidence interval; HMB: heavy menstrual bleeding. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded two levels for very serious imprecision. fDowngraded two levels for very serious risk of bias of the direct evidence.
The network is connected; Figure 40 presents the NMA diagram for any adverse events.
38.
Network for any adverse event for second‐line treatments. The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
There is no evidence of inconsistency using the node‐split or inconsistency models (P = 0.43). See consistency details in Appendix 12. We used the fixed‐effect model.
See the forest plot in Figure 41.
39.
Forest plot for any adverse events for second‐line treatments
Figure 42 shows the cumulative probabilities for each treatment at each possible rank for having any adverse events of any degree (having less).
40.
Cumulative rankogram comparing each treatment for any adverse events for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in terms of adverse events (that is, the least likely to produce any), was open hysterectomy (SUCRA 70%, mean rank 2.6, OR 0.21, 95% CI 0.30 to 1.41; very low certainty evidence).
Second, NREA (SUCRA 70%, mean rank 2.9, OR 0.28, 95% CI 0.11 to 0.69; very low certainty evidence).
Third, EA (any or unspecified) (SUCRA 70%, mean rank 3.0, OR 0.23, 95% CI 0.01 to 4.00; very low certainty evidence).
Fourth, REA (SUCRA 60%, mean rank 3.2, OR 0.29, 95% CI 0.15 to 0.59; very low certainty evidence).
Fifth, minimally invasive hysterectomy (SUCRA 50%, mean rank 3.9, OR 0.36, 95% CI 0.11 to 1.16; very low certainty evidence).
Sixth, LNG‐IUS (SUCRA 10%, mean rank 6.1, reference comparator).
Worst, hysterectomy (unspecified or surgeon's choice of route of hysterectomy) (SUCRA 10%, mean rank 6.2, OR 1.30, 95% CI 0.20 to 8.37; very low certainty evidence).
Figure 43 summarises the risk of bias of each trial contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), attrition, reporting and other bias, but they were all at high risk of performance and detection bias (blinding), as blinding is feasible only between different endometrial ablation and resection procedures, less likely with LNG‐IUS compared to EAs procedures, and impossible with hysterectomies. The outcome is likely to be influenced by the lack of blinding.
41.
Risk of bias of studies contributing with direct evidence for any adverse events for second‐line treatments.
3. Requirement of further surgery
For this outcome, we grouped all hysterectomy routes as one intervention (hysterectomy).
The requirement of further surgery for HMB was reported in 22 trials, comparing four treatments: REA, NREA, EA (unspecified), LNG‐IUS and hysterectomy. See Table 19 for the summary of findings table for this outcome.
19. Summary of findings for second‐line treatments: requirement of further surgery.
| Estimated effects, confidence intervals, and certainty of the evidence for requirement of further surgery (ablation or hysterectomy) with second‐line treatment for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: resectoscopic endometrial ablation (REA), non‐resectoscopicendometrial ablation (NREA), endometrial ablation (any/unspecified) and hysterectomya (all routes of hysterectomy) Comparator (reference): LNG‐IUS Outcome: requirement of further surgery (ablation or hysterectomy) Setting: clinical Network geometry plot: Figure 44 | |||||||
| Total studies: 22 trials, 2859 women with direct evidence for all comparisons; 8 trials with 515 women compared the interventions versus LNG‐IUS the comparator | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effectb (95% CI) (from network) |
Anticipated absolute effectc (95% CI) | Certainty of evidence |
Mean rank (1 to 5) and SUCRAd |
||
|
Without intervention (with comparator, LNG‐IUS) |
With intervention | Difference | |||||
| NREA Other direct evidence. 8 trials, 1389 women Comparing NREA to REA and hysterectomya |
3 trials, 214 women |
0.52 (0.28 to 0.97) |
149 per 1000 | 83 per 1000 |
66 fewer (102 to 4 fewer) |
⊕⊕⊕⊝ Moderatee |
2.3 (SUCRA 70%) |
| REA Other direct evidence 5 trials, 695 women Comparing REA to hysterectomya |
5 trials, 301 women |
0.61 (0.32 to 1.15) |
149 per 1000 | 96 per 1000 | 53 fewer (96 fewer to 19 more) |
⊕⊝⊝⊝ Very lowf,g |
3.0 (SUCRA 30%) |
| EA (unspecified) Other direct evidence 1 trial, 260 women Comparing EA (unspecified) to hysterectomya |
No direct evidence with comparator |
3.42 (0.14 to 78.19) |
149 per 1000 | 374 per 1000 | 225 more (125 fewer to 783 more) |
⊕⊝⊝⊝ Very lowe‐h | 4.5 (SUCRA 10%) |
| Hysterectomya |
No direct evidence with comparator |
0.03 (0.01 to 0.13) |
149 per 1000 | 5 per 1000 |
144 fewer (147 to 127 fewer) |
⊕⊕⊕⊝ Moderatef |
1.0 (SUCRA 100%) |
| LNG‐IUS | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.1 (SUCRA 20%) |
| CI: confidence interval; HMB: heavy menstrual bleeding. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aAs bleeding control and requirement of further surgery for HMB are related to total or subtotal hysterectomy and to to the route of the hysterectomy, in this case all hysterectomies are grouped as one intervention. bEstimates are reported as odds ratio (OR) and confidence interval (CI). cAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. dMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). eDowngraded one level for serious imprecision. fDowngraded one level for serious risk of bias of the direct evidence. gDowngraded two levels for very serious risk of bias of the direct evidence. hDowngraded two levels for very serious risk of imprecision.
The network is connected: Figure 44 presents the NMA diagram for requirement of further surgery for HMB treatment. There is no evidence of inconsistency using the node‐split or inconsistency models (P value for inconsistency model 0.73). See consistency details in Appendix 13. We used the fixed‐effect model.
42.
Network for requirement of further surgery (22 studies, seven treatments) for second‐line treatments. The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
See the forest plot in Figure 45. Figure 46 presents the cumulative probabilities for each treatment at each possible rank for requirement of further surgery (requiring less).
43.
Forest plot for the requirement of further surgery for the treatment of heavy menstrual bleeding. Second‐line treatments. Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
44.
Cumulative rankogram comparing each treatment for requirement of further surgery for HMBat one year follow‐up for second‐line treatments. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface under the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in terms of the requirement of further surgery for HMB is hysterectomy (SUCRA 100%, mean rank 1.0, OR 0.03, 95% CI 0.01 to 0.13; moderate certainty evidence).
Second, NREA (SUCRA 70%, mean rank 2.3, OR 0.52, 95% CI 0.28 to 0.97; moderate certainty evidence).
Third, REA (SUCRA 50%, mean rank 3, OR 0.61, 95% CI 0.32 to 1.15; very low certainty evidence).
Fourth, LNG‐IUS (SUCRA 20%, mean rank 4.1, reference comparator).
Worst, EA (any/unspecified) (SUCRA 10%, mean rank 4.5, OR 3.42, 95% CI 0.14 to 78.19; very low certainty evidence).
Figure 47 summarises the risk of bias of trials contributing with data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), reporting and other bias, but over 90% were at high risk of performance and detection bias (blinding), as blinding is feasible only between different endometrial ablation and resection procedures, less likely with LNG‐IUS compared to EA procedures, and impossible with hysterectomies. Lack of blinding may be less likely to interfere with the requirement of further surgery compared to other outcomes.
45.
Risk of bias of studies included on the network for requirement of further surgery for second‐line treatments.
Subgroup analysis
Endometrial resection and ablation procedures (EA)
1. Menstrual blood loss reduction (subgroup EA)
Pictorial blood assessment chart (PBAC) improvement
Ten studies reported PBAC improvement, comparing seven EA procedures: TCRE with or without rollerball, other REA, microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA. See Table 20 for the summary of findings table for this outcome.
20. Summary of findings for endometrial ablation subgroup: bleeding (pictorial blood assessment chart) improvement.
| Estimated effects, confidence intervals, and certainty of the evidence for bleeding (PBAC) improvement with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: TCRE/rollerball, other resectoscopic endometrial ablation (other REA), microwave non‐resectoscopicendometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: bleeding (PBAC) improvement Setting: clinical Network geometry plot: Figure 48 | |||||||
| Total studies: 10 trials, 1852 women with direct evidence for all comparisons; 7 trials with 1602 women compared the interventions versus TCRE/rollerball (the comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 7) and SUCRAc | ||
|
Without intervention (with comparator, TCRE/rollerball) |
With intervention | Difference | |||||
| Other REA Other direct evidence 1 trial, 70 women Comparing other REA with balloon NREA |
1 trial, 91 women |
1.24 (0.43 to 3.61) |
788 per 1000 | 822 per 1000 | 34 more (173 fewer to 142 more) |
⊕⊝⊝⊝ Very lowd | 3.4 (SUCRA 60%) |
| Microwave NREA Other direct evidence 1 trial, 66 women Comparing microwave with bipolar NREA |
2 trials, 562 women |
1.57 (1.03 to 2.39) |
788 per 1000 | 854 per 1000 |
66 more (5 to 111 more) |
⊕⊕⊝⊝ Lowd,e | 2.0 (SUCRA 80%) |
| Hydrothermal ablation NREA | 1 trial, 250 women |
0.72 (0.37 to 1.41) |
788 per 1000 | 728 per 1000 | 60 fewer (209 fewer to 52 more) |
⊕⊝⊝⊝ Very lowd,f,g | 5.9 (SUCRA 20%) |
| Bipolar NREA Other direct evidence 1 trial, 104 women Comparing Bipolar end balloon NREA |
1 trial, 236 women |
1.33 (0.76 to 2.31) |
788 per 1000 | 832 per 1000 | 44 more (49 fewer to 108 more) |
⊕⊝⊝⊝ Very lowd,e,f | 2.9 (SUCRA 70%) |
| Balloon NREA | 1 trial, 239 women |
0.64 (0.39 to 1.05) |
788 per 1000 | 704 per 1000 | 84 fewer (196 fewer to 8 more) |
⊕⊝⊝⊝ Very lowd,g | 6.3 (SUCRA 10%) |
| Other NREA |
1 trial, 234 women |
1.51 (0.90 to 2.53) |
788 per 1000 | 849 per 1000 | 61 more (18 fewer to 116 more) |
⊕⊝⊝⊝ Very lowd,f,g | 3.0 (SUCRA 70%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.5 (SUCRA 40%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). b Anticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for inconsistency. eDowngraded one level for serious imprecision. fDowngraded one level for serious risk of bias of the direct evidence. gDowngraded two levels for very serious risk of imprecision.
The network is connected. See Figure 48 for network diagram.
46.
Subgroup analysis EA. Network, bleeding (PBAC) improvement at one year follow‐up.
The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
We used the fixed‐effect model. There were only a few studies per comparison. There was evidence of inconsistency using the node‐split model in two comparisons (TCRE/rollerball versus microwave NREA and microwave NREA versus bipolar NREA); and in the inconsistency model (P value for inconsistency 0.002). Results should be interpreted with caution due to inconsistency. See Appendix 14.
See forest plot in Figure 49.
47.
Subgroup analysis EA. Forest plot bleeding (PBAC) improvement. Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
Figure 50 presents the cumulative probabilities for each treatment at each possible rank for bleeding (PBAC) improvement at one year follow‐up.
48.
Subgroup analysis EA. Cumulative rankogram comparing each treatment for menstrual bleeding (PBAC) reduction. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in the subanalysis of endometrial resection and ablation techniques for bleeding improvement was microwave NREA (SUCRA 80%, mean rank 2.0, OR 1.57, 95% CI 1.03 to 2.39; low certainty evidence).
Second, bipolar NREA (SUCRA 70%, mean rank 2.9, OR 1.33, 95% CI 0.76 to 2.31; very low certainty evidence).
Third, other NREA (SUCRA 70%, mean rank 3.0, OR 1.51, 95% CI 0.90 to 2.53; very low certainty evidence).
Fourth, other REA (SUCRA 60%, mean rank 3.4, OR 1.24, 95% CI 0.43 to 3.61; very low certainty evidence).
Fifth, TCRE/rollerball (SUCRA 40% and mean rank 4.5, reference comparator).
Sixth, hydrothermal ablation (SUCRA 20%, mean rank 5.9, OR 0.72, 95% CI 0.37 to 1.41; very low certainty evidence).
Worst, balloon NREA (SUCRA 10%, mean rank 6.3, OR 0.64, 95% CI 0.39 to 1.05; very low certainty evidence).
Figure 51 summarises the risk of bias of trials contributing data to the NMA. All trials were at either low or unknown risk of bias for all types of bias except blinding bias, as 65% of the trials were at high risk of performance and detection bias. The lack of blinding could have influenced bleeding improvement.
49.
Subgroup analysis EA. Risk of bias of trials contributing with direct evidence for bleeding (PBAC) improvement.
Amenorrhoea (subgroup EA)
Twenty‐two studies reported menorrhoea at one year follow‐up, comparing seven interventions: TCRE with or without rollerball, other REA, microwave NREA, hydrothermal ablation NREA, bipolar NREA, ballooon NREA and other NREA. See Table 21 for the summary of findings table for this outcome.
21. Summary of findings for endometrial ablation subgroup: amenorrhoea.
| Estimated effects, confidence intervals, and certainty of the evidence for amenorrhoea with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: TCRE/rollerball, other resectoscopic endometrial ablation (other REA), microwave non‐resectoscopicendometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: amenorrhoea at one year follow‐up Setting: clinical Network geometry plot: Figure 52 | |||||||
| Total studies: 22 trials, 3403 women with direct evidence for all comparisons; 14 trials with 2507 women compared the interventions versus TCRE/rollerball (the comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 7) and SUCRAc | ||
|
Without intervention (with TCRE/rollerball) |
With intervention | Difference | |||||
| Other REA Other direct evidence 1 trial, 67 women Comparing other REA with balloon NREA |
3 trials, 419 women |
1.30 (0.46 to 3.69) |
434 per 1000 | 499 per 1000 | 65 more (173 fewer to 303 more) |
⊕⊕⊝⊝ Very lowd,e,f | 3.6 (SUCRA 60%) |
| Microwave NREA Other direct evidence 2 trials, 348 women Comparing microwave NREA with balloon and bipolar NREA |
2 trials, 562 women |
1.84 (0.69 to 4.93) |
434 per 1000 | 585 per 1000 | 151 more (88 fewer to 357 more) |
⊕⊝⊝⊝ Very lowe,f | 2.3 (SUCRA 80%) |
| Hydrothermal ablation NREA Other direct evidence 1 trial, 146 women Comparing hydrothermal ablation NREA and bipolar NREA |
1 trial, 250 women |
0.66 (0.17 to 2.56) |
434 per 1000 | 336 per 1000 | 98 fewer (319 fewer to 229 more) |
⊕⊝⊝⊝ Very lowd,e,f | 5.6 (SUCRA 20%) |
| Bipolar NREA Other direct evidence 4 trials, 335 women Comparing bipolar to balloon NREA |
2 trials, 420 women |
1.86 (0.79 to 4.36) |
434 per 1000 | 588 per 1000 | 154 more (57 fewer to 336 more) |
⊕⊝⊝⊝ Very lowd,e,f | 2.2 (SUCRA 80%) |
| Balloon NREA | 3 trials, 304 women |
0.64 (0.28 to 1.45) |
434 per 1000 | 329 per 1000 | 105 fewer (257 fewer to 92 more) | ⊕⊝⊝⊝ Very lowe,f | 6.0 (SUCRA 20%) |
| Other NREA |
3 trials, 552 women |
1.11 (0.38 to 3.20) |
434 per 1000 | 460 per 1000 | 32 more (208 fewer to 276 more) |
⊕⊝⊝⊝ Very lowd,e,f | 4.0 (SUCRA 50%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.5 (SUCRA 40%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded two levels for very serious risk of imprecision. fDowngraded one level for incoherence (inconsistency).
The network is connected: Figure 52 presents the NMA diagram for amenorrhoea.
50.
Subgroup analysis EA. Network: women with amenorrhoea at one year follow‐up. The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
We used the random‐effects model, as a Tau value of 0.88 suggests moderate/large heterogeneity in the log‐odds ratio scale, and there was evidence of inconsistency using the node‐split model for comparisons (microwave NREA versus bipolar NREA and bipolar NREA versus balloon NREA). Evidence for inconsistency was unclear using the inconsistency model (P = 0.089), with a Tau of 0.76, which is lower than for the consistency model and suggests some inconsistency. Results should be interpreted with caution. See consistency details in Appendix 15.
See the forest plot in Figure 53.
51.
Subgroup EA. Forest plot for amenorrhoea.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
Figure 54 shows the cumulative probabilities for each treatment at each possible rank for achieving amenorrhoea at one year follow‐up.
52.
Subgroup analysis EA. Cumulative rankogram comparing each treatment amenorrhoea at one year follow‐up. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment among endometrial resection and ablation techniques for achieving amenorrhoea at one year follow‐up is bipolar NREA (SUCRA 80%, mean rank 2.2, OR 1.86, 95% CI 0.79 to 4.36; very low certainty evidence).
Second, microwave NREA (SUCRA 80%, mean rank 2.3, OR 1.84, 95% CI 0.69 to 4.93; very low certainty evidence).
Third, other REA (SUCRA 60%, mean rank 3.5, OR 1.30, 95% CI 0.46 to 3.69; very low certainty evidence).
Fourth, other NREA (SUCRA 50%, mean rank 4.0, OR 1.11, 95% CI 0.38 to 3.20; very low certainty evidence).
Fifth, TCRE/rollerball (SUCRA 40%, mean rank 4.5, reference comparator).
Sixth, hydrothermal ablation (SUCRA 20%, mean rank 5.6, OR 0.66, 95% CI 0.17 to 2.56; very low certainty evidence).
Worst, balloon NREA (SUCRA 20%, mean rank 6.0, OR 0.64, 95% CI 0.28 to 1.45; very low certainty evidence).
Figure 55 summarises the risk of bias of trials contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), reporting and other bias, but 60% were at high risk of performance and detection bias (blinding). We are uncertain whether the lack of blinding may influence the perception of amenorrhoea.
53.
Subgroup analysis EA. Risk of bias amenorrhoea at one year follow‐up.
2. Satisfaction (subgroup EA)
Twenty‐seven studies reported satisfaction at one year follow‐up, comparing seven interventions: TCRE with or without rollerball, other REA, microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA. See Table 22 for the summary of findings table for this outcome.
22. Summary of findings for endometrial ablation subgroup: satisfaction.
| Estimated effects, confidence intervals, and certainty of the evidence for satisfaction with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: other resectoscopic endometrial ablation (other REA), microwave non‐resectoscopicendometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: satisfaction Setting: clinical Network geometry plot: Figure 56 | |||||||
| Total studies: 22 trials, 3316 women with direct evidence for all comparisons; 14 trials with 2431 women compared the interventions versus TCRE/rollerball | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) |
Certainty of evidence |
Mean rank (1 to 7) and SUCRAc |
||
|
Without intervention (with TCRE/ Rollerball) |
With intervention | Difference | |||||
| Other REA Other direct evidence 1 trial, 57 women Comparing other REA with balloon NREA |
2 trials, 412 women |
1.01 (0.53 to 1.93) |
894 per 1000 | 895 per 1000 | 1 more (77 fewer to 48 more) |
⊕⊕⊝⊝ Lowd,e | 4.7 (SUCRA 40%) |
| Microwave NREA Other direct evidence 2 trials, 346 women Comparing microwave NREA with balloon and bipolar NREA |
2 trials, 533 women |
1.02 (0.69 to 1.61) |
894 per 1000 | 896 per 1000 | 2 more (41 fewer to 37 more) |
⊕⊕⊝⊝ Lowd,e | 4.6 (SUCRA 40%) |
| Hydrothermal ablation NREA Other direct evidence 1 trial, 146 women Comparing hydrothermal ablation NREA and bipolar NREA |
1 trial, 203 women |
0.66 (0.16 to 2.64) |
894 per 1000 | 848 per 1000 | 46 fewer (320 fewer to 63 more) |
⊕⊝⊝⊝ Very lowd,f | 5.5 (SUCRA 20%) |
| Bipolar NREA Other direct evidence 4 trials, 336 women Comparing bipolar NREA to Balloon NREA |
2 trials, 389 women |
2.37 (1.39 to 4.05) |
894 per 1000 | 952 per 1000 |
58 more (27 more to 78 more) |
⊕⊕⊝⊝ Lowd,e | 1.3 (SUCRA 100%) |
| Balloon NREA |
5 trials, 504 women |
1.05 (0.67 to 1.64) |
894 per 1000 | 898 per 1000 | 5 more (44 fewer to 39 more) |
⊕⊕⊝⊝ Lowd,e | 4.5 (SUCRA 40%) |
| Other NREA | 2 trials, 390 women |
1.70 (0.85 to 3.40) |
894 per 1000 | 935 per 1000 | 41 more (16 fewer to 72 more) |
⊕⊕⊝⊝ Lowd,e | 2.3 (SUCRA 80%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.8 (SUCRA 40%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). d Downgraded one level for serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded two levels for very serious risk of imprecision.
The network is connected: Figure 56 presents the NMA diagram for satisfaction.
54.
Subgroup analysis EA. Network: satisfaction at one year follow‐up.The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
We used the fixed‐effect model. There is no evidence of inconsistency using node‐split or inconsistency models (P value for inconsistency 0.49). See consistency details in Appendix 16.
See forest plot is in Figure 57.
55.
Subgroup analysis EA. Forest plot satisfaction.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
Figure 58 shows the cumulative probabilities for each treatment at each possible rank for achieving amenorrhoea at one year follow‐up.
56.
Subgroup analysis EA. Cumulative rankogram comparing each treatment for satisfaction. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in the subanalysis of endometrial resection and ablation techniques for satisfaction is bipolar NREA (SUCRA 100%, mean rank 1.3, OR 2.37, 95% CI 1.39 to 4.05; low certainty evidence).
Second, other NREA (SUCRA 80%, mean rank 2.3, OR 1.70, 95% CI 0.85 to 3.40; low certainty evidence).
Third, balloon NREA (SUCRA 40%, mean rank 4.5, OR 1.05, 95% CI 0.67 to 1.64; low certainty evidence).
Fourth, microwave NREA (SUCRA 40%, mean rank 4.6, OR 1.02, 95% CI 0.69 to 1.61; low certainty evidence).
Fifth, other REA (SUCRA 40%, mean rank 4.7, OR 1.01, 95% CI 0.53 to 1.93; low certainty evidence).
Sixth, TCRE/rollerball (SUCRA 40%, mean rank 4.8, reference comparator).
Worst, hydrothermal ablation (SUCRA 20%, mean rank 5.5, OR 0.66, 95% CI 0.16 to 2.64; very low certainty evidence).
Figure 59 summarises the risk of bias of trials contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment) and reporting bias, but 60% were at high risk of performance and detection bias (blinding). We are uncertain whether the lack of blinding may influence the perception of amenorrhoea. The risk of attrition and other bias was low or unknown in 80% of the trials.
57.
Subgroupanalysis EA. Risk of bias of trials contributing with direct evidence for Satisfaction.
3. Adverse events (subgroup EA)
There were no data available for any adverse event.
Perforation
Nine studies reported perforation, comparing seven treatments: TCRE with or without rollerball, other REA, microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA. See Table 23 for the summary of findings table for this outcome. The network was connected; see Figure 60 for network diagram.
23. Summary of findings for endometrial ablation subgroup: adverse events, perforation.
| Estimated effects, confidence intervals, and certainty of the evidence for perforation with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: other resectoscopic endometrial ablation (REA), microwave non‐resectoscopicendometrial ablation (NREA), balloon NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: perforation Setting: clinical Network geometry plot: Figure 60 | |||||||
| Total studies: 10 trials, 2421 women compared the interventions versus TCRE/rollerball; 9 trials with 2265 women compared the interventions to the comparator (TCR/rollerball) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) |
Certainty of evidence |
Mean rank (1 to 5) and SUCRA d | ||
|
Without intervention (with comparator, TCRE/rollerball) |
With intervention | Difference | |||||
| Other REA | 1 trial, 366 women |
1.40 (0.01 to 2.71) |
18 per 1000 | 25 per 1000 | 7 more (18 fewer to 29 more) |
⊕⊕⊝⊝ Lowd,e | 3.0 (SUCRA 70%) |
| Microwave NREA |
2 trials, 587 women |
1.55 (0.20 to 12.12) |
18 per 1000 | 27 per 1000 | 9 more (14 fewer to 162 more) |
⊕⊝⊝⊝ Very lowd,f | 6.1 (SUCRA 10%) |
| Hydrothermal ablation Other direct evidence 1 trial, 156 participants Comparing it to bipolar NREA |
No direct evidence with comparator |
0.24 (0.01 to 10.38) |
18 per 1000 | 4 per 1000 | 13 fewer (18 fewer to 140 more) |
⊕⊝⊝⊝ Very lowd,f | 3.9 (SUCRA 50%) |
| Bipolar NREA | 1 trial, 267 women | 0.07 (0.00 to 1.39) |
18 per 1000 | 1 per 1000 | 13 fewer (18 fewer to 7 more) |
⊕⊕⊕⊝ Moderated |
2.1 (SUCRA 80%) |
| Balloon NREA |
2 trials, 382 women |
0.30 (0.04 to 2.16) |
18 per 1000 | 5 per 1000 | 12 fewer (17 fewer to 20 more) |
⊕⊕⊝⊝ Lowd,e | 3.2 (SUCRA 60%) |
| Other NREA | 3 trials, 663 women |
0.19 (0.38 to 1.00) |
18 per 1000 | 3 per 1000 | 14 fewer (11 fewer to 0 fewer ) |
⊕⊕⊕⊝ Moderated |
3.9 (SUCRA 50%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 5.8 (SUCRA 20%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded two levels for very serious risk of imprecision.
58.
Subgroup analysis EA. Subnetwork connected. Perforation. The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
See forest plot in Figure 61.
59.
Subgroup analysis EA. Forest plot for perforation.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
We used the fixed‐effect model. There is potential for inconsistency as there were no loops in the network.
Figure 62 shows the cumulative probabilities for each treatment at each possible rank for perforation (lower rate of perforation has a higher rank).
60.
Subgroup analysis EA.Cumulative rankogram comparing each treatment for perforation. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments..
Higher rank reflects a lower chance of perforation.
The best treatment in the subanalysis of endometrial resection and ablation techniques for perforation (lower rate) was bipolar NREA (SUCRA 80%, mean rank 2.1, OR 0.07, 95% CI 0.00 to 1.39; moderate certainty evidence).
Second, other REA (SUCRA 70%, mean rank 3.0, OR 1.40, 95% CI 0.01 to 2.71; low certainty evidence).
Third, balloon NREA (SUCRA 60%, mean rank, 3.2, OR 0.30, 95% CI 0.04 to 2.16; low certainty evidence).
Fourth, other NREA (SUCRA 50%, mean rank 3.9, OR 0.19, 95% CI 0.38 to 1.00; moderate certainty evidence).
Fifth, hydrothermal ablation NREA (SUCRA 50%, mean rank 3.9, OR 0.24, 95% CI 0.01 to 10.38; very low certainty evidence).
Sixth, TCRE/rollerball (SUCRA 20%, mean rank 5.8, reference comparator).
Worst, microwave NREA (SUCRA 10%, mean rank 6.1, OR 1.55, 95% CI 0.20 to 12.12; very low certainty evidence).
Figure 63 summarises the risk of bias of trials contributing data to the NMA. All trials were at either low or unknown risk of selection bias (random sequence generation and allocation concealment), reporting and other bias, but they were all at high risk of performance and detection bias (blinding). The lack of blinding probably did not influence the diagnosis of perforation, but we cannot be certain. The risk of attrition bias was low or unknown in 90% of trials.
61.
Subgroup analysis EA. Risk of bias of trials contributing with direct evidence to perforation.
4. Requirement of further surgery for HMB (subgroup EA)
Requirement of any surgery for HMB treatment (endometrial ablation/resection or hysterectomy), subgroup EA
Thirteen studies reported the requirement of any further hysterectomy to treat persistent HMB, comparing seven treatments: TCRE with or without rollerball, other REA, microwave NREA, hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA. See Table 24 for the summary of findings table for this outcome.
24. Summary of findings for endometrial ablation subgroup: further surgery (endometrial ablation or hysterectomy) for heavy menstrual bleeding.
| Estimated effects, confidence intervals, and certainty of the evidence for further surgery with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: TCRE/rollerball, other resectoscopic endometrial ablation (other REA), microwave non‐resectoscopicendometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: further (endometrial ablation or hysterectomy) surgery for HMB Setting: clinical Network geometry plot: Figure 64 | |||||||
| Total studies: 13 trials, 1966 women with direct evidence for all comparisons; 7 trials with 1378 women compared the interventions versus TCRE/rollerball (the comparator) | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 7) and SUCRAc | ||
|
Without intervention (with the comparator, TCRE/rollerball) |
With intervention | Difference | |||||
| Other REA Other direct evidence 1 trial, 67 women Comparing other REA with balloon NREA |
2 trials, 388 women |
0.80 (0.49 to 1.32) |
144 per 1000 | 119 per 1000 | 25 fewer (68 fewer to 38 more) |
⊕⊕⊝⊝ Lowd,e | 3.5 (SUCRA 60%) |
| Microwave NREA | 1 trial, 249 women |
0.97 0.49 to 1.90) |
144 per 1000 | 141 per 1000 | 4 fewer (68 fewer to 98 more) |
⊕⊕⊝⊝ Lowd,e | 4.5 (SUCRA 40%) |
| Hydrothermal ablation NREA Other direct evidence 1 trial, 160 women Comparing hydrothermal ablation NREA and bipolar NREA |
No direct evidence with comparator |
2.25 (0.52 to 9.77) |
144 per 1000 | 275 per 1000 | 131 more (64 fewer to 478 more) |
⊕⊕⊝⊝ Lowf | 6.4 (SUCRA 10%) |
| Bipolar NREA Other direct evidence 4 trials, 361 women Comparing |
No direct evidence with comparator |
0.52 (0.19 to 1.46) |
144 per 1000 | 81 per 1000 | 64 fewer (113 fewer to 53 more) |
⊕⊕⊝⊝ Lowf | 2.1 (SUCRA 80%) |
| Balloon NREA | 3 trials, 462 women |
0.59 (0.31 to 1.10) |
144 per 1000 | 91 per 1000 | 54 fewer (95 fewer to 12 more) |
⊕⊕⊕⊝ Moderated | 2.2 (SUCRA 80%) |
| Other NREA |
1 trial, 279 women |
1.00 (0.42 to 2.40) |
144 per 1000 | 144 per 1000 | 0 fewer (78 fewer to 144 more) |
⊕⊝⊝⊝ Very lowf,g | 4.5 (SUCRA 40%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator |
4.8 (SUCRA 40%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; MD: mean difference; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as MDs. bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded one level for serious imprecision. fDowngraded two levels for very serious risk of imprecision. gDowngraded two levels for very serious risk of bias of the direct evidence.
The network is connected. See Figure 64 for the network diagram.
62.
Subgroup analysis EA. Network: requirement of further surgery (ablation or hysterectomy) for HMB.The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
There is no evidence of inconsistency using the node‐split or inconsistency model (P value for inconsistency 0.97). See Appendix 17.
See the forest plot in Figure 65. Figure 66 presents the cumulative probabilities for each treatment at each possible rank for requirement of further surgery for HMB at up to one year follow‐up.
63.
Subgroup analysis EA. Forest plot for requirement of any further surgery for HMB treatment.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
64.
Subgroup analysis EA. Cumulative rankogram comparing each treatment for requirement of further surgery for HMB treatment. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in the subanalysis of endometrial resection and ablation techniques for requirement of further surgery (endometrial ablation or hysterectomy) for persistent bleeding (lower rate) was bipolar NREA (SUCRA 80%, mean rank 2.1, OR 0.52, 95% CI 0.19 to 1.46; low certainty evidence).
Second, balloon NREA (SUCRA 80%, mean rank 2.2, OR 0.59, 95% CI 0.31 to 1.10; moderate certainty evidence).
Third, other REA (SUCRA 60%, mean rank 3.5, OR 0.80, 95% CI 0.49 to 1.32; very low certainty evidence).
Fourth and fifth (tied), microwave NREA (SUCRA 40%, mean rank 4.5, OR 0.97, 95% CI 0.49 to 1.90; low certainty evidence) and other NREA (SUCRA 40%, mean rank 4.5, OR 1.00, 95% CI 0.42 to 2.40; very low certainty evidence).
Sixth, TCRE/rollerball (SUCRA 40% mean rank 4.8, reference comparator).
Worst, hydrothermal ablation (SUCRA 10%, mean rank 6.4, OR 2.25, 95% CI 0.52 to 9.77; low certainty evidence).
Figure 67 summarises the risk of bias of trials contributing with data to the NMA. All trials were at either low or unknown risk of bias for all types of bias except blinding and attrition. Over half were at high risk of blinding bias, and 80% were at low or unclear risk of attrition bias. The lack of blinding is not likely to influence the requirement of further surgery for persistent bleeding.
65.
Subgroup analysis EA. Risk of bias of trials contributing with direct evidence to any further surgery for HMB treatment.
Requirement of further hysterectomy for HMB treatment (subgroup EA)
Fourteen studies reported the requirement of further hysterectomy to treat persistent HMB, comparing six treatments: TCRE with or without rollerball, microwave NREA, hydrothermal ablation NREA, bipolar NREA, ballooon NREA and other NREA. See Table 25 for the summary of findings table for this outcome.
25. Summary of findings for endometrial ablation subgroup: further hysterectomy for heavy menstrual bleeding.
| Estimated effects, confidence intervals, and certainty of the evidence for further hysterectomy with endometrial ablation for heavy menstrual bleeding | |||||||
|
BENEFITS Patient or population: women with HMB, surgical candidates (with medical treatment failure, have completed their families or hysterectomy candidates) Interventions: microwave non‐resectoscopicendometrial ablation (NREA), hydrothermal ablation NREA, bipolar NREA, balloon NREA and other NREA Comparator (reference): TCRE/rollerball Outcome: further hysterectomy for HMB Setting: clinical Network geometry plot: Figure 68 | |||||||
| Total studies: 14 trials, 2412 women with direct evidence for all comparisons. 9 trials with 1675 women compared the interventions versus TCRE/Rollerball (comparator). | |||||||
| Intervention | N trials and N women (direct evidence with comparator) |
Relative effecta (95% CI) (from network) |
Anticipated absolute effectb (95% CI) | Certainty of evidence | Mean rank (1 to 5) and SUCRAc | ||
|
Without intervention (with comparator) |
With intervention | Difference | |||||
| Microwave NREA Other direct evidence 2 trials, 353 women Comparing Microwave NREA with bipolar and balloon NREA |
2 trials, 571 women |
0.92 (0.49 to 1.70) |
103 per 1000 | 95 per 1000 | 7 fewer (5 fewer to 6 more) |
⊕⊕⊝⊝ Moderated | 3.9 (SUCRA 40%) |
| Hydrothermal ablation NREA Other direct evidence 1 trial, 160 women Comparing Hydrothermal ablation NREA with bipolar NREA |
Only indirect evidence |
1.39 (0.30 to 6.41) |
103 per 1000 | 138 per 1000 | 35 more (70 fewer to 321 more) |
⊕⊕⊝⊝ Very lowd‐g | 4.9 (SUCRA 20%) |
| Bipolar NREA Other direct evidence 2 trials, 224 women Comparing bipolar and balloon NREA |
1 trial, 153 women |
0.55 (0.22 to 1.37) |
103 per 1000 | 59 per 1000 | 44 fewer (78 fewer to 33 more) |
⊕⊝⊝⊝ Lowd,f | 1.9 (SUCRA 80%) |
| Balloon NREA | 3 trials, 306 women |
0.86 (0.51 to 1.48) |
103 per 1000 | 90 per 1000 | 13 fewer (48 fewer to 42 more) |
⊕⊝⊝⊝ Lowd,f | 3.6 (SUCRA 50%) |
| Other REA | 3 trials, 645 women |
0.64 (0.33 to 1.23) |
103 per 1000 | 68 per 1000 | 35 fewer (66 fewer to 21 more) |
⊕⊝⊝⊝ Lowd,f | 2.3 (SUCRA 70%) |
| TCRE/rollerball | Reference comparator |
Not estimable | Not estimable | Not estimable | Not estimable | Reference comparator | 4.5 (SUCRA 30%) |
| CI: confidence interval; HMB: heavy menstrual bleeding; MD: mean difference; PBAC: pictorial blood assessment chart; TCRE: transcervical endometrial resection. | |||||||
| Grade Working Group grades of evidence (or certainty of evidence) High certainty: we are very confident that the true lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. | |||||||
aEstimates are reported as odds ratio (OR) and confidence interval (CI). bAnticipated absolute effect. Anticipated absolute effects compares two risks by calculating the difference between the risks of the intervention groups with the risk of the control group. cMean rank and SUCRA. Rank statistics is defined as the probabilities that a treatment out of N treatments in a network meta‐analysis is the best, the second‐best, the third‐best, and so on until the least effective treatment. SUCRA (Surface Under the Cumulative RAnking curve) is a numeric presentation of the overall ranking and presents a single number associated with each treatment; SUCRA values range from 0 (worst) to 100% (best). dDowngraded one level for serious risk of bias of the direct evidence. eDowngraded two levels for very serious risk of bias of the direct evidence. fDowngraded one level for serious imprecision. gDowngraded two levels for very serious risk of imprecision.
The network was connected. See Figure 68 for network diagram.
66.
Subgroup analysis EA. Network: requirement of further hysterectomy for HMB. The nodes represent an intervention, and their size is proportional to the number of trials comparing this intervention to any other intervention in the network. The lines connecting each pair of interventions represent a direct comparison and are drawn proportional to the number of trials making each direct comparison. Multiarm trials contribute to more than one comparison.
We used the fixed‐effect model. There is no evidence of inconsistency using node‐split or inconsistency modelw (P value for inconsistency 0.7). See Appendix 18 for details.
See the forest plot in Figure 69. Figure 70 presents the cumulative probabilities for each treatment at each possible rank for requirement of further hysterectomy for HMB treatment at one year follow‐up.
67.
Subgroup EA. Forest plot requirement of further hysterectomy for HMB.
Colours of the lines are according to the certainty of the evidence: dark green, high certainty evidence (HCE); light green, moderate certainty evidence (MCE); yellow, low certainty evidence (LCE); and red, very low certainty evidence (VLCE).
68.
Subgroup analysis EA. Cumulative rankogram comparing each treatment for requirement of further hysterectomy for HMB. Ranking indicates the cumulative probability of being the best treatment, the second best, the third best, etc. The x axis shows the relative ranking and the y‐axis, the cumulative probability of each ranking. The surface underneath the curve is a cumulative ranking line (SUCRA); the larger the SUCRA the higher its rank among all available treatments.
The best treatment in the subanalysis of endometrial resection and ablation techniques for requirement of further hysterectomy for persistent bleeding (lower rate) was bipolar NREA (SUCRA 80%, mean rank 1.9, OR 0.55, 95% CI 0.22 to 1.37; low certainty evidence).
Second, other REA (SUCRA 70%, mean rank 2.3, OR 0.64, 95% CI 0.33 to 1.23; low certainty evidence).
Third, balloon NREA (SUCRA 50%, mean rank 3.6, OR 0.86, 95% CI 0.51 to 1.48; low certainty evidence).
Fourth, microwave NREA (SUCRA 40%, mean rank 3.9, OR 0.92, 95% CI 0.49 to 1.70; moderate certainty evidence).
Fifth, TCRE/rollerball (SUCRA 30%, mean rank 4.5, reference comparator).
Sixth, hydrothermal ablation (SUCRA 20%, mean rank 4.9, OR 1.39, 95% CI 0.30 to 6.41; very low certainty evidence).
Figure 71 summarises the risk of bias of trials contributing with data to the NMA. All trials were at either low or unknown risk of all biases except blinding. However, the lack of blinding is not likely to influence the requirement of further hysterectomy for persistent bleeding.
69.
Subgroup analysis EA. Risk of bias of trials contributing with direct evidence to further hysterectomy for HMB.
Discussion
Summary of main results
This overview included nine Cochrane Reviews (covering 104 studies) of interventions to treat heavy menstrual bleeding (HMB) in women of reproductive age. Eighty‐five of these studies (9950 participants) provided direct evidence for the overview network meta‐analysis (NMA). The overview shows that first‐line treatments for HMB vary widely in efficacy, safety and satisfaction, while the differences are less pronounced for second‐line treatments. We assessed the certainty of the evidence for specific interventions as being from moderate to very low, with most being low to very low certainty.
First‐line treatments
For women with HMB, evidence suggests LNG‐IUS is the best intervention to reduce menstrual blood loss. Antifibrinolytics are probably second‐best. Evidence suggests that long‐cycle progestogens reduce menstrual blood loss, and NSAIDs slightly reduce HMB; we are uncertain of the true effect of the remaining interventions, such as combined oral contraceptives (COC), combined vaginal ring (CVR), cyclical luteal progestogens and ethamsylate, as the evidence was very low certainty; see Table 5.
The sensitivity analysis for mean blood loss (without combining data, see Table 6), the perception of bleeding improvement (Table 7), and satisfaction (Table 8) provided very low certainty evidence; thus, we are uncertain of the true effect of the interventions for these outcomes.
The evidence suggests that there is little to no difference in quality of life improvement when comparing tranexamic acid versus long‐cycle progestogens, or comparing LNG‐IUS versus COC. We are uncertain of the true effect on quality of life when comparing antifibrinolytics versus short‐cycle progestogens (very low certainty evidence; Table 9).
The evidence suggests that the use of antifibrinolytics, NSAIDs, LNG‐IUS or placebo results in little to no difference on the rate of any adverse events, and also that COC increases the rate of any side effect. We are uncertain of the true effect of the remaining interventions (CVR, luteal and long‐cycle progestogens and ethamsylate), as the evidence is very low certainty (Table 10).
There may be little to no difference in serious adverse events between antifibrinolytics and placebo (Table 11).
The evidence suggests that women using an LNG‐IUS are less likely to require further treatment for HMB compared to women receiving COC (Table 12).
Second‐line treatments
Evidence suggests that hysterectomy of any type is the best intervention to reduce menstrual blood loss (with and without imputed data). NREA probably reduces menstrual blood loss and is the second best (with and without imputed data). Evidence suggests that REA reduces menstrual blood loss without imputed data, but we are uncertain of the true effect with imputed data (very low certainty evidence; Table 13 Table 14).
Evidence on amenorrhoea was of very low certainty, mainly because of very serious risk of bias of the direct evidence and serious or very serious imprecision; thus, we are uncertain of the true effect of the interventions on this outcome (Table 15).
Evidence suggests that minimally invasive hysterectomy (laparoscopic or vaginal) results in a large increase in the proportion of women satisfied with treatment, and NREA results in an increase in the proportion of women satisfied with treatment. We are uncertain of the true effect of the remaining interventions (very low certainty evidence; Table 16).
Women with minimally invasive hysterectomy probably have a better quality of life at up to two years follow‐up compared to women receiving NREA. There may be little to no difference in the general health for the following pairwise comparisons: unspecified endometrial ablation (EA) versus hysterectomy (unspecified or at surgeon's discretion), and NREA versus LNG‐IUS. The evidence is very uncertain about the effect on quality of life of the following comparisons: REA versus minimally invasive hysterectomy or open hysterectomy, and LNG‐IUS versus minimally invasive hysterectomy or open hysterectomy (very low certainty evidence; Table 17).
Evidence on any adverse event was very low certainty, so we are uncertain of the true effect of the interventions on this outcome. This uncertainty may be related to the wide variety of adverse events being reported (Table 18).
Hysterectomy (all routes) and NREA probably reduce the rate of the requirement of further surgery for HMB. We are uncertain of the true effect of REA and unspecified EA (very low certainty evidence; Table 19).
EA subgroup analysis
For women with HMB treated with EA, evidence suggests that microwave NREA is the best option for reducing menstrual blood loss, and bipolar NREA is probably the best option for achieving amenorrhoea. We are uncertain of the true effect of the remaining interventions on these outcomes, as the evidence was very low certainty (Table 20; Table 21).
For the outcome of satisfaction, evidence suggests bipolar NREA has the best satisfaction rate; other REA, microwave NREA, balloon NREA, and other NREA may not be different compared to TCRE/rollerball, and we are uncertain of the true effect of hydrothermal NREA, as the evidence was very low certainty (Table 22).
For the adverse event of perforation, bipolar NREA probably shows little to no difference in the perforation rate compared to TCRE/rollerball. Evidence suggests that REA and balloon NREA are little to no different than TCRE/rollerball in the perforation rate. We are uncertain of the true effect of hydrothermal ablation NREA and microwave NREA (very low certainty evidence; Table 23).
Evidence suggests bipolar NREA lowers the rate of requirement of further surgery for HMB. Balloon NREA probably lowers the rate of requirement of further surgery. Evidence suggests that other REA, microwave NREA and hydrothermal NREA produce little to no difference in the rate of requirement of further surgery compared to TCRE. We are uncertain of the true effect of other REA (Table 24).
Evidence suggests little to no difference in the requirement of further hysterectomy for treating HMB between bipolar NREA, other REA, balloon NREA and TCRE/rollerball; and probably little to no difference between microwave NREA and TCRE/rollerball. We are uncertain of the true effect of hydrothermal NREA (very low certainty evidence; Table 25).
Overall completeness and applicability of evidence
We updated all the reviews whose last search was over 18 months old when we started the overview. We summarised all the data available in the Cochrane Reviews, extracting data by studies. We consider it to be complete, although we acknowledge that our searches may have missed relevant studies, and the evidence had limitations.
This overview provides three sets of network meta‐analyses reporting on the relative effectiveness of first and second‐line HMB treatments, plus a subgroup analysis by type of endometrial ablation technique, in a coherent and methodologically robust way across clinical outcomes, combining direct and indirect evidence to increase the statistical power and confidence in the results. Most of the trials included in the reviews reported the bleeding outcome, but they did so in a wide variety of ways, limiting our ability to combine them. The network meta‐analysis results were mostly consistent, and where there was significant inconsistency, this was likely to be due to unstable estimates from single studies.
We excluded trials where participants' baseline characteristics did not match with the overview, that is, studies that included women with HMB due to an intrauterine device, ovulation disorders, or in treatment with anticoagulant drugs. All trials were carried out in clinical settings.
Most trials compared first‐line treatments to placebo, enabling us to set placebo as the reference treatment in the networks. Trials did not compare second‐line treatments to placebo or no treatment. We decided to use LNG‐IUS as the reference treatment for second‐line treatments and TCRE/rollerball as the reference treatment for the subgroup analysis of endometrial ablation techniques.
Unfortunately, we were unable to draw conclusions regarding quality of life, as network meta‐analysis was not possible.
Another limitation was that we did not have enough studies to use a funnel plot to explore the possibility of small‐study effects (a tendency for estimates of the intervention effect to be more beneficial in smaller studies).
We consider the information provided by this overview to be valuable for women with HMB, healthcare providers, and policymakers.
Methodologies for both overviews and NMAs have been developed since the publication of the protocol in 2018. It may be appropriate for the update to consider evolving the overview to a review of interventions, which would facilitate the process by allowing a single search with all the NMA details, instead of updating each review.
Quality of the evidence
We recognise that there is no single established approach for assessing the certainty of the effect estimates generated by the network meta‐analysis. We applied the GRADE Working Group method for appraising quality of network evidence. Overall, the evidence presented was low, and our confidence in the effect estimates ranged from very low to low. In general, due to the lack of blinding in most of the trials, and some indirectness arising from small studies, most of the comparisons started at moderate certainty. Either because of wide intervals or inconsistency on the network, or both, we then downgraded them to low or very low certainty.
Potential biases in the overview process
Two overview authors (AL and CF) were involved in one study contributing data to the overview ((D) Talis 2006). They did not participate in data extraction or quality assessment for their study. Several authors have been involved on the reviews providing data for the overview, but they did not participate in the assessment of the quality of the reviews. One review contributing studies to the overview was declared stable in 2007, when the only risk of bias assessed was allocation concealment. The earliest trial included was from 1970, and since then the standard of care and even the HMB diagnosis has changed.
Agreements and disagreements with other studies or reviews
In the first‐line treatments network, the LNG‐IUS appears to be the best for the outcome of bleeding reduction compared to the remaining interventions. This finding is consistent with the evidence summary reported by NICE 2018b which describes the LNG‐IUS as effective or more effective than other treatments for HMB in terms of improving health‐related quality of life, treatment satisfaction, discontinuation rates and blood loss. In addition, it is widely used in clinical practice and routinely available within primary care, and it should be the first option for treating HMB in women with no identified pathology.
NICE 2018b did not report on adverse events, but instead on discontinuation of treatment due to adverse events, with NSAIDs and tranexamic acid ranking first (1 to 3) and second (1 to 5), and COC ranking 7 (6 to 7, in a ranking from 1 to 7); danazol was not included in the study. This difference maybe due to the first‐line treatments network of this overview using any adverse event as an outcome, which may not reflect the presence of serious adverse events.
With respect to second‐line treatments for blood loss, NICE 2018b agreed that for women who do not wish to have pharmacological treatment and who do not want to conserve their fertility, surgical options could be considered as a first‐line treatment option, which is similar to the second‐line treatment results reporting hysterectomy as the best treatment for menstrual blood reduction and avoiding further surgery. The NICE report used a different division of the endometrial ablation procedures, which made it difficult to compare. In terms of blood loss, they reported that NREA was more effective and had higher satisfaction than REA techniques. The second‐line treatment network of this overview reported hysterectomy and NREA as being the two best options for menstrual blood reduction, yielding higher rates of satisfaction and avoiding future surgery for HMB.
Daniels 2012 published a network meta‐analysis of EA techniques, including bipolar NREA, microwave NREA, thermal balloon NREA, hydrothermal NREA, laser NREA, cryoablation NREA and REA. The authors assessed bleeding reduction and satisfaction, reporting that bipolar and microwave ablation devices are more effective than thermal balloon and free fluid ablation for treating HMB with NREA devices, which is consistent with the numerical finding of the subgroup analysis for endometrial ablation techniques.
Authors' conclusions
Implications for practice.
This overview provides up‐to‐date evidence on HMB treatments. HMB is a condition that affects a significant proportion of women of reproductive age. There is moderate to very low certainty evidence that treatments for HMB are effective and improve satisfaction. It is important to highlight that the best treatment may not be the same for all women. Clinicians should assess the patient characteristics before applying the result of this NMA, starting with the first‐line treatments network, and then, in case of treatment failure or in women who have completed their families, moving to the second‐line treatments network.
In women who are not candidates for surgery, LNG‐IUS is the most effective first‐line treatment to reduce menstrual blood loss. LNG‐IUS is probably little to no different than placebo in the rate of adverse events, and the requirement for further treatment may be less likely than in women receiving COC. Antifibrinolytics are probably better than placebo at reducing blood loss, with little to no difference in the rate of adverse events. Long‐cycle progestogens may be better than placebo at reducing blood loss. Unfortunately, we are unable to draw conclusions about perceived improvement and satisfaction due to very low certainty evidence.
In women who are candidates for surgery, hysterectomy is the most effective second‐line treatment for reducing menstrual blood loss and avoiding further surgery for HMB. Minimally invasive hysterectomy may be better than the LNG‐IUS for achieving satisfaction. REA and NREA may be better than LNG‐IUS at reducing blood loss. NREA may have higher satisfaction rates than LNG‐IUS and may have lower rates of further surgery for HMB. Unfortunately, we were unable to pool the wide variety of adverse events reported, and most were specific to one technique.
For endometrial ablation or resection, the certainty of the evidence for bleeding outcomes was mostly very low. Evidence suggests microwave NREA is better than TCRE for reducing blood loss, but it may not be very different in the satisfaction rate or the rate of further surgery. Evidence suggests that other REA produces a higher amenorrhoea rate than TCRE/rollerball, with little to no difference in the rate of satisfaction, perforation rate and requirement for further surgery than TCRE, with or without rollerball. Evidence suggests that bipolar NREA is better than TCRE/rollerball at achieving satisfaction, with little to no difference in perforation rate and a lower requirement of further surgery. Evidence suggests that balloon NREA produces little to no difference in the satisfaction and perforation rate compared to TCRE/rollerball. Balloon NREA probably has a lower requirement of further surgery than TCRE/rollerball. Evidence suggests that other NREA produces similar satisfaction and perforation rates compared to TCRE, with or without rollerball. Other factors such as the ability to use local (as opposed to general anaesthesia), surgical time, and other adverse events (not considered in this network meta‐analysis) may influence the decision by clinician and woman regarding their choice of endometrial ablation options for the treatment of heavy menstrual bleeding.
Implications for research.
LNG‐IUS is the most effective first‐line treatment to reduce menstrual blood loss in women requiring contraception, but there is no clear ranking for the remaining options. There is an overlap between contraception and heavy menstrual bleeding treatment, as hormonal contraceptives reduce menstrual blood loss, and in some cases cause amenorrhoea; progestin‐only contraceptives (pill, injections and implants) have been used off‐label as HMB treatment. Future research should assess the efficacy and safety of progestogen‐only contraceptives and compare different preparations of combined hormonal contraceptives for HMB treatment.
Quality of life should be the primary outcome for future research, as the current HMB definition is based on quality of life.
We had difficulties trying to combine data from different studies, as quality of life was reported in a wide range of ways. This highlights the importance of developing a core outcome set for heavy menstrual bleeding trials that would facilitate more effective analysis, avoid (or restrict) the use of imputed data, and reduce data waste. The COMET initiative (Core Outcome Measures in Effectiveness Trial) has an ongoing project, 'Defining core outcomes for clinical trials of heavy menstrual bleeding. A Core Outcome sets for Gynaecological conditions (COGS) project' that aims to develop a core outcome set covering all the aspects of heavy menstrual bleeding (COMET 2018).
A core outcomes set should also improve the ability to combine quality of life data. There should be specified safety outcomes, encouraging reporting of any adverse event, as it makes it simpler to compare, avoids the risk of double counting and allows fair comparisons. Comparing adverse events between interventions if they are specific to the intervention (as perforation for endometrial ablation and wound infection for hysterectomy) could lead to false conclusions.
Future research should assess interventions that have been used off‐label to treat HMB, such as progestin‐only contraceptives, and focus on quality of life improvement.
What's new
| Date | Event | Description |
|---|---|---|
| 9 February 2023 | Amended | Correction of links in Table 2 |
History
Protocol first published: Issue 11, 2018 Review first published: Issue 5, 2022
| Date | Event | Description |
|---|---|---|
| 21 June 2022 | Amended | Corrections to caption of figure 1 and to Table 5 |
Acknowledgements
The authors acknowledge the contribution of Helen Nagels (Managing Editor) at the Cochrane Gynaecology and Fertility Group's editorial base for her time and support. They thank Abha Maheshwari, Selma Mourad, Kirsten Duckitt and Rui Wang for providing peer review comments. They also thank Malcolm Munro for enlightening them about the history of endometrial ablation and Juan Jose Yepes‐Nunez for responding to questions about the NMA summary of findings.
Appendices
Appendix 1. Search strategy for progestin‐only contraceptives for HMB
Embase
1 exp Menorrhagia/ (9472)
2 hypermenorrhea.tw. (306)
3 Menorrhagia.tw. (5205)
4 heavy menstru$.tw. (1761)
5 heavy period$.tw. (175)
6 dysfunctional uter* bleeding.tw. (1070)
7 or/1‐6 (11760)
8 progestogen implant*.tw. (33)
9 (contracepti$ adj2 implant$).tw. (1384)
10 (Implanon$ or Jadelle$).tw. (1014)
11 Levonorgestrel implant$.tw. (173)
12 etonorgestrel$.tw. (50)
13 exp etonogestrel/ (2534)
14 or/8−13 (3706)
15 7 and 14 (174)
16 Clinical Trial/ (956560)
17 Randomized Controlled Trial/ (589457)
18 exp randomization/ (86255)
19 Single Blind Procedure/ (38135)
20 Double Blind Procedure/ (167151)
21 Crossover Procedure/ (62281)
22 Placebo/ (333233)
23 Randomi?ed controlled trial$.tw. (222624)
24 Rct.tw. (35951)
25 random allocation.tw. (1979)
26 randomly.tw. (431237)
27 randomly allocated.tw. (34337)
28 allocated randomly.tw. (2504)
29 (allocated adj2 random).tw. (810)
30 Single blind$.tw. (24161)
31 Double blind$.tw. (199587)
32 ((treble or triple) adj blind$).tw. (1105)
33 placebo$.tw. (297889)
34 prospective study/ (584539)
35 or/16‐34 (2384555)
36 case study/ (67249)
37 case report.tw. (393966)
38 abstract report/ or letter/ (1084454)
39 or/36‐38 (1535481)
40 35 not 39 (2331210)
41 (exp animal/ or animal.hw. or nonhuman/) not (exp human/ or human cell/ or (human or humans).ti.) (5924391)
42 40 not 41 (2169223)
43 15 and 42 (58)
Database: Ovid MEDLINE(R) Epub Ahead of Print, In Process & Other Non‐Indexed Citations, Ovid MEDLINE (R) Daily, and Ovid MEDLINE (R) 1946‐Present
Search Strategy:
1 exp Menorrhagia/ (4216)
2 hypermenorrhea.tw. (245)
3 Menorrhagia.tw. (3216)
4 heavy menstru$.tw. (956)
5 heavy period$.tw. (103)
6 dysfunctional uter* bleeding.tw. (838)
7 or/1‐6 (6952)
8 progestogen implant*.tw. (19)
9 (contracepti$ adj2 implant$).tw. (1107)
10 (Implanon$ or Jadelle$).tw. (304)
11 Levonorgestrel implant$.tw. (156)
12 etonorgestrel$.tw. (21)
13 8 or 9 or 10 or 11 or 12 (1404)
14 7 and 13 (31)
15 randomized controlled trial.pt. (501215)
16 controlled clinical trial.pt. (93562)
17 randomized.ab. (471788)
18 randomised.ab. (94356)
19 placebo.tw. (211267)
20 clinical trials as topic.sh. (190299)
21 randomly.ab. (328458)
22 trial.ti. (214342)
23 (crossover or cross‐over or cross over).tw. (83681)
24 or/15‐23 (1337759)
25 exp animals/ not humans.sh. (4675662)
26 24 not 25 (1231078)
27 14 and 26 (4)
Database: Embase
Search Strategy:
1 exp Menorrhagia/ (9472)
2 hypermenorrhea.tw. (306)
3 Menorrhagia.tw. (5205)
4 heavy menstru$.tw. (1761)
5 heavy period$.tw. (175)
6 dysfunctional uter$ bleeding.tw. (1070)
7 or/1‐6 (11760)
8 exp Medroxyprogesterone Acetate/ (16251)
9 medroxyprogesterone$.tw. (6889)
10 depoprovera$.tw. (309)
11 depo provera$.tw. (1808)
12 8 or 9 or 10 or 11 (17867)
13 7 and 12 (530)
14 Clinical Trial/ (956560)
15 Randomized Controlled Trial/ (589457)
16 exp randomization/ (86255)
17 Single Blind Procedure/ (38135)
18 Double Blind Procedure/ (167151)
19 Crossover Procedure/ (62281)
20 Placebo/ (333233)
21 Randomi?ed controlled trial$.tw. (222624)
22 Rct.tw. (35951)
23 random allocation.tw. (1979)
24 randomly.tw. (431237)
25 randomly allocated.tw. (34337)
26 allocated randomly.tw. (2504)
27 (allocated adj2 random).tw. (810)
28 Single blind$.tw. (24161)
29 Double blind$.tw. (199587)
30 ((treble or triple) adj blind$).tw. (1105)
31 placebo$.tw. (297889)
32 prospective study/ (584539)
33 or/14‐32 (2384555)
34 case study/ (67249)
Database: Ovid MEDLINE(R) Epub Ahead of Print, In Process & Other Non‐Indexed Citations, Ovid MEDLINE (R) Daily, and Ovid MEDLINE (R) 1946‐Present
Search Strategy:
1 exp Menorrhagia/ (4216)
2 hypermenorrhea.tw. (245)
3 Menorrhagia.tw. (3216)
4 heavy menstru$.tw. (956)
5 heavy period$.tw. (103)
6 dysfunctional uter$ bleeding.tw. (838)
7 or/1‐6 (6952)
8 exp Medroxyprogesterone Acetate/ (4840)
9 medroxyprogesterone$.tw. (6154)
10 depoprovera$.tw. (25)
11 depo provera$.tw. (781)
12 8 or 9 or 10 or 11 (8049)
13 7 and 12 (102)
14 randomized controlled trial.pt. (501215)
15 controlled clinical trial.pt. (93562)
16 randomized.ab. (471788)
17 randomised.ab. (94356)
18 placebo.tw. (211267)
19 clinical trials as topic.sh. (190299)
20 randomly.ab. (328458)
21 trial.ti. (214342)
22 (crossover or cross‐over or cross over).tw. (83681)
23 or/14‐22 (1337759)
24 exp animals/ not humans.sh. (4675662)
25 23 not 24 (1231078)
26 13 and 25 (24)
Appendix 2. AMSTAR 2
| 1. Did the research questions and inclusion criteria for the review include the components of PICO? YES: PICO Time frame No |
| 2. Did the report of the review contain an explicit statement that the review methods were established prior to the conduct of the review and did the report justify any significant deviations from the protocol? Partial yes (ALL):
Yes (partial + all)
No |
| 3. Did the review authors explain their selection of the study designs for inclusion in the review? Yes (ONE)
No |
| 4. Did the review authors use a comprehensive literature search strategy? Partial yes (ALL)
YES (partial + ALL)
No |
| 5. Did the review authors perform study selection in duplicate? Yes (ONE)
No |
| 6. Did the review authors perform data extraction in duplicate? Yes (ONE)
No |
| 7. Did the review authors provide a list of excluded studies and justify the exclusions? Partial yes
Yes (also)
No |
| 8. Did the review authors describe the included studies in adequate detail? For partial yes (ALL the following)
For yes, should also have ALL the following
No |
| 9. Did the review authors use a satisfactory technique for assessing the risk of bias in individual studies that were included in the review? RCT For partial yes, must have assessed risk of bias from:
For yes, must also have assessed risk of bias from:
No |
| 10. Did the review authors report on the sources of funding for the studies included in the review? For yes
No |
| 11. If meta‐analysis was performed did the review authors use appropriate methods for statistical combination of results? RCT For yes
No No meta analysis conducted |
| 12. If meta‐analysis was performed, did the review authors assess the potential impact of risk of bias in individual studies on the results of the meta‐analysis or other evidence synthesis? For yes:
No No meta analysis conducted |
| 13. Did the review authors account for RoB in individual studies when interpreting/ discussing the results of the review? For yes
No |
| 14. Did the review authors provide a satisfactory explanation for, and discussion of, any heterogeneity observed in the results of the review? For yes
No |
| 15. If they performed quantitative synthesis did the review authors carry out an adequate investigation of publication bias (small study bias) and discuss its likely impact on the results of the review? For yes
No No meta analysis conducted |
| 16. Did the review authors report any potential sources of conflict of interest, including any funding they received for conducting the review? For yes
No |
Appendix 3. Node‐split results. First‐line. Menstrual blood loss. (Combined)
| COMBINED MEAN BLOOD LOSS (ENDPOINT AND CHANGE FROM BASELINE) | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Mean Diff | Std. Err. | Mean Diff | Std. Err. | Mean Diff | Std. Err. | |||
| 1 | 6 | Placebo | NSAIDs | −49.9811 | 25.89066 | −11.2014 | 46.26587 | −38.7798 | 53.00106 | 0.464 |
| 1 | 7 | Placebo | Antifibrinolytics | −68.6471 | 29.19427 | −107.379 | 44.2173 | 38.73227 | 53.00226 | 0.465 |
| 2 | 3 | CVR | COC | 27.3509 | 41.09786 | 20.18564 | 72.18571 | 7.165267 | 83.14633 | 0.931 |
| 2 | 9 | CVR | Long‐cycle Pg | 2.100004 | 53.35153 | 9.22418 | 63.76601 | −7.12418 | 83.14139 | 0.932 |
| 3 | 4 | COC | Danazol | −19.1684 | 57.07587 | −65.7967 | 74.38362 | 46.62828 | 93.55745 | 0.618 |
| 3 | 6 | COC | NSAIDs | 15.70302 | 57.24792 | 15.7019 | 61.29866 | 0.001122 | 83.53626 | 1 |
| 3 | 10 | COC | LNG‐IUS | −52.9658 | 40.29542 | −36.1379 | 71.8173 | −16.8278 | 82.3826 | 0.838 |
| 4 | 6 | Danazol | NSAIDs | 48.92792 | 39.86311 | 60.84555 | 65.92257 | −11.9176 | 77.01708 | 0.877 |
| 4 | 8 | Danazol | Luteal Pg | 86.4587 | 57.65394 | 63.49177 | 53.01963 | 22.96694 | 78.22073 | 0.769 |
| 5 | 7 | Ethamsylate | Antifibrinolytics | −101.31 | 56.73824 | −76.3184 | 64.7377 | −24.9918 | 85.50179 | 0.77 |
| 6 | 7 | NSAIDs | Antifibrinolytics | −72.6287 | 57.95434 | −30.647 | 30.11349 | −41.9817 | 65.58373 | 0.522 |
| 6 | 8 | NSAIDs | Luteal Pg | 7.499999 | 54.89114 | 30.44654 | 42.10213 | −22.9465 | 69.17822 | 0.74 |
| 7 | 8 | Antifibrinolytics | Luteal Pg | 63.47069 | 42.17429 | 58.9224 | 51.59067 | 4.548289 | 66.5435 | 0.946 |
| 7 | 9 | Antifibrinolytics | Long‐cycle Pg | 8.709065 | 38.78747 | −16.4963 | 75.18427 | 25.20541 | 84.59993 | 0.766 |
| 9 | 10 | Long‐cycle Pg | LNG‐IUS | −21.42 | 52.94217 | −38.2336 | 63.12108 | 16.81357 | 82.38413 | 0.838 |
Appendix 4. Node‐split results. First‐line. Menstrual blood loss (sensitivity analysis)
| MEAN BLOOD LOSS AT ENDPOINT | ||||||||||
| Node‐split results ‐ random‐effects model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Mean Diff | Std. Err. | Mean Diff | Std. Err. | Mean Diff | Std. Err. | |||
| 1 | 6 | Placebo | NSAIDs | −48.6964 | 18.49676 | −28.2365 | 46.34697 | −20.4599 | 49.89525 | 0.682 |
| 1 | 7 | Placebo | Antifibrinolytics | −95.736 | 38.58291 | −116.152 | 31.66106 | 20.41588 | 49.90111 | 0.682 |
| 2 | 3 | CVR | COC | 19.54857 | 25.13099 | 93.31438 | 43.19615 | −73.7658 | 50.53359 | 0.144 |
| 2 | 9 | CVR | Long‐cycle Pg | 2.100004 | 29.45055 | −71.6779 | 41.06098 | 73.77786 | 50.53056 | 0.144 |
| 3 | 4 | COC | Danazol | −19.5755 | 39.37223 | −93.9778 | 51.22275 | 74.40232 | 64.30718 | 0.247 |
| 3 | 6 | COC | NSAIDs | 14.71929 | 40.18335 | −11.4579 | 43.8228 | 26.17721 | 58.97513 | 0.657 |
| 3 | 10 | COC | LNG‐IUS | −173.6 | 18.32955 | −33.6154 | 25.0968 | −139.985 | 31.07766 | 0 |
| 4 | 6 | Danazol | NSAIDs | 48.91868 | 27.65232 | 53.23686 | 48.42096 | −4.31818 | 55.73304 | 0.938 |
| 4 | 8 | Danazol | Luteal Pg | 73.75836 | 42.16143 | 43.6648 | 37.20711 | 30.09356 | 56.12898 | 0.592 |
| 5 | 7 | Ethamsylate | Antifibrinolytics | −101.358 | 39.86295 | −111.91 | 51.16417 | 10.5518 | 64.18923 | 0.869 |
| 6 | 7 | NSAIDs | Antifibrinolytics | −71.3622 | 42.25772 | −58.3115 | 26.39803 | −13.0507 | 50.32471 | 0.795 |
| 6 | 8 | NSAIDs | Luteal Pg | 7.499996 | 37.07959 | 6.69817 | 32.13147 | 0.801826 | 49.06452 | 0.987 |
| 7 | 8 | Antifibrinolytics | Luteal Pg | 59.54995 | 30.5242 | 83.67885 | 38.17113 | −24.1289 | 48.82551 | 0.621 |
| 7 | 9 | Antifibrinolytics | Long‐cycle Pg | 8.720853 | 25.55254 | −50.6457 | 53.22126 | 59.36651 | 59.03729 | 0.315 |
| 9 | 10 | Long‐cycle Pg | LNG‐IUS | −21.42 | 16.07161 | −161.402 | 26.59992 | 139.9818 | 31.07818 | 0 |
Appendix 5. Node‐split results. First‐line. Perception of bleeding improvement
| PERCEPTION OF BLEEDING IMPROVEMENT | ||||||||||
| Node‐split results ‐ random‐effects model | ||||||||||
| Side | Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | ||||
| X | Y | Ln‐odds ratio | Std. Err. | Ln‐odds ratio | Std. Err. | Ln‐odds ratio | Std. Err. | |||
| 1 | 6 | Placebo | NSAIDs | 3.120895 | 1.077231 | 0.285072 | 1.324783 | 2.835824 | 1.707476 | 0.097 |
| 1 | 7 | Placebo | Antifibrinolytics | 1.150863 | 1.143165 | 3.986551 | 1.268326 | −2.83569 | 1.707476 | 0.097 |
| 4 | 6 | Danazol | NSAIDs | −0.18232 | 1.431655 | −0.94035 | 1.303436 | 0.758024 | 1.936125 | 0.695 |
| 4 | 8 | Danazol | Luteal Pg | −1.59003 | 0.99435 | −0.83198 | 1.661421 | −0.75805 | 1.93612 | 0.695 |
| 5a | 6 | Ethamsylate | NSAIDs | 0.664976 | 1.263524 | 0.484471 | 2.820899 | 0.180505 | 3.101458 | 0.954 |
| 5a | 7 | Ethamsylate | Antifibrinolytics | 1.034074 | 1.262739 | 1.214601 | 2.821953 | −0.18053 | 3.101458 | 0.954 |
| 6a | 7 | NSAIDs | Antifibrinolytics | 1.336838 | 0.715685 | −1.25733 | 0.987732 | 2.59417 | 1.241331 | 0.037 |
| 6 | 8 | NSAIDs | Luteal Pg | −0.37067 | 1.17565 | −1.10953 | 1.012907 | 0.738863 | 1.551687 | 0.634 |
| 7 | 8 | Antifibrinolytics | Luteal Pg | −1.19803 | 0.630909 | −1.43008 | 1.744104 | 0.23205 | 1.859741 | 0.901 |
| 7 | 9 | Antifibrinolytics | Long‐cycle Pg | −1.82924 | 1.147346 | 1.092652 | 1.414133 | −2.9219 | 1.821037 | 0.109 |
| 7 | 10 | Antifibrinolytics | LNG‐IUS | 2.281122 | 1.111473 | −1.10914 | 1.139989 | 3.390259 | 1.556894 | 0.029 |
| 8 | 10 | Luteal Pg | LNG‐IUS | 2.025794 | 1.358828 | 1.593935 | 1.59729 | 0.431859 | 2.073284 | 0.835 |
| 9 | 10 | Long‐cycle Pg | LNG‐IUS | 0.58783 | 0.895417 | 3.509751 | 1.586125 | −2.92192 | 1.821043 | 0.109 |
Footnote
aThe evidence for these comparisons comes from trials that directly compare them.
Appendix 6. Node‐split results. First‐line. Any adverse event
| ANY ADVERSE EFFECT | ||||||||||
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Ln‐odds ratio | Std. Err. | Ln‐odds ratio | Std. Err. | Ln‐odds ratio | Std. Err. | |||
| 1 | 2 | Placebo | COC | 0.772 | 0.226639 | 1.326314 | 1.19228 | −0.55431 | 1.21363 | 0.648 |
| 1 | 5 | Placebo | Antifibrinolytics | 0.308553 | 0.384999 | −0.24577 | 1.150933 | 0.554326 | 1.213618 | 0.648 |
| 2 | 8 | COC | LNG‐IUS | −0.40547 | 0.975182 | 0.14885 | 0.722438 | −0.55432 | 1.21363 | 0.648 |
| 4 | 5 | NSAIDs | Antifibrinolytics | −0.01835 | 1.42713 | 0.172479 | 0.833552 | −0.19083 | 1.652728 | 0.908 |
| 4 | 6 | NSAIDs | Luteal Pg | 0.654926 | 0.764046 | 0.4641 | 1.465517 | 0.190826 | 1.652727 | 0.908 |
| 5 | 6 | Antifibrinolytics | Luteal Pg | 0.503178 | 0.335882 | 0.276984 | 1.343592 | 0.226195 | 1.382645 | 0.87 |
| 5 | 8 | Antifibrinolytics | LNG‐IUS | 0.167295 | 0.619065 | 1.095312 | 0.842583 | −0.92802 | 1.039433 | 0.372 |
| 6 | 8 | Luteal Pg | LNG‐IUS | 0.609616 | 0.647781 | −1.27515 | 0.942485 | 1.884769 | 1.150687 | 0.101 |
Appendix 7. Node‐split results. Second‐line. Bleeding improvement (no imputed data)
| PBAC IMPROVEMENT (NO IMPUTED VALUES) | ||||||||||
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 3 | 4 | NREA | LNG‐IUS | −2.35608 | 0.892178 | −0.9155 | 0.443133 | −1.44058 | 0.996166 | 0.148 |
| 2 | 3 | REA | NREA | 0.173453 | 0.152816 | 1.614023 | 0.984374 | −1.44057 | 0.996165 | 0.148 |
| 2 | 4 | REA | LNG‐IUS | −0.74204 | 0.41595 | −2.18262 | 0.90517 | 1.440576 | 0.996166 | 0.148 |
Appendix 8. Imputation details. Second‐line. Pictorial blood assessment chart (PBAC)
INTRODUCTION
The outcome of interest is the probability of bleeding improvement measured by PBAC being under 75 points after treatment. The PBAC < 75 threshold was chosen by review authors as it was the limited used in most trials to define bleeding improvement; some trials also used it as similar to acceptable bleeding improvement (reporting it together as one or the other). We decided to use the same outcome.
However, six studies ((D) Athanatos 2015; (D) Hawe 2003; (D) Meyer 1998; (D) Sesti 2011; (D) Sesti 2012; (D) Vercellini 1999) only reported the mean PBAC score at one year for individuals in each arm (with its standard deviation). In order to include as many relevant studies as possible, we imputed individuals with PBAC improvement for theses studies based on the mathematical relationship between the mean score and the probability of exceeding a given threshold.
METHODS
To transform the data we assumed that n ik individuals are randomised to each arm k of study i, where for individual j =1,…, n ik in arm k of trial i, represents the PBAC score at follow‐up. Let represent the PBAC improvement status at follow‐up for individual j in arm k of trial i, defined as having a follow‐up PBAC score below pre‐defined threshold h, i.e.
Equation 1: Figure 72.
70.
Formula 1 second‐line treatments imputation.
Using equation 1 and assuming all individuals have the same underlying distribution of PBAC scores represented by random variable with representing the mean and the variance of the PBAC score, we can write the probability of PBAC improvement for individuals in arm k of trial i as
Equation 2: Figure 73.
71.
Formula 2 second‐line treatments imputation
Therefore, studies providing information on the mean PBAC at 1 year, also provide information on the probability of a PBAC improvement, defined as having a PBAC score lower than the selected threshold h. We estimated the number of individuals with PBAC improvement for arm k of trial i as
Equation 3: Figure 74.
72.
Formula 3 second‐line treatments imputation.
Where the observed mean and standard deviation of PBAC score at follow‐up in arm k of trial i and represents the ‘floor’ function, meaning values are rounded down to the nearest integer.
We transformed the mean PBAC measurement at follow‐up into the number of people with PBAC improvement, defined as PBAC score under 75 points, using equation where h =75 and the other parameters are as reported in the trial.
ASSUMPTIONS AND LIMITATIONS
We assumed that our methods for converting continuous PBAC means would give reliable estimates of the number of patients with PBAC improvement. These methods are based on a mathematical relationship with the assumption of normality of the underlying continuous data (PBAC score). However, it may be that the distribution of PBAC scores is skewed and the normality assumption may not be reasonable. Given the available data, it was not possible to empirically check this assumption.
Appendix 9. Node‐split results. Second‐line. Bleeding improvement (with imputed data)
|
Bleeding improvement (PBAC) with imputed data Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 3 | 4 | NREA | LNG‐IUS | −2.35608 | 0.892178 | −0.76842 | 0.41947 | −1.58766 | 0.985868 | 0.107 |
| 1 | 3 | Hysterectomy (*) | NREA | −1.95606 | 1.110263 | −1.22445 | 1.156875 | −0.73161 | 1.603447 | 0.648 |
| 1 | 4 | Hysterectomy (*) | LNG‐IUS | −2.32389 | 1.088635 | −3.0555 | 1.17725 | 0.731611 | 1.603447 | 0.648 |
| 2 | 3 | REA | NREA | 0.057813 | 0.129425 | 1.121585 | 0.878548 | −1.06377 | 0.88803 | 0.231 |
| 2 | 4 | REA | LNG‐IUS | −0.74204 | 0.41595 | −1.80582 | 0.784592 | 1.063776 | 0.888031 | 0.231 |
Appendix 10. Node‐split results. Second‐line. Amenorrhoea
| AMENORRHOEA | ||||||||||
| Node‐split results ‐ random‐effects model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 2 | 3 | REA | NREA | 0.059995 | 0.22538 | −1.05225 | 1.012026 | 1.112248 | 1.037476 | 0.284 |
| 2 | 4 | REA | LNG‐IUS | −0.42946 | 0.629722 | 0.682783 | 0.824839 | −1.11225 | 1.037475 | 0.284 |
| 3 | 4 | NREA | LNG‐IUS | 0.622791 | 0.792538 | −0.48946 | 0.668774 | 1.11225 | 1.037477 | 0.284 |
Appendix 11. Node‐split results. Second‐line. Satisfaction
| SATISFACTION | ||||||||||
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 6 | 7 | NREA | LNG‐IUS | −0.47151 | 0.212347 | −0.45806 | 0.45638 | −0.01345 | 0.503363 | 0.979 |
| 1 | 5 | Hysterectomy (any/unspecified) | REA | −1.02962 | 0.781228 | 0.471634 | 0.584988 | −1.50125 | 0.975976 | 0.124 |
| 1 | 7 | Hysterectomy (any/unspecified) | LNG‐IUS | 0.161183 | 0.535064 | −1.34007 | 0.816231 | 1.501253 | 0.975975 | 0.124 |
| 5 | 6 | REA | NREA | 0.245228 | 0.171773 | 0.258675 | 0.473147 | −0.01345 | 0.503363 | 0.979 |
| 5 | 7 | REA | LNG‐IUS | −0.56241 | 0.472548 | −0.11744 | 0.262426 | −0.44497 | 0.540527 | 0.410 |
Appendix 12. Node‐split results. Second‐line. Any adverse event
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 5 | 6 | REA | NREA | 1.021979 | 1.162058 | −0.3586 | 0.600226 | 1.38058 | 1.307918 | 0.291 |
| 5 | 7 | REA | LNG‐IUS | 1.060618 | 0.370122 | 2.548414 | 1.092378 | −1.4878 | 1.153378 | 0.197 |
| 3 | 5 | Minimally invasive hysterectomy | REA | −1.051545 | 1.647347 | −0.05968 | 0.686901 | −0.99186 | 1.784821 | 0.578 |
| 3 | 6 | Minimally invasive hysterectomy | NREA | −0.2258164 | 0.395163 | −1.21768 | 1.740526 | 0.991862 | 1.78482 | 0.578 |
| 6 | 7 | NREA | LNG‐IUS | 1.589235 | 0.524404 | 0.101436 | 1.02727 | 1.487799 | 1.153379 | 0.197 |
Appendix 13. Node‐split results. Second‐line. Any further surgery
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 3 | 4 | REA | NREA | −0.13281 | 0.186886 | −0.5079 | 0.596236 | 0.375082 | 0.624839 | 0.548 |
| 3 | 5 | REA | LNG‐IUS | 0.430223 | 0.451275 | 0.554857 | 0.465895 | −0.12463 | 0.648619 | 0.848 |
| 1 | 3 | Hysterectomy (*) | REA | 3.151015 | 0.658708 | 0.1451 | 2.022371 | 3.005914 | 2.126941 | 0.158 |
| 1 | 4 | Hysterectomy (*) | NREA | 0 | 2.014441 | 3.005928 | 0.682587 | −3.00593 | 2.126945 | 0.158 |
| 4 | 5 | NREA | LNG‐IUS | 0.710958 | 0.427085 | 0.586324 | 0.488165 | 0.124634 | 0.648619 | 0.848 |
Appendix 14. Node‐splits results. Second‐line. Endometrial ablation, network meta‐analysis. Bleeding (pictorial blood assessment chart) improvement
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 1 | 2 | TCRE/rollerball (REA) | Other REA | 1.774518 | 1.116862 | −0.2664 | 0.622647 | 2.040921 | 1.278698 | 0.11 |
| 1 | 3 | TCRE/rollerball (REA) | Microwave NREA | 0.2204712 | 0.224719 | 2.771021 | 0.709214 | −2.55055 | 0.743965 | 0.001 |
| 1 | 5 | TCRE/rollerball (REA) | Bipolar NREA | 0.5256407 | 0.380132 | −0.02212 | 0.424038 | 0.547757 | 0.569481 | 0.336 |
| 1 | 6 | TCRE/rollerball (REA) | Balloon NREA | −0.2484614 | 0.268901 | −0.83357 | 0.527121 | 0.585112 | 0.591746 | 0.323 |
| 2 | 6 | Other REA | Balloon NREA | −0.1772065 | 0.572839 | −2.21813 | 1.143208 | 2.04092 | 1.278698 | 0.11 |
| 3 | 5 | Microwave NREA | Bipolar NREA | −2.041626 | 0.637029 | 0.508924 | 0.384288 | −2.55055 | 0.743965 | 0.001 |
| 5 | 6 | Bipolar NREA | Balloon NREA | −1.315283 | 0.477079 | −0.12868 | 0.422944 | −1.18661 | 0.637562 | 0.063 |
Appendix 15. Node‐split results. Subgroup: endometrial ablation, network meta‐analysis. Amenorrhoea
| Node‐split results ‐ random‐effects model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 1 | 2 | TCRE/rollerball (REA) | Other REA | 0.3445016 | 0.6185282 | −0.0537665 | 1.142229 | 0.3982681 | 1.299354 | 0.759 |
| 1 | 3 | TCRE/rollerball (REA) | Microwave NREA | 0.1960558 | 0.6579212 | 1.249303 | 0.817412 | −1.053247 | 1.049358 | 0.316 |
| 1 | 4 | TCRE/rollerball (REA) | Hydrothermal ablation NREA | −0.4245752 | 0.9597677 | −0.3967262 | 1.100154 | −0.027849 | 1.459963 | 0.985 |
| 1 | 5 | TCRE/rollerball (REA) | Bipolar NREA | 0.5899178 | 0.688005 | 0.6471101 | 0.6022771 | −0.0571923 | 0.9142737 | 0.95 |
| 1 | 6 | TCRE/rollerball (REA) | Balloon NREA | −0.1236863 | 0.6413865 | −0.7183315 | 0.5852692 | 0.5946452 | 0.8701993 | 0.494 |
| 2 | 6 | Other REA | Balloon NREA | −0.444686 | 1.045237 | −0.8429552 | 0.7718783 | 0.3982692 | 1.299353 | 0.759 |
| 3 | 5 | Microwave NREA | Bipolar NREA | −2.278869 | 0.8679116 | 0.8963039 | 0.5187321 | −3.175172 | 1.011115 | 0.002 |
| 3 | 6 | Microwave NREA | Balloon NREA | −0.1555177 | 0.8962274 | −1.556801 | 0.6717775 | 1.401283 | 1.120048 | 0.211 |
| 4 | 5 | Hydrothermal ablation NREA | Bipolar NREA | 1.022239 | 0.9894131 | 1.050089 | 1.073567 | −0.0278493 | 1.45996 | 0.985 |
| 5 | 6 | Bipolar NREA | Balloon NREA | −1.690421 | 0.5038039 | −0.030568 | 0.6470731 | −1.659853 | 0.820491 | 0.043 |
Appendix 16. Node‐split results. Subgroup: endometrial ablation, network meta‐analysis. Satisfaction
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 1 | 2 | TCRE/rollerball (REA) | Other REA | −0.0429798 | 0.343345 | 0.687731 | 1.276226 | −0.73071 | 1.321604 | 0.58 |
| 1 | 3 | TCRE/rollerball (REA) | Microwave NREA | 0.0628475 | 0.2924 | −0.05468 | 0.389666 | 0.117524 | 0.487173 | 0.809 |
| 1 | 4 | TCRE/rollerball (REA) | Hydrothermal ablation NREA | 0.2876821 | 0.925235 | −1.43955 | 1.10975 | 1.727233 | 1.444855 | 0.232 |
| 1 | 5 | TCRE /rollerball (REA) | Bipolar NREA | 0.4961681 | 0.376293 | 1.271592 | 0.395433 | −0.77542 | 0.545861 | 0.155 |
| 1 | 6 | TCRE /rollerball (REA) | Balloon NREA | 0.2543976 | 0.350426 | −0.11181 | 0.304707 | 0.36621 | 0.464376 | 0.43 |
| 2 | 6 | Other REA | Balloon NREA | −0.6190392 | 1.254662 | 0.111672 | 0.415286 | −0.73071 | 1.321604 | 0.58 |
| 3 | 5 | Microwave NREA | Bipolar NREA | 2.559537 | 1.498918 | 0.766708 | 0.318482 | 1.792829 | 1.53238 | 0.242 |
| 3 | 6 | Microwave NREA | Balloon NREA | 0.0040057 | 0.282037 | 0.067346 | 0.395626 | −0.06334 | 0.485865 | 0.896 |
| 4 | 5 | Hydrothermal ablation NREA | Bipolar NREA | 2.240372 | 1.074424 | 0.513138 | 0.966033 | 1.727234 | 1.444855 | 0.232 |
| 5 | 6 | Bipolar NREA | Balloon NREA | −0.9305043 | 0.332309 | −0.62823 | 0.43405 | −0.30228 | 0.546652 | 0.58 |
Appendix 17. Node‐split results. Subgroup: endometrial ablation, network meta‐analysis. Requirement of further surgery for heavy menstrual bleeding
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 1 | 2 | TCRE /rollerball (REA) | Other REA | −0.2141179 | 0.267122 | −0.24197 | 0.801564 | 0.027855 | 0.844902 | 0.974 |
| 1 | 6 | TCRE /rollerball (REA) | Balloon NREA | −0.5341095 | 0.351366 | −0.50625 | 0.768375 | −0.02786 | 0.844901 | 0.974 |
| 2 | 6 | Other REA | Balloon NREA | −0.2921364 | 0.72045 | −0.31999 | 0.441375 | 0.027855 | 0.844902 | 0.974 |
Appendix 18. Node‐split results. Subgroup: endometrial ablation, network meta‐analysis. Requirement of further hysterectomy for heavy menstrual bleeding
| Node‐split results ‐ fixed‐effect model | ||||||||||
| Side | Comparison of Y vs X | Direct | Indirect | Difference | P>|z| | |||||
| X | Y | Coef. | Std. Err. | Coef. | Std. Err. | Coef. | Std. Err. | |||
| 1 | 2 | TCRE/rollerball (REA) | Microwave NREA | −0.1034404 | 0.3665263 | −0.0416137 | 0.625926 | −0.06183 | 0.725345 | 0.932 |
| 1 | 4 | TCRE/rollerball (REA) | Bipolar NREA | −1.139434 | 0.9296057 | −0.4212295 | 0.53929 | −0.7182 | 1.074709 | 0.504 |
| 1 | 5 | TCRE/rollerball (REA) | Balloon NREA | −0.07298 | 0.3106057 | −0.3852061 | 0.570816 | 0.312226 | 0.649852 | 0.631 |
| 2 | 4 | Microwave NREA | Bipolar NREA | −1.670996 | 1.568948 | −0.3711225 | 0.552263 | −1.29987 | 1.663307 | 0.435 |
| 2 | 5 | Microwave NREA | Balloon NREA | 0.065958 | 0.5899163 | −0.1301057 | 0.456586 | 0.196064 | 0.74597 | 0.793 |
| 4 | 5 | Bipolar NREA | Balloon NREA | 0.1934443 | 0.4920656 | 1.238682 | 0.845778 | −1.04524 | 0.978503 | 0.285 |
Appendix 19. GRADE wording
| Size of the effect estimate | Suggested statements (replace X with intervention, replace ‘reduce/increase’ with direction of effect, replace ‘outcome’ with name of outcome, include ‘when compared with Y’ when needed) |
| HIGH certainty of the evidence | |
| Large effect | X results in a large reduction/increase in outcome |
| Moderate effect | X reduces/increases outcome X results in a reduction/increase in outcome |
| Small important effect | X reduces/increases outcome slightly X results in a slight reduction/increase in outcome |
| Trivial, small unimportant effect or no effect | X results in little to no difference in outcome X does not reduce/increase outcome |
| MODERATE certainty of the evidence | |
| Large effect | X likely results in a large reduction/increase in outcome X probably results in a large reduction/increase in outcome |
| Moderate effect | X likely reduces/increases outcome X probably reduces/increases outcome X likely results in a reduction/increase in outcome X probably results in a reduction/increase in outcome |
| Small important effect | X probably reduces/increases outcome slightly X likely reduces/increases outcome slightly X probably results in a slight reduction/increase in outcome X likely results in a slight reduction/increase in outcome |
| Trivial, small unimportant effect or no effect | X likely results in little to no difference in outcome X probably results in little to no difference in outcome X likely does not reduce/increase outcome X probably does not reduce/increase outcome |
| LOW certainty of the evidence | |
| Large effect | X may result in a large reduction/increase in outcome The evidence suggests X results in a large reduction/increase in outcome |
| Moderate effect | X may reduce/increase outcome The evidence suggests X reduces/increases outcome X may result in a reduction/increase in outcome The evidence suggests X results in a reduction/increase in outcome |
| Small important effect | X may reduce/increase outcome slightly The evidence suggests X reduces/increases outcome slightly X may result in a slight reduction/increase in outcome The evidence suggests X results in a slight reduction/increase in outcome |
| Trivial, small unimportant effect or no effect | X may result in little to no difference in outcome The evidence suggests that X results in little to no difference in outcome X may not reduce/increase outcome The evidence suggests that X does not reduce/increase outcome |
| VERY LOW certainty of the evidence | |
| Any effect | The evidence is very uncertain about the effect of X on outcome X may reduce/increase/have little to no effect on outcome but the evidence is very uncertain |
Differences between protocol and review
In the protocol we stated that we intended to perform a network meta‐analysis of the primary outcomes and two of the secondary outcomes (quality of life and adverse events). In the overview we intended to perform network meta‐analysis for all three secondary outcomes to add completeness.
We planned to present subgroup analyses for medical and surgical interventions. As the LNG‐IUS was compared to medical and surgical interventions, using different participant characteristics, we decided to subgroup into first‐ and second‐line treatments according to those characteristics.
We added a provision to impute data where missing for primary outcomes and conduct a sensitivity analysis with and without the imputed data.
In the protocol we stated we were going to conduct Bayesian network meta‐analysis, but we used Stata, as recommended by one of the statistician authors (SD), because a more user‐friendly software programme became available between publishing the protocol and performing the overview.
We changed the original title of the overview to highlight the network meta‐analysis, adding 'Overview of Cochrane Reviews and network meta‐analysis'.
Contributions of authors
All the authors contributed either to the study design or the original protocol. Two overview authors searched for reviews (JB and MBR), extracted data from the studies on the review (MBR and VJ) and analysed the data (SD and MBR). Three overview authors performed the AMSTAR (two per review, to ensure review's authors were not involved on the quality assessment of their own reviews) (SL, MW, MBR). MBR wrote all the versions of the overview and the graphs. All review authors critically reviewed the manuscript and approved the final version.
Sources of support
Internal sources
-
None, Other
None
-
The Liggins Institute, The University of Auckland, New Zealand
Stipend for the first year of MBR's PhD
External sources
-
None, Other
None
Declarations of interest
Magdalena Bofill Rodriguez, Julie Brown, Michelle Wise, Cindy Farquhar, Vanessa Jordan and Anne Lethaby have been involved as authors or co‐authors of Cochrane Reviews that may be included in this overview of reviews. Magdalena Bofill, Michelle Wise and Vanessa Jordan became involved updating the already selected reviews.
Julie Brown has been employed as a medical writer; we are satisfied that her employer had no financial interest in any intervention of interest to this overview.
Sofia Dias, Sarah Lensen and Jack Wilkinson have no conflict of interest.
Cindy Farquhar, Vanessa Jordan and Anne Lethaby are editors of Cochrane Gynecology and Fertility. They took no part in the editorial process.
Edited (no change to conclusions)
References
References to included reviews
Beaumont 2007
- Beaumont H, Augood C, Duckitt K, Lethaby A. Danazol for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2007, Issue 3. Art. No: CD001017. [DOI: 10.1002/14651858.CD001017.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bofill Rodriguez 2019a
- Bofill Rodriguez M, Lethaby A, Grigore M, Brown J, Hickey M, Farquhar C. Endometrial resection and ablation techniques for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2019, Issue 1. Art. No: CD001501. [DOI: 10.1002/14651858.CD001501.pub5] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bofill Rodriguez 2019b
- Bofill Rodriguez M, Lethaby A, Low C, Cameron I. Cyclical progestogens for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2019, Issue 8. Art. No: CD001016. [DOI: 10.1002/14651858.CD001016.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bofill Rodriguez 2019c
- Bofill Rodriguez M, Lethaby A, Farquhar C. Non-steroidal anti-inflammatory drugs for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2019, Issue 9. Art. No: CD000400. [DOI: 10.1002/14651858.CD000400.pub4] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bofill Rodriguez 2020
- Bofill Rdriguez M, Lethaby A, Jordan V. Progestogen-releasing intrauterine systems for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2020, Issue 4. Art. No: CD002126. [DOI: 10.1002/14651858.CD002126.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bofill Rodriguez 2021
- Bofill Rodriguez M, Lethaby A, Fergusson R. Endometrial resection and ablation versus hysterectomy for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2021, Issue 2. Art. No: CD000329. [DOI: 10.1002/14651858.CD000329.pub4] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bryant‐Smith 2018
- Bryant‐Smith AC, Lethaby A, Farquhar C, Hickey M. Antifibrinolytics for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2018, Issue 4. Art. No: CD000249. [DOI: 10.1002/14651858.CD000249.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Lethaby 2019
- Lethaby A, Wise MR, Weterings MAJ, Bofill Rodriguez M, Brown J. Combined hormonal contraceptives for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2019, Issue 2. Art. No: CD000154. [DOI: 10.1002/14651858.CD000154.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Marjoribanks 2016
- Marjoribanks J, Lethaby A, Farquhar C. Surgery versus medical therapy for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2016, Issue 1. Art. No: CD003855. [DOI: 10.1002/14651858.CD003855.pub3] [DOI] [PubMed] [Google Scholar]
References to excluded reviews
Grimes 2006
- Grimes DA, Hubacher D, Lopez LM, Schulz KF. Non‐steroidal anti-inflammatory drugs for heavy bleeding or pain associated with intrauterine‐device use. Cochrane Database of Systematic Reviews 2006, Issue 4. Art. No: CD006034. [DOI: 10.1002/14651858.CD006034.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Gupta 2014
- Gupta JK, Sinha A, Lumsden MA, Hickey M. Uterine artery embolization for symptomatic uterine fibroids. Cochrane Database of Systematic Reviews 2014, Issue 12. Art. No: CD005073. [DOI: 10.1002/14651858.CD005073.pub4] [DOI] [PMC free article] [PubMed] [Google Scholar]
Hickey 2012
- Hickey M, Higham JM, Fraser I. Progestogens with or without oestrogen for irregular uterine bleeding associated with anovulation. Cochrane Database of Systematic Reviews 2012, Issue 9. Art. No: CD001895. [DOI: 10.1002/14651858.CD001895.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu 2013
- Liu JP, Yang H, Xia Y, Cardini F. Herbal preparations for uterine fibroids. Cochrane Database of Systematic Reviews 2013, Issue 4. Art. No: CD005292. [DOI: 10.1002/14651858.CD005292.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Murji 2017
- Murji A, Whitaker L, Chow TL, Sobel ML. Selective progesterone receptor modulators (SPRMs) for uterine fibroids. Cochrane Database of Systematic Reviews 2017, Issue 4. Art. No: CD010770. [DOI: 10.1002/14651858.CD010770.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Ping Liu 2013
- Liu JP, Yang H, Xia Y, Cardini F. Herbal preparations for uterine fibroids. Cochrane Database of Systematic Reviews 2013, Issue 4. Art. No: CD005292. [DOI: 10.1002/14651858.CD005292.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Ray 2016
- Ray S, Ray A. Non‐surgical interventions for treating heavy menstrual bleeding (menorrhagia) in women with bleeding disorders. Cochrane Database of Systematic Reviews 2016, Issue 11. Art. No: CD010338. [DOI: 10.1002/14651858.CD010338.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]
Sangkomkamhang 2013
- Sangkomkamhang U, Lumbiganon P, Laopaiboon M, Mol BWJ. Progestogens or progestogen‐releasing intrauterine systems for uterine fibroids. Cochrane Database of Systematic Reviews 2013, Issue 2. Art. No: CD008994. [DOI: 10.1002/14651858.CD008994.pub2] [DOI] [PubMed] [Google Scholar]
Song 2013
- Song H, Lu D, Navaratnam K, Shi G. Aromatase inhibitors for uterine fibroids. Cochrane Database of Systematic Reviews 2013, Issue 10. Art. No: CD009505. [DOI: 10.1002/14651858.CD009505.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Tan 2013
- Tan YH, Lethaby A. Pre‐operative endometrial thinning agents before endometrial destruction for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2013, Issue 11. Art. No: CD010241. [DOI: 10.1002/14651858.CD010241.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Tristan 2012
- Tristan M, Orozco LJ, Steed A, Ramírez-Morera A, Stone P. Mifepristone for uterine fibroids. Cochrane Database of Systematic Reviews 2012, Issue 8. Art. No: CD007687. [DOI: 10.1002/14651858.CD007687.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Additional references
(D) Abbott 2003
- Abbott J, Hawe J, Hunter D, Garry R. A double-blind randomized trial comparing the Cavaterm and the Novasure endometrial ablation systems for the treatment of dysfunctional uterine bleeding. Fertility and Sterility 2003;80(1):203-8. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Agarwal 2016
- Agarwal N, Gupta M, Kriplani A, Bhatla N, Singh N. Comparison of combined hormonal vaginal ring with ultralow-dose combined oral contraceptive pills in the management of heavy menstrual bleeding: a pilot study. Journal of Obstetrics & Gynecology 2016;36:71-5. [Included in Lethaby 2019] [DOI] [PubMed] [Google Scholar]
(D) Ashraf 2017
- Ashraf MN, Habib-Ur-Rehman A, Shehzad Z, AlSharari SD, Murtaza G. Clinical efficacy of levonorgestrel and norethisterone for the treatment of chronic abnormal uterine bleeding. Journal of Pakistan Medical Association 2017;67(9):1331-8. [Included in Bofill Rodriguez 2019b and Bofill Rodriguez 2019d] [PMID: 28924270] [PubMed] [Google Scholar]
(D) Athanatos 2015
- Athanatos D, Pados G, Venetis C, Stamatopoulos P, Rousso D, Tsolakidis D, et al. Novasure impedance control system versus microwave endometrial ablation for the treatment of dysfunctional uterine bleeding: a double-blind, randomized controlled trial. Clinical and Experimental Obstetrics & Gynecology 2015;42(3):347-51. [Included in Bofill Rodriguez 2019a] [PMID: 26152008] [PubMed] [Google Scholar]
(D) Barrington 2002
- Barrington JW, Arunkalaivanan AS, Abdel-Fattah M. Comparison between the levonorgestrel intrauterine system (LNG-IUS) and thermal balloon ablation in the treatment of menorrhagia. European Journal of Obstetrics and Gynaecology and Reproductive Biology 2003;108:72-4. [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Bhattacharya 1997
- Bhattacharya S, Cameron IM, Parkin DE, Abramovich DR, Mollison J, Pinion SB, et al. A pragmatic randomised comparison of transcervical resection of the endometrium with endometrial laser ablation for the treatment of menorrhagia. British Journal of Obstetrics and Gynaecology 1997;104:601-7. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Bonduelle 1991
- Bonduelle M, Walker JJ, Calder AA. A comparative study of danazol and norethisterone in dysfunctional uterine bleeding presenting as menorrhagia. Postgraduate Medical Journal 1991;67:833-6. [Included in Bofill Rodriguez 2019b and Beaumont 2007] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Bongers 2004
- Bongers MY, Bourdrez P, Mol BWJ, Heintz APM, Brolmann HAM. Randomised controlled trial of bipolar radio-frequency endometrial ablation and balloon endometrial ablation. British Journal of Obstetrics and Gynaecology 2004;111:1095-102. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Bonnar 1996
- Bonnar J, Sheppard B. Treatment of menorrhagia during menstruation: randomized controlled trial of ethamsylate, mefenamic acid and tranexamic acid. BMJ 1996;313(7057):579-82. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019c] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Brun 2006
- Brun J-L, Raynal J, Burlet G, Galand B, Quereux C, Bernard P. Cavaterm thermal balloon endometrial ablation versus hysteroscopic endometrial resection to treat menorrhagia: the French, multicenter, randomized study. Journal of Minimally Invasive Gynecology 2006;13:424-30. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Buyru 1995
- Buyru F, Yalcin O, Kovanci E, Turfanda A. Danazol therapy in dysfunctional uterine bleeding. Istanbul Tip Fakultesi Mecmuasi 1995;58(3):37-40. [Included in Bofill Rodriguez 2019b and Beaumont 2007] [Google Scholar]
(D) Callender 1970
- Callender ST, Warner GT, Cope E. Treatment of menorrhagia with tranexamic acid. A double blind trial. British Medical Journal 1970;4(5729):214-6. [Included in Bryant-Smith 2018] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Cameron 1990
- Cameron IT, Haining R, Lumsden MA, Thomas VR, Smith SK. The effects of mefenamic acid and norethisterone on measured menstrual blood loss. Obstetrics and Gynecology 1990;76:85-8. [Included in Bofill Rodriguez 2019b and Bofill Rodriguez 2019c] [PubMed] [Google Scholar]
(D) Chamberlain
- Chamberlain G, Freeman R, Price F, Kennedy A, Green D, Eve L. Raw data from RCT (as supplied prior to 28 April 2022). Lorex Synthélabo Laboratory, UK. [Included in Bofill Rodriguez 2019c]
(D) Clark 2011
- Clark TJ, Samuel N, Malick S, Middleton LJ, Daniels J, Gupta JK. Bipolar radiofrequency compared with thermal balloon endometrial ablation in the office. Obstetrics and Gynecology 2011;117:109-18. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Cooper 1999
- Cooper KG, Bain C, Parkin DE. Comparison of microwave endometrial ablation and transcervical resection of the endometrium for treatment of heavy menstrual loss: a randomised trial. Lancet 1999;354:1859-63. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Cooper 2002
- Cooper J, Gimpelson R, Laberge P, Galen D, Garza-Leal JG, Scott J, et al. A randomized, multicenter trial of safety and efficacy of the NovaSure system in the treatment of menorrhagia. Journal of the American Association of Gynecologic Laparoscopists 2002;9(4):418-28. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Cooper 2004
- Cooper JM, Anderson TL, Fortin CA, Jack SA, Plentl MB. Microwave endometrial ablation vs rollerball electroablation for menorrhagia: a multicenter randomized trial. Journal of the American Association of Gynecologic Laparoscopists 2004;11(3):394-403. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Cooper 2019
- Cooper K, Breeman S, Scott N, Scotland G, Clark J, et al, on behalf of the HEALTH Study Group. Laparoscopic supracervical hysterectomy versus endometrial ablation for women with heavy menstrual bleeding (HEALTH). Lancet 2019;394(10207):1425-36. [DOI: ] [Included in Ferguson 2020] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Corson 2000
- Corson SL, Brill AI, Brooks PG, Cooper JM, Indman PD, Liu JH, et al. One-year results of the Vesta system for endometrial ablation. Journal of the American Association of Gynecologic Laparoscopists 2000;7(4):489-97. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Corson 2001
- Corson SL. A multicenter evaluation of endometrial ablation by hydrothermablator and rollerball for treatment of menorrhagia. Journal of the American Association of Gynecologic Laparoscopists 2001;8(3):359-67. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Crosignani 1997
- Crosignani PG, Vercellini P, Apolone G, De Giorgi O, Cortesi I, Meschia M. Endometrial resection versus vaginal hysterectomy for menorrhagia: long-term clinical and quality-of-life outcomes. American Journal of Obstetrics and Gynecology 1997;177(1):95-101. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(D) Crosignani 1997a
- Crosignani PG, Vercellini P, Mosconi P, Oldani S, Cortesi I, Giorgi O. Levonorgestrel-releasing intrauterine device versus hysteroscopic endometrial resection in the treatment of dysfunctional uterine bleeding. Obstetrics and Gynecology 1997;90:257-63. [CRSREF 286340] [Included in Marjoribanks 2016 and Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Dahiya 2016
- Dahiya P, Dalal M, Yadav A, Dahiya K, Jain S, Silan V. Efficacy of combined hormonal vaginal ring in comparison to combined hormonal pills in heavy menstrual bleeding. European Journal of Obstetrics & Gynecology and Reproductive Biology 2016;203:147-51. [DOI: 10.1016/j.ejogrb.2016.05.009] [Included in Lethaby 2019] [DOI] [PubMed] [Google Scholar]
(D) De Souza 2010
- De Souza SS, Camargos AF, Rezende CP, Pereira FA, Araujo CA, Silva Filho AL. A randomized prospective trial comparing the levonorgestrel-releasing intrauterine system with thermal balloon ablation for the treatment of heavy menstrual bleeding [Contraception]. 2010 81;3:226-31. [CRSREF 286343] [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Dickersin 2007
- Dickersin K, Munro MG, Clark M, Langenberg P, Scherer R, Frick K, et al. Hysterectomy compared with endometrial ablation for dysfunctional uterine bleeding. Obstetrics and Gynecology 2007;110(6):1279-89. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(D) Dockeray 1989
- Dockeray CJ, Sheppard BL, Bonnar J. Comparison between mefenamic acid and danazol in the treatment of established menorrhagia. British Journal of Obstetrics and Gynaecology 1989;96:840-4. [Included in Beaumont 2007] [DOI] [PubMed] [Google Scholar]
(D) Duleba 2003
- Duleba AJ, Heppard MC, Soderstrom RM, Townsend DE. A randomized study comparing endometrial cryoablation and rollerball electroablation for treatment of dysfunctional uterine bleeding. Journal of the American Association of Gynecologic Laparoscopists 2003;10(1):17-26. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Dunphy 1998
- Dunphy BC, Goerzen J, Greene CA, De La Ronde S, Seidel J, Ingelson B. A double blind randomised study comparing danazol and medroxyprogesterone acetate in the management of menorrhagia. Journal of Obstetrics and Gynaecology 1998;18(6):553-5. [Included in Beaumont 2007] [DOI] [PubMed] [Google Scholar]
(D) Dwyer 1993
- Dwyer N, Hutton J, Stirrat GM. Randomised controlled trial comparing endometrial resection with abdominal hysterectomy for the surgical treatment of menorrhagia. British Journal of Obstetrics and Gynaecology 1993;100(3):237-43. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(D) Edlund 1995
- Edlund M, Andersson K, Rybo G, Lindoff CC, Astedt B, Schoult BV. Reduction of menstrual blood loss in women suffering from idiopathic menorrhagia with a novel antifibrinolytic drug (Kabi2161). British Journal of Obstetrics and Gynaecology 1995;102(11):913-7. [Included in Bryant-Smith 2018] [DOI] [PubMed] [Google Scholar]
(D) Endrikat 2009
- Endrikat J, Shapiro H, Lukkari-Lax E, Kunz M, Schmidt W, Fortier M. A Canadian, multicentre study comparing efficacy of a levonorgestrel-releasing intrauterine system to an oral contraceptive in women with idiopathic menorrhagia. Journal of Obstetrics and Gynaecology Canada 2009;31(4):340-7. [Included in Lethaby 2019 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Ergun 2012
- Ergun B, Kuru O, Sen S, Kilic Y, Bastu E. Roller-ball endometrial ablation versus levonorgestrel releasing intrauterine system in the management of abnormal uterine bleeding. Gineco.eu 2012;8(4):199-201. [CRSREF 286346] [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [Google Scholar]
(D) Fraser 1981
- Fraser IS, Pearse C, Shearman RP, Elliott PM, McIlveen J, Markham R. Efficacy of mefenamic acid in patients with a complaint of menorrhagia. Obstetrics & Gynaecology 1981;58(5):543-51. [Included in Bofill Rodriguez 2019c] [PubMed] [Google Scholar]
(D) Fraser 1991
- Fraser IS, McCarron G. Randomized trial of 2 hormonal and 2 prostaglandin-inhibiting agents in women with a complaint of menorrhagia. Australian & New Zealand Journal of Obstetrics & Gynaecology 1991;31:66-70. [Included in Lethaby 2019, Bofill Rodriguez 2019c and Beaumont 2007] [DOI] [PubMed] [Google Scholar]
(D) Fraser 2011
- Fraser IS, Romer T, Parke S, Zeun S, Mellinger U, Machlitt A, Jet al. Effective treatment of heavy and/or prolonged menstrual bleeding with an oral contraceptive containing estradiol valerate and dienogest: a randomized double-blind Phase III trial. Human Reproduction 2011;26(10):2698-708. [DOI: 10.1093/humrep/der224] [Included in Lethaby 2019] [DOI] [PubMed] [Google Scholar]
(D) Freeman 2011
- Freeman EW, Lukes A, VanDrie D, Mabey RG, Gersten J, Adomako TL. A dose-response study of a novel, oral tranexamic formulation for heavy menstrual bleeding. American Journal of Obstetrics and Gynecology 2011;205(4):319.e1-319.e7. [Included in Bryant-Smith 2018] [DOI] [PubMed] [Google Scholar]
(D) Gannon 1991
- Gannon MJ, Holt EM, Fairbank J, Fitzgerald M, Milne MA, Crystal AM, et al. A randomised trial comparing endometrial resection and abdominal hysterectomy for the treatment of menorrhagia. BMJ (Clinical Research Ed.) 1991;303(6814):1362-4. [Included in Ferguson 2020] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Ghazizadeh 2014
- Ghazizadeh S, Panahi Z, Ghanbari Z, Menshadi AT, Farahmandian T, Javadian P. Comparative efficacy of NovaSure, the levonorgestrel-releasing intrauterine system, and hysteroscopic endometrial resection in the treatment of menorrhagia: a randomized clinical trial. Journal of Gynecologic Surgery 2014;30(4):215-8. [Data reported in a way that do not allowed comparisons] [Included in Marjoribanks 2016, Bofill Rodriguez 2019a and Bofill Rodriguez 2019d] [Google Scholar]
(D) Goshtasebi 2013
- Goshtasebi A, Moukhah S, Gandevani SB. Treatment of heavy menstrual bleeding of endometrial origin: randomized controlled trial of medroxyprogesterone acetate and tranexamic acid. Archives of Gynecology and Obstetrics 2013;288(5):1055-60. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019b] [DOI] [PubMed] [Google Scholar]
(D) Grover 1990
- Grover V, Usha R, Gupta U, Kalra S. Management of cyclical menorrhagia with prostaglandin synthetase inhibitor. Asia-Oceania Journal of Obstetrics & Gynaecology 1990;16:255-9. [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(D) Hashim 2012
- Hashim HA, Alsherbini W, Bazeed M. Contraceptive vaginal ring treatment of heavy menstrual bleeding: a randomized controlled trial with norethisterone. Contraception 2012;85:246-52. [Included in Lethaby 2019 and Bofill Rodriguez 2019b] [DOI] [PubMed] [Google Scholar]
(D) Hawe 2003
- Hawe J, Abbott J, Hunter D, Phillips G, Garry R. A randomised controlled trial comparing the Cavaterm endometrial ablation system with the Nd:YAG laser for the treatment of dysfunctional uterine bleeding. British Journal of Obstetrics and Gynaecology 2003;110:350-7. [Included in Bofill Rodriguez 2019a] [PubMed] [Google Scholar]
(D) Herman 2013
- Beelen P, Van Den Brink MJ, Geomini P, Dekker J, Herman M, Duijnhoven R, Mol B, Berger M, Bongers M. RCT comparing the LNG-IUS versus endometrial ablation in women with heavy menstrual bleeding (MIRA). International Journal of Gynecology and Obstetrics 2018;143(3):318-319. [10.1002/ijgo.12582] [Included in Bofill Rodriguez 2019d] [Google Scholar]
(D) Higham 1993
- Higham JM, Shaw RW. A comparative study of danazol, a regimen of decreasing doses of danazol, and norethindrone in the treatment of objectively proven unexplained menorrhagia. American Journal of Obstetrics and Gynecology 1993;169:1134-9. [Included in Bofill Rodriguez 2019b and Beaumont 2007] [DOI] [PubMed] [Google Scholar]
(D) Hurskainen 2004
- Hurskainen R, Teperi J, Rissanen P, Aalto AM, Grenman S, Kivela A, et al. Clinical outcomes and costs with the levonorgestrel-releasing intrauterine system or hysterectomy for the treatment of menorrhagia. JAMA 2004;291(12):1456-63. [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Irvine 1988
- Irvine GA, Campbell-Brown MB, Lumsden MA, Heikkila A, Walker JJ, Cameron IT. Randomised comparative trial of the levonorgestrel intrauterine system and norethisterone for treatment of idiopathic menorrhagia. British Journal of Obstetrics and Gynaecology 1988;105(6):592-8. [Included in Bofill Rodriguez 2019b] [DOI] [PubMed] [Google Scholar]
(D) Istre 1998
- Rauramo I, Elo I, Istre O. Long-term treatment of menorrhagia with levonorgestrel intrauterine system versus endometrial resection. Obstetrics and Gynecology 2004;104(6):1314-21. [Included in Marjoribanks 2016] [DOI] [PubMed] [Google Scholar]
(D) Jain 2016
- Jain P, Rajaram S, Gupta B, Goel N, Srivastava H. Randomized controlled trial of thermal balloon ablation versus vaginal hysterectomy for leiomyoma-induced heavy menstrual bleeding. International Journal of Gynaecology and Obstetrics 2016;135(2):140-4. [Included in Ferguson 2020] [https://doi.org/10.1016/j.ijgo.2016.04.020] [DOI] [PubMed] [Google Scholar]
(D) Jaisamrarn 2006
- Jaisamrarn U, Sitavarin S, Suwikrom S, Kamolpornwijit W. Randomized trial of medical therapies for menorrhagia: clinical efficacy and effect on quality of life. In: XVIII FIGO World Congress of Gynecology and Obstetrics; 2006 Nov 5-10; Kuala Lumpur (Malaysia). Vol. 3. 2006:83. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019c]
(D) Jensen 2011
- Jensen JT, Parke S, Mellinger U, Machlitt A, Fraser IS. Effective treatment of heavy menstrual bleeding with estradiol valerate and dienogest. Obstetrics & Gynecology 2011;117:777-87. [DOI: 10.1097/AOG.Ob013e3182118ac3] [Included in Lethaby 2019] [DOI] [PubMed] [Google Scholar]
(D) Kaunitz 2010
- Kaunitz AM, Bissonnette F, Monteiro I, Lukkari-Lax E, Muysers C, Jensen JT. Levonorgestrel-releasing intrauterine system or medroxyprogesterone for heavy menstrual bleeding. Obstetrics and Gynecology 2010;116:3625-32. [CRSREF: 328939] [Included in Bofill Rodriguez 2019b and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Kiseli 2016
- Kiseli M, Kayikcioglu F, Evliyaoglu O, Haberal A. Comparison of therapeutic efficacies of norethisterone, tranexamic acid and levonogestrel-releasing intrauterine system for the treatment of heavy menstrual bleeding: a randomized controlled study. Gynecologic and Obstetric Investigation 2016;81(5):447-53. [Included in Bryant-Smith 2018, Bofill Rodriguez 2019b and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Kittelsen 1998
- Kittelsen N, Istre O. A randomized study comparing levonorgestrel intrauterine system (LNG IUS) and transcervical resection of the endometrium (TCRE) in the treatment of menorrhagia: preliminary results. Gynecological Endocrinology 1998;7:61-5. [Included in Bofill Rodriguez 2019d] [Google Scholar]
(D) Kriplani 2006
- Kriplani A, Kulshrestha V, Agarwal N, Diwakar S. Role of tranexamic acid in management of dysfunctional uterine bleeding in comparison with medroxyprogesterone acetate. Journal of Obstetrics and Gynaecology 2006;26(7):673-8. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019b] [DOI] [PubMed] [Google Scholar]
(D) Laberge 2017
- Laberge P, Garza-Leal J, Fortin C, Grainger D, Johns DA, Adkins RT, et al. A randomized controlled multicenter US Food and Drug Administration trial of the safety and efficacy of the Minerva Endometrial Ablation System: one-year follow-up results. Journal of Minimally Invasive Gynecology 2017;24(1):124-32. [DOI: 10.1016/j.jmig.2016.09.009] [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Lukes 2010
- Lukes AS, Moore KA, Muse KN, Gersten JK, Hecht BR, Edlund M, et al. Tranexamic acid treatment for heavy menstrual bleeding. Obstetrics and Gynecology 2010;116(4):865-75. [Included in Bryant-Smith 2018] [DOI] [PubMed] [Google Scholar]
(D) Malak 2006
- Malak KA, Shawki O. Management of menorrhagia with the levonorgestrel intrauterine system versus endometrial resection. Gynecological Surgery 2006;3:275-80. [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [Google Scholar]
(D) McClure 1992
- McClure N, Marners M, Healy DL, Hill DJ, Lawrence AS, Wingfield M, et al. A quantitative assessment of endometrial electrocautery in the management of menorrhagia and a comparative report of argon laser endometrial ablation. Gynaecological Endoscopy 1992;1:199-202. [Included in Bofill Rodriguez 2019a] [Google Scholar]
(D) Meyer 1998
- Meyer WR, Walsh BW, Grainger DA, Peacock LM, Loffer FD, Steege JF. Thermal balloon and rollerball ablation to treat menorrhagia: a multicenter comparison. Obstetrics and Gynecology 1998;92:98-103. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Muggeridge 1983
- Muggeridge J, Elder MG. Mefenamic acid in the treatment of menorrhagia. Research and Clinical Forums 1983;5:83-8. [Included in Bofill Rodriguez 2019c] [Google Scholar]
(D) O'Connor 1997
- O'Connor H, Broadbent JA, Magos AL, McPherson K. Medical Research Council randomised trial of endometrial resection versus hysterectomy in management of menorrhagia. Lancet 1997;349(9056):897-901. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(D) Ozdegirmenci 2011
- Ozdegirmenci O, Kayikcioglu F, Akgul MA, Kaplan M, Karcaaltincaba M, Haberal A, et al. Comparison of levonorgestrel intrauterine system versus hysterectomy on efficacy and quality of life in patients with adenomyosis. Fertility and Sterility 2011;95(2):497-502. [Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Pellicano 2002
- Pellicano M, Guida M, Acunzo G, Cirillo D, Bifulco G, Nappi C. Hysteroscopic transcervical endometrial resection versus thermal destruction for menorrhagia: a prospective randomized trial on satisfaction rate. American Journal of Obstetrics and Gynecology 2002;187:545-50. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Penninx 2010
- Penninx JPM, Mol BW, Engels R, Rumste MME, Kleijn C, Koks CAM, et al. Bipolar radiofrequency endometrial ablation compared with hydrothermablation for dysfunctional uterine bleeding. Obstetrics and Gynecology 2010;116:819-26. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Penninx 2016
- Penninx J, Herman M, Kruitwagen R, Ter Haar A, Mol B, Bongers M. Bipolar versus balloon endometrial ablation in the office: a randomized controlled trial. European Journal of Obstetrics & Gynecology and Reproductive Biology 2016;196:52-6. [DOI: 10.1016/j.ejogrb.2015.10.010] [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Perino 2004
- Perino A, Castelli A, Cucinella G, Biondo A, Pane A, Venezia R. A randomized comparison of endometrial laser intrauterine thermotherapy and hysteroscopic endometrial resection. Fertility and Sterility 2004;82:731-4. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Pinion 1994
- Pinion SB, Parkin DE, Abramovich DR, Naji A, Alexander DA, Russell IT, et al. Randomised trial of hysterectomy, endometrial laser ablation, and transcervical endometrial resection for dysfunctional uterine bleeding. British Medical Journal (Clinical Research Ed.) 1994;309(6960):979-83. [Included in Ferguson 2020] [DOI] [PMC free article] [PubMed] [Google Scholar]
(D) Preston 1995
- Preston JT, Cameron IT, Adams EJ, Smith SK. Comparative study of tranexamic acid and norethisterone in the treatment of ovulatory menorrhagia. British Journal of Obstetrics and Gynaecology 1995;102(5):401-6. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019b] [DOI] [PubMed] [Google Scholar]
(D) Romer 1998
- Romer T. The treatment of recurrent menorrhagias - Cavaterm-balloon-coagulation versus Rollerball-endometrial ablation - a prospective randomized comparative study [Die therapie rezidivierender Menorrhagien - Cavaterm-Ballon-Koagulatioon versus Roller-Ball-Endometriumkoagulation - eine prospektive randomisierte Vergleichsstudie]. Zentralblatt fur Gynakologie 1998;120(10):511-4. [Included in Bofill Rodriguez 2019a] [PubMed] [Google Scholar]
(D) Sambrook 2009
- Sambrook AM, Cooper KG, Campbell MK, Cook JA. Clinical outcomes from a randomised comparison of microwave endometrial ablation with thermal balloon endometrial ablation for the treatment of heavy menstrual bleeding. British Journal of Obstetrics and Gynaecology 2009;116:1038-45. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Sayed 2011
- Sayed GH, Zakherah MS, El-Nashar SA, Shaaban MM. A randomised clinical trial of a levonorgestrel-releasing intrauterine system and a low-dose combined oral contraceptive for fibroid-related menorrhagia. International Journal of Gynecology and Obstetrics 2011;112:126-30. [Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Sesti 2011
- Sesti F, Ruggeri V, Pietropolli A, Piancatelli R, Piccione E. Thermal balloon ablation versus laparoscopic supracervical hysterectomy for the surgical treatment of heavy menstrual bleeding: a randomized study. Journal of Obstetrics and Gynaecology Research 2011;37(11):1650-7. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(D) Sesti 2012
- Sesti F, Piancatelli R, Pietropolli A, Ruggeri V, Piccione E. Levonorgestrel-releasing intrauterine system versus laparoscopic supracervical hysterectomy for the treatment of heavy menstrual bleeding: a randomized study. Journal of Women's Health 2012;21(8):851-7. [Included in Marjoribanks 2016 and Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Shabaan 2011
- Shabaan MM, Zakherah MS, El-Nashar SA, Sayed GH. Levonorgestrel-releasing intrauterine system compared to low dose combined oral contraceptive pills for idiopathic menorrhagia: a randomized clinical trial. Contraception 2011;83:48-54. [DOI: 10.1016/j.contraception.2010.06.011] [Included in Lethaby 2019 and Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Shaw 2007
- Shaw R, Symonds IM, Tamizian O, Chaplain J, Mukhopadhyay S. Randomised comparison of thermal balloon ablation and levonorgestrel intrauterine system in patients with idiopathic menorrhagia. Australian and New Zealand Journal of Obstetric and Gynaecology 2007;47:335-40. [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Soysal 2002
- Soysal M, Soysal S, Ozer S. A randomized controlled trial of levonorgestrel releasing IUD and thermal balloon ablation in the treatment of menorrhagia. Zentralblatt fur Gynakologie 2002;124(4):213-9. [Included in Marjoribanks 2016 Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Talis 2006
- Busfield RA, Farquhar CM, Sowter MC, Lethaby A, Sprecher M, Yu Y, et al. A randomised trial comparing the levonorgestrel intrauterine system and thermal balloon ablation for heavy menstrual bleeding. BJOG: an International Journal of Obstetrics and Gynaecology 2006;113:257-63. [Included in Marjoribanks 2016 and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Tam 2006
- Tam WH, Yuen PM, Ng DPS, Lung PL, Lok IH, Rogers MS. Health status function after treatment with thermal balloon endometrial ablation and levonorgestrel intrauterine system for idiopathic menorrhagia: a randomised study. Gynecologic and Obstetric Investigation 2006;62:84-8. [Included in Marjoribanks 2016 Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(D) Tsang 1987
- Tsang BK, Domingo MT, Spence JE, Garner PR, Dudley DK, Oxorn H. Endometrial prostaglandins and menorrhagia: influence of a prostaglandin synthetase inhibitor in vivo. Canadian Journal of Physiology and Pharmacology 1987;65:2081-4. [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(D) van Eijkeren 1992
- Eijkeren MA, Christiaens GC, Geuze HJ, Haspels AA, Sixma JJ. Effects of mefenamic acid on menstrual hemostasis in essential menorrhagia. American Journal of Obstetrics and Gynecology 1992;166:1419-28. [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(D) van Zon‐Rabelink 2003
- Zon-Rabelink IAA, Vleugels MPH, Merkus HMWM, Graaf R. Endometrial ablation by rollerball electrocoagulation compared to uterine balloon thermal ablation. Technical and safety aspects. European Journal of Obstetrics, Gynecology, and Reproductive Biology 2003;110:220-3. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Vercellini 1999
- Vercellini P, Oldani S, Yaylayan L, Zaina B, De Giorgi O, Crosignani PG. Randomised comparison of vaporising electrode and cutting loop for endometrial ablation. Obstetrics and Gynecology 1999;94:521-7. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(D) Ylikorkala 1986
- Ylikorkala O, Pekonen F. Naproxen reduces idiopathic but not fibromyoma-induced menorrhagia. Obstetrics & Gynaecology 1986;68:10-2. [Included in Bofill Rodriguez 2019c] [PubMed] [Google Scholar]
(D) Zhang 2008
- Zhang Y, He F, Sun Z, LI S, Bi S, Huang X, et al. A multicenter, prospective, randomized, open comparator study on the treatment of ovulatory menorrhagia with tranexamic acid and norethisterone in China. Chinese Journal of Obstetrics and Gynaecology 2008;43(4):247-50. [Included in Bryant-Smith 2018 and Bofill Rodriguez 2019b] [PubMed] [Google Scholar]
(D) Zupi 2003
- Zupi E, Zullo F, Marconi D, Sbracia M, Pellicano M, Solima E, et al. Hysteroscopic endometrial resection versus laparoscopic supracervical hysterectomy for menorrhagia: a prospective randomised trial. American Journal of Obstetrics and Gynecology 2003;188(1):7-12. [Included in Ferguson 2020] [DOI] [PubMed] [Google Scholar]
(N) Andersch 1988
- Andersch B, Milsom I, Rybo G. An objective evaluation of flurbiprofen and tranexamic acid in the treatment of idiopathic menorrhagia. Acta Obstetricia et Gynecologica Scandinavica 1988;67:645-8. [Data reported in a way that do not allow comparisons] [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(N) Boujida 2002
- Boujida VH, Philipsen T, Pelle J, Joergensen JC. Five-year follow-up of endometrial ablation: endometrial coagulation versus endometrial resection. Obstetrics and Gynecology 2002;99:988-92. [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(N) Cameron 1987
- Cameron IT, Leask R, Kelly RW, Baird DT. The effects of danazol, mefenamic acid, norethisterone and a progesterone-impregnated coil on endometrial prostaglandin concentrations in women with menorrhagia. Prostaglandins 1987;34:99-110. [Included in Bofill Rodriguez 2019b, Bofill Rodriguez 2019c and Beaumont 2007] [DOI] [PubMed] [Google Scholar]
(N) Chimbira 1980
- Chimbira T, Andreson A, Naish C, Cope E, Turnbull A. Reduction of menstrual blood loss by danazol in unexplained menorrhagia: lack of effect of placebo. British Journal of Obstetrics and Gynaecology 1980;87(12):1152-8. [Excluded as is a comparison between different dosages of danazol] [Included in Beaumont 2007] [PMID: ] [DOI] [PubMed] [Google Scholar]
(N) Cooper 1997
- Cooper KG, Parkin DE, Garratt AM, Grant AM. A randomised comparison of medical and hysteroscopic management in women consulting a gynaecologist for treatment of heavy menstrual loss. British Journal of Obstetrics and Gynaecology 1997;104:1360-6. [CRSREF 2868337] [Included in Marjoribanks 2016] [No randomization on the medical treatment arm] [DOI] [PubMed] [Google Scholar]
(N) Fathima 2012
- Fathima A, Sultana A. Clinical efficacy of a Unani formulation ‘Safoof Habis’ in menorrhagia: a randomized controlled trial. European Journal of Integrative Medicine 2012;4(3):e315-22. [Excluded as comparison is not medical treatment] [Included in Bryant-Smith 2018] [Google Scholar]
(N) Goshtasebi 2015
- Goshtasebi A, Mazari Z, Behboudi Gandevani S, Naseri M. Anti-hemorrhagic activity of Punica granatum L. flower (Persian Golnar) against heavy menstrual bleeding of endometrial origin: a double-blind, radomized controlled trial. Medical Journal of the Islamic Republic of Iran 2015;29:199. [Comparison was not a medical treatment] [Included in Bryant-Smith 2018] [PMC free article] [PubMed] [Google Scholar]
(N) Gupta 2013
- Gupta J, Kai J, Middleton L, Pattison H, Gray R, Daniels J. Levonorgestrel intrauterine system versus medical therapy for menorrhagia. New England Journal of Medicine 2013;368:128-37. [Included in Bofill Rodriguez 2019d] [one arm was not randomized] [DOI] [PubMed] [Google Scholar]
(N) Hall 1987
- Hall P, MacLachlan N, Thorn N, Nudd MW, Taylor CG, Garrioch DB. Control of menorrhagia by the cyclo-oxygenase inhibitors naproxen sodium and mefenamic acid. British Journal of Obstetrics and Gynaecology 1987;94:554-8. [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(N) Khajehei 2013
- Khajehei N, Abdali K, Tabatabaee. The effect of mefenamic acid and naproxen on heavy menstrual bleeding: A placebo-controlled study. South African Journal of Obstetricsand gynaecology 2013;19(2):31-34. [DOI: 10.7196/SAJOG.587] [DOI] [Google Scholar]
(N) Kilic 2009
- Kilic S, Yuksel B, Doganay M, Bardakci H, Akinsu F, Uzunlar O, et al. The effect of levonorgestrel-releasing intrauterine device on menorrhagia in women taking anticoagulant medication after cardiac valve replacement. Contraception 2009;80(2):152-7. [Included in Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(N) Kupperman 2004
- Kupperman M, Varner RE, Summitt RL, Learman LA, Ireland C, Vittinghoff E, et al. Effect of hysterectomy vs medical treatment on health-related quality of life and sexual functioning. JAMA 2004;291(12):1447-55. [Included in Marjoribanks 2016] [No randomization on the medical treatment arm] [DOI] [PubMed] [Google Scholar]
(N) Lamb 1987
- Lamb MP. Danazol in menorrhagia: a double blind placebo controlled trial. Journal of Obstetrics and Gynaecology 1987;7:212-16. [Compares Danazol and placebo] [Included in Beaumont 2007] [Google Scholar]
(N) Makarainen 1986
- Makarainen L, Ylikorkala O. Primary and myoma-associated menorrhagia: role of prostaglandin and effects of ibuprofen. British Journal of Obstetrics and Gynaecology 1986;93:974-8. [Included in Bofill Rodriguez 2019c] [DOI] [PubMed] [Google Scholar]
(N) Najam 2010
- Najam R, Agarwald D, Tyagi R, Singh S. Comparison of tranexamic acid with a combination of tranexamic acid and mefenamic acid in reducing menstrual blood loss in ovulatory dysfunctional uterine bleeding. Journal of Clinical and Diagnostic Research 2010;4:3020-5. [Included in Bofill Rodriguez 2019c] [No data available for the comparisons] [Google Scholar]
(N) Onoglu 2007
- Onoglu A, Taskin O, Inal M, Sadik S, Simsek M, Akar M, et al. Comparison of the long-term histopathologic and morphologic changes after endometrial rollerball ablation and resection: a prospective randomized trial. Journal of Minimally Invasive Gynecology 2007;14:39-42. [(E) The only review outcome reported was surgical time] [Included in Bofill Rodriguez 2019a] [DOI] [PubMed] [Google Scholar]
(N) Reid 2005
- Reid PC, Virtanen-Kari S. Randomised comparative trial of the levonorgestrel intrauterine system and mefenamic acid for the treatment of idiopathic menorrhagia: a multiple analysis using total menstrual fluid loss, menstrual blood loss and pictorial blood loss assessment charts. British Journal of Obstetrics and Gynaecology 2005;112:1121-5. [Data reported in ranges] [Included in Bofill Rodriguez 2019c and Bofill Rodriguez 2019d] [DOI] [PubMed] [Google Scholar]
(N) Shravage 2011
- Shravage J, Mekhala D, Bellad MB, Ganachari MS, Dhumale HA. Ormeloxifene versus medroxyprogesterone acetate (MPA) in the treatment of dysfunctional uterine bleeding. A double blind randomized controlled trial. Journal of South Asian Federation of Obstetrics and Gynaecology 2011;3(1):21-4. [10.5005/jp-journals-10006-1116] [Data was reported in a format that do not allow comparisons] [Included in Bofill Rodriguez 2019b] [Google Scholar]
(N) Thabet 2010
- Thabet SMA. New attempt using ablative curettage technique for managing benign premenopausal uterine bleeding. Obstetrics and Gynaecology Research 2010;36(4):803-9. [Included in Bofill Rodriguez 2019a] [No data for comaprisons (overcouretage)] [DOI] [PubMed] [Google Scholar]
Alvarado 2001
- Alvarado RG, Liu JY, Zwolak RM. Danazol and limb-threatening arterial thrombosis: two case reports. Journal of Vascular Surgery 2001;34(6):1123-6. [DOI: 10.1067/mva.2001.118078] [DOI] [PubMed] [Google Scholar]
Alvaro 1996
- Alvaro D, Piat C, Francia C, Franchitto A, Furfaro S, Valente C et al. Ultrastructural features of danazol-induced cholestasis: a case study. Ultrastructural Pathology 1996;20(5):491-5. [DOI: 10.3109/01913129609016353] [DOI] [PubMed] [Google Scholar]
Audia 2016
- Audia S, Godeau B, Bonnotte B. Is there still a place for "old therapies" in the management of immune thrombocytopenia? La Revue de Medecine Interne 2016;37(1):43-9. [DOI: 10.1016/j.revmed.2015.08.007] [DOI] [PubMed] [Google Scholar]
Bacchi 2012
- Bacchi S, Palumbo P, Sponta A, Coppolino M. Clinical pharmacology of non-steroidal anti-inflammatory drugs: a review. Anti-Inflammatory & Anti-Allergy Agents in Medicinal Chemistry 2012;11(1):52-64. [DOI: 10.2174/187152312803476255] [DOI] [PubMed] [Google Scholar]
Blacker 1902
- Blacker JF. Vaporization of the uterus. British Journal of Obstetrics and Gynaecologu: An International Journal of Obstetrics & Gynaecology 1902;1(5):488-511. [Google Scholar]
Bofill Rodriguez 2020a
- Bofill Rodriguez M, Lethaby A, Farquhar C, Duffy JM. Interventions commonly available during pandemics for heavy menstrual bleeding: an overview of Cochrane Reviews.. Cochrane Database of Systematic Reviews 2020, Issue 7. Art. No: CD013651. [DOI: 10.1002/14651858.CD013651.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Bongers 2004
- Bongers MY, Mol BW, Brölmann HA. Current treatment of dysfunctional uterine bleeding. Maturitas 2004;47(3):159-74. [DOI: 10.1016/j.maturitas.2003.08.002. PMID: 15036486] [DOI] [PubMed] [Google Scholar]
Bradley 2016
- Bradley L, Guyen N. The medical management of abnormal uterine bleeding in reproductive-aged women. American Journal of Obstetrics and Gynecology 2015;214(1):31-44. [DOI: 10.1016/j.ajog.2015.07.044] [DOI] [PubMed] [Google Scholar]
Brignardello‐Petersen2017
- Brignardello-Petersen R, Bonner A, Alexander PE, Siemieniuk RA, Furukawa TA, Rochwerg B, et al. GRADE Working Group. Advances in the GRADE approach to rate the certainty in estimates from a network meta-analysis.. Journal of Clinical Epidemiology 2018;93:36-44. [DOI: 10.1016/j.jclinepi.2017.10.005] [Erratum in: J Clin Epidemiol. 2018 Jun;98:162. PMID: 29051107.] [DOI] [PubMed] [Google Scholar]
Brignardello‐Petersen 2019
- Brignardello-Petersen R, Mustafa RA, Siemieniuk RAC, Murad MH, Agoritsas T, Izcovich A et al, GRADE Working Group. GRADE approach to rate the certainty from a network meta-analysis: addressing incoherence. Journal of Clinical Epidemiology 2019;108:77-85. [DOI: 10.1016/j.jclinepi.2018.11.025] [DOI] [PubMed] [Google Scholar]
Brignardello‐Petersen 2020
- Brignardello-Petersen R, Izcovich, A, Rochwerg B, Florez I, Hazlewood G, et al. GRADE approach to drawing conclusions from a network meta-analysis using a partially contextualised framework. BMJ 2020;371:m3907. [DOI: 10.1136/bmj.m3907] [DOI] [PubMed] [Google Scholar]
Brunskill 1992
- Brunskill P. The effects of fetal exposure to danazol. British Journal of Obstetetrics and Gynaecology 1992;99(3):212-5. [PMID: ] [DOI] [PubMed] [Google Scholar]
Caldwell 2014
- Caldwell DM. An overview of conducting systematic reviews with network meta-analysis. Systematic Reviews Journal 2014;3:109. [DOI: 10.1186/2046-4053-3-109] [DOI] [PMC free article] [PubMed] [Google Scholar]
Clarke‐Pearson 2013
- Clarke-Pearson D, Geller E. Complications of hysterectomy. Obstetrics & Gynecology 2013;121(3):654-73. [DOI: 10.1097/AOG.0b013e3182841594] [DOI] [PubMed] [Google Scholar]
Cole 1971
- Cole S, Billewicz W, Thomson A. Sources of variation in menstrual blood loss. Journal of Obstetrics and Gynaecology of the British Commonwealth 1971;78:933-9. [PMID: ] [DOI] [PubMed] [Google Scholar]
COMET 2018
- Cooper N, Khan K, Rivas C. Defining core outcomes for clinical trials of heavy menstrual bleeding. A Core Outcome sets for Gynaecological conditions (COGS) project. www.comet-initiative.org/studies/details/789 (accessed prior to 28 April 2022).
Coulter 1995
- Coulter A, Kelland J, Peto V, Rees M. Treating menorrhagia in primary care. An overview of drug trials and a survey of prescribing practice. International Journal of Technology Assessment in Health Care 1995;11(3):456-71. [PMID: ] [DOI] [PubMed] [Google Scholar]
Daniels 2012
- Daniels J, Middleton L, Champaneria R, Khan K, Cooper K, Bhattacharya S, et al. Second generation endometrial ablation techniques for heavy menstrual bleeding: network meta-analysis. BMJ 2012;344:e2564. [DOI: ] [DOI] [PMC free article] [PubMed] [Google Scholar]
De Bastos 2014
- De Bastos M, Stegeman BH, Rosendaal FR, Van Hylckama Vlieg A, Helmerhorst FM, Stijnen T, et al. Combined oral contraceptives: venous thrombosis. Cochrane Database of Systematic Reviews 2014, Issue 3. Art. No: CD010813. [DOI: 10.1002/14651858.CD010813.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]
Dias 2010
- Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in mixed treatment comparison meta-analysis. Statistics in Medicine 2010;28(7-8):932-44. [DOI: 10.1002/sim.3767] [DOI] [PubMed] [Google Scholar]
Dias 2018
- Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network Meta-analysis for Decision Making. Wiley, 2018. [DOI: 10.1002/9781118951651] [Online ISBN:9781118951651] [DOI] [Google Scholar]
Dias 2019
- Dias S, Caldwell D. Network meta-analysis explained. Archives of Disease in Childhood. Fetal and Neonatal Edition 2019;104(1):F8-F12. [DOI: 10.1136/archdischild-2018-315224] [DOI] [PMC free article] [PubMed] [Google Scholar]
Di Carlo, 2015
- Di Carlo C, Guida M, De Rosa N, Sansone A, Gargano V, Cagnacci A, et al. Bleeding profile in users of an etonogestrel sub-dermal implant: effects of anthropometric variables. An observational uncontrolled preliminary study in Italian population. Gynaecological Endocrinology 2015;31(6):491-4. [DOI: 10.3109/09513590.2015.1018163] [DOI] [PubMed] [Google Scholar]
Edlund 2011
- Edlund M. Nonhormonal treatments for heavy menstrual bleeding. Journal of Women's Health 2011;20(11):1645-53. [DOI: 10.1089/jwh.2010.2696] [DOI] [PubMed] [Google Scholar]
Famuyide 2018
- Famuyide A. Endometrial Ablation. Journal of Minimally Invasive Gynecoly 2018;25(2):299-307. [DOI: 10.1016/j.jmig.2017.08.656] [DOI] [PubMed] [Google Scholar]
Fraser 2009
- Fraser I, Langham S, Uhl-Hochgraeber K. Health-related quality of life and economic burden of abnormal uterine bleeding. Expert Review of Obstetrics & Gynecology 2009;4(2):179-89. [DOI: 10.1586/17474108.4.2.179] [DOI] [Google Scholar]
Friberg 2006
- Friberg B, Orno A, Lindgren A, Lethagen S. Bleeding disorders among young women: a population-based prevalence study. Acta Obstetetrica et Gynecologica Scandinavica 2006;85(2):200-6. [DOI: ] [DOI] [PubMed] [Google Scholar]
Fritz 2012
- Fritz M, Speroff L. Clinical Gynecologic Endocrinology and Infertility. Wolters Kluwer Health/Lippincott Williams & Wilkins, 2012. [https://slycksyxr01.storage.googleapis.com/MDc4MTc3OTY4NQ==01.pdf] [Google Scholar]
Garay 2006
- Garay RP, Chiavaroli C, Hannaert P. Therapeutic efficacy and mechanism of action of ethamsylate, a long-standing hemostatic agent. American Journal of Therapeutics 2006;13(3):236-47. [DOI: 10.1097/01.mjt.0000158336.62740.54.] [DOI] [PubMed] [Google Scholar]
Gleeson 1994
- Gleeson N. Cyclic changes in endometrial tissue plasminogen activator and plasminogen activator inhibitor type 1 in women with normal menstruation and essential menorrhagia. American Journal of Obstetrics and Gynecology 1994;171(1):178-83. [PMID: ] [DOI] [PubMed] [Google Scholar]
GRADEpro GDT 2015 [Computer program]
- GRADEpro GDT. Hamilton (ON): McMaster University (developed by Evidence Prime), accessed prior to 28 April 2022. Available at gradepro.org.
Hallberg 1966
- Hallberg L, Hogdahl A, Nilsson LM, Rybo G. Menstrual blood loss - a population study. Variation at different ages and attempts to define normality. Acta Obstetetrica et Gynecologica Scandinava 1966;45(3):320-51. [DOI: 10.3109/00016346609158455] [DOI] [PubMed] [Google Scholar]
Harlow 2004
- Harlow S, Campbell OM. Epidemiology of menstrual disorders in developing countries: a systematic review. British Journal of Obstetricas and Gynecology 2004;111(1):6-16. [PMID: ] [DOI] [PubMed] [Google Scholar]
Harrison 1976
- Harrison RF Campbell S. A double-blind trial of ethamsylate in the treatment of primary and intrauterine-device menorrhagia. Lancet 1976;308(7980):283-5. [DOI: ] [DOI] [PubMed] [Google Scholar]
Haththotuwa 2011
- Haththotuwa R, Goonewardene M, Desai S, Senanayake L, Tank J, Fraser I. Management of abnormal uterine bleeding in low- and high-resource settings: consideration of cultural issues. Seminars in Reproductive Medicine 2011;29(5):446-58. [DOI: 10.1055/s-0031-1287668] [DOI] [PubMed] [Google Scholar]
Herman 2013
- Herman MC, den Brink MJ, Geomini PM, Meurs HS, Huirne JA, Eising HP, et al. Levonorgestrel releasing intrauterine system (Mirena) versus endometrial ablation (Novasure) in women with heavy menstrual bleeding: a multicentre randomised controlled trial. BioMed Central Womens Health 2013;13:32. [DOI: 10.1186/1472-6874-13-32] [DOI] [PMC free article] [PubMed] [Google Scholar]
Higgins 2011
- Higgins JP, Green S, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (updated March 2011). The Cochrane Collaboration, 2011. Available from handbook.cochrane.org.
Higgins 2021
- Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page MJ. Cochrane Handbook for Systematic Reviews of Interventions Version 6.2 (updated February 2021). Cochrane, 2021. Available from training.cochrane.org/handbook. [www.training.cochrane.org/handbook]
Hurskainen 2007
- Hurskainen R, Grenman S, Komi I, Kujansuu E, Luoto R, Orrainen M, et al. Diagnosis and treatment of menorrhagia. Acta Obstetetrica et Gynecologica Scandinava 2007;86(6):749-57. [DOI: 10.1080/00016340701415400] [DOI] [PubMed] [Google Scholar]
Jacobsten 2014
- Jacobstein R, Polis CB. Progestin-only contraception: injectables and implants. Best Practice & Research: Clinical Obstetrics & Gynaecology 2014;28(6):795-806. [DOI: 10.1016/j.bpobgyn.2014.05.003] [DOI] [PubMed] [Google Scholar]
Jick 1995
- Jick SS, Myers MW. A study of danazol's safety. Pharmacotherapy 1995;15(6):740-1. [PMID: ] [PubMed] [Google Scholar]
Karlsson 2014
- Karlsson T, Marions L, Edlund M. Heavy menstrual bleeding significantly affects quality of life. Acta Obstetrica et Gynecologica Scandinava 2014;93(1):52-7. [DOI: 10.1111/aogs.12292] [DOI] [PubMed] [Google Scholar]
Kumar 2016
- Kumar V, Chodankar R, Gupta J. Endometrial ablation for heavy menstrual bleeding. Women's Health 2016;12:45-52. [DOI: 10.2217/whe.15.86] [DOI] [PMC free article] [PubMed] [Google Scholar]
Laberge 2015
- Laberge P, Leyland N, Murji A, Fortin C, Martyn P, Vilos G, et al. Endometrial ablation in the management of abnormal uterine bleeding. Journal of Obstetrics and Gynaecology Canada 2015;37(4):362-79. [DOI: 10.1016/s1701-2163(15)30288-7] [DOI] [PubMed] [Google Scholar]
Lecker I 2016
- Lecker I, Wang D, Whissell P, Avramescu S, Mazer D, Orser B. Tranexamic acid–associated seizures: causes and treatment. Annals of Neurology 2016;79:18-26. [PMID: ] [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu 2007
- Liu Z, Doan Q, Blumenthal P, Dubois R. A systematic review evaluating health-related quality of life, work impairment, and health-care costs and utilization in abnormal uterine bleeding. Value in Health 2007;10(3):183-94. [DOI: 10.1111/j.1524-4733.2007.00168.x] [DOI] [PubMed] [Google Scholar]
Micks 2013
- Micks EA, Jensen JT. Treatment of heavy menstrual bleeding with the estradiol valerate and dienogest oral contraceptive pill. Advances in Therapy 2013;30(1):1-13. [DOI: 10.1007/s12325-012-0071-3] [DOI] [PubMed] [Google Scholar]
Miller 2015
- Miller J, Lenhart G, Bonafede M, Basinski C, Lukes A, Troeger K. Cost effectiveness of endometrial ablation with the NovaSure (R) system versus other global ablation modalities and hysterectomy for treatment of abnormal uterine bleeding: US commercial and Medicaid payer perspectives. International Journal of Women's Health 2015;7:59-73. [DOI: 10.2147/ijwh.s75030] [DOI] [PMC free article] [PubMed] [Google Scholar]
Morrison 2008
- Morrison J, Patel S, Watson W, Zaidi Q, Mangione A, Goss T. Assessment of the prevalence and impact of anemia on women hospitalized for gynecologic conditions associated with heavy uterine bleeding. Journal of Reproductive Medicine 2008;53(5):323-30. [PMID: ] [PubMed] [Google Scholar]
Munro 2012
- Munro M, Critchley H, Fraser I. The FIGO systems for nomenclature and classification of causes of abnormal uterine bleeding in the reproductive years: who needs them? American Journal of Obstetetrics and Gynecology 2012;207(4):259-65. [DOI: 10.1016/j.ajog.2012.01.046] [DOI] [PubMed] [Google Scholar]
Munro 2018
- Munro M. Endometrial ablation. Best Practice & Research Clinical Obstetrics & Gynaecology 2018;46:120-39. [DOI: ] [DOI] [PubMed] [Google Scholar]
NICE 2007
- National Institute for Health and Clinical Excellence. Heavy Menstrual Bleeding. Clinical Guideline No. 44. National Collaborating Centre for Women's and Children’s Health Commissioned by the National Institute for Health and Clinical Excellence 2007. [https://www.nice.org.uk/guidance/cg44]
NICE 2018a
- National Institute for Health and Clinical Excellence. Heavy Menstrual Bleeding. Clinical Guideline No. 88. National Collaborating Centre for Women's and Children’s Health Commissioned by the National Institute for Health and Clinical Excellence 2018. [https://www.nice.org.uk/guidance/ng88/evidence]
NICE 2018b
- RCOG, National Guideline Alliance, hosted by the Royal College of Obstetrician and Gynaecologists. Evidence reviews for management of heavy menstrual bleeding. National Institute for Health and Clinical Excellence 2018. [file:///P:/FMHSfiles/network%20meta-analisis/NICE%20NMA%20HMB.pdf] [PubMed]
Pincus 1899a
- Pincus L. Response to the article of Mr Duhrssen about "autmocausis" [Erwiderung auf den Aufsatz des Herrn Duhrssen uber “Atmocausis”]. Centralblatt für Gynäkologie 1899;13:352. [Google Scholar]
Pincus 1899b
- Pincus L. Some remarks on the article on Zestocausis the Treub's Clinic [Einege Bemerkungen zu dem Aufsatz Zestocausis aus der Treub'schen Klinic]. Centralblatt für Gynäkologie 1899;4:113. [Google Scholar]
Puhan 2014
- Puhan M, Schünemann H, Hasan Murad, M, Li T, Brigardello-Petersen R et al, GRADE working group. A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. BMJ 2014;349:g5630. [DOI: ] [DOI] [PubMed] [Google Scholar]
RCOG 2012
- Royal College of Obstetricians and Gynaecologists. National Heavy Menstrual Bleeding Audit. London: RCOG; 2012. Second Annual Report. [https://www.rcog.org.uk/en/news/rcog-release-four-year-national-audit-shows-improved-treatment-for-women-with-heavy-menstrual-bleeding-hmb/]
Rees 1987
- Rees M, DiMarzo V, Tippins J, Morris H, Turnbull A. Leukotriene release by endometrium and myometrium throughout the menstrual cycle in dysmenorrhoea and menorrhagia. Journal of Endocrinology 1987;113(2):291-5. [PMID: ] [DOI] [PubMed] [Google Scholar]
Review Manager 2020 [Computer program]
- Review Manager 5 (RevMan 5). Version 5.4. Copenhagen: Nordic Cochrane Centre, The Cochrane Collaboration, 2020. [https://training.cochrane.org/online-learning/core-software-cochrane-reviews/revman]
Salanti 2014
- Salanti G, Del Giovane C, Chaimani A, Caldwell DM, Higgins JP. Evaluating the quality of evidence from a network meta-analysis. PLOS ONE 2014;9(7):e99682. [DOI: 10.1371/journal.pone.0099682] [DOI] [PMC free article] [PubMed] [Google Scholar]
Santesso 2020
Shaw 1972
- Shaw S, Aaronson D, Moyer D. Quantitation of menstrual blood loss- further evaluation of the alkaline hematin method. Contraception 1972;5(6):497-513. [DOI: 10.1016/0010-7824(72)90015-7] [DOI] [PubMed] [Google Scholar]
Shea 2007
- Shea B, Grimshaw J, Wells G, Boers M, Andersson N, Hamel C, et al. Development of AMSTAR: a measurement tool to assess the methodological quality of systematic reviews. Bio Med Central Medical Research Methodology 2007;7:10. [DOI: 10.1186/1471-2288-7-10] [DOI] [PMC free article] [PubMed] [Google Scholar]
Shea 2017
- Shea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J, et al. AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ 2017;257:j4008. [DOI: ] [DOI] [PMC free article] [PubMed] [Google Scholar]
Smith 1981
- Smith SK, Abel MH, Kelly RW, Baird DT. Prostaglandin synthesis in the endometrium of women with ovular dysfunctional uterine bleeding. British Journal Obstetrics & Gynaecology 1981;88(4):434-42. [PMID: ] [DOI] [PubMed] [Google Scholar]
Speroff 2010
- Speroff L, Darney P. Clinical Guide for Contraception. Philadelphia: Lippincott Williams & Wilkins (LWW), 2010. [ISBN: 978-1-60-831610-6] [Google Scholar]
Tanaka 2013
- Tanaka E, Momoeda M, Osuga Y, Rossi B, Nomoto K, Hayakawa M, et al. Burden of menstrual symptoms in Japanese women: results from a survey-based study. Journal of Medical Economics 2013;16(11):1255-66. [DOI: 10.3111/13696998.2013.830974] [DOI] [PubMed] [Google Scholar]
Uhm 2014
- Uhm S, Perriera L. Hormonal contraception as treatment for heavy menstrual bleeding: a systematic review. Clinical Obstetrics and Gynecology 2014;57(4):694-717. [DOI: 10.1097/grf.0000000000000061] [DOI] [PubMed] [Google Scholar]
Van der Heijden 2017
- Van der Heijden P, Geomini P, Herman M, Veersema S, Bongers M. Timing of insertion of levonorgestrel-releasing intrauterine system: a randomised controlled trial. British Journal of Obstetrics and Gynecology 2016;124(2):299-305. [DOI: 10.1111/1471-0528.14445] [DOI] [PubMed] [Google Scholar]
Van Valkenhoef 2016
- Van Valkenhoef G, Dias S, Ades AE, Welton N. Automated generation of node-splitting models for assessment of inconsistency in network meta-analysis. Research Synthesis Methods 2016;7(1):80-93. [DOI: 10.1002/jrsm.1167] [DOI] [PMC free article] [PubMed] [Google Scholar]
White 2015
- White IR. Network meta analysis. Stata Journal 2015;15:951-85. [https://www.stata-journal.com/article.html?article=st0410] [Google Scholar]
Yepes‐Nunez 2019
- Yepes-Nunez J, Lic SA, Guyatta G, Jacka SM, Brozek JL, Beyenea J, et al. Development of the summary of findings table for network meta-analysis. Journal of Clinical Epidemiology 2019;115:1-13. [DOI: ] [DOI] [PubMed] [Google Scholar]