Pre-operative endometrial thinning agents before endometrial destruction for heavy menstrual bleeding

  • Review
  • Intervention

Authors


Abstract

Background

Heavy menstrual bleeding is one of the most common reasons for referral of premenopausal women to a gynaecologist. Although medical therapy is generally first line, many women eventually will require further treatment. Endometrial ablation by hysteroscopic and more recent "second-generation" devices such as balloon, radiofrequency or microwave ablation offers a day-case surgical alternative to hysterectomy. Complete endometrial destruction is one of the main determinants of treatment success. Surgery is most effective if undertaken when endometrial thickness is less than four millimeters. One option is to perform the surgery in the immediate postmenstrual phase, which is not always practical. The other option is to use hormonal agents that induce endometrial thinning pre-operatively. The most commonly evaluated agents are goserelin (a gonadotrophin-releasing hormone analogue, or GnRHa) and danazol. Other GnRH analogues and progestogens have also been studied, although fewer data are available. It has been suggested that these agents will reduce operating time, improve the intrauterine operating environment and reduce absorption of fluid used for intraoperative uterine cavity distension. They may also improve long-term outcomes, including menstrual loss and dysmenorrhoea.

Objectives

To investigate the effectiveness and safety of pre-operative endometrial thinning agents (GnRH agonists, danazol, estrogen-progestins and progestogens) versus another agent or placebo when given before endometrial destruction in premenopausal women with heavy menstrual bleeding.

Search methods

The following electronic databases were searched to April 2013 for published and unpublished randomised controlled trials that met the inclusion criteria: the Menstrual Disorders and Subfertility Group (MDSG) Specialised Register of controlled trials, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL and PsycINFO.

Other electronic sources of trials included trial registers for ongoing and registered trials; citation indexes; conference abstracts in the Web of Knowledge; the LILACS database for trials from the Portuguese- and Spanish-speaking world; PubMed; and the OpenSIGLE database and Google for grey literature.

All searches were performed in consultation with the MDSG Trials Search Co-ordinator.

Selection criteria

Randomised controlled trials (RCTs) were included if they compared the effects of these agents with one other, or with placebo or no treatment, on relevant intraoperative and postoperative treatment outcomes. Selection of trials was carried out independently by two review authors.

Data collection and analysis

Two review authors independently assessed studies for risk of bias and extracted data on surgical outcomes, effectiveness outcomes, proportion of women requiring further surgical therapy during follow-up, endometrial outcome measures, acceptability of use outcomes and quality of life. Data were analysed on an intention-to-treat basis. Dichotomous data were combined for meta-analysis with RevMan software using the Mantel-Haenszel method to estimate pooled risk ratios (RRs). Continuous data were combined for meta-analysis with RevMan software using an inverse variance method to estimate the pooled mean difference (MD) with 95% confidence interval (CI). The overall quality of evidence for the main findings was assessed with the use of GRADE working group methods.

Main results

Twenty studies with 1969 women were included in this review. These studies compared GnRHa, danazol and progestogens versus placebo or no treatment; GnRHa versus danazol, progestogens, GnRH antagonists or dilatation & curettage; and danazol versus progestogens. Four studies performed more than one comparison.

When compared with no treatment, GnRHa used before hysteroscopic resection were associated with a higher rate of postoperative amenorrhoea at 12 months (RR 1.6, 95% CI 1.2 to 2.0, 7 RCTs, 605 women, moderate heterogeneity; I2 = 40%) and at 24 months (RR 1.62, 95% CI 1.04 to 2.52, 2 RCTs, 357 women, no heterogeneity; I2 = 0%), a slightly shorter duration of surgery (-3.5 minutes, 95% CI -4.7 to -2.3, 5 RCTs, 156 women, substantial heterogeneity; I2 = 72%) and greater ease of surgery (RR 0.32, 95% CI 0.22 to 0.46, 2 RCTs, 415 women, low heterogeneity; I2 = 4%). Postoperative dysmenorrhoea was reduced (RR 0.59, 95% CI 0.40 to 0.87, 2 RCTs, 133 women, no heterogeneity; I2 = 0%). The use of GnRHa had no effect on intraoperative complication rates (RR 1.47, 95% CI 0.35 to 6.06, 5 RCTs, 592 women, no heterogeneity; I2 = 0%), and participant satisfaction with this surgery was high irrespective of the use of pre-operative endometrial thinning agents (RR 0.99, 95% CI 0.93 to 1.05, 6 RCTs, 599 women, low heterogeneity; I2 = 11%). GnRHa produced more consistent endometrial atrophy than was produced by danazol (RR 1.84, 95% CI 1.23 to 2.75, 2 RCTs, 142 women, no heterogeneity; I2 = 0%). For other intraoperative and postoperative outcomes, any differences were minimal, and no benefits of GnRHa pretreatment were noted in studies in which women underwent second-generation ablation techniques. Both GnRHa and danazol produced side effects in a significant proportion of women, although few studies reported these in detail. Few randomised data were available to allow assessment of the effectiveness of progestogens as endometrial thinning agents. When reported, the long-term effects of endometrial thinning agents on benefits such as postoperative amenorrhoea were reduced with time.

The main study weaknesses were that most participants received no follow-up beyond 24 months and that the studies used a small sample size. Heterogeneity for outcomes reported ranged from none to substantial. More than half the trials had no blinding of participants or outcome assessment. Most of the trials were determined to have uncertain selection and reporting bias, as they did not report allocation concealment and evidence of selective reporting was noted. The quality of reporting of adverse events was generally poor, but, when described in the studies, they included menopausal symptoms such as hot flushes, vaginal dryness, hirsutism, decreased libido and voice changes, as well as other side effects such as headache and weight gain.

Authors' conclusions

Low-quality evidence suggests that endometrial thinning with GnRHa and danazol before hysteroscopic surgery improves operating conditions and short-term postoperative outcomes. GnRHa produced slightly more consistent endometrial thinning than was produced by danazol, although both achieved satisfactory results. The effect of these agents on longer-term postoperative outcomes was reduced with time. No benefits of GnRHa pretreatment were apparent with second-generation ablation techniques. Also, side effects were more common when these agents were used.

Résumé scientifique

Agents anticoagulants de l'endomètre préopératoire avant la destruction de l'endomètre pour saignements menstruels abondants

Contexte

Selon les gynécologues, les saignements menstruels abondants sont l'un des motifs les plus courants chez les femmes pré-ménopausées. Bien qu’un traitement médical soit généralement de première ligne, de nombreuses femmes exigeront éventuellement un traitement supplémentaire. L'ablation de l'endomètre par hystéroscopie et les appareils plus récents de deuxième génération, tels que l'ablation par micro-ondes, par ballonnet ou par radiofréquence, offrent une alternative chirurgicale de jour comparée à l'hystérectomie. La destruction complète de l'endomètre est l'un des principaux facteurs déterminants le succès du traitement. La chirurgie est plus efficace lorsque l'épaisseur de l'endomètre est inférieure à quatre millimètres. Une option consiste à réaliser la chirurgie dans la phase post-menstruelle immédiate, ce qui n'est pas toujours pratique. L'autre option consiste à utiliser des agents hormonaux qui amincissent l'endomètre avant l'opération. Les agents les plus couramment évalués sont la goséréline (un analogue de l'hormone de libération des gonadotrophines, ou GnRHa) et le danazol. D'autres analogues de la GnRH et des progestatifs ont également été étudiés, bien que moins de données soient disponibles. Il a été suggéré que ces agents permettent de réduire la durée de l'opération, d’améliorer l’environnement de l'opération intra-utérine et de réduire l'absorption de liquide utilisé pour la distension de la cavité utérine peropératoire. Ils peuvent également améliorer les résultats à long terme, y compris les pertes menstruelles et la dysménorrhée.

Objectifs

Étudier l'efficacité et l'innocuité préopératoire des agents anticoagulants de l'endomètre (agonistes de la GnRH, danazol, œstrogènes - progestérone et progestatifs) par rapport à un autre agent ou à un placebo lorsqu' ils sont administrés avant la destruction de l'endomètre chez les femmes pré-ménopausées présentant des saignements menstruels abondants.

Stratégie de recherche documentaire

Les bases de données électroniques suivantes ont fait l'objet de recherches jusqu'en avril 2013 pour identifier des essais contrôlés randomisés publiés et non publiés qui remplissaient les critères d'inclusion : le registre spécialisé des essais contrôlés du groupe sur les troubles menstruels et la fertilité, le registre Cochrane des essais contrôlés (CENTRAL), MEDLINE, EMBASE, CINAHL et PsycINFO.

D'autres sources électroniques d'essais incluaient les registres des essais en cours et les essais enregistrés; les indices de citation; les actes de conférences dans le Web of Knowledge; la base de données LILACS pour les essais en Portugais et en Espagnol; PubMed; et la base de données OpenSIGLE et Google pour la littérature grise.

Toutes les recherches ont été effectuées en consultation avec le coordinateur du groupe sur les troubles menstruels et la fertilité.

Critères de sélection

Les essais contrôlés randomisés (ECR) ont été inclus s'ils comparaient les effets de ces agents à un autre, ou à un placebo ou à l'absence de traitement, concernant les critères de jugement du traitement peropératoire et postopératoire pertinents. La sélection des essais a été réalisée indépendamment par deux auteurs de la revue.

Recueil et analyse des données

Deux auteurs de la revue ont indépendamment évalué le risque de biais des études et extrait les données des résultats chirurgicaux, l'efficacité des critères de jugement, la proportion de femmes nécessitant un traitement chirurgical au cours du suivi, d'autres mesures de résultats de l'endomètre, les critères de jugement concernant l’acceptabilité de l'utilisation et la qualité de vie. Les données ont été analysées en intention de traiter. Les données dichotomiques ont été combinées en une méta-analyse à l'aide du logiciel RevMan utilisant la méthode de Mantel-Haenszel pour estimer les rapports de risque (RR) combinés. Les données en continu ont été combinées en une méta-analyse à l'aide du logiciel RevMan utilisant une méthode de variance inverse afin d'estimer la différence moyenne (DM) avec un intervalle de confiance (IC) à 95%. La qualité globale des preuves pour les principaux résultats a été évaluée avec les méthodes du groupe de travail GRADE.

Résultats principaux

Vingt études totalisant 1 969 femmes ont été inclues dans cette revue. Ces études comparaient les GnRHa, le danazol et les progestatifs par rapport à un placebo ou à l'absence de traitement; les GnRHa versus le danazol, les progestatifs, les antagonistes de la GnRH ou la dilatation & le curetage; et le danazol versus les progestatifs. Quatre études ont réalisé plus d'une comparaison.

Par rapport à l'absence de traitement, les GnRHa utilisés avant la résection hystéroscopique étaient associées à un taux plus élevé d'aménorrhée postopératoire à 12 mois (RR de 1,6, IC à 95% de 1,2 à 2,0, 7 ECR, 605 femmes, hétérogénéité modérée; I 2 =40%) et à 24 mois (RR 1,62, IC à 95% de 1,04 à 2,52, 2 ECR, 357 femmes, absence d'hétérogénéité; I 2 =0%), durée de la chirurgie légèrement plus courte (-3,5 minutes, IC à 95% de -4,7 à -2,3, 5 ECR, 156 femmes, hétérogénéité substantielle; I 2 =72%) et chirurgie plus facile (RR 0,32, IC à 95% de 0,22 à 0,46, 2 ECR, 415 femmes, hétérogénéité faible; I 2 =4%). La dysménorrhée postopératoire a été réduite (RR 0,59, IC à 95% de 0,40 à 0,87, 2 ECR, 133 femmes, absence d'hétérogénéité; I 2 =0%). L'utilisation de GnRHa n'avait aucun effet sur les taux de complications peropératoires (RR de 1,47, IC à 95% de 0,35 à 6,06, 5 ECR; 592 femmes, absence d'hétérogénéité; I 2 =0%) et la satisfaction des participants envers cette intervention était élevée, indépendamment de l'utilisation des agents préopératoires anticoagulants de l'endomètre (RR 0,99, IC à 95% de 0,93 à 1,05, 6 ECR, 599 femmes, hétérogénéité faible; I 2 =11%). Les GnRHa produisaient une atrophie endométriale plus constante que le danazol (RR 1,84, IC à 95% de 1,23 à 2,75, 2 ECR, 142 femmes, absence d'hétérogénéité; I 2 =0%). Pour les autres résultats peropératoires et postopératoires, les différences étaient minimes et aucun bénéfice de prétraitement par GnRHa n’a été observé dans les études dans lesquelles les femmes avaient subi les techniques d'ablation de deuxième génération. Les GnRHa et le danazol produisaient des effets secondaires chez une proportion significative de femmes, bien que peu d'études aient fourni un compte rendu détaillé de ces effets. Peu de données randomisées était disponibles pour permettre une évaluation de l'efficacité des progestatifs comme agents anticoagulants de l'endomètre. Lorsque rapportés, les effets bénéfiques à long terme des agents anticoagulants de l'endomètre comme aménorrhée postopératoire ont été réduits avec le temps.

La principale faiblesse de l'étude était que la plupart des participants n’avaient reçu aucun suivi au-delà de 24 mois et que les études utilisaient une taille d'échantillon réduite. L'hétérogénéité pour les critères de jugement indiqués variait de aucune à importante. Plus de la moitié des essais ne présentaient ni de participant en aveugle, ni d'évaluation des résultats. La plupart des essais avaient une sélection et un rapport de biais incertains, car ils n'avaient pas rendu compte de l'assignation secrète et des preuves de notification sélective étaient observées. La qualité de notification d'effets indésirables était généralement médiocre. Toutefois, lorsqu' ils étaient décrits dans les études, ils incluaient les symptômes de la ménopause, tels que les bouffées de chaleur, la sécheresse vaginale, l'hirsutisme, une diminution de la libido, les modifications de la voix et d'autres effets secondaires, tels que des céphalées et une prise de poids.

Conclusions des auteurs

Les preuves de faible qualité suggèrent que l'amincissement de l'endomètre par GnRHa et par danazol avant la chirurgie hystéroscopique offre de meilleures conditions opératoires et améliore les résultats postopératoires à court terme. Les GnRHa produisaient un amincissement de l'endomètre légèrement plus constant, comparés au danazol, bien que les deux résultats obtenus soient satisfaisants. L'effet de ces agents sur les résultats postopératoires à plus long terme a été réduit avec le temps. Aucun bénéfice de prétraitement par GnRHa n’a été observé avec les techniques d'ablation de deuxième génération. De même, les effets secondaires étaient plus fréquents lorsque ces agents étaient utilisés.

Plain language summary

Pre-operative endometrial thinning agents before endometrial destruction for heavy menstrual bleeding

Review Question

Cochrane authors reviewed evidence for the effectiveness and safety of medications used to thin the lining of the womb before surgery performed to destroy (ablate) this lining (endometrium) in premenopausal women with heavy menstrual bleeding.

Background

Heavy menstrual bleeding is one of the most common reasons why premenopausal women are referred to a gynaecologist; this condition can cause significant physical, emotional and social stress in a woman's life. Surgery to ablate the endometrium is a treatment option available for this condition that is less invasive than removal of the womb (i.e. hysterectomy). We wanted to discover whether using medications to thin the lining of the womb before endometrial destruction enhanced the effectiveness of surgery in reducing symptoms and improved operating conditions for the surgeon. We also wanted to evaluate the safety of these medications (i.e. observe whether side effects or surgical complications were increased). These medications included gonadotrophin-releasing hormone analogues (GnRH analogues), danazol and progestogens. Endometrial destruction surgery included either the older 'hysteroscopic' technique, whereby the lining of the womb is destroyed under direct vision, or the newer second-generation techniques, which include balloon, radiofrequency and microwave ablation.

Study Characteristics

The evidence is current to April 2013. The review included 20 randomised controlled trials with a total of 1969 premenopausal women with heavy menstrual bleeding for whom non-surgical treatment had not worked. Studies compared GnRH analogues, danazol and progestogens versus placebo or no treatment; GnRH analogues versus danazol, progestogens, GnRH antagonists or dilatation & curettage; and danazol versus progestogens. Four studies performed more than one comparison. Three studies used the newer second-generation surgical techniques for endometrial destruction.

Key Results

GnRH analogues and danazol used before hysteroscopic surgery improve both operating conditions for the surgeon and short-term bleeding symptoms for women (up to 24 months after surgery). GnRH analogues thin the lining of the womb better and more consistently than danazol, although both agents produce satisfactory results. Adverse effects were more common in women taking GnRH analogues or danazol compared with no treatment, and this was especially true with danazol. Adverse effects included menopausal symptoms such as hot flushes, vaginal dryness, hirsutism, decreased libido and voice changes, as well as other side effects such as headache and weight gain. The use of medications to thin the lining of the womb before surgery does not appear to improve heavy menstrual bleeding in the long term (i.e. longer than 24 months). However, only a few small studies followed up with women for longer than 24 months. Also, medications given to thin the womb lining do not provide additional benefit when used with the newer second-generation endometrial destruction techniques, which are being performed increasingly in hospitals.

Quality of Evidence

Overall, the quality of the evidence was very low because of risk of bias in the included studies and differences between the studies. The quality of reporting of adverse events was generally poor.

Résumé simplifié

Agents anticoagulants de l'endomètre préopératoire avant la destruction de l'endomètre pour saignements menstruels abondants

Question de la revue

Des auteurs de la revue Cochrane ont examiné les preuves concernant l'efficacité et l'innocuité de médicaments utilisés afin d’amincir la muqueuse utérine avant la chirurgie réalisée pour détruire (par ablation) cette muqueuse (endomètre) chez les femmes pré-ménopausées présentant des saignements menstruels abondants.

Contexte

Les saignements menstruels abondants sont l'un des motifs les plus courants pour lesquels les femmes pré-ménopausées sont orientées vers un gynécologue; cette pathologie peut entraîner un stress physique, émotionnel et social significatif dans la vie d'une femme. La chirurgie pour l’ablation de l'endomètre est une option de traitement disponible pour cette affection qui est moins invasive que l'ablation de l'utérus (hystérectomie). Nous avons voulu déterminer si l'utilisation de médicaments pour amincir la muqueuse utérine avant la destruction de l'endomètre améliorait l'efficacité de la chirurgie dans la réduction des symptômes et améliorait les conditions opératoires pour le chirurgien. Nous avons également voulu évaluer l'innocuité de ces médicaments (par exemple, savoir si les effets secondaires ou les complications chirurgicales augmentaient). Ces médicaments incluaient des analogues de libération de gonadotrophine (GnRH analogues), du danazol et des progestatifs. La chirurgie de destruction de l'endomètre incluait soit l’ancienne technique « hystéroscopique », dans laquelle la muqueuse utérine est détruite sous vision directe, soit la technique plus récente de deuxième génération qui comprend l'ablation par micro-ondes, par ballonnet et par radiofréquence.

Caractéristiques de l’étude

Les preuves sont à jour en avril 2013. La revue a inclus 20 essais contrôlés randomisés avec un total de 1 969 femmes pré-ménopausées présentant des saignements menstruels abondants, pour lesquelles un traitement non chirurgical n'avait pas fonctionné. Les études comparaient des analogues de la GnRH, du danazol et des progestatifs par rapport à un placebo ou à l'absence de traitement; des analogues de la GnRH versus le danazol, des progestatifs, des antagonistes de la GnRH ou dilatation & curetage; et du danazol versus des progestatifs. Quatre études ont réalisé plus d'une comparaison. Trois études utilisaient les techniques de chirurgie plus récentes de deuxième génération pour la destruction de l'endomètre.

Résultats principaux

Les analogues de la GnRH et le danazol utilisés avant l’hystéroscopie améliorent les conditions opératoires pour le chirurgien et les symptômes de saignement chez les femmes à court terme (jusqu' à 24 mois après l'opération). Les analogues de la GnRH amincissent la muqueuse utérine plus efficacement et de façon plus constante que le danazol, bien que ces deux agents produisent des résultats satisfaisants. Les effets indésirables étaient plus fréquents chez les femmes prenant des analogues de la GnRH ou du danazol par rapport à l'absence de traitement et cela était particulièrement vrai pour le danazol. Les effets indésirables incluaient les symptômes de la ménopause, tels que les bouffées de chaleur, la sécheresse vaginale, l'hirsutisme, une diminution de la libido et des modifications de la voix, ainsi que d'autres effets secondaires, tels que des céphalées et une prise de poids. L'utilisation de médicaments pour amincir la muqueuse utérine avant la chirurgie ne semblait pas améliorer les saignements menstruels abondants sur le long terme (plus de 24 mois). Cependant, seules quelques petites études ont suivi les femmes pendant plus de 24 mois. En outre, des médicaments administrés pour amincir la muqueuse utérine ne fournissaient pas de bénéfice supplémentaire lorsqu' ils étaient utilisés par les nouvelles techniques de deuxième génération pour l’ablation de l'endomètre, qui sont de plus en plus réalisées dans les hôpitaux.

La qualité de l’évidence

Dans l'ensemble, la qualité des preuves était très faible en raison du risque de biais dans les études incluses et des différences entre les études. La qualité de notification d'effets indésirables était généralement médiocre.

Notes de traduction

Traduit par: French Cochrane Centre 1st November, 2013
Traduction financée par: Financeurs pour le Canada : Instituts de Recherche en Santé du Canada, Ministère de la Santé et des Services Sociaux du Québec, Fonds de recherche du Québec-Santé et Institut National d'Excellence en Santé et en Services Sociaux; pour la France : Ministère en charge de la Santé

Summary of findings(Explanation)

Summary of findings for the main comparison. GnRH analogues compared with placebo or no treatment for heavy menstrual bleeding
  1. 1Few studies were blinded, and only one study had good quality.
    2Moderate heterogeneity.
    3Mostly small studies.

GnRH analogues compared with placebo or no treatment for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRH analogues
Comparison: placebo or no treatment
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Placebo or no treatment GnRH analogues
Postoperative amenorrhoea at 12 months or less Study population RR 1.58
(1.24 to 2.01)
605
(seven studies)
⊕⊝⊝⊝
very low 1,2,3
 
246 per 1000 389 per 1000
(305 to 494)
*The basis for the assumed risk is the median control group risk across studies. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 2 GnRH analogues compared with danazol for heavy menstrual bleeding

Summary of findings 2. GnRH analogues compared with danazol for heavy menstrual bleeding
  1. 1No trials were blinded, and only one had adequate allocation concealment.
    2Substantial heterogeneity.

GnRH analogues compared with danazol for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRH analogues
Comparison: danazol
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Danazol GnRH analogues
Postoperative amenorrhoea at 12 months or less 348 per 1000 408 per 1000
(314 to 530)
RR 1.17
(0.9 to 1.52)
340
(five studies)
⊕⊝⊝⊝
very low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 3 GnRH analogues compared with progestogens for heavy menstrual bleeding

Summary of findings 3. GnRH analogues compared with progestogens for heavy menstrual bleeding
  1. 1One trial had adequate allocation concealment, and none reported blinding.
    2Wide confidence intervals with limited number of participants.

GnRH analogues compared with progestogens for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRH analogues
Comparison: progestogens
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Progestogens GnRH analogues
Postoperative amenorrhoea at 12 months or less 258 per 1000 408 per 1000
(248 to 674)
RR 1.58
(0.96 to 2.61)
146
(three studies)
⊕⊝⊝⊝
very low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 4 GnRH analogue compared with GnRH antagonist for heavy menstrual bleeding

Summary of findings 4. GnRH analogue compared with GnRH antagonist for heavy menstrual bleeding
  1. 1One trial with small population.

GnRH analogue compared with GnRH antagonist for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRH analogue
Comparison: GnRH antagonist
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
GnRH antagonist GnRH analogue
Postoperative amenorrhoea at 12 months or less 840 per 1000 798 per 1000
(664 to 958)
RR 0.95
(0.79 to 1.14)
100
(one study)
⊕⊕⊕⊝
moderate 1
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 5 GnRH analogue compared with dilatation & curettage for heavy menstrual bleeding

Summary of findings 5. GnRH analogue compared with dilatation & curettage for heavy menstrual bleeding
  1. 1No blinding of participants; selective reporting and small sample size in only one study.

GnRH analogue compared with dilatation & curettage for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRH analogue
Comparison: dilatation & curettage
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Dilatation & curettage GnRH analogue
Postoperative amenorrhoea at 12 months or less 0 per 1000 0 per 1000
(0 to 0)
RR 20.12
(1.21 to 333.63)
92
(one study)
Very low quality 1 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 6 Danazol compared with no pretreatment for heavy menstrual bleeding

Summary of findings 6. Danazol compared with no pretreatment for heavy menstrual bleeding
  1. 1One trial had adequate description of randomisation, the other had adequate allocation concealment, but neither reported blinding.
    2Limited number of participants.

Danazol compared with no pretreatment for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: danazol
Comparison: no pretreatment
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
No pretreatment Danazol
Postoperative amenorrhoea at 12 months or less 517 per 1000 543 per 1000
(414 to 719)
RR 1.05
(0.8 to 1.39)
179
(two studies)
⊕⊕⊝⊝
low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 7 Progestogens compared with no pretreatment for heavy menstrual bleeding

Summary of findings 7. Progestogens compared with no pretreatment for heavy menstrual bleeding
  1. 1Adequate allocation concealment but not clear if blinded.
    2One trial with few participants.

Progestogens compared with no pretreatment for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: progestogens
Comparison: no pretreatment
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
No pretreatment Progestogens
Postoperative amenorrhoea at 12 months or less 520 per 1000 281 per 1000
(135 to 582)
RR 0.54
(0.26 to 1.12)
50
(one study)
⊕⊕⊝⊝
low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 8 Danazol compared with progestogens for heavy menstrual bleeding

Summary of findings 8. Danazol compared with progestogens for heavy menstrual bleeding
  1. 1Only one of two trials had adequate allocation concealment, and neither trial reported blinding.
    2Both trials had limited numbers of participants.

Danazol compared with progestogens for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: danazol
Comparison: progestogens
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Progestogens Danazol
Postoperative amenorrhoea at 12 months or less 269 per 1000 509 per 1000
(302 to 856)
RR 1.89
(1.12 to 3.18)
103
(two studies)
⊕⊕⊝⊝
low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 9 GnRHa or danazol compared with no pretreatment for heavy menstrual bleeding

Summary of findings 9. GnRHa or danazol compared with no pretreatment for heavy menstrual bleeding
  1. 1One of two trials had adequate allocation concealment, but neither trial reported blinding.
    2Only two trials with limited numbers of participants.

GnRHa or danazol compared with no pretreatment for heavy menstrual bleeding
Patient or population: women with heavy menstrual bleeding
Settings:
Intervention: GnRHa or danazol
Comparison: no pretreatment
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
No pretreatment GnRHa or danazol
Postoperative amenorrhoea at 12 months or less 428 per 1000 483 per 1000
(381 to 611)
RR 1.13
(0.89 to 1.43)
280
(two studies)
⊕⊕⊝⊝
low 1,2
 
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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).
CI: Confidence interval; RR: Risk ratio.
GRADE Working Group grades of evidence.
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Background

Description of the condition

Heavy menstrual bleeding or menorrhagia is one of the most common reasons why premenopausal women are referred to a gynaecologist; it is a significant cause of morbidity in these women, resulting in symptoms of anaemia and reduced quality of life. Although medical therapy is generally the first approach, many women eventually require or request further treatment, with endometrial ablation offering a less invasive surgical approach compared with hysterectomy. Each year in the UK, one in 20 women between the ages of 30 and 49 years consults her general practitioner because of menorrhagia (Grant 2000). In a 2009 study, the English rate of surgery for menorrhagia, including both hysterectomy and endometrial ablation, was shown to be 143 procedures per 100,000 women, and 40% of these procedures were hysterectomy (Cromwell 2009). The rate of hysterectomy for premenopausal women in Australia was 312 per 100,000 in 2004 to 2005, and similar numbers were reported in New Zealand during the same period (Hill 2010; McPherson 2011). A recent nationwide analysis in Germany (Stang 2011) showed that hysterectomy rates for benign reasons varied from 213.8 to 361.9 per 100,000 women. Hysterectomy is associated with a significant inpatient hospital stay and a period of recuperation that makes it an unappealing and unnecessarily invasive choice for many women. Recent studies have reported increased risk of pelvic floor prolapse and urinary stress incontinence, as well as increased association with cardiovascular disease following hysterectomy (Cooper 2011; Ingelsson 2011). Hysteroscopic endometrial ablation or resection (and more recently balloon or microwave ablation) offers a day-case surgical alternative to hysterectomy for many of these women. In addition, studies have suggested that these procedures are more cost-effective than hysterectomy in terms of short- and long-term costs to the health system and the shortened recovery period for the woman (Hidlebaugh 1998). If effective in the long term, this procedure could result in savings in health expenditures.

Description of the intervention

A variety of endometrial ablation methods may be used to treat heavy menstrual bleeding. Neodymium-doped yttrium aluminium garnet (Nd:YAG) laser ablation was first performed in 1981 (Goldrath 1990), and use of the resectoscope was described shortly afterwards (DeChenery 1983). Rollerball ablation was reported in 1988 (Lin 1988) in Japan, and in 1989 in Europe (Vancaillie 1989). These are known as the 'first-generation' or the more historical hysteroscopic endometrial ablation techniques. More recently, second-generation endometrial ablation techniques (Neuwirth 1994) have been described that produce thermal damage to the endometrium without the need for skills in operative hysteroscopy. Second-generation endometrial ablation techniques include balloon ablation, microwave energy, bipolar radiofrequency and diffused laser energy or hot water. These 'blind' ablation techniques are technically easier to perform, take less time and achieve similar results when compared with hysteroscopic resection and ablation or first-generation techniques (Munro 2001; Lethaby 2009), which they have largely replaced. A recent meta-analysis has suggested that second-generation endometrial ablation techniques such as bipolar radiofrequency (Novasure) and microwave ablative devices (Microwave Endometrial Ablation), which now are commonly used, are more effective than the thermal balloon (ThermaChoice) and free fluid ablation (Hydro ThermAblator) in the treatment of menorrhagia (Daniels 2012).

Observational follow-up studies (Erian 1994; Krogh 2009), randomised studies (Dwyer 1993; Pinion 1994; Sesti 2011) and systematic reviews (Bhattacharya 2011) comparing these techniques with hysterectomy have reported high levels of postoperative participant satisfaction, especially with second-generation endometrial ablation techniques. However, long-term rates of repeat surgery needed with endometrial ablation range from 12.4% at one year to 38.2% at four years, showing that some women will experience little benefit from the procedure or will subsequently request further surgery, usually a hysterectomy, thus narrowing the cost difference over time (Longinotti 2008; Lethaby 2010). The proportion of women who experience amenorrhoea (no periods) after surgery varies in different series from 30% to 60% (Dwyer 1993; Erian 1994; Pinion 1994), although the proportion experiencing improvement in menstrual symptoms is considerably higher, ranging from 86.8% within one year to 97.3% within four years (Lethaby 2010).

Complete removal or destruction of the endometrium is one of the most important determinants of treatment success. The aim of endometrial ablation techniques is for the operating surgeon to remove or destroy the entire thickness of the uterine endometrium (lining of the uterus) and the basal endometrial glands present in the superficial myometrium (underlying muscle layer) (Garry 1995). During the menstrual cycle, endometrial thickness varies from as little as 1 mm in the immediate postmenstrual phase to 10 mm or greater in the late secretory phase (Weingold 1990). The radius of a standard electrosurgery loop used for endometrial resection is about 4 mm, and the depth of tissue destruction with an Nd:YAG laser or with a rollerball electrode is 4 to 6 mm (Goldrath 1990; Duffy 1992; El-Nashar 2009). Best results with ablation can be achieved if the endometrium is first rendered thin and inactive, either during the immediate postmenstrual phase or with administration of hormonal agents that induce endometrial thinning or atrophy, to increase the likelihood that ablation will include the basal layer of the endometrium, which can lie 4 to 6 mm beneath the surface, depending on the phase of the cycle.

How the intervention might work

Difficulty in timing surgery for the immediate postmenstrual phase and the unpredictable thickness of the unprepared endometrium have resulted in a focus on the use of endometrial thinning agents before surgery. A number of randomised studies have been undertaken to compare different hormonal agents with one other or with no pre-operative treatment or placebo. Agents studied include goserelin (a gonadotrophin-releasing hormone (GnRH) analogue), danazol, progestogens, estrogen-progestin contraceptives and other GnRH agonists.

It has been suggested that use of these agents, particularly GnRH analogues, reduces operating time, improves the intrauterine operating environment and reduces distension medium absorption (this is the fluid used to distend the uterine cavity during surgery). Greater improvements in long-term outcomes such as menstrual blood loss and dysmenorrhoea may be achieved (Donnez 2001). An improved operating environment might also reduce the rate of complications associated with the procedure. Improved postoperative outcomes might increase patient satisfaction and reduce the proportion of women undergoing subsequent hysterectomy.

These agents add a significant additional cost to any hysteroscopic procedure. However, if they have any effect on the proportion of women undergoing a second procedure (usually hysterectomy), then the additional costs and potential for uncomfortable side effects associated with these agent are likely to be outweighed by the advantages they offer (Banu 2005).

Why it is important to do this review

The aim of this review is to evaluate evidence from randomised controlled trials to assess whether endometrial thinning agents have a beneficial effect on postoperative and long-term outcomes of endometrial resection and ablation, and to determine whether their use significantly improves the intrauterine operating environment. Endometrial ablation agents have the potential to improve both short- and long-term effectiveness of endometrial ablation in women with heavy menstrual bleeding, thus reducing the number of women requiring further surgery and possibly improving the cost-to-benefit ratio. It is important to do this review because the evidence gathered will evaluate whether endometrial thinning agents should be used more consistently in clinical practice.

Objectives

To investigate the effectiveness and safety of pre-operative endometrial thinning agents (GnRH agonists, danazol, estrogen-progestins and progestogens) versus another agent or placebo when given before endometrial destruction in premenopausal women with heavy menstrual bleeding.

Methods

Criteria for considering studies for this review

Types of studies

Both published and unpublished randomised controlled trials were considered for inclusion in this review.

Types of participants

Premenopausal women with heavy menstrual bleeding undergoing endometrial resection or ablation with first-generation or hysteroscopic techniques (i.e. resectoscope, rollerball diathermy, Nd:YAG laser) or any second-generation technique (i.e. bipolar radiofrequency, balloon ablation, diffused laser energy, free fluid ablation, microwave ablation).

Types of interventions

Studies comparing the effects of the following interventions before endometrial resection or ablation with any of the first- or second-generation techniques used for heavy menstrual bleeding were considered suitable for inclusion in this review.

1. Any medical therapy (GnRH analogues, danazol, estrogen-progestin contraceptives or progestogens) compared with placebo or no treatment.
2. Any specific type of medical therapy (GnRH analogues, danazol, estrogen-progestin contraceptives or progestogens) compared with any other specific type of medical therapy (GnRH analogues, danazol, estrogen-progestin contraceptives or progestogens).

Types of outcome measures

Outcome measures were divided into primary and secondary outcomes for the purposes of the review.

Studies were considered suitable for inclusion in this review if they evaluated any of the following outcome measures.

Primary outcomes

1. Number of complications during surgery.

2. Proportion of women with amenorrhoea (no periods) postoperatively, at 12 months or less and at 24 months or longer.

3. Proportion of women requiring or requesting further surgical therapy during follow-up.

Secondary outcomes

4. Endometrial outcome measures
a. Endometrial thickness after therapy (ultrasound or histological measurement or other type of measurement).
b. Cavity length of the uterus after therapy (ultrasound or hysteroscopic measurement at the time of surgery).

5. Acceptability of use of endometrial thinning agents
a. Continuation of treatment with endometrial thinning agent before endometrial resection or ablation surgery and participation levels.
b. Antiestrogenic and androgenic side effects (e.g. hot flushes, oily skin).
c. Other side effects.

6. Other surgical/post-operative outcome measures
a. Duration of operation.
b. Proportion of procedures during which operative difficulty was encountered.
c. Distension medium absorption during surgery.
d. Blood loss during surgery.
e. Postoperative blood loss (measured objectively or subjectively).
f. Proportion of women with dysmenorrhoea (painful periods) postoperatively.

7. Proportion of women expressing satisfaction with surgery, with quality of life scores on validated scales.

Search methods for identification of studies

We searched for all published and unpublished randomised controlled trials that met the inclusion criteria, with no language restriction, in consultation with the Menstrual Disorders and Subfertility Group (MDSG) Trials Search Co-ordinator.

Electronic searches

We searched the following electronic databases for trials meeting the inclusion criteria: Menstrual Disorders and Subfertility Group (MDSG) Specialised Register of Controlled Trials, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL and PsycINFO.

Other electronic sources of trials included the following.

  1. Trial registers for ongoing and registered trials.

  2. Citation indexes.

  3. Conference abstracts in the Web of Knowledge.

  4. LILACS database, for trials from the Portuguese- and Spanish-speaking world.

  5. PubMed.

  6. OpenSIGLE database and Google for grey literature.

The search strategies used for these searches are itemised in Appendix 1.

Searching other resources

We handsearched reference lists of articles retrieved by the search and contacted experts in the field to obtain additional data. We also handsearched relevant journals and conference abstracts that were not covered in the MDSG register, in liaison with the Trials Search Co-ordinator.

Data collection and analysis

Selection of studies

After an initial screen of the titles and abstracts retrieved by the search, which was conducted independently by two review authors (YHT and AL), the full texts of all potentially eligible studies were retrieved. Two review authors (YHT and AL) independently examined the full-text articles for compliance with the inclusion criteria and selected eligible studies for inclusion in this review. We corresponded with study investigators as required to clarify study eligibility. No disagreements arose as to study eligibility. The selection process was documented with a 'PRISMA' flow chart (Figure 1).

Figure 1.

Study flow diagram for 2002 to 2013 searches.

Data extraction and management

Two review authors (YHT and AL) independently extracted data from eligible studies using a data extraction form designed and pilot-tested by the review authors, in accordance with Cochrane guidelines. Disagreements were resolved by discussion. When necessary, we corresponded with the principal study investigators of eligible trials to ask for further data on trial methodology or actual trial data. When studies were the subject of multiple publications, the main trial report was used as the reference, and additional details were derived from secondary papers. Included trials were analysed for the following quality criteria and methodological details.

Trial characteristics

1. Method of randomisation.
2. Multi-centre or single-centre design.
3. Presence or absence of blinding to treatment allocation.
4. Numbers of women randomly assigned, excluded or lost to follow-up.
5. Whether an 'intention-to-treat' analysis was done.
6. The presence of a power calculation.
7. Duration, timing and location of the study.
8. Sources of funding.

Characteristics of study participants

1. Age and any other recorded characteristics of the women in the study.
2. Methods used to define heavy menstrual bleeding.
3. Exclusion criteria.
4. Type of hysteroscopic surgery performed.

Interventions used

1. Types of therapeutic agents used.
2. Dose, duration and timing of administration of therapeutic agents.

Outcomes

1. Intraoperative and postoperative outcome measures, as defined above.
2. Methods used to evaluate postoperative symptoms and patient satisfaction.
3. Methods used to measure postoperative menstrual blood loss.

We extracted data in the most accurate way possible. In cases for which results were presented in graphs, and no actual data were given, the data were extracted from the graphs.

Assessment of risk of bias in included studies

Two review authors (YHT and AL) independently assessed the included studies for risk of bias using the Cochrane 'Risk of bias assessment tool' (www.cochrane-handbook.org) to assess selection bias (random sequence generation and allocation concealment); performance bias (blinding of participants and personnel); detection bias (blinding of outcome assessors); attrition bias (incomplete outcome data); reporting bias (selective outcome reporting); and other sources of bias. Disagreements were resolved by discussion. We fully described all judgements and presented conclusions in the 'Summary of risk of bias' (Figure 2).

Figure 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Each domain was scored as follows:

  1. 'Low risk' of bias;

  2. 'High risk' of bias; or

  3. 'Unclear risk' of bias.

Scoring of each domain was assessed in conjunction with The Cochrane Collaboration's tool for assessing risk of bias.

Measures of treatment effect

For dichotomous data, we used the numbers of events in the control and intervention groups of each study to calculate Mantel-Haenszel risk ratios (RRs) with 95% confidence intervals (CIs). For continuous data, if all studies reported exactly the same outcomes, we calculated the mean difference (MD) between treatment groups. If similar outcomes were reported on different scales, we calculated the standardised mean difference (SMD). We reversed the direction of effects of individual studies, if required, to ensure consistency across trials. When ordinal data were used to measure outcomes, for example, satisfaction rates, the categories were collapsed and the data dichotomised and presented as RRs with 95% CIs. When data needed to calculate RRs or MDs were not available, we used the most detailed numerical data provided, such as test statistics or P values, to facilitate similar analyses of included studies. We compared the magnitude and direction of effects reported by studies with how they were presented in the review, while taking account of legitimate differences.

Dealing with missing data

The data were analysed as far as possible on an intention-to-treat basis. When data for eligible outcomes were missing from the publications, we attempted to contact authors of the trials for clarification. When these data could not be obtained, imputation of individual values was undertaken for the primary outcomes only. For other outcomes, only available data were analysed. When studies reported details sufficient for calculating the MD but did not provide information on the associated standard deviation (SD), the outcome was assumed to have an SD equal to the highest SD reported by other studies within the same analysis.

Assessment of heterogeneity

We considered whether the clinical and methodological characteristics of the included studies were sufficiently similar for meta-analysis to provide a clinically meaningful summary. Included studies were carefully inspected for evidence of clinical heterogeneity in characteristics of participants, interventions provided, outcomes or trial duration. We assessed statistical heterogeneity by using the I2 statistic to determine the percentage of total variation across studies that was due to heterogeneity rather than to chance. An I2 measurement greater than 50% was taken to indicate substantial heterogeneity (Higgins 2003). In cases in which extreme statistical heterogeneity could not be explained by differences between studies, the estimates were not pooled in the meta-analyses. We considered any statistical heterogeneity when interpreting the results, especially if any variation in the direction of effect was noted.

Assessment of reporting biases

In view of the difficulty of detecting and correcting for publication bias and other reporting biases, the review authors aimed to minimise their potential impact by ensuring a comprehensive search for eligible studies and by staying alert for duplication of data. We planned to assess publication bias by constructing funnel plots when 10 or more studies were identified for outcomes. However, fewer than 10 studies were identified for all outcomes, so funnel plots were not presented.

Data synthesis

Outcome data were pooled in a meta-analysis when no significant clinical heterogeneity and no evidence of any major skew were found. In studies with evidence of major skew, outcome data were not included in the main meta-analyses. However, we assessed the outcome data when evidence of major skew was noted, and we commented narratively on the results.

Dichotomous data were combined for meta-analysis with RevMan software using the Mantel-Haenszel method to estimate pooled risk ratios. For negative outcomes (e.g. proportion requiring further surgery), an increase in the risk of a particular outcome for the experimental group is displayed graphically in the meta-analyses to the right of the centre-line, and a decrease in the risk of an outcome is displayed graphically to the left of the centre-line. For positive outcomes (e.g. satisfaction with treatment), an increase in risk is shown on the reverse axis. Graphs were labelled for ease of interpretation.

Continuous data were combined for meta-analysis with RevMan software using an inverse variance method to estimate the pooled MD with 95% CI. Fixed-effect models were used in the meta-analyses, but when evidence indicated I2 > 50%, results were compared with those obtained using a random-effects model, which produced a more conservative estimation of the summary effect measure.

A 'Summary of findings' table was generated with the use of GRADEPRO software. This table evaluated the overall quality of the body of evidence for main review outcomes by using GRADE criteria (study limitations, i.e. risk of bias, consistency of effect, imprecision, indirectness and publication bias). Judgements about evidence quality (high, moderate or low) for each comparison were justified, documented and incorporated into reporting of results for the primary outcome of postoperative amenorrhoea.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis was undertaken according to:

  1. months of pretreatment with thinning agents (i.e. one month, two months, three months or more than three months of treatment); and

  2. first- or second-generation endometrial ablation.

Sensitivity analysis

We planned to perform sensitivity analyses for the following quality features.

  1. Adequate allocation concealment versus all included trials.

  2. A placebo group as the control group versus all included trials, with a no pretreatment group as the control.

  3. Dropouts < 20% versus all included trials.

  4. Pharmaceutical funding excluded.

  5. Continuous outcomes for which no estimation of variance has had to be made for pooling in the meta-analysis versus all included trials (for MD).

However, the number of identified studies was insufficient to allow performance of the above prespecified sensitivity analyses.

Overall quality of the body of evidence: Summary of findings tables

We generated Summary of findings tables with the use of GRADEPRO software. These tables evaluated the overall quality of the body of evidence for main review outcomes, when data were available, by using GRADE criteria (study limitations (i.e. risk of bias), consistency of effect, imprecision, indirectness and publication bias). Judgements about evidence quality (high, moderate or low) were justified, documented and incorporated into reporting of results for each outcome.

Results

Description of studies

Results of the search

For the initial publication of the review in 1998, a total of 17 trials were identified that evaluated the use of potential endometrial thinning agents administered before endometrial resection or rollerball/laser ablation. The review in 2001 included 12 trials that met the inclusion criteria (Sutton 1994; Fraser 1996; Garry 1996; Romer 1996; Taskin 1996; Vercellini 1996; Donnez 1997; Sorensen 1997; Sowter 1997; Lissak 1999; Rai 2000; Kriplani 2001). The updated search in 2013 located an additional 11 studies that appeared to meet the inclusion criteria, of which eight trials were included in this review (English 1998; Taskin 1998; Alborzi 2002; Kriplani 2002; Shawki 2002; Jack 2005; Bhatia 2008; Vilos 2010). Twenty studies (31 articles) were suitable for inclusion in the review and were retrieved in full text. Four studies were followed by more than one publication reporting on follow-up after treatment (Donnez 1997; Jack 2005; Romer 1996; Vercellini 1996). Five publications, which are awaiting classification, were identified as conference abstracts, and the authors were contacted (Gannon 1994; Mayonda 1994; McDonald 1994; Nagele 1995; Watermeyer 2005). No replies have been received. No trials still in progress were identified. See study tables: Characteristics of included studies, Characteristics of excluded studies, Characteristics of studies awaiting classification.

The search process is documented by a PRISMA flow chart (Figure 1).

Included studies

Study design and setting

Twenty parallel-design randomised controlled trials were included in the review. Sixteen studies were single-centre studies. One Australian study had two centres, two studies had three centres (one in Egypt and two in the United States for one study, and three centres in Italy for the other study) and one study was a multi-centre study (37 centres in 12 countries, from Europe, South Africa, the UK and Canada).

Participants

The studies included a total of 1969 women. In all, 1029 women were randomly assigned to an intervention group and 940 women to control groups. All were premenopausal women with dysfunctional uterine bleeding not responding to medical treatment. Three studies including 345 women used second-generation endometrial ablation techniques.

Interventions
Outcomes

Excluded studies

Sixteen studies were excluded from the review, for the following reasons.

  • Seven of 16 were not randomised controlled trials.

  • Nine of 16 did not meet the inclusion criteria for participants or treatment groups or outcome measures.

Risk of bias in included studies

Allocation

Eleven studies were at low risk of selection bias related to sequence generation, as they used computer randomisation or a random numbers table (Fraser 1996; Taskin 1996; Vercellini 1996; Donnez 1997; Sorensen 1997; Taskin 1998; Lissak 1999; Kriplani 2001; Kriplani 2002; Jack 2005; Bhatia 2008). The other nine studies did not describe the randomisation method used and were at unclear risk of this bias (Sutton 1994; Garry 1996; Romer 1996; Sowter 1997; English 1998; Rai 2000; Alborzi 2002; Shawki 2002; Vilos 2010).

Seven studies were at low risk of selection bias related to allocation concealment, as they used opaque sealed envelopes or a centralised telephone system (Vercellini 1996; Donnez 1997; English 1998; Rai 2000; Jack 2005; Bhatia 2008; Vilos 2010). The other 13 studies did not describe the method used and were at unclear risk of this bias.

Blinding

We did not consider that blinding was likely to have a major influence on findings for most of the surgical outcomes (duration of operation, distension medium absorption during surgery, blood loss during surgery, number of complications during surgery, proportion of women with dysmenorrhoea postoperatively) or on endometrial outcome measures (endometrial thickness, proportion of women with atrophic endometrial glands). For the outcome of proportion in which operative difficulty was encountered, potential for performance bias depended on how operative difficulty was defined, especially if the surgeon was not blinded to the interventions (Vercellini 1996). However, for the other outcomes (proportion of women with amenorrhoea postoperatively at 12 months or less, proportion of women with amenorrhoea postoperatively, at 24 months or longer, postoperative blood loss, proportion of women requiring further treatment at follow-up, proportion with optimal endometrial thinning by operator assessment, side effects, proportion of women satisfied with outcome), blinding status could potentially affect findings.

Five studies were deemed to be at low risk of performance bias, as they described a placebo identical to the intervention or treatment that was administered by staff who were not involved in the surgery or in the analysis of results (Taskin 1996; Donnez 1997; English 1998; Taskin 1998; Bhatia 2008). Thirteen studies were deemed to be at high risk of performance bias, as they were open studies that did not involve a placebo and that used different treatment protocols. Two studies were single-blinded and so were at unclear risk of this bias.

Five studies were deemed to be at low risk of detection bias, as they were double-blinded with blinding of outcome assessors (Taskin 1996; Donnez 1997; English 1998; Taskin 1998; Bhatia 2008). Twelve studies were assessed to be at high risk of detection bias, as they did not have blinding of outcome assessors, or they were open studies with different treatment protocols. Three studies did not mention whether outcome assessors were blinded and so were at unclear risk of detection bias.

Incomplete outcome data

Fifteen studies were at low risk of attrition bias, as all or most (> 95%) women who were randomly assigned were analysed. Two studies were deemed to be at high risk of attrition bias, as withdrawals were not balanced between the two groups or women were not analysed in their randomly assigned groups (Sutton 1994; Sorensen 1997). Three studies were considered to be at unclear risk of attrition bias, as reasons for dropouts were not given.

Selective reporting

Two studies were considered to be at low risk of reporting bias, as all prespecified outcomes were reported. Two studies were deemed to be at high risk of reporting bias, as they failed to report on outcomes that had been looked at (Alborzi 2002; Vilos 2010). Sixteen studies were at unclear risk of reporting bias, as prior published protocols were not available or investigators did not fully report on all prespecified outcomes.

Other potential sources of bias

Five studies were considered to be at low risk of other bias, as they were balanced at baseline with no other potential sources of bias identified. Fifteen studies were at unclear risk of bias, as it was unclear whether the groups were balanced at baseline, whether participants were not evenly distributed amongst the groups or whether the study was funded by a pharmaceutical company.

Risk of bias for all studies is summarised in Figure 2 and Figure 3.

Figure 3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Effects of interventions

See: Summary of findings for the main comparison GnRH analogues compared with placebo or no treatment for heavy menstrual bleeding; Summary of findings 2 GnRH analogues compared with danazol for heavy menstrual bleeding; Summary of findings 3 GnRH analogues compared with progestogens for heavy menstrual bleeding; Summary of findings 4 GnRH analogue compared with GnRH antagonist for heavy menstrual bleeding; Summary of findings 5 GnRH analogue compared with dilatation & curettage for heavy menstrual bleeding; Summary of findings 6 Danazol compared with no pretreatment for heavy menstrual bleeding; Summary of findings 7 Progestogens compared with no pretreatment for heavy menstrual bleeding; Summary of findings 8 Danazol compared with progestogens for heavy menstrual bleeding; Summary of findings 9 GnRHa or danazol compared with no pretreatment for heavy menstrual bleeding

1. Comparison of GnRH analogues versus placebo or no treatment

Primary outcomes
1.1 Surgical outcome measures

These included number of complications during surgery (Vercellini 1996; Donnez 1997; Sorensen 1997; Sowter 1997; English 1998).

Five studies reported on intraoperative complications (Vercellini 1996; Donnez 1997; Sorensen 1997; Sowter 1997; English 1998). Three uterine perforations were reported in the 306 women who received GnRH analogues, and one perforation was decribed in the 286 women who had received no treatment or had been given placebo (RR 1.47, 95% CI 0.35 to 6.06, 5 RCTs, 592 women, I2 = 0%) (Figure 4).

Figure 4.

Forest plot of comparison: 1 GnRH analogues versus placebo or no treatment, outcome: 1.1 Intraoperative complications—uterine perforation.

1.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

1.2 Women with amenorrhoea within 12 months (Romer 1996; Vercellini 1996; Donnez 1997; Sorensen 1997; Sowter 1997; Lissak 1999; Rai 2000).

The estimated overall proportion of women with amenorrhoea in the GnRH analogue group within 12 months was significantly greater than in the control group when data from one and two months of therapy were combined (RR 1.6, 95% CI 1.2 to 2.0, 7 RCTs, 605 women, I2 = 40%) (Figure 5). The risk difference for this outcome is 0.14 (95% CI 0.06 to 0.21), giving a number needed to treat to produce one additional case of amenorrhoea of 7.1. The number of women receiving only one month of therapy is probably too small to determine whether the duration of therapy has any effect on the proportion of women with postoperative amenorrhoea.

Figure 5.

Forest plot of comparison: 1 GnRH analogues versus placebo or no treatment, outcome: 1.2 Postoperative amenorrhoea at 12 months or less.

1.3 Women with amenorrhoea at 24 or more months

Two studies report amenorrhoea rates at 24 months or longer postsurgery (Donnez 1997; Romer 1996). The benefits of GnRH analogue administration appear to be reduced (RR 1.62, 95% CI 1.04 to 2.52, 2 RCTs, 357 women, I2 = 0%) (Figure 6). No benefit of GnRH analogue pretreatment was found in the single small study in which balloon (second-generation) ablation was used (Lissak 1999).

Figure 6.

Forest plot of comparison: 1 GnRH analogues versus placebo or no treatment, outcome: 1.3 Postoperative amenorrhoea at 24 months or longer.

1.4 Women requiring or requesting further surgical therapy during follow-up

These outcomes included the following.

1.4 Women requiring further surgery within 12 months of follow-up (Donnez 1997; Sorensen 1997; Sowter 1997; Lissak 1999).

1.5 Women requiring surgery at 24 or more months of follow-up (Donnez 1997; Romer 1996)

The number of women undergoing further surgery within 12 months of follow-up in the four studies that reported this was 17 of the 241 women followed up who were given GnRH analogues, and 18 in the 247 women given placebo or no treatment (RR 0.99, 95% CI 0.53 to 1.86, 4 RCTs, 488 women, I2 = 0%) (Figure 7). These figures do not suggest that treatment with GnRH analogues reduces the proportion of women who will subsequently request further surgery. This is confirmed by two studies in follow-up greater than 24 months (RR 1.51, 95% CI 0.97 to 2.35, 2 RCTs, 319 women, moderate quality evidence) (Figure 8).

Figure 7.

Forest plot of comparison: 1 GnRH analogues versus placebo or no treatment, outcome: 1.4 Requiring further surgery within 12 months of follow-up.

Figure 8.

Forest plot of comparison: 1 GnRH analogues versus placebo or no treatment, outcome: 1.5 Requiring further surgery ≥ 24 months of follow-up.

The duration of follow-up in these studies ranges from six months (Romer 1996; Donnez 1997) to 12 months (Sorensen 1997; Sowter 1997) or longer (Romer 1996; Donnez 1997).

Secondary outcomes
1.6 Endometrial outcome measures

These outcomes included the following.

1.6 Endometrial thickness (ultrasound) (Romer 1996).

1.7 Endometrial thickness (descriptive data) (Donnez 1997; Rai 2000).

Donnez 1997 showed a significant reduction in endometrial thickness on ultrasound measurement when GnRH analogue rather than placebo was used (1.61 mm vs 3.53 mm; estimated ratio 0.46; 95% CI 0.41 to 0.52), and this is confirmed by Rai 2000. These results are reported in the 'Other data' section of the review. A similar result was reported by Romer 1996 in the meta-analysis (MD -2.7 mm, 95% CI -3.5 to -1.9, 1 RCT, 20 women).

1.8 Proportion with atrophic endometrial glands (Romer 1996; Vercellini 1996; Donnez 1997; Rai 2000)

GnRHa pretreatment effectively induced atrophic endometrial glands when compared with either placebo or no pretreatment (RR 6.02, 95% CI 4.11 to 8.81, 4 RCTs, 483 women, I2 = 68%, moderate quality evidence).

1.9 Proportion with optimal endometrial thinning by operator assessment (Romer 1996)

In one small study (Romer 1996), nine out of ten women given triptorelin had optimal endometrial preparation (as assessed by the operating surgeon) compared with six out of ten who had received no treatment, although this was not a significant difference (RR 6.0, 95% CI 0.87 to 41.21, 1 RCT, 20 women).

1.10 Acceptability of use of endometrial thinning agents

Outcomes assessed included the following.

1.10 Side effects (Donnez 1997; English 1998). However, English 1998 reported side effects only for women receiving GnRH analogues. Donnez 1997 reported that a significantly higher proportion of women receiving GnRH analogues experienced hot flushes, sweating and headaches than those receiving placebo, although only one woman withdrew (out of 174) because of treatment side effects (headache).

1.11 Other surgical/postoperative outcome measures

These included the following.

1.11 Duration of operation (minutes) (Taskin 1996; Vercellini 1996; Donnez 1997; Sorensen 1997; English 1998; Taskin 1998).

The duration of surgery (in minutes) was shorter in the GnRH analogue (GnRHa) group in all included studies that evaluated this outcome, although this finding did not reach significance in two small studies (Vercellini 1996; English 1998). In two other studies included in the meta-analysis (Sorensen 1997; English 1998), standard deviations were not available and estimates of variation were imputed, with the highest SD value from other included studies used in the analysis. The data reported in Donnez 1997 as a geometric range have been converted and an SD found. The difference in mean operating times in the meta-analysis varied from two minutes (Vercellini 1996) to 14 minutes (Sorensen 1997). When the data for all studies were combined, the pooled estimate for the mean difference is -3.5 minutes (95% CI -4.7 to -2.3, 5 RCTs, 156 women, I2 = 72%) in favour of GnRH analogues. In these studies, the mean duration of operation in the GnRHa group ranged from 12.1 to 32.8 minutes, and the mean duration of operation in the placebo group ranged from 14.6 to 46.4 minutes, thus showing that the studies had significant heterogeneity and that the mean difference of -3.5 minutes may not be clinically significant.

1.12 Procedures in which operative difficulty was encountered (Vercellini 1996; Donnez 1997).

The degree of operative difficulty was graded by two studies either as the proportion of operations reported as more difficult than usual (Donnez 1997) or on an ordinal scale from none to severe difficulty (Vercellini 1996). The data have been combined as the proportion of procedures in which some difficulty or more difficulty than usual was encountered. In these studies, GnRH analogues significantly reduced the proportion of procedures in which difficulty was encountered (RR 0.32, 95% CI 0.22 to 0.46, 2 RCTs, 415 women, I2 = 4%).

1.13 Procedures with distension medium absorption during surgery (Taskin 1996; Vercellini 1996; Sorensen 1997; English 1998).

The volume of distension medium absorbed was lower in the GnRH analogue group in the studies that measured this outcome, although the differences were relatively small (MD -154.5, 95% CI -211.4 to -97, 4 RCTs, 137 women, I2 = 0%). These differences reached statistical significance in three of the studies, although the authors commented that the differences probably were not clinically significant. Only in one study (Sorensen 1997) did a woman suffer from fluid overload, and this occurred in the no treatment group. Standard deviations were imputed for two studies as above, as these were not available from the publications or from the study authors.

1.14 Procedures with distension medium absorption during surgery (descriptive data) (Donnez 1997).

One study (Donnez 1997) reported distension medium absorption during surgery as median fluid absorption with no range, and this is reported in the 'Other data' section of this review.

From these studies, it seems likely that treatment with GnRH analogues will reduce distension medium absorption, although the treatment effect is small.

1.15 Procedures with postoperative blood loss (Romer 1996; Donnez 1997; Sorensen 1997; Sowter 1997).

No objective measures were used to assess menstrual blood loss postoperatively. The combined estimate of four studies suggested that women were more likely to report reduced menstrual loss and were less likely to report moderate or heavy menstrual bleeding if given GnRH analogues (RR 0.74, 95% CI 0.59 to 0.92, 4 RCTs, 480 women, I2 = 39%, moderate quality evidence).

1.16 Women with dysmenorrhoea (painful periods) postoperatively (Donnez 1997; Sorensen 1997; Sowter 1997).

Dysmenorrhoea was reduced by surgery in both treated and untreated women in the three studies that recorded this (Donnez 1997; Sorensen 1997; Sowter 1997). Data were combined in meta-analysis for two studies as the proportion of women experiencing any postoperative dysmenorrhoea (Sorensen 1997; Sowter 1997). This shows a significant difference in favour of GnRH analogues (RR 0.59, 95% CI 0.40 to 0.87, 2 RCTs, 133 women, I2 = 0%). However, these data should be treated cautiously, as by far the largest study (Donnez 1997) showed no difference in postoperative pain scores between women given GnRH analogues and those given placebo, although no actual patient data were reported. It is likely that any reduction in postoperative dysmenorrhoea is small.

1.17 Quality of life

The outcome assessed was proportion of women satisfied with outcome (Vercellini 1996; Donnez 1997; Sorensen 1997; Sowter 1997; Lissak 1999; Rai 2000). Participant satisfaction was high in both groups. When data were combined, no difference was seen between women treated with GnRH analogues and those given no treatment or placebo (RR 0.99, 95% CI 0.93 to 1.05, 6 RCTs, 599 women, I2 = 11%).

2. Comparison of GnRH analogues with danazol

Primary outcomes
2.1 Surgical outcome measures

None of the studies have described any intraoperative complications. It is not possible to comment on whether GnRH analogues or danazol is associated with a lower complication rate during surgery.

2.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

2.2 Women with amenorrhoea within 12 months (Fraser 1996; Garry 1996; Romer 1996; Rai 2000; Shawki 2002).

2.3 Women with amenorrhoea at 24 or more months (Romer 1996).

The proportion of women experiencing postoperative amenorrhoea after two months of GnRH analogues was significantly higher in one study (Garry 1996), significantly lower in another study (Rai 2000) and not significantly different in the other three studies, regardless of duration of pretreatment (Fraser 1996; Romer 1996; Shawki 2002). Sutton 1994 assessed only whether there had been an overall improvement in menstrual symptoms at three months. When data are combined, no significant difference in the postoperative amenorrhoea rate is apparent when GnRH analogues rather than danazol are used. For postoperative amenorrhoea at 12 months or less, the RR is 1.17 (95% CI 0.90 to 1.52, 5 RCTs, 340 women, I2 = 52%) (Figure 9). In Romer 1996, an identical amenorrhoea rate was reported in the two groups at three-year follow up (RR 1.00, 95% CI 0.49 to 2.05, 1 RCT, 20 women) (Analysis 2.3).

Figure 9.

Forest plot of comparison: 2 GnRH analogues versus danazol, outcome: 2.2 Postoperative amenorrhoea at 12 months or less.

2.4 Women requiring or requesting further surgical therapy during follow-up

This included women requiring further surgery during follow-up (Sutton 1994; Romer 1996). No evidence suggested a significant difference between the number of women requiring further surgery in the small Romer 1996 study, which was included in the meta-analysis (RR 0.33, 95% CI 0.02 to 7.32, 1 RCT, 20 women). Follow-up ranged from three months (Sutton 1994) to 24 months (Romer 1996).

Secondary outcomes
2.5 Endometrial outcome measures

Outcomes included the following.

2.5 Endometrial thickness (ultrasound) (Romer 1996).

2.6 Endometrial thickness (descriptive data) (Sutton 1994; Garry 1996; Rai 2000).

Median endometrial thickness on ultrasound measurement or histology after GnRH analogues compared with danazol therapy was significantly lower in one study (Sutton 1994) and similar in the other three studies (Garry 1996; Romer 1996; Rai 2000). Combining data on endometrial thickness was not possible, so data have been presented as medians with no useable measure of variance in the 'Other data' section of the review for three of the four studies.

2.7 Women with atrophic endometrial glands (Sutton 1994; Garry 1996; Romer 1996; Rai 2000)

Four studies compared the proportion of women in each group with atrophic endometrial glands on histology. Two of the four studies reported that the proportion of women with atrophic or inactive endometrial glands or stroma was greater in the GnRH analogue–treated group (Sutton 1994; Garry 1996). These studies reported their histology slightly differently. Garry 1996 makes no distinction between inactive and atrophic glands, and data have been combined using the proportion of women with inactive glands for one study (Garry 1996) and with atrophic glands for the other (Sutton 1994). Garry 1996 also makes no distinction between one and two months of therapy in reporting endometrial histology. Two other studies reported similar proportions in each group. When data are combined, the overall proportion of women with atrophic or inactive glands appears to be greatest in women treated with GnRH analogues (RR 1.28, 95% CI 1.03 to 1.58, 4 RCTs, 318 women, I2 = 43%).

2.8 Women with satisfactory thinning at hysteroscopy (Garry 1996; Romer 1996)

2.9 Women with optimal endometrial thinning by operator assessment (Romer 1996)

2.10 Women with complete atrophy at hysteroscopy

A subjective hysteroscopic assessment of either the degree of endometrial atrophy (Sutton 1994; Fraser 1996) or how satisfactory endometrial suppression appeared (Sutton 1994; Garry 1996; Romer 1996) was made in four studies. When reported data have been combined for the former outcome, the meta-analysis suggests that on hysteroscopic assessment, the proportion of women with complete endometrial atrophy or with a very thin endometrium is slightly greater with GnRH analogues than with danazol (RR 1.84, 95% CI 1.23 to 2.75, 2 RCTs, 142 women, I2 = 0%). The proportion of cases in which satisfactory suppression is obtained is slightly greater with GnRH analogues, but the differences are likely to be minimal (RR 1.09, 95% CI 1.00 to 1.19, 2 RCTs, 165 women, I2 = 0%). An identical rate of optimal endometrial thinning was noted in the two groups (RR 1.00, 95% CI 0.49 to 2.05, 1 RCT, 20 women).

2.11 Acceptability of use of endometrial thinning agents

Outcomes included the following.

2.11 Side effects (Sutton 1994; Fraser 1996; Garry 1996). The spectrum of side effects reported reflects the different modes of action of these two drugs, and no study used a psychometric scale to evaluate the impact of these side effects on "quality of life", so only the proportion of women experiencing each particular side effect can be reported. Garry 1996 has not reported side effects in sufficient detail for results to be combined with those of any other study, but the proportion of women reporting various side effects is shown for Fraser 1996 and Sutton 1994. Hot flushes, reduced libido, depression, headaches and vaginal dryness were significantly more common with GnRHa.

A greater proportion of women withdrew from the studies because of side effects from danazol (RR 0.08 favouring GnRHa, 95% CI 0.01 to 0.59, 3 RCTs, 285 women, I2 = 0%). However, this apparent difference should be treated with caution because once inserted, GnRH analogues cannot be removed, ensuring complete compliance unless side effects within the first few weeks of therapy are intolerable.

2.12 Weight gain was more common with danazol.

2.13 Other surgical/postoperative outcome measures

Outcomes included the following.

2.13 Duration of operation (Fraser 1996; Garry 1996; English 1998; Shawki 2002).

Operating time was not significantly different in three studies (Sutton 1994—no data given; Fraser 1996—data obtained from author; and English 1998) and was significantly reduced in favour of GnRH analogues in two other studies (Garry 1996; Shawki 2002). Data were combined for four studies (Fraser 1996; Garry 1996; English 1998; Shawki 2002) for one, two and three months of therapy and showed a reduction in operating time in favour of GnRH analogues (MD -5.02, 95% CI -6.25 to -3.78, 4 RCTs, 325 women, I2 = 42%). This result should be treated cautiously, as results were inconsistent between trials, and data were not available from Sutton 1994.

2.14 Procedures in which operative difficulty was encountered (Fraser 1996).

In the two studies in which a subjective assessment of operative difficulty was made (Sutton 1994; Fraser 1996), no significant difference was noted in the proportion of women in whom operative difficulty was encountered. Complete data were available only for Fraser 1996 (RR 0.68, 95% CI 0.35 to 1.51, 1 RCT, 56 women) for GnRHa compared with danazol (i.e. a non-significant difference).

2.15 Distension medium absorption during surgery (Fraser 1996; Garry 1996; English 1998).

A similar pattern was seen for distension medium absorption, although the reported data were heterogeneous, and no difference in fluid absorption is apparent when data are combined (MD -22.38, 95% CI -87.07 to 42.32, 3 RCTs, 223 women, I2 = 78%). A more useful measure might be the proportion of women with any distension medium absorption in each group. This has been reported by Garry 1996, where 54% of women who had received danazol had distension medium absorption compared with 32% of women who had received GnRH analogues.

2.16 Postoperative blood loss measured objectively (Fraser 1996) or assessed subjectively (Sutton 1994; Garry 1996; Romer 1996).

The only study that objectively measured postoperative blood loss (Fraser 1996) showed at both three and six months postoperatively a significantly lower mean loss and a shorter duration of bleeding in women given GnRH analogues compared with those given danazol (MD -6.40, 95% CI -9.19 to -3.61, 1 RCT, 56 women). However, it is a small study, and results should be interpreted cautiously.

2.17 Quality of life

The outcome assessed is proportion satisfied with outcome (Sutton 1994; Garry 1996; Rai 2000). Participant satisfaction after endometrial resection was high in all these studies, and although one study reported a significant difference between groups, the overall combined estimate was non-significant (RR 0.94, 95% CI 0.82 to 1.08, 3 RCTs, 286 women, I2 = 61%) (Analysis 2.17).

3. Comparison of GnRH analogues with progestogens

Primary outcomes
3.1 Surgical outcome measures

Intraoperative complications were not assessed in these studies.

3.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

3.2 Women with amenorrhoea within 12 months (Romer 1996; Rai 2000; Shawki 2002).

3.3 Women with amenorrhoea at 24 or more months (Romer 1996).

No evidence of a significant difference in postoperative amenorrhoea rates up to 12 months after surgery was found between groups either in individual trial results or overall (combining one, two and three months of pretreatment with GnRHa), although the summary effect estimate was close to the cutoff P value for significance overall, in favour of GnRHa pretreatment (RR 1.58, 95% CI 0.96 to 2.61, 3 RCTs, 146 women, I2 = 0%). No evidence of a significant difference in these amenorrhoea rates was noted at longer follow-up, two years after surgery, in a very small trial (Romer 1996) (RR 3.00, 95% CI 0.79 to 11.44, 1 RCT, 20 women) (Figure 10).

Figure 10.

Forest plot of comparison: 3 GnRH analogues versus progestogens, outcome: 3.2 Postoperative amenorrhoea at 12 months or less.

3.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery during follow-up (Romer 1996). No evidence of differences between groups was found in the requirement for further surgical intervention up to two years after surgery (RR 0.33, 95% CI 0.02 to 7.32, 1 RCT, 20 women) (Analysis 3.4).

Secondary outcomes
3.5 Endometrial outcome measures

Outcomes included the following.

3.5 Endometrial thickness (ultrasound) (Romer 1996).

3.6 Endometrial thickness (descriptive data) (Rai 2000).

Romer 1996 reported a significant difference in the thickness of the endometrium (MD -2.8 mm, 95% CI -3.6 to -2.0, 1 RCT, 20 women), but this was not confirmed by another small trial, for which results are described in narrative form with median and range (Rai 2000 in 'Other data' section of the review).

3.7 Women with atrophic endometrium on histology (Romer 1996; Rai 2000)

3.8 Women with optimal endometrial thinning by operator assessment (Romer 1996).

GnRHa pretreatment was associated with a greater likelihood of endometrial atrophy (as measured by histology) compared with progestogen pretreatment when the effects of one and two months of treatment were combined (RR 2.25, 95% CI 1.13 to 4.49, 2 RCTs, 70 women, I2 = 0%). No evidence indicated a significant difference in the proportion with optimal endometrial thinning as measured by operator assessment (RR 3.00, 95% CI 0.79 to 11.44, 1 RCT, 20 women).

Acceptability of use of endometrial thinning agents

Side effects were not measured in these studies.

3.9 Other surgical/postoperative outcome measures

Outcomes included duration of operation (Shawki 2002). Proportion of procedures in which operative difficulty was encountered, distension medium absorption during surgery, postoperative blood loss and postoperative dysmenorrhoea were not looked at in these studies.

The duration of surgery was significantly shorter after GnRHa pretreatment for one and three months before surgery in Shawki 2002 (one month MD -15 minutes, 95% CI -19.44 to -10.56, 1 RCT, 50 women; three months MD -17 minutes, 95% CI -20.90 to -13.10, 1 RCT, 53 women).

3.10 Quality of life

Outcomes assessed include proportion of women satisfied with outcome (Rai 2000). Participants were less likely to be satisfied (pleased or very pleased) with the outcome if they had undergone GnRHa pretreatment when compared with progestogen pretreatment (RR 0.8, 95% CI 0.65 to 0.99, 1 RCT, 20 women).

4. Comparison of GnRH analogues with GnRH antagonists

Primary outcomes
4.1 Surgical outcome measures

Intraoperative complications were assessed (Bhatia 2008).

The same number of intraoperative complications was reported for the two groups (Analysis 4.1).

4.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

4.2 Women with amenorrhoea within 12 months (Bhatia 2008).

4.3 Women with amenorrhoea and/or hypomenorrhoea at 12 months or less (Bhatia 2008).

No statistically significant difference was noted between the two groups in terms of the proportion of women with amenorrhoea within 12 months of treatment (RR 0.95, 95% CI 0.79 to 1.14, 1 RCT, 100 women) (Analysis 4.3) or the proportion of women with amenorrhoea and/or hypomenorrhoea within 12 months of treatment (RR 0.94, 95% CI 0.83 to 1.06, 1 RCT, 100 women) (Analysis 4.3).

4.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery during follow-up (Bhatia 2008). No statistically significant difference was noted between the two groups in terms of proportion of women requiring further surgery at six-month follow-up (RR 1.33, 95% CI 0.31 to 5.65, 1 RCT, 100 women) (Analysis 4.4).

Secondary Outcomes
4.5 Endometrial outcome measures

Outcomes included endometrial thickness (ultrasound) (Bhatia 2008). Proportion with atrophic endometrium on histology and proportion with optimal endometrial thinning by operator assessment were not assessed.

The endometrium was significantly thinner after GnRH analogue pretreatment than after GnRH antagonist pretreatment, both when hyperplasia was included (MD 0.54 mm, 95% CI 0.17 to 0.91, 1 RCT, 88 women) and when hyperplasia was excluded (MD 0.53 mm, 95% CI 0.22 to 0.84, 1 RCT, 83 women).

4.6 Acceptability of use of endometrial thinning agents

Side effects were measured in one study (Bhatia 2008). More episodes of hot flushes were reported in the GnRH analogue group (15/50) than in the GnRH anatagonist group (6/50), and this was statistically significant (RR 0.40, 95% CI 0.17 to 0.95, 1 RCT, 100 women). Differences between the groups in terms of skin irritation and non-specific adverse effects were not statistically significant.

4.7 Other surgical/postoperative outcomes measures

Outcomes included the following.

4.7 Duration of operation (Bhatia 2008).

No significant difference was noted in the duration of operation between GnRH analogues compared with GnRH antagonists (MD 2.26, 95% CI -1.11 to 5.63, 1 RCT, 100 women).

4.8 Distension medium absorption during surgery (Bhatia 2008).

Distension medium absorption during surgery was provided as median values with a range, and no significant difference was noted between the two treatments.

4.9 Procedures with good operative view (Bhatia 2008).

No significant difference was noted in the proportion of women with a good operative view between the two treatments (RR 0.96, 95% CI 0.89 to 1.04, 1 RCT, 100 women).

4.10 Women with postoperative dysmenorrhoea (Bhatia 2008). Postoperative blood loss was not assessed.

The same proportion with postoperative dysmenorrhoea was reported in both groups.

4.11 Quality of life

Outcomes assessed include proportion of women satisfied with outcome (Bhatia 2008). No statistically significant difference was noted between the two groups in terms of the proportion of women satisfied with outcome (RR 0.93, 95% CI 0.81 to 1.07, 1 RCT, 100 women). Follow-up was provided at six months after surgery.

5. Comparison of GnRH analogues with dilatation & curettage

Primary outcomes
5.1 Surgical outcome measures

Intraoperative complications were not looked at in these studies.

5.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

5.2 Women with normal bleeding or less at 12 months (Vilos 2010).

5.3 Women with amenorrhoea at 12 months or less (Vilos 2010).

Evidence shows that an increased number of women in the GnRH analogue group had amenorrhoea at 12 months or less (RR 20.13, 95% CI 1.21 to 333.63, 1 RCT, 92 women, low quality evidence) (Analysis 5.3). No evidence of a significant difference between groups was found for the proportion of women with normal bleeding or less at 12 months (RR 1.03, 95% CI 0.91 to 1.16, 1 RCT, 92 women).

5.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery within 12 months of follow-up (Vilos 2010). No evidence of a significant difference was found between groups for this outcome (RR 0.75, 95% CI 0.18 to 3.18, 1 RCT, 100 women) (Analysis 5.4).

Secondary outcomes
5.5 Endometrial outcome measures

Endometrial thickness, proportion with atrophic endometrium on histology and proportion with optimal endometrial thinning by operator assessment were not assessed.

5.6 Acceptability of use of endometrial thinning agents

The proportion of women with premenstrual syndrome (PMS) symptoms was assessed, and no evidence of a statistically significant difference between the two groups was found in Vilos 2010 (RR 1.03, 95% CI 0.67 to 1.58, 1 RCT, 93 women).

5.7 Other surgical/postoperative outcome measures

Outcomes assessed included proportion with postoperative dysmenorrhoea (Vilos 2010). Duration of operation, distension medium absorption during surgery postoperative blood loss and proportion of procedures with good operative view were not looked at.

No statistically significant difference was reported between the two pretreatment groups in terms of the proportion of women with postoperative dysmenorrhoea (RR 1.00, 95% CI 0.66 to 1.52, 1 RCT, 92 women).

5.8 Quality of life

Outcomes assessed include proportion of women satisfied with outcome (Vilos 2010). No statistically significant difference in this outcome was observed between the two groups (RR 0.94, 95% CI 0.83 to 1.05, 1 RCT, 92 women).

6. Comparison of danazol with placebo or no pretreatment

Primary outcomes
6.1 Surgical outcome measures

Intraoperative complications were not assessed in these studies.

6.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

6.2 Women with amenorrhoea at 12 months or less (Rai 2000; Kriplani 2002).

No evidence of a significant difference was noted between groups for proportion of women with amenorrhoea at 12 months or less (RR 1.05, 95% CI 0.80 to 1.39, 2 RCTs, 179 women, I2 = 14%) (Figure 11).

Figure 11.

Forest plot of comparison: 6 Danazol versus no pretreatment, outcome: 6.2 Postoperative amenorrhoea at 12 months or less.

6.3 Women with amenorrhoea at two years or more (Romer 1996; Kriplani 2002).

Findings were similar for proportions of women with amenorrhoea at two years or more (RR 1.27, 95% CI 0.79 to 2.04, 2 RCTs, 81 women, I2 = 55%) (Figure 12).

Figure 12.

Forest plot of comparison: 6 Danazol versus no pretreatment, outcome: 6.3 Postoperative amenorrhoea at two years or more.

6.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery at two years or more (Romer 1996; Kriplani 2002). No evidence revealed a significant difference between the two groups for proportion of women requiring further surgery at two years or more (RR 1.37, 95% CI 0.28 to 6.69, 2 RCTs, 152 women, I2 = 0%) (Figure 13).

Figure 13.

Forest plot of comparison: 6 Danazol versus no pretreatment, outcome: 6.4 Requiring further surgery at two years or more.

Secondary outcomes
6.5 Endometrial outcome measures

6.5 Endometrial thickness (by ultrasound) (Romer 1996; Kriplani 2002).

6.6 Endometrial thickness (descriptive data) (Rai 2000).

All studies separately reported a significantly thinner endometrium after danazol pretreatment compared with no pretreatment (trials unable to be pooled because of extreme heterogeneity and non-normal distribution of data) with differences between groups ranging between 0.5 mm and 8 mm.

6.7 Women with atrophic endometrial glands (Romer 1996; Rai 2000)

Participants taking danazol pretreatment were more likely to have atrophic endometrial glands (RR 3.15, 95% CI 1.46 to 6.80, 2 RCTs, 70 women, I2 = 63%).

6.8 Women with optimal endometrial thinning by operator assessment (Romer 1996)

A trend towards greater optimal endometrial thinning (by operator assessment) was observed in this one small study (RR 6.00, 95% CI 0.87 to 41.21, 1 RCT, 20 women).

6.9 Acceptability of use of endometrial thinning agents

Side effects were not assessed.

6.10 Other surgical/postoperative outcome measures

Outcomes assessed included the following.

6.10 Duration of operation (English 1998; Kriplani 2002).

With overall results pooling, duration of surgery was significantly shorter with danazol pretreatment but with significant heterogeneity (MD -6.84 minutes, 95% CI -7.97 to -5.72, 2 RCTs, 157 women, I2 = 94%).

6.11 Distension medium absorption during surgery (English 1998; Kriplani 2002).

Overall benefit was noted for danazol in fluid absorption during surgery (MD-109.45 mL, 95% CI -193.3 to -25.7, 2 RCTs, 156 women, I2 = 0%).

Proportion of procedures with good operative view, proportion with postoperative dysmenorrhoea and postoperative blood loss were not assessed.

6.12 Quality of life

Outcomes assessed include proportion of women satisfied with outcome (Rai 2000). No statistically significant difference in this outcome was noted between the two groups (RR 1.00, 95% CI 0.93 to 1.08, 1 RCT, 50 women).

7. Comparison of progestogens with placebo or no pretreatment

Primary outcomes
7.1 Surgical outcome measures

Intraoperative complications were not assessed in the included studies.

7.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

7.2 Women with amenorrhoea at 12 months or less (Rai 2000).

7.3 Women with postoperative amenorrhoea at two to four years (Romer 1996; Kriplani 2001).

No evidence of a significant difference was seen between groups in either of the above outcomes—proportion of women with amenorrhoea at 12 months or less (RR 0.54, 95% CI 0.26 to 1.12, 1 RCT, 50 women) (Analysis 7.3) and proportion of women with amenorrhoea at two to four years (RR 0.75, 95% CI 0.36 to 1.54, 2 RCTs, 70 women, I2 = 0%) (Figure 14).

Figure 14.

Forest plot of comparison: 7 Progestogens versus no pretreatment, outcome: 7.3 Postoperative amenorrhoea at two to four years.

7.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery at two to four years of follow-up (Romer 1996; Kriplani 2001). No evidence of a significant difference between groups was noted for this outcome (RR 3.00, 95% CI 0.50 to 17.95, 2 RCTs, 70 women, I2 = 0.23) (Figure 15).

Figure 15.

Forest plot of comparison: 7 Progestogens versus no pretreatment, outcome: 7.4 Requiring further surgery at two to four years.

Secondary outcomes
7.5 Endometrial outcome measures

7.5 Endometrial thickness (by ultrasound) (Romer 1996).

7.6 Endometrial thickness (descriptive data) (Rai 2000).

7.7 Women with atrophic endometrial glands (Romer 1996; Rai 2000).

7.8 Women with optimal endometrial thinning by operator assessment (Romer 1996).

These were assessed, and no evidence was found of a significant difference in any of the endometrial outcome measures.

7.9 Acceptability of use of endometrial thinning agents

Side effects were not assessed.

7.10 Other surgical/postoperative outcomes

None of the included studies assessed any of these outcomes.

7.11 Quality of life

Outcome assessed was women satisfied with outcome (Rai 2000). No statistically significant difference in this outcome was noted between the two groups (RR 1.00, 95% CI 0.93 to 1.08, 1 RCT, 50 women).

8. Comparison of danazol with progestogens

Primary outcomes
8.1 Surgical outcome measures

Intraoperative complications were not looked at.

8.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

8.2 Women with amenorrhoea at 12 months or less (Rai 2000; Shawki 2002).

Participants having danazol pretreatment were more likely to have amenorrhoea up to 12 months after surgery (RR 1.89, 95% CI 1.12 to 3.18, 2 RCTs, 103 women, I2 = 16%) (Figure 16).

Figure 16.

Forest plot of comparison: 8 Danazol versus progestogens, outcome: 8.2 Postoperative amenorrhoea at 12 months or less.

8.3 Women with postoperative amenorrhoea at two years (Romer 1996).

One small study showed no evidence of a significant difference between the two groups for proportion of women with postoperative amenorrhoea at two years (RR 3.00, 95% CI 0.79 to 11.44, 1 RCT, 20 women) (Figure 17).

Figure 17.

Forest plot of comparison: 9 GnRHa or danazol versus no pretreatment, outcome: 9.2 Postoperative amenorrhoea at 12 months or less.

8.4 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the outcome of proportion of women requiring further surgery at two years of follow-up (Romer 1996). No evidence of a significant difference between groups for this outcome was found in one small study (RR 1.00, 95% CI 0.07 to 13.87, 1 RCT, 20 women) (Analysis 8.3).

Secondary outcomes
8.5 Endometrial outcome measures

8.5 Endometrial thickness (by ultrasound) (Romer 1996).

8.6 Endometrial thickness (descriptive data) (Rai 2000).

The endometrium was significantly thinner after danazol pretreatment when compared with after progestogen pretreatment in one very small study (Romer 1996) (MD -3.00, 95% CI -3.76 to -2.24, 1 RCT, 20 women), but this was not confirmed by another study (Rai 2000), with data showing no significant difference (in the 'Other data' section of the review).

8.7 Women with atrophic endometrial glands (Romer 1996; Rai 2000)

Participants given danazol pretreatment were more likely to have atrophic endometrial glands than those given progestogen pretreatment (RR 2.50, 95% CI 1.27 to 4.92, 2 RCTs, 70 women, I2 = 0%).

8.8 Women with optimal endometrial thinning by operator assessment (Romer 1996)

No evidence indicated that optimal endometrial thinning rates differed between groups in one very small trial (RR 3.00, 95% CI 0.79 to 11.44, 1 RCT, 20 women).

8.9 Acceptability of use of endometrial thinning agents

Side effects were not assessed.

8.10 Other surgical/postoperative outcomes

Outcomes assessed included duration of operation (Shawki 2002).

Distension medium absorption during surgery, proportion of procedures with good operative view, proportion with postoperative dysmenorrhoea and postoperative blood loss were not assessed.

Duration of surgery was significantly shorter after danazol pretreatment for three months than after progestogen pretreatment (MD -11.00, 95% CI -14.59 to -7.41, 1 RCT, 53 women).

8.11 Quality of life

Outcomes assessed include proportion of women satisfied with outcome (Rai 2000). No statistically significant difference in this outcome was noted between the two groups (RR 1.00, 95% CI 0.93 to 1.08, 1 RCT, 50 women).

9. Comparison of GnRH analogues or danazol with placebo or no pretreatment

Primary outcomes
9.1 Surgical outcome measures

Intraoperative complications were not looked at.

9.2 Effectiveness of endometrial thinning agents on outcome of endometrial resection or ablation as a surgical treatment for menorrhagia

These outcomes included the following.

9.2 Women with amenorrhoea at 12 months or less (Alborzi 2002; Jack 2005).

9.3 Women with postoperative amenorrhoea at five years (Jack 2005).

No significant difference was noted in the proportion of women with postoperative amenorrhoea at 12 months or less (RR 1.13, 95% CI 0.89 to 1.43, 2 RCTs, 280 women, I2 = 0%) (Analysis 9.3) or in the proportion with postoperative amenorrhoea at five years (RR 1.05, 95% CI 0.91 to 1.20, 1 RCT, 154 women) between women given a GnRH analogue or danazol versus no pretreatment (Analysis 9.3).

9.4 Improvement at five years (Jack 2005)

No difference between groups was observed in the proportion of women with bleeding improvement (light bleeding/amenorrhoea) at five years (RR 1.01, 95% CI 0.98 to 1.05, 1 RCT, 154 women).

9.5 Proportion of women requiring or requesting further surgical therapy during follow-up

This included the following.

9.5 Women requiring further surgery at one-year follow-up (Jack 2005).

9.6 Women requiring further surgery at five-year follow-up (Jack 2005). No evidence showed a significant difference between groups for proportion of women requiring further surgery at one year (RR 0.97, 95% CI 0.06 to 15.29, 1 RCT, 197 women) (Analysis 9.5) or for proportion of women requiring further surgery at five years (RR 0.73, 95% CI 0.32 to 1.65, 1 RCT, 197 women) (Analysis 9.6).

Secondary outcomes
9.7 Endometrial outcome measures

Endometrial thickness (by ultrasound) (Jack 2005) was assessed. Proportion with atrophic endometrial glands and proportion with optimal endometrial thinning by operator assessment were not assessed.

The endometrium was significantly thinner after GnRH analogues or danazol pretreatment than with no pretreatment (MD -1.9 mm, 95% CI -2.54 to -1.26, 1 RCT, 197 women).

9.8 Acceptability of use of endometrial thinning agents

Side effects, including hot flushes, nausea, rashes or itch and weight gain, were assessed (Jack 2005). Side effects were significantly more likely with endometrial thinning treatment with danazol or GnRH analogue than with no pretreatment.

9.9 Other surgical/postoperative outcome measures

Outcomes assessed included the following.

9.9 Duration of operation (Alborzi 2002; Jack 2005).

Duration of surgery was compared in subgroups according to whether first- or second-generation ablation was undertaken. In the trial in which women underwent TCRE and rollerball surgery, participants given pretreatment with GnRH analogues or danazol had significantly shorter surgery than those not given pretreatment (MD -15 minutes, 95% CI -16.87 to -13.13, 1 RCT, 90 women). In the trial in which women underwent microwave surgery, no significant difference in procedure time was noted between groups (MD 0.40 minutes, 95% CI -0.89 to 1.69, 1 RCT, 197 women).

9.10 Postoperative blood loss at five years (Jack 2005).

No difference was noted between groups in the proportion of women with postoperative menorrhagia (RR 0.81, 95% CI 0.22 to 2.93, 2 RCTs, 280 women, I2 = 0%).

Distension medium absorption during surgery, proportion of procedures with good operative view and proportion with postoperative dysmenorrhoea were not assessed.

9.11 Quality of life

Outcomes assessed included the following.

9.11 Women satisfied with outcome at 12 months (Jack 2005).

9.12 Women satisfied with outcome at five years of follow-up (Jack 2005).

No statistically significant difference in this outcome was observed between the two groups at 12 months (RR 0.96, 95% CI 0.87 to 1.05, 1 RCT, 188 women) or at five years (RR 0.94, 95% CI 0.83 to 1.06, 1 RCT, 154 women).

Sensitivity Analysis

Sensitivity analysis could not be performed because insufficient studies were entered into the meta-analyses. As the review is updated with additional trials, it may be possible in the future to assess the robustness of the results by sensitivity analysis.

Discussion

Summary of main results

We are unable to determine whether the use of pre-operative endometrial thinning agents reduces the risk of complications, as the complication risk, specifically, uterine perforation, is so uncommon that further randomised trials would have to be of an unfeasibly large size to demonstrate any small differences. Useful information can be drawn from the MISTLETOE study, the largest published audit of more than 10,000 endometrial resections (Overton 1997), in which the use of endometrial thinning agents was not associated with any reduction in complication rates.

One possible advantage of being able to operate with a consistently thinned endometrium is a more reliable postoperative result. Combining the results of all postoperative outcome measures is difficult because several methods of evaluating postoperative results have been used, and some studies have not reported non-significant data in any detail. However, postoperative amenorrhoea is one measure of postoperative outcome that has been consistently reported by different studies, irrespective of the method used to evaluate postoperative menstrual loss. Good evidence suggests that administration of GnRH analogues before surgery results in higher rates of postoperative amenorrhoea at 12 and 24 months compared with placebo or no treatment. Whether this difference is maintained after two or three years is uncertain, as few women taking part in randomised trials have been followed up for this length of time.

Analysis of the randomised controlled trials included in this review shows that use of GnRH analogues before endometrial destruction surgery results in slightly shorter operating time and a reduction in intraoperative distension medium absorption. The reduction in operating time is relatively small in terms of the effect it may have on the cost of the procedure (i.e. mean reduction of 3.5 minutes) and may not be clinically significant, but it is a reflection of the increased ease of surgery produced by the use of GnRH analogues. The use of GnRH analogues also more reliably results in a thinner endometrium than when surgery is timed to the immediate postmenstrual phase of the menstrual cycle.

Using GnRH analogues rather than no therapy also reduces the proportion of women who continue to experience heavy menstrual loss postoperatively. It is interesting to note that these postoperative differences are not apparent when participant satisfaction rates are compared. This may be true because the proportion of women expressing satisfaction with the surgery is high, irrespective of whether endometrial preparation is used, or because the determinants of participant satisfaction are rather more complex than just amenorrhoea alone. Similarly, the proportion of women undergoing further surgery seems to be unaffected by the use of GnRH analogues. More long-term follow up studies are needed to assess the effects of endometrial thinning with GnRH analogues on outcomes after three years.

Only published trials of small sample sizes have compared danazol with no pretreatment or placebo, as its value can be assessed only from trials that have compared it with GnRH analogues. In these comparisons, some differences are seen in favour of GnRH analogues, but their clinical significance is small. More consistent induction of endometrial atrophy is seen, although this does not appear to have any great impact on operating time, distension medium absorption, the proportion of procedures in which difficulty is encountered or participant satisfaction. GnRH analogues are shown to increase the proportion of women with postoperative amenorrhoea to a greater extent than danazol, but, as previously mentioned, the study populations for danazol are small, and it is uncertain whether further trials with larger populations will show danazol to be equivalent to GnRH analogues in increasing postoperative amenorrhoea.

Given these small differences, compliance, side effects and costs may be more important. Depot preparations of GnRH analogues are certainly associated with better patient compliance, although this is at least in part a reflection of the impossibility of removing them, whereas an oral preparation can be easily discontinued. Women should be advised that although the spectrum of side effects will vary, both GnRH analogues and danazol are likely to produce side effects in a significant proportion of women, particularly if used for two months. As GnRH analogue use appears to be beneficial, add-back estrogen therapy may have a role in enabling women to tolerate the side effects of GnRH analogues. Danazol appears to have a worse side effect profile than GnRH analogues. Evidence for the use of GnRH analogues is more substantial and is of better quality than evidence for any of the other medications that have been looked at in this review.

Cost would be one factor that would make the use of progestogens attractive, as they are significantly cheaper than both GnRH analogues and danazol. However, pretreatment with lynestrenol or MPA had no beneficial effects on thinning of the endometrium or on postoperative amenorrhoea or satisfaction rates in this review. Few data are available from randomised trials to support their use, and no trial, either unpublished or in progress, is evaluating their use. Observational studies that have included women treated with progestogens have reported disappointing effects. If progestogens are to be used at all, this should occur only within the context of a trial undertaken to formally evaluate them. No studies have investigated estrogen-progestin contraception as a pre-operative endometrial thinning agent; a future trial could be considered to evaluate this.

Only 3 studies with a total of 345 women used second-generation endometrial ablation techniques; although the study sizes were small, the results suggest that pre-operative endometrial thinning agents given before second-generation endometrial ablation techniques are performed do not significantly reduce operation time or the proportion of women with postoperative amenorrhoea. This is important, as many hospitals are moving towards these techniques for endometrial ablation rather than using the hysteroscopic techniques.

Overall completeness and applicability of evidence

Overall, the included studies addressed the review question, which was to look at effectiveness and safety of pre-operative endometrial thinning agents (GnRH agonists, danazol, estrogen-progestin contraceptives and progestogens) versus another agent or versus placebo before endometrial destruction in premenstrual women with heavy menstrual bleeding. All studies included relevant participants, interventions and outcomes. Current practice and guidelines do not recommend routine use of pre-operative endometrial thinning agents before endometrial destruction. The findings of this review show that the use of GnRH analogues and danazol before endometrial resection and ablation can improve the intrauterine operating environment and postoperative amenorrhoea at 12 and 24 months, but that their benefits in the long term are less certain, and side effects may limit their usefulness. Use of these agents should be assessed on a case-by-case basis by the clinician involved in the care of the woman. This review should inform future practice and guidelines. Insufficient studies on progestogens were identified, and no studies on the estrogen-progestin contraceptive were found; thus, there is room to further update this review in the future.

Quality of the evidence

Low-quality evidence from clinical trials shows the effects of GnRH analogues used before endometrial thinning, as outlined above, as more than half the trials included no blinding of participants or outcome assessment. Most of the trials had uncertain selection and reporting bias, as they did not report allocation concealment and had selective reporting. Moderate heterogeneity was noted across the studies, and only one trial was of good quality. No RCTs of any significant size have compared danazol with no treatment versus placebo; this reduces the quality of evidence obtained when danazol was looked at as pretreatment. Substantial heterogeneity was observed in the studies that included danazol as pretreatment. Studies looking at progestogens as pretreatment were not of significant size, and none of the studies were blinded or reported blinding; thus, evidence was of low quality. Two studies reported a pharmaceutical company as a source of funding, and this was an uncertain source of bias.

Potential biases in the review process

No potential biases were identified in the review process, as any disagreements between the two review authors (YHT and AL) involved in screening and analysing studies were resolved by discussion.

Agreements and disagreements with other studies or reviews

This updated review agrees with the previous version of the same review published in 2002 but with some additional conclusions. However, larger studies with longer-term follow-up than three years could make a difference in the results of this review in the future.

Authors' conclusions

Implications for practice

Low-quality evidence suggests that GnRH analogues or danazol given before endometrial resection and ablation has a beneficial effect on the intrauterine operating environment and on short-term postoperative outcomes. GnRH analogues may produce greater endometrial inhibition than is produced by danazol, but the clinical effects of these differences are small, and any decision regarding which agent to use may be best based on cost and side effects. The benefits of these agents for treatment outcomes in the longer term (i.e. greater than 24 months) is uncertain. Any benefit may be outweighed by the side effects of the endometrial thinning agents used, especially in the case of danazol. Few data are available from randomised trials to allow assessment of the value of progestogens, and, given the lack of supporting data from observational studies, progestogens should be used only if they are being evaluated within such a trial. Pre-operative endometrial thinning agents do not appear to be beneficial when used before second-generation endometrial ablation techniques, but further evaluation is needed. Thus, there is currently no place for routine use of endometrial thinning agents before endometrial resection and ablation.

Implications for research

Uncertainty continues regarding whether any postoperative benefits derived from the use of endometrial thinning agents are sustained with follow-up beyond 24 months, or whether the proportion of women requiring a second surgical procedure is reduced by their use. Few studies with follow-up beyond 24 months have been published. It would be helpful for the length of follow-up to extend beyond 24 months in future trials. It is uncertain whether a single GnRH analogue implant would provide adequate endometrial thinning; this could be further assessed in future studies. Progestogens have been only inadequately evaluated within a randomised setting, and there may be scope for reassessing their use, particularly as they are cheaper than other agents. Newer techniques such as balloon or microwave ablation, which are being used increasingly in clinical practice, have been evaluated in only three studies with small populations; these too can be evaluated in larger future trials.

Acknowledgements

We would like to acknowledge the support of the editorial base of the Cochrane Menstrual Disorders and Subfertility Group, which ha provided encouragement and assistance with development of the protocol.

We also would like to acknowledge Martin Sowter, who was the first author of the previous version of this review (Sowter 2002). In addition, we acknowledge the contributions of Dr Amita Singla to Sowter 2002.

Data and analyses

Download statistical data

Comparison 1. GnRH analogues versus placebo or no treatment
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Intraoperative complications—uterine perforation5592Risk Ratio (M-H, Fixed, 95% CI)1.47 [0.35, 6.06]
2 Postoperative amenorrhoea at 12 months or less7605Risk Ratio (M-H, Fixed, 95% CI)1.58 [1.24, 2.01]
2.1 one month of therapy389Risk Ratio (M-H, Fixed, 95% CI)1.80 [0.95, 3.40]
2.2 two months of therapy4516Risk Ratio (M-H, Fixed, 95% CI)1.54 [1.19, 2.01]
3 Postoperative amenorrhoea at 24 months or longer2357Risk Ratio (M-H, Fixed, 95% CI)1.62 [1.04, 2.52]
3.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
3.2 two months of therapy1337Risk Ratio (M-H, Fixed, 95% CI)1.50 [0.94, 2.41]
4 Requiring further surgery within 12 months of follow-up4488Risk Ratio (M-H, Fixed, 95% CI)0.99 [0.53, 1.86]
4.1 one month of therapy269Risk Ratio (M-H, Fixed, 95% CI)0.64 [0.21, 1.97]
4.2 two months of therapy2419Risk Ratio (M-H, Fixed, 95% CI)1.20 [0.55, 2.61]
5 Requiring further surgery ≥ 24 months of follow-up2319Risk Ratio (M-H, Fixed, 95% CI)1.51 [0.97, 2.35]
5.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
5.2 two months of therapy1299Risk Ratio (M-H, Fixed, 95% CI)1.51 [0.97, 2.35]
6 Endometrial thickness (ultrasound, mm)120Mean Difference (IV, Fixed, 95% CI)-2.70 [-3.49, -1.91]
6.1 one month of therapy120Mean Difference (IV, Fixed, 95% CI)-2.70 [-3.49, -1.91]
7 Endometrial thickness (descriptive data)  Other dataNo numeric data
7.1 two months of therapy  Other dataNo numeric data
8 Atrophic endometrial glands4483Risk Ratio (M-H, Fixed, 95% CI)6.02 [4.11, 8.81]
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)15.0 [0.97, 231.84]
8.2 two months of therapy3463Risk Ratio (M-H, Fixed, 95% CI)5.84 [3.97, 8.58]
9 Optimal endometrial thinning (operator assessment)120Risk Ratio (M-H, Fixed, 95% CI)6.0 [0.87, 41.21]
9.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)6.0 [0.87, 41.21]
10 Side effects2 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
10.1 hot flushes1346Risk Ratio (M-H, Fixed, 95% CI)3.16 [2.24, 4.45]
10.2 sweating1346Risk Ratio (M-H, Fixed, 95% CI)3.08 [1.50, 6.32]
10.3 headaches1346Risk Ratio (M-H, Fixed, 95% CI)1.46 [1.02, 2.08]
10.4 withdrawal from treatment due to side effects2372Risk Ratio (M-H, Fixed, 95% CI)3.20 [0.35, 29.75]
11 Duration of operation (minutes)6502Mean Difference (IV, Fixed, 95% CI)-3.62 [-4.76, -2.49]
11.1 one month of therapy139Mean Difference (IV, Fixed, 95% CI)-13.60 [-19.44, -7.76]
11.2 two months of therapy5463Mean Difference (IV, Fixed, 95% CI)-3.23 [-4.39, -2.07]
12 Where operative difficulty encountered2415Risk Ratio (M-H, Fixed, 95% CI)0.32 [0.22, 0.46]
12.1 one month of therapy00Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
12.2 two months of therapy2415Risk Ratio (M-H, Fixed, 95% CI)0.32 [0.22, 0.46]
13 Distension medium absorption during surgery4137Mean Difference (IV, Fixed, 95% CI)-154.53 [-211.41, -97.65]
13.1 one month of therapy139Mean Difference (IV, Fixed, 95% CI)-149.0 [-291.32, -6.68]
13.2 two months of therapy398Mean Difference (IV, Fixed, 95% CI)-155.59 [-217.64, -93.53]
14 Distension medium absorption during surgery (descriptive data)  Other dataNo numeric data
14.1 two months of therapy  Other dataNo numeric data
15 Women with moderate/heavy postoperative blood loss4480Risk Ratio (M-H, Fixed, 95% CI)0.74 [0.59, 0.92]
15.1 one month of therapy259Risk Ratio (M-H, Fixed, 95% CI)0.58 [0.22, 1.51]
15.2 two months of therapy2421Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.60, 0.94]
16 Postoperative dysmenorrhoea2133Risk Ratio (M-H, Fixed, 95% CI)0.59 [0.40, 0.87]
16.1 one month of therapy158Risk Ratio (M-H, Fixed, 95% CI)0.52 [0.25, 1.07]
16.2 two months of therapy175Risk Ratio (M-H, Fixed, 95% CI)0.63 [0.41, 0.98]
17 Satisfaction with outcome6599Risk Ratio (M-H, Fixed, 95% CI)0.99 [0.93, 1.05]
17.1 one month of therapy268Risk Ratio (M-H, Fixed, 95% CI)0.91 [0.71, 1.16]
17.2 two months of therapy4531Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.94, 1.06]
Analysis 1.1.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 1 Intraoperative complications—uterine perforation.

Analysis 1.2.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 1.3.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 3 Postoperative amenorrhoea at 24 months or longer.

Analysis 1.4.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 4 Requiring further surgery within 12 months of follow-up.

Analysis 1.5.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 5 Requiring further surgery ≥ 24 months of follow-up.

Analysis 1.6.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 6 Endometrial thickness (ultrasound, mm).

Analysis 1.7.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 7 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyGnRHaControlMann-Whitney U test
two months of therapy
Donnez 1997

n=176

Median thickness: 1.61 mm

No range reported

n=173

Median thickness: 3.53 mm

No range reported

Estimated ratio: 0.46, 95% CI 0.41, 0.52; p=0.0001
Rai 2000

n=25

Median: 1.5 mm

Range: 1 - 4

n=25

Median:2 mm

Range: 0.5 - 4

p=0.02
Analysis 1.8.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 8 Atrophic endometrial glands.

Analysis 1.9.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 9 Optimal endometrial thinning (operator assessment).

Analysis 1.10.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 10 Side effects.

Analysis 1.11.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 11 Duration of operation (minutes).

Analysis 1.12.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 12 Where operative difficulty encountered.

Analysis 1.13.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 13 Distension medium absorption during surgery.

Analysis 1.14.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 14 Distension medium absorption during surgery (descriptive data).

Distension medium absorption during surgery (descriptive data)
StudyGnRHaPlacebo or no treatmentTest results
two months of therapy
Donnez 1997

n=177

Median fluid absorption: 150mL

No range reported

n=175

Median fluid absorption: 225mL

No range reported

Adjustments made for effect of centre.

GnRHa treated patients absorbed on average a median of 40mL less fluid than placebo treated patients, p=0.0325

Analysis 1.15.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 15 Women with moderate/heavy postoperative blood loss.

Analysis 1.16.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 16 Postoperative dysmenorrhoea.

Analysis 1.17.

Comparison 1 GnRH analogues versus placebo or no treatment, Outcome 17 Satisfaction with outcome.

Comparison 2. GnRH analogues versus danazol
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less5340Risk Ratio (M-H, Fixed, 95% CI)1.17 [0.90, 1.52]
2.1 one month of therapy291Risk Ratio (M-H, Fixed, 95% CI)1.60 [0.83, 3.11]
2.2 two months of therapy3174Risk Ratio (M-H, Fixed, 95% CI)1.07 [0.78, 1.47]
2.3 three months of therapy141Risk Ratio (M-H, Fixed, 95% CI)1.15 [0.49, 2.74]
2.4 three months of therapy with danazol and one month with GnRHa134Risk Ratio (M-H, Fixed, 95% CI)1.08 [0.42, 2.74]
3 Postoperative amenorrhoea at 24 months1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
3.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.49, 2.05]
4 Requesting further surgery during follow-up120Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.02, 7.32]
4.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.02, 7.32]
5 Endometrial thickness (ultrasound)120Mean Difference (IV, Fixed, 95% CI)0.20 [-0.46, 0.86]
5.1 one month of therapy120Mean Difference (IV, Fixed, 95% CI)0.20 [-0.46, 0.86]
6 Endometrial thickness (descriptive data)  Other dataNo numeric data
6.1 one month of therapy  Other dataNo numeric data
6.2 two months of therapy  Other dataNo numeric data
7 Atrophic endometrial glands4318Risk Ratio (M-H, Fixed, 95% CI)1.28 [1.03, 1.58]
7.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.53, 1.46]
7.2 two months of therapy3298Risk Ratio (M-H, Fixed, 95% CI)1.33 [1.06, 1.67]
8 Satisfactory thinning (hysteroscopy)2165Risk Ratio (M-H, Fixed, 95% CI)1.09 [1.00, 1.19]
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.75, 1.34]
8.2 two months of therapy1145Risk Ratio (M-H, Fixed, 95% CI)1.10 [1.00, 1.21]
9 Optimal endometrial thinning (operator assessment)120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.49, 2.05]
9.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.49, 2.05]
10 Complete atrophy (hysteroscopy)2142Risk Ratio (M-H, Fixed, 95% CI)1.84 [1.23, 2.75]
10.1 two months of therapy2142Risk Ratio (M-H, Fixed, 95% CI)1.84 [1.23, 2.75]
11 Side effects3 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
11.1 hot flushes2168Risk Ratio (M-H, Fixed, 95% CI)1.99 [1.50, 2.64]
11.2 vaginal dryness2168Risk Ratio (M-H, Fixed, 95% CI)2.16 [1.30, 3.59]
11.3 depression2168Risk Ratio (M-H, Fixed, 95% CI)1.32 [0.89, 1.97]
11.4 voice changes2168Risk Ratio (M-H, Fixed, 95% CI)0.79 [0.37, 1.65]
11.5 hirsutism2168Risk Ratio (M-H, Fixed, 95% CI)0.78 [0.29, 2.08]
11.6 decrease in libido2168Risk Ratio (M-H, Fixed, 95% CI)2.15 [1.08, 4.28]
11.7 headaches1110Risk Ratio (M-H, Fixed, 95% CI)1.89 [1.23, 2.91]
11.8 withdrawal from treatment due to side effects3285Risk Ratio (M-H, Fixed, 95% CI)0.08 [0.01, 0.59]
12 Weight gain158Mean Difference (IV, Fixed, 95% CI)-2.3 [-3.70, -0.90]
13 Duration of operation (minutes)4325Mean Difference (IV, Fixed, 95% CI)-5.02 [-6.25, -3.78]
13.1 one month of therapy with GnRHa and danazol174Mean Difference (IV, Fixed, 95% CI)-6.40 [-9.16, -3.64]
13.2 two months of therapy with GnRHa and danazol3150Mean Difference (IV, Fixed, 95% CI)-3.78 [-5.92, -1.64]
13.3 three months of therapy with GnRHa and danazol152Mean Difference (IV, Fixed, 95% CI)-6.0 [-8.24, -3.76]
13.4 three months of danazol and one month of GnRHa149Mean Difference (IV, Fixed, 95% CI)-4.0 [-7.08, -0.92]
14 Where operative difficulty encountered156Risk Ratio (M-H, Fixed, 95% CI)0.68 [0.31, 1.51]
14.1 two months of therapy156Risk Ratio (M-H, Fixed, 95% CI)0.68 [0.31, 1.51]
15 Distension medium absorption during surgery3223Mean Difference (IV, Fixed, 95% CI)-22.38 [-87.07, 42.32]
15.1 one month of therapy174Mean Difference (IV, Fixed, 95% CI)-504.0 [-776.80, -231.20]
15.2 two months of therapy3149Mean Difference (IV, Fixed, 95% CI)6.32 [-60.27, 72.92]
16 Postoperative blood loss—objective156Mean Difference (IV, Fixed, 95% CI)-6.4 [-9.19, -3.61]
16.1 two months of therapy156Mean Difference (IV, Fixed, 95% CI)-6.4 [-9.19, -3.61]
17 Satisfaction with outcome3286Risk Ratio (M-H, Fixed, 95% CI)0.92 [0.83, 1.01]
17.1 one month of therapy171Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.90, 1.13]
17.2 two months of therapy3215Risk Ratio (M-H, Fixed, 95% CI)0.89 [0.78, 1.01]
Analysis 2.2.

Comparison 2 GnRH analogues versus danazol, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 2.3.

Comparison 2 GnRH analogues versus danazol, Outcome 3 Postoperative amenorrhoea at 24 months.

Analysis 2.4.

Comparison 2 GnRH analogues versus danazol, Outcome 4 Requesting further surgery during follow-up.

Analysis 2.5.

Comparison 2 GnRH analogues versus danazol, Outcome 5 Endometrial thickness (ultrasound).

Analysis 2.6.

Comparison 2 GnRH analogues versus danazol, Outcome 6 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyGnRHaDanazolTest results
one month of therapy
Garry 1996

n=39

Median: 1.0

Range: 0.5-2.0

n=37

Median: 1.0

Range: 0.5-3.0

Results not reported
two months of therapy
Garry 1996

n=38

Median: 0.5 mm

Range: 0.5-1.5

n=32

Median: 0.5

Range 0.4-1.5

Results not reported
Rai 2000

n=25

Median 1.5 mm

Range: 1-4

n=25

Median 1.5 mm

Range 1 - 3.5 mm

P value not reported

Not significantly different

Sutton 1994

n=55

Median: 1.5 mm

No other data reported

n=55

Median: 2.0 mm

No other data reported

Difference between the 2 treatments: 0.9 mm [-1.5, -0.3]

Wilcoxon Rank sum test: p=0.0015

Analysis 2.7.

Comparison 2 GnRH analogues versus danazol, Outcome 7 Atrophic endometrial glands.

Analysis 2.8.

Comparison 2 GnRH analogues versus danazol, Outcome 8 Satisfactory thinning (hysteroscopy).

Analysis 2.9.

Comparison 2 GnRH analogues versus danazol, Outcome 9 Optimal endometrial thinning (operator assessment).

Analysis 2.10.

Comparison 2 GnRH analogues versus danazol, Outcome 10 Complete atrophy (hysteroscopy).

Analysis 2.11.

Comparison 2 GnRH analogues versus danazol, Outcome 11 Side effects.

Analysis 2.12.

Comparison 2 GnRH analogues versus danazol, Outcome 12 Weight gain.

Analysis 2.13.

Comparison 2 GnRH analogues versus danazol, Outcome 13 Duration of operation (minutes).

Analysis 2.14.

Comparison 2 GnRH analogues versus danazol, Outcome 14 Where operative difficulty encountered.

Analysis 2.15.

Comparison 2 GnRH analogues versus danazol, Outcome 15 Distension medium absorption during surgery.

Analysis 2.16.

Comparison 2 GnRH analogues versus danazol, Outcome 16 Postoperative blood loss—objective.

Analysis 2.17.

Comparison 2 GnRH analogues versus danazol, Outcome 17 Satisfaction with outcome.

Comparison 3. GnRH analogues versus progestogens
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less3146Risk Ratio (M-H, Fixed, 95% CI)1.58 [0.96, 2.61]
2.1 one month of therapy257Risk Ratio (M-H, Fixed, 95% CI)1.84 [0.85, 3.98]
2.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.29 [0.57, 2.91]
2.3 three months of therapy139Risk Ratio (M-H, Fixed, 95% CI)1.67 [0.55, 5.03]
3 Postoperative amenorrhoea at 24 months120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
3.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
4 Requiring further surgery at two years of follow-up120Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.02, 7.32]
4.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.02, 7.32]
5 Endometrial thickness (ultrasound)120Mean Difference (IV, Fixed, 95% CI)-2.80 [-3.59, -2.01]
5.1 one month of therapy120Mean Difference (IV, Fixed, 95% CI)-2.80 [-3.59, -2.01]
6 Endometrial thickness (descriptive data)  Other dataNo numeric data
6.1 two months of therapy  Other dataNo numeric data
7 Endometrial atrophy (histology)270Risk Ratio (M-H, Fixed, 95% CI)2.25 [1.13, 4.49]
7.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.5 [0.95, 12.90]
7.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.83 [0.80, 4.19]
8 Optimal endometrial thinning (operator assessment)1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
8.2 two months of therapy00Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
9 Duration of operation (minutes)1 Mean Difference (IV, Fixed, 95% CI)Subtotals only
9.1 one month of therapy150Mean Difference (IV, Fixed, 95% CI)-15.0 [-19.44, -10.56]
9.2 three months of therapy153Mean Difference (IV, Fixed, 95% CI)-17.0 [-20.90, -13.10]
10 Satisfaction with outcome150Risk Ratio (M-H, Fixed, 95% CI)0.80 [0.65, 0.99]
10.1 one month of therapy00Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
10.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)0.80 [0.65, 0.99]
Analysis 3.2.

Comparison 3 GnRH analogues versus progestogens, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 3.3.

Comparison 3 GnRH analogues versus progestogens, Outcome 3 Postoperative amenorrhoea at 24 months.

Analysis 3.4.

Comparison 3 GnRH analogues versus progestogens, Outcome 4 Requiring further surgery at two years of follow-up.

Analysis 3.5.

Comparison 3 GnRH analogues versus progestogens, Outcome 5 Endometrial thickness (ultrasound).

Analysis 3.6.

Comparison 3 GnRH analogues versus progestogens, Outcome 6 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyGnRHaProgestogensMann-Whitney U test
two months of therapy
Rai 2000

n=25

Median: 1.5 mm

Range: 1 - 4

n=25

Median: 2 mm

Range: 0.5 - 3

No p value reported

Not significantly different

Analysis 3.7.

Comparison 3 GnRH analogues versus progestogens, Outcome 7 Endometrial atrophy (histology).

Analysis 3.8.

Comparison 3 GnRH analogues versus progestogens, Outcome 8 Optimal endometrial thinning (operator assessment).

Analysis 3.9.

Comparison 3 GnRH analogues versus progestogens, Outcome 9 Duration of operation (minutes).

Analysis 3.10.

Comparison 3 GnRH analogues versus progestogens, Outcome 10 Satisfaction with outcome.

Comparison 4. GnRH analogue versus GnRH antagonist
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Intraoperative complications1300Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.25, 3.92]
1.1 uterine perforation1100Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.13, 71.92]
1.2 traumatic dilatation of cervix1100Risk Ratio (M-H, Fixed, 95% CI)0.33 [0.01, 7.99]
1.3 fluid deficit > 1500 mL1100Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.15, 6.82]
2 Postoperative amenorrhoea at 12 months or less1100Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.79, 1.14]
2.1 six months after surgery1100Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.79, 1.14]
3 Postoperative amenorrhoea and/or hypomenorrhoea at 12 months or less1100Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.83, 1.06]
3.1 six months after surgery1100Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.83, 1.06]
4 Requiring further surgery within 12 months of follow-up1100Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.31, 5.65]
4.1 at six months of follow-up1100Risk Ratio (M-H, Fixed, 95% CI)1.33 [0.31, 5.65]
5 Endometrial thickness (ultrasound)1171Mean Difference (IV, Fixed, 95% CI)0.53 [0.29, 0.77]
5.1 hyperplasia included188Mean Difference (IV, Fixed, 95% CI)0.54 [0.17, 0.91]
5.2 hyperplasia excluded183Mean Difference (IV, Fixed, 95% CI)0.53 [0.22, 0.84]
6 Side effects1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
6.1 hot flushes1100Risk Ratio (M-H, Fixed, 95% CI)0.4 [0.17, 0.95]
6.2 skin irritation1100Risk Ratio (M-H, Fixed, 95% CI)2.67 [0.75, 9.47]
6.3 nonspecific adverse effects1100Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.15, 6.82]
7 Duration of operation (minutes)1100Mean Difference (IV, Fixed, 95% CI)2.26 [-1.11, 5.63]
8 Good operative view1100Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.89, 1.04]
9 Distension medium absorption during surgery  Other dataNo numeric data
10 Postoperative dysmenorrhoea1100Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.48, 2.09]
11 Satisfaction with outcome1100Risk Ratio (M-H, Fixed, 95% CI)0.93 [0.81, 1.07]
11.1 at six months of follow-up1100Risk Ratio (M-H, Fixed, 95% CI)0.93 [0.81, 1.07]
Analysis 4.1.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 1 Intraoperative complications.

Analysis 4.2.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 4.3.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 3 Postoperative amenorrhoea and/or hypomenorrhoea at 12 months or less.

Analysis 4.4.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 4 Requiring further surgery within 12 months of follow-up.

Analysis 4.5.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 5 Endometrial thickness (ultrasound).

Analysis 4.6.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 6 Side effects.

Analysis 4.7.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 7 Duration of operation (minutes).

Analysis 4.8.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 8 Good operative view.

Analysis 4.9.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 9 Distension medium absorption during surgery.

Distension medium absorption during surgery
StudyGnRH antagonistGnRH analogueMann-Whitney U test
Bhatia 2008

n=50

Median: 300 mL

Range 0 to 2000

n=50

Median: 250 mL

Range 0 to 2000

Mann-Whitney test

P = 0.641

Analysis 4.10.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 10 Postoperative dysmenorrhoea.

Analysis 4.11.

Comparison 4 GnRH analogue versus GnRH antagonist, Outcome 11 Satisfaction with outcome.

Comparison 5. GnRH analogue versus dilatation & curettage
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Normal bleeding or less at 12 months192Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.91, 1.16]
2.1 at 12 months of follow-up192Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.91, 1.16]
3 Postoperative amenorrhoea at 12 months or less192Risk Ratio (M-H, Fixed, 95% CI)20.13 [1.21, 333.63]
3.1 at 12 months of follow-up192Risk Ratio (M-H, Fixed, 95% CI)20.13 [1.21, 333.63]
4 Requiring further surgery within 12 months of follow-up1100Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.18, 3.18]
4.1 at one year of follow-up1100Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.18, 3.18]
6 Side effects—presence of PMS symptoms193Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.67, 1.58]
6.1 at one year of follow-up193Risk Ratio (M-H, Fixed, 95% CI)1.03 [0.67, 1.58]
7 Postoperative dysmenorrhoea192Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.66, 1.52]
7.1 at one year of follow-up192Risk Ratio (M-H, Fixed, 95% CI)1.00 [0.66, 1.52]
8 Satisfaction with outcome192Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.83, 1.05]
8.1 at 12 months of follow-up192Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.83, 1.05]
Analysis 5.2.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 2 Normal bleeding or less at 12 months.

Analysis 5.3.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 3 Postoperative amenorrhoea at 12 months or less.

Analysis 5.4.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 4 Requiring further surgery within 12 months of follow-up.

Analysis 5.6.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 6 Side effects—presence of PMS symptoms.

Analysis 5.7.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 7 Postoperative dysmenorrhoea.

Analysis 5.8.

Comparison 5 GnRH analogue versus dilatation & curettage, Outcome 8 Satisfaction with outcome.

Comparison 6. Danazol versus no pretreatment
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less2179Risk Ratio (M-H, Fixed, 95% CI)1.05 [0.80, 1.39]
2.1 one month of therapy1129Risk Ratio (M-H, Fixed, 95% CI)0.95 [0.68, 1.34]
2.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.31 [0.82, 2.08]
3 Postoperative amenorrhoea at two years or more281Risk Ratio (M-H, Fixed, 95% CI)1.27 [0.79, 2.04]
3.1 one month of therapy281Risk Ratio (M-H, Fixed, 95% CI)1.27 [0.79, 2.04]
4 Requiring further surgery at two years or more2152Risk Ratio (M-H, Fixed, 95% CI)1.37 [0.28, 6.69]
4.1 one month of therapy2152Risk Ratio (M-H, Fixed, 95% CI)1.37 [0.28, 6.69]
5 Endometrial thickness (ultrasound)2 Mean Difference (IV, Random, 95% CI)Totals not selected
5.1 one month of therapy2 Mean Difference (IV, Random, 95% CI)0.0 [0.0, 0.0]
6 Endometrial thickness (descriptive data)  Other dataNo numeric data
6.1 two months of therapy  Other dataNo numeric data
7 Atrophic endometrial glands270Risk Ratio (M-H, Fixed, 95% CI)3.15 [1.46, 6.80]
7.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)17.0 [1.11, 259.87]
7.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)2.0 [0.89, 4.49]
8 Optimal endometrial thinning (operator assessment)120Risk Ratio (M-H, Fixed, 95% CI)6.0 [0.87, 41.21]
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)6.0 [0.87, 41.21]
10 Duration of operation (minutes)2157Mean Difference (IV, Fixed, 95% CI)-6.84 [-7.97, -5.72]
10.1 one month of therapy1132Mean Difference (IV, Fixed, 95% CI)-7.90 [-9.13, -6.67]
10.2 two months of therapy125Mean Difference (IV, Fixed, 95% CI)-1.30 [-4.12, 1.52]
11 Distension medium absorption during surgery (mL)2156Mean Difference (IV, Fixed, 95% CI)-109.45 [-193.25, -25.65]
11.1 one month of therapy1132Mean Difference (IV, Fixed, 95% CI)-119.90 [-209.00, -28.80]
11.2 two months of therapy124Mean Difference (IV, Fixed, 95% CI)-52.0 [-265.65, 161.65]
12 Satisfaction with outcome150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
12.1 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
Analysis 6.2.

Comparison 6 Danazol versus no pretreatment, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 6.3.

Comparison 6 Danazol versus no pretreatment, Outcome 3 Postoperative amenorrhoea at two years or more.

Analysis 6.4.

Comparison 6 Danazol versus no pretreatment, Outcome 4 Requiring further surgery at two years or more.

Analysis 6.5.

Comparison 6 Danazol versus no pretreatment, Outcome 5 Endometrial thickness (ultrasound).

Analysis 6.6.

Comparison 6 Danazol versus no pretreatment, Outcome 6 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyDanazolControlMann-Whitney U test
two months of therapy
Rai 2000

n=25

Median: 1.5 mm

Range: 1 - 3.5

n=25

Median: 2 mm

Range: 0.5 - 4

p=0.02
Analysis 6.7.

Comparison 6 Danazol versus no pretreatment, Outcome 7 Atrophic endometrial glands.

Analysis 6.8.

Comparison 6 Danazol versus no pretreatment, Outcome 8 Optimal endometrial thinning (operator assessment).

Analysis 6.10.

Comparison 6 Danazol versus no pretreatment, Outcome 10 Duration of operation (minutes).

Analysis 6.11.

Comparison 6 Danazol versus no pretreatment, Outcome 11 Distension medium absorption during surgery (mL).

Analysis 6.12.

Comparison 6 Danazol versus no pretreatment, Outcome 12 Satisfaction with outcome.

Comparison 7. Progestogens versus no pretreatment
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less150Risk Ratio (M-H, Fixed, 95% CI)0.54 [0.26, 1.12]
2.1 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)0.54 [0.26, 1.12]
3 Postoperative amenorrhoea at two to four years270Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.36, 1.54]
3.1 one month of therapy270Risk Ratio (M-H, Fixed, 95% CI)0.75 [0.36, 1.54]
4 Requiring further surgery at two to four years270Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.50, 17.95]
4.1 one month of therapy270Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.50, 17.95]
5 Endometrial thickness (ultrasound)120Mean Difference (IV, Fixed, 95% CI)0.10 [-0.78, 0.98]
5.1 one month of therapy120Mean Difference (IV, Fixed, 95% CI)0.10 [-0.78, 0.98]
6 Endometrial thickness (descriptive data)  Other dataNo numeric data
6.1 two months of therapy  Other dataNo numeric data
7 Atrophic endometrial glands270Risk Ratio (M-H, Fixed, 95% CI)1.31 [0.53, 3.25]
7.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)5.0 [0.27, 92.62]
7.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.37, 2.68]
8 Optimal endometrial thinning (operator)120Risk Ratio (M-H, Fixed, 95% CI)2.0 [0.21, 18.69]
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)2.0 [0.21, 18.69]
11 Satisfaction with outcome150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
11.1 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
Analysis 7.2.

Comparison 7 Progestogens versus no pretreatment, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 7.3.

Comparison 7 Progestogens versus no pretreatment, Outcome 3 Postoperative amenorrhoea at two to four years.

Analysis 7.4.

Comparison 7 Progestogens versus no pretreatment, Outcome 4 Requiring further surgery at two to four years.

Analysis 7.5.

Comparison 7 Progestogens versus no pretreatment, Outcome 5 Endometrial thickness (ultrasound).

Analysis 7.6.

Comparison 7 Progestogens versus no pretreatment, Outcome 6 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyProgestogensNo treatmentMann-Whitney U test
two months of therapy
Rai 2000

n=25

Median: 2 mm

Range: 0.5 - 3

n=25

Median: 2 mm

Range: 0.5 - 4

No p value reported

Not significantly different

Analysis 7.7.

Comparison 7 Progestogens versus no pretreatment, Outcome 7 Atrophic endometrial glands.

Analysis 7.8.

Comparison 7 Progestogens versus no pretreatment, Outcome 8 Optimal endometrial thinning (operator).

Analysis 7.11.

Comparison 7 Progestogens versus no pretreatment, Outcome 11 Satisfaction with outcome.

Comparison 8. Danazol versus progestogens
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less2103Risk Ratio (M-H, Fixed, 95% CI)1.89 [1.12, 3.18]
2.1 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)2.43 [1.23, 4.81]
2.2 three months of therapy153Risk Ratio (M-H, Fixed, 95% CI)1.34 [0.58, 3.06]
3 Postoperative amenorrhoea at two years120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
3.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
4 Requiring further surgery at two years of follow-up120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.07, 13.87]
4.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.07, 13.87]
5 Endometrial thickness (ultrasound)120Mean Difference (IV, Fixed, 95% CI)-3.00 [-3.76, -2.24]
5.1 one month of therapy120Mean Difference (IV, Fixed, 95% CI)-3.00 [-3.76, -2.24]
6 Endometrial thickness (descriptive data)  Other dataNo numeric data
6.1 two months of therapy  Other dataNo numeric data
7 Atrophic endometrial glands270Risk Ratio (M-H, Fixed, 95% CI)2.5 [1.27, 4.92]
7.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)4.0 [1.11, 14.35]
7.2 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)2.0 [0.89, 4.49]
8 Optimal endometrial thinning (operator)120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
8.1 one month of therapy120Risk Ratio (M-H, Fixed, 95% CI)3.0 [0.79, 11.44]
10 Duration of operation (minutes)153Mean Difference (IV, Fixed, 95% CI)-11.0 [-14.59, -7.41]
10.1 three months of therapy153Mean Difference (IV, Fixed, 95% CI)-11.0 [-14.59, -7.41]
11 Satisfaction with outcome150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
11.1 two months of therapy150Risk Ratio (M-H, Fixed, 95% CI)1.0 [0.93, 1.08]
Analysis 8.2.

Comparison 8 Danazol versus progestogens, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 8.3.

Comparison 8 Danazol versus progestogens, Outcome 3 Postoperative amenorrhoea at two years.

Analysis 8.4.

Comparison 8 Danazol versus progestogens, Outcome 4 Requiring further surgery at two years of follow-up.

Analysis 8.5.

Comparison 8 Danazol versus progestogens, Outcome 5 Endometrial thickness (ultrasound).

Analysis 8.6.

Comparison 8 Danazol versus progestogens, Outcome 6 Endometrial thickness (descriptive data).

Endometrial thickness (descriptive data)
StudyDanazolProgestogensMann-Whitney U test
two months of therapy
Rai 2000

n=25

Median: 1.5 mm

Range: 1 - 3.5

n=25

Median: 2 mm

Range: 0.5 - 3

No p value reported

Not significantly different

Analysis 8.7.

Comparison 8 Danazol versus progestogens, Outcome 7 Atrophic endometrial glands.

Analysis 8.8.

Comparison 8 Danazol versus progestogens, Outcome 8 Optimal endometrial thinning (operator).

Analysis 8.10.

Comparison 8 Danazol versus progestogens, Outcome 10 Duration of operation (minutes).

Analysis 8.11.

Comparison 8 Danazol versus progestogens, Outcome 11 Satisfaction with outcome.

Comparison 9. GnRHa or danazol versus no pretreatment
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
2 Postoperative amenorrhoea at 12 months or less2280Risk Ratio (M-H, Fixed, 95% CI)1.13 [0.89, 1.43]
3 Amenorrhoea at five years1154Risk Ratio (M-H, Fixed, 95% CI)1.05 [0.91, 1.20]
4 Improvement (light bleeding/amenorrhoea) at five years1154Risk Ratio (M-H, Fixed, 95% CI)1.01 [0.98, 1.05]
5 Requiring further surgery at one year1197Risk Ratio (M-H, Fixed, 95% CI)0.97 [0.06, 15.29]
6 Requiring further surgery at five years1197Risk Ratio (M-H, Fixed, 95% CI)0.73 [0.32, 1.65]
7 Endometrial thickness (mm)1197Mean Difference (IV, Fixed, 95% CI)-1.9 [-2.54, -1.26]
8 Side effects1 Risk Ratio (M-H, Fixed, 95% CI)Subtotals only
8.1 hot flushes1197Risk Ratio (M-H, Fixed, 95% CI)2.95 [2.02, 4.31]
8.2 nausea1197Risk Ratio (M-H, Fixed, 95% CI)2.06 [1.22, 3.48]
8.3 rashes/itch1197Risk Ratio (M-H, Fixed, 95% CI)3.88 [1.34, 11.19]
8.4 weight gain1197Risk Ratio (M-H, Fixed, 95% CI)2.36 [1.42, 3.94]
9 Duration of operation (minutes)2 Mean Difference (IV, Fixed, 95% CI)Subtotals only
9.1 TCRE + rollerball190Mean Difference (IV, Fixed, 95% CI)-15.0 [-16.87, -13.13]
9.2 microwave1197Mean Difference (IV, Fixed, 95% CI)0.40 [-0.89, 1.69]
10 Postoperative menorrhagia2280Risk Ratio (M-H, Fixed, 95% CI)0.81 [0.22, 2.93]
11 Satisfaction with outcome at 12 months of follow-up1188Risk Ratio (M-H, Fixed, 95% CI)0.96 [0.87, 1.05]
12 Satisfaction with outcome at five years1154Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.83, 1.06]
Analysis 9.2.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 2 Postoperative amenorrhoea at 12 months or less.

Analysis 9.3.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 3 Amenorrhoea at five years.

Analysis 9.4.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 4 Improvement (light bleeding/amenorrhoea) at five years.

Analysis 9.5.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 5 Requiring further surgery at one year.

Analysis 9.6.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 6 Requiring further surgery at five years.

Analysis 9.7.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 7 Endometrial thickness (mm).

Analysis 9.8.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 8 Side effects.

Analysis 9.9.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 9 Duration of operation (minutes).

Analysis 9.10.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 10 Postoperative menorrhagia.

Analysis 9.11.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 11 Satisfaction with outcome at 12 months of follow-up.

Analysis 9.12.

Comparison 9 GnRHa or danazol versus no pretreatment, Outcome 12 Satisfaction with outcome at five years.

Appendices

Appendix 1. Search strategies

Cochrane Menstrual Disorders and Subfertility Group Specialised Register

Keywords CONTAINS "Goserelin" or "Gonadorelin" or "gonadotrophin" or "GnRHa-gonadotropin" or "GnRHa" or "gonadotropin releasing hormone agonist" or  "goserelin acetate" or "goserelin pretreatment" or "Gosereline " or "danazol" or "progestagen" or "progestin" or "progestins" or "progestogens" or "cyproterone" or "cyproterone acetate" or "medroxyprogesterone" or "Medroxyprogesterone Acetate" or "*Medrogestone" or "norethindrone" or "norethisterone" or "Norethisterone" or "norethisterone acetate" or "endometrial preparation" or "endometrial priming" or "endometrial stripping" or "endometrial thinning" or "endometrioma wall stripping" or "DMPA" or "Zoladex" or "Pretreatment" or "Androgen Antagonists" or "androgen priming"

 AND

 Keywords CONTAINS "endometrial ablation" or "endometrial ablation, bipolar radiofrequency" or "Endometrial ablation, chemical" or "endometrial ablation, laser" or "endometrial ablation, microwave" or "endometrial ablation, Novasure" or "endometrial ablation, rollerball" or "endometrial ablation, thermal balloon" or "endometrial atrophy" or "endometrial cryoablation" or "endometrial resection" or "endometrial resection, transcervical" or "endometrioma wall stripping" or "Laser Ablation" or "microwave endometrial ablation" or "microwave endometrial ablation" or "microwave" or "rollerball" or "rollerball electroablation" or "balloon endometrial ablation" or "thermal balloon" or "thermal balloon ablation" or "thermal endometrial ablation" or "cryoablation therapy" or "transcervical resection" or "transcervical endometrial resection" or "electrosurgery" or "electrosurgical" or "electroresection" or "photoablation" or "photodynamic therapy" or  "Thermochoice balloon ablation" or "Thermachoice" or "NovaSure"

Cochrane Central Register of Controlled Trials (CENTRAL) to March 2013

1 exp Gonadotropin-Releasing Hormone/ (1776)
2 exp Goserelin/ (340)
3 exp Danazol/ (182)
4 exp Progestins/ (1731)
5 buserelin/ or leuprolide/ or nafarelin/ or triptorelin/ (744)
6 exp cyproterone/ or exp cyproterone acetate/ (280)
7 cyproterone$.tw. (363)
8 exp medroxyprogesterone/ or exp medroxyprogesterone 17-acetate/ (993)
9 medroxyprogesterone$.tw. (1254)
10 exp Norethindrone/ (668)
11 (norethisterone or Norethindrone).tw. (759)
12 GnRH analog$.tw. (268)
13 Gonadotropin-Releasing Hormone$.tw. (699)
14 Gonadotrophin-Releasing Hormone$.tw. (296)
15 (Goserelin or Danazol).tw. (608)
16 (buserelin or leuprolide or nafarelin or triptorelin).tw. (869)
17 (endometri$ adj2 prepar$).tw. (61)
18 (endometri$ adj2 pretreat$).tw. (19)
19 progestogen$.tw. (588)
20 progestagen$.tw. (121)
21 DMPA.tw. (102)
22 Zoladex.tw. (219)
23 (delay$ adj2 treat$).tw. (1048)
24 (hormon$ adj2 pretreat$).tw. (42)
25 (hormon$ adj2 inhibit$).tw. (169)
26 GnRH agonist$.tw. (655)
27 GnRH a.tw. (1093)
28 GnRHa.tw. (191)
29 (endometri$ adj2 thin$).tw. (37)
30 (preoperative adj2 endometri$).tw. (18)
31 (preoperative$ adj2 treatment$).tw. (447)
32 or/1-31 (8807)
33 (endometri$ adj2 ablat$).tw. (173)
34 (laser adj2 ablat$).tw. (263)
35 (endometr$ adj2 resect$).tw. (116)
36 (microwave adj2 ablat$).tw. (38)
37 rollerball.tw. (28)
38 (balloon adj2 ablat$).tw. (49)
39 (thermal adj2 ablat$).tw. (68)
40 cryoablat$.tw. (39)
41 (endometri$ adj2 inhibit$).tw. (21)
42 (endometri$ adj2 destr$).tw. (8)
43 (endometri$ adj2 removal).tw. (8)
44 TCRE.tw. (21)
45 (transcerv$ adj3 resect$).tw. (55)
46 electrosurg$.tw. (148)
47 (hypertherm$ or photodynam$).tw. (1200)
48 thermotherap$.tw. (151)
49 (phototherap$ or radiofreq$).tw. (1489)
50 (laser adj3 interstit$).tw. (28)
51 Thermachoice.tw. (18)
52 Cavaterm.tw. (13)
53 (Elitt or Vesta).tw. (8)
54 (Novasure or Cryogen).tw. (31)
55 (electrode$ adj2 ablat$).tw. (13)
56 (saline adj2 irrigat$).tw. (197)
57 (heat$ adj2 balloon).tw. (0)
58 hysteroscop$.tw. (419)
59 or/33-58 (4071)
60 59 and 32 (140)
61 limit 60 to yr="2012 -Current" (2)

MEDLINE to March 2013

1 exp Gonadotropin-Releasing Hormone/ (27996)
2 exp Goserelin/ (1401)
3 exp Danazol/ (2142)
4 exp Progestins/ (59516)
5 buserelin/ or leuprolide/ or nafarelin/ or triptorelin/ (6046)
6 exp cyproterone/ or exp cyproterone acetate/ (2505)
7 cyproterone$.tw. (2452)
8 exp medroxyprogesterone/ or exp medroxyprogesterone 17-acetate/ (6417)
9 medroxyprogesterone$.tw. (5201)
10 exp Norethindrone/ (3995)
11 (norethisterone or Norethindrone).tw. (2968)
12 GnRH analog$.tw. (2012)
13 Gonadotropin-Releasing Hormone$.tw. (10654)
14 Gonadotrophin-Releasing Hormone$.tw. (2504)
15 (Goserelin or Danazol).tw. (2864)
16 (buserelin or leuprolide or nafarelin or triptorelin).tw. (3324)
17 (endometri$ adj2 prepar$).tw. (348)
18 (endometri$ adj2 pretreat$).tw. (44)
19 progestogen$.tw. (4632)
20 progestagen$.tw. (1916)
21 DMPA.tw. (827)
22 Zoladex.tw. (368)
23 (delay$ adj2 treat$).tw. (9151)
24 (hormon$ adj2 pretreat$).tw. (246)
25 (hormon$ adj2 inhibit$).tw. (3818)
26 GnRH agonist$.tw. (3274)
27 GnRH a.tw. (884)
28 GnRHa.tw. (1048)
29 (endometri$ adj2 thin$).tw. (201)
30 (preoperative adj2 endometri$).tw. (121)
31 (preoperative$ adj2 treatment$).tw. (2957)
32 or/1-31 (122177)
33 (endometri$ adj2 ablat$).tw. (1012)
34 (laser adj2 ablat$).tw. (4875)
35 (endometr$ adj2 resect$).tw. (529)
36 (microwave adj2 ablat$).tw. (586)
37 rollerball.tw. (129)
38 (balloon adj2 ablat$).tw. (211)
39 (thermal adj2 ablat$).tw. (1550)
40 cryoablat$.tw. (1903)
41 (endometri$ adj2 inhibit$).tw. (319)
42 (endometri$ adj2 destr$).tw. (103)
43 (endometri$ adj2 removal).tw. (120)
44 TCRE.tw. (69)
45 (transcerv$ adj3 resect$).tw. (283)
46 electrosurg$.tw. (2531)
47 (hypertherm$ or photodynam$).tw. (38880)
48 thermotherap$.tw. (1787)
49 (phototherap$ or radiofreq$).tw. (24636)
50 (laser adj3 interstit$).tw. (580)
51 Thermachoice.tw. (40)
52 Cavaterm.tw. (17)
53 (Elitt or Vesta).tw. (54)
54 (Novasure or Cryogen).tw. (325)
55 (electrode$ adj2 ablat$).tw. (237)
56 (saline adj2 irrigat$).tw. (812)
57 (heat$ adj2 balloon).tw. (20)
58 hysteroscop$.tw. (4415)
59 or/33-58 (80426)
60 59 and 32 (811)
61 randomized controlled trial.pt. (346887)
62 controlled clinical trial.pt. (85740)
63 randomized.ab. (264972)
64 placebo.tw. (147356)
65 clinical trials as topic.sh. (163937)
66 randomly.ab. (192897)
67 trial.ti. (113202)
68 (crossover or cross-over or cross over).tw. (56472)
69 or/61-68 (853070)
70 exp animals/ not humans.sh. (3801093)
71 69 not 70 (786500)
72 60 and 71 (120)
73 (2012$ or 2013$).ed. (1334173)
74 72 and 73 (4)

EMBASE to March 2013

1 exp Goserelin/ (5456)
2 exp Danazol/ (7023)
3 exp Progestins/ (132528)
4 buserelin/ or leuprolide/ or nafarelin/ or triptorelin/ (14189)
5 cyproterone$.tw. (2647)
6 exp medroxyprogesterone/ or exp medroxyprogesterone 17-acetate/ (17018)
7 medroxyprogesterone$.tw. (5654)
8 (norethisterone or Norethindrone).tw. (2732)
9 GnRH analog$.tw. (2669)
10 Gonadotropin-Releasing Hormone$.tw. (11415)
11 Gonadotrophin-Releasing Hormone$.tw. (2658)
12 (Goserelin or Danazol).tw. (3684)
13 (buserelin or leuprolide or nafarelin or triptorelin).tw. (4237)
14 (endometri$ adj2 prepar$).tw. (471)
15 (endometri$ adj2 pretreat$).tw. (53)
16 progestogen$.tw. (4956)
17 progestagen$.tw. (2014)
18 DMPA.tw. (883)
19 Zoladex.tw. (1907)
20 (delay$ adj2 treat$).tw. (11681)
21 (hormon$ adj2 pretreat$).tw. (261)
22 (hormon$ adj2 inhibit$).tw. (3898)
23 GnRH agonist$.tw. (4275)
24 GnRH a.tw. (1036)
25 GnRHa.tw. (1329)
26 (endometri$ adj2 thin$).tw. (276)
27 (preoperative adj2 endometri$).tw. (171)
28 (preoperative$ adj2 treatment$).tw. (3749)
29 exp Gonadorelin/ (27707)
30 exp Androgen Antagonists/ (40578)
31 Androgen Antagonist$.tw. (248)
32 or/1-31 (225366)
33 (endometri$ adj2 ablat$).tw. (1473)
34 (laser adj2 ablat$).tw. (5366)
35 (endometr$ adj2 resect$).tw. (770)
36 (microwave adj2 ablat$).tw. (855)
37 rollerball.tw. (173)
38 (balloon adj2 ablat$).tw. (345)
39 (thermal adj2 ablat$).tw. (2107)
40 cryoablat$.tw. (2781)
41 (endometri$ adj2 inhibit$).tw. (390)
42 (endometri$ adj2 destr$).tw. (141)
43 (endometri$ adj2 removal).tw. (178)
44 TCRE.tw. (111)
45 (transcerv$ adj3 resect$).tw. (395)
46 electrosurg$.tw. (2994)
47 (hypertherm$ or photodynam$).tw. (45179)
48 thermotherap$.tw. (2304)
49 (phototherap$ or radiofreq$).tw. (33024)
50 (laser adj3 interstit$).tw. (753)
51 Thermachoice.tw. (79)
52 Cavaterm.tw. (25)
53 (Elitt or Vesta).tw. (63)
54 (Novasure or Cryogen).tw. (428)
55 (electrode$ adj2 ablat$).tw. (345)
56 (saline adj2 irrigat$).tw. (1017)
57 (heat$ adj2 balloon).tw. (22)
58 hysteroscop$.tw. (6580)
59 or/33-58 (100112)
60 59 and 32 (1821)
61 Clinical Trial/ (876616)
62 Randomized Controlled Trial/ (339824)
63 exp randomization/ (61095)
64 Single Blind Procedure/ (17193)
65 Double Blind Procedure/ (113926)
66 Crossover Procedure/ (36574)
67 Placebo/ (215599)
68 Randomi?ed controlled trial$.tw. (85221)
69 Rct.tw. (11175)
70 random allocation.tw. (1224)
71 randomly allocated.tw. (18514)
72 allocated randomly.tw. (1873)
73 (allocated adj2 random).tw. (717)
74 Single blind$.tw. (13139)
75 Double blind$.tw. (135239)
76 ((treble or triple) adj blind$).tw. (308)
77 placebo$.tw. (186716)
78 prospective study/ (229387)
79 or/61-78 (1317872)
80 case study/ (19126)
81 case report.tw. (241441)
82 abstract report/ or letter/ (863495)
83 or/80-82 (1119046)
84 79 not 83 (1281725)
85 84 and 60 (456)
86 (2012$ or 2013$).em. (1611640)
87 85 and 86 (34)

CINAHL

1     exp Gonadotropin-Releasing Hormone/

2     exp Goserelin/

3     exp Danazol/

4     exp Progestins/

5     buserelin/ or leuprolide/ or nafarelin/ or triptorelin/

6     exp cyproterone/ or exp cyproterone acetate/

7     cyproterone$.tw.

8     exp medroxyprogesterone/ or exp medroxyprogesterone 17-acetate/

9     medroxyprogesterone$.tw.

10     exp Norethindrone/

11     (norethisterone or Norethindrone).tw.

12     GnRH analog$.tw.

13     Gonadotropin-Releasing Hormone$.tw.

14     Gonadotrophin-Releasing Hormone$.tw.

15     (Goserelin or Danazol).tw.

16     (buserelin or leuprolide or nafarelin or triptorelin).tw.

17     (endometri$ adj2 prepar$).tw.

18     (endometri$ adj2 pretreat$).tw.

19     progestogen$.tw.

20     progestagen$.tw.

21     DMPA.tw.

22     Zoladex.tw.

23     (delay$ adj2 treat$).tw.

24     (hormon$ adj2 pretreat$).tw.

25     (hormon$ adj2 inhibit$).tw.

26     GnRH agonist$.tw.

27     GnRH a.tw.

28     GnRHa.tw.

29     (endometri$ adj2 thin$).tw.

30     (preoperative adj2 endometri$).tw.

31     (preoperative$ adj2 treatment$).tw.

32     or/1-31

33     (endometri$ adj2 ablat$).tw.

34     (laser adj2 ablat$).tw.

35     (endometr$ adj2 resect$).tw.

36     (microwave adj2 ablat$).tw.

37     rollerball.tw.

38     (balloon adj2 ablat$).tw.

39     (thermal adj2 ablat$).tw.

40     cryoablat$.tw.

41     (endometri$ adj2 inhibit$).tw.

42     (endometri$ adj2 destr$).tw.

43     (endometri$ adj2 removal).tw.

44     TCRE.tw.

45     (transcerv$ adj3 resect$).tw.

46     electrosurg$.tw.

47     (hypertherm$ or photodynam$).tw.

48     thermotherap$.tw.

49     (phototherap$ or radiofreq$).tw.

50     (laser adj3 interstit$).tw.

51     Thermachoice.tw.

52     Cavaterm.tw.

53     (Elitt or Vesta).tw.

54     (Novasure or Cryogen).tw.

55     (electrode$ adj2 ablat$).tw.

56     (saline adj2 irrigat$).tw.

57     (heat$ adj2 balloon).tw.

58     hysteroscop$.tw.

59     or/33-58

60     59 and 32

61     limit 60 to yr="2001 - 2008"

62     exp clinical trials/

63     Clinical trial.pt.

64     (clinic$ adj trial$1).tw.

65     ((singl$ or doubl$ or trebl$ or tripl$) adj (blind$3 or mask$3)).tw.

66     Randomi?ed control$ trial$.tw.

67     Random assignment/

68     Random$ allocat$.tw.

69     Placebo$.tw.

70     Placebos/

71     Quantitative studies/

72     Allocat$ random$.tw.

73     or/62-72

74     73 and 61

75     from 74 keep 1-7

PsycINFO to March 2013

1 pretreatment.tw. (11303)
2 gonadotropin releasing hormone agonist.tw. (31)
3 GnRH analog$.tw. (30)
4 gonadotropin-releasing hormone$.tw. (480)
5 Gonadotrophin-Releasing Hormone$.tw. (158)
6 (endometri$ adj2 prepar$).tw. (1)
7 GnRH agonist$.tw. (48)
8 (preoperative adj2 endometri$).tw. (0)
9 (preoperative$ adj2 treatment$).tw. (24)
10 (endometri$ adj2 thin$).tw. (2)
11 or/1-10 (11975)
12 endometri$.tw. (363)
13 rollerball.tw. (1)
14 ablat$.tw. (3766)
15 hysteroscop$.tw. (10)
16 electrosurg$.tw. (9)
17 TCRE.tw. (1)
18 or/12-17 (4136)
19 11 and 18 (40)
20 limit 19 to yr="2012 -Current" (5)

Contributions of authors

Yu Hwee Tan: wrote the protocol, performed searches, selected trials and extracted data for this review. Also entered the data and wrote this review.
Amita Singla: commented on the draft protocol and the review.
Anne Lethaby: supervised development of the protocol and commented on the draft. Was a co-author on this review and also selected trials, extracted and entered data from selected trials and commented on the review.

Declarations of interest

Martin Sowter is the principal author of one of the studies included in the review. No other conflicts of interest have been reported.

Sources of support

Internal sources

  • Department of Obstetrics and Gynaecology, University of Auckland, New Zealand.

External sources

  • None, Not specified.

Differences between protocol and review

Some of the primary outcomes in the protocol were moved to secondary outcomes.

The prespecified sensitivity analyses in the protocol could not be carried out in the review, as an insufficient number of studies was entered into the review. Otherwise no differences were identified between the protocol and the review.

Notes

This review updates and replaces the review Sowter 2002, with the same title 'Pre-operative endometrial thinning agents before endometrial destruction for heavy menstrual bleeding'.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Alborzi 2002

Methods

A single-centre, open, randomised study

Randomisation method not reported

No dropouts after randomisation

Power calculation not reported

Source of funding not reported

Participants

90 women

Inclusion criteria: chief complaint of menorrhagia; reproductive age; no response to medication or to D & C
Exclusion criteria: active pelvic inflammatory disease, malignant or premalignant endometrial pathology; extensive uterine cavities; distorted large fibroid; future desire to maintain fertility

Resectoscope used for surgery

Interventions

1. Preoperative thinning of the endometrium by danazol 600 mg/d for four to six weeks or GnRHa (Decapeptyl) one ampoule/mo for two months

2. No preoperative thinning of the endometrium

Timing of surgery not reported

Outcomes

1. Duration of operation

2. Menstrual blood loss at 12 months (amenorrhoea and failed treatment)

NotesThe number of participants randomly assigned not specifically stated—only the number of endometrial ablations performed
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described. Stated that the trial was randomised only in the abstract
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskThe two groups were given different preoperative treatment
Blinding of outcome assessment (detection bias)
All outcomes
High riskThe two groups were given different preoperative treatment
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo dropouts reported
Selective reporting (reporting bias)High riskData on side effects were collected but are not reported in the publication
Other biasUnclear riskNo table describes the characteristics of the randomly assigned groups, so it is not possible to assess whether the groups were similar at baseline

Bhatia 2008

Methods

A single-centre randomised controlled trial

Randomly assigned in 1:1 ratio

Computer-generated randomisation schedule

Intention-to-treat analysis

Power calculation for sample size performed—sample size of 100 women (i.e. 50 women per arm) to provide the study 80% power to detect 20% fall in the 'no bleed' rate from standard treatment

Serono UK Ltd (manufacturer of Cetrorelix) stated as providing unconditional grant to pay for trial nurse

Participants

106 premenopausal women
Inclusion criteria: selected transcervical resection of endometrium after failed medical management of dysfunctional uterine bleeding; normal smear history; o concurrent gynaecological problems; no desire for additional children; if submucous fibroids less than 5 cm; if on hormonal therapy or Mirena, had to discontinue treatment three months before surgery

Exclusion criteria: uterine size equivalent 12 weeks or more of pregnancy; previous transcervical resection of endometrium (TCRE); those with menopausal symptoms and significant medical problems or severe allergic reactions

Interventions

1. Cetrorelix 3 mg subcutaneous injection at four to seven days before TCRE and placebo injection at three to four weeks before TCRE

2. Leuprorelin acetate 3.75 mg subcutaneous injection at three to four weeks before TCRE and placebo injection at four to seven days before TCRE

Outcomes

Primary:

1. Amenorrhoea rate at six months after surgery

2. Endometrial thickness as measured on TVS on day of surgery

Secondary:

1. Intraoperative and histological assessment of the endometrium

2. Drug adverse effect profile

3. Biochemical evidence of suppression

4. Duration of operation

5. Operative view

6. Blood loss

7. Fluid deficit

8. Complication rate

9. Overall participant satisfaction in menstrual flow at six months after surgery

Notes

Three withdrawals in cetrorelix group for medical reasons

Three withdrawals in leuprorelin group—2 for medical reasons, 1 for personal reasons

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandomisation list was prepared with use of a computer-generated, 1:1 block randomisation schedule from North Staffordshire University. The list was kept by the trial nurse and was forwarded to the local ethics committee 
Allocation concealment (selection bias)Low riskThe injection was prepared by the trial nurse in a separate treatment room from the room where an investigator was seeing the participant. Once prepared, the injection was placed behind a curtain, and the participant was then brought into the treatment room. The woman was advised to lie down in semi-prone position on a couch while facing the wall, and the injection was administered in the gluteal region by the trial nurse
Blinding of participants and personnel (performance bias)
All outcomes
Low riskInjections given were blinded to participants, investigators and surgeons. Apart from helping with blood samples, the trial nurse was not involved in the measurement of any outcomes or analysis
Blinding of outcome assessment (detection bias)
All outcomes
Low riskApart from helping with blood samples, the trial nurse was not involved in the measurement of any outcomes or analysis
The randomisation list was forwarded to the main investigator (KB) only after completion of follow-up process 
Incomplete outcome data (attrition bias)
All outcomes
Low riskOf 110 consecutive eligible women, four declined to participate because they were not keen to have an extra injection. A total of 106 consecutive women undergoing TCRE were randomly assigned to receive either leuprorelin or cetrorelix before their operation. Six dropouts were reported, three in each group: Operation was cancelled in five women who had received both injections, because of newly detected medical problems (three raised blood pressure, one hyperglycaemia and one acute illness), and one participant from the leuprorelin group decided not go ahead with the operation for personal reasons a few days after her first injection. The minimum required total of 100 women underwent surgery, with 50 in each group and none lost to follow-up. All 100 women were analysed
Selective reporting (reporting bias)Low riskAll prespecified outcomes were reported
Other biasUnclear riskBaseline characteristics of both groups were similar. However, uterine cavity mean length was > 10% longer in cetrorelix group

Donnez 1997

Methods

A multi-centre double-blind randomised trial (37 centres in 12 countries)

(AZTEC—Adjunctive Zoladex for Thinning the Endometrium Comparison Study)

Randomly assigned in a 1:1 ratio

Computer-generated randomisation scheme produced for each centre

Intention-to-treat and per-protocol analyses of data performed.

Power calculation not reported.

Pharmaceutical company stated as source of funding

Participants

358 women

Inclusion criteria: over 30 years old, dysfunctional uterine bleeding (diagnosed clinically), recognisable menstrual cycles, uterine volume less than seven-week gestation, cavity length less than 10 cm, negative cervical smear in previous 12 months, negative endometrial biopsy, no pelvic inflammatory disease in previous two months

Exclusion criteria: pregnancy or breast feeding, previous endometrial resection, any hormonal agent used in previous two months, submucous fibroids, use of intrauterine device in previous three months

Resectoscope used for surgery

Interventions

1. GnRH analogue—two goserelin implants (3.6 mg) four weeks apart preoperatively (180 women)

2. Placebo—two placebo implants four weeks apart preoperatively (178 women)

In both groups, treatment was timed to allow surgery six weeks after the first injection and on day 7 of the menstrual cycle

Outcomes

1. Endometrial thickness at surgery (ultrasound and histological measurement)

2. Endometrial atrophy (on histology) at surgery

3. Duration of operation

4. Difficulty of operation (subjective assessment by surgeon as routine, easier or more difficult than usual)

5. Intraoperative distension medium absorption

6. Menstrual symptoms at 24 weeks postoperatively (using diary card to record blood loss and pain)

7. Participant satisfaction at 24 weeks postoperatively (assessed by participant questionnaire)

8. Side effects during treatment

Notes

One participant withdrew before treatment, five withdrew during treatment, six withdrew at or after surgery; six withdrawals (three in each group) were caused by adverse events (three considered treatment-related—one perforated uterus, a case of headache and nervousness in another participant)

Authors stated that outcomes were analysed on an intention-to-treat basis, but withdrawals were minimal after randomisation for some outcomes

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated randomisation scheme
Allocation concealment (selection bias)Low riskSequential allocation with central control
Blinding of participants and personnel (performance bias)
All outcomes
Low riskStated as double-blind—presumably participants and surgeons
Blinding of outcome assessment (detection bias)
All outcomes
Low riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskMinimal exclusions and lost to follow-up—unlikely to affect assessment of outcomes
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskSome baseline imbalances between groups

English 1998

Methods

A single-centre double-blind randomised trial

Method of randomisation not reported

Allocation concealment by opaque sealed envelopes

Minimal dropouts after randomisation: 3/39 for one outcome, 5/39 for another outcome

Power calculation not reported

Source of funding not reported

Participants

39 women

Inclusion criteria: diagnosis of dysfunctional uterine bleeding following hysteroscopy

Exclusion criteria: not reported

Resectoscope used for surgery

Interventions

1. GnRH analogue (Decapeptyl SR) 3.75 mg intramuscular injection (six and two weeks before surgery)

2. Danazol 200 mg capsules orally twice daily (six weeks before surgery until two weeks postsurgery)

3. Placebo capsules orally twice daily (six weeks before surgery until two weeks after surgery)

Follow-up for seven menstrual cycles

Outcomes

Primary:

Duration of surgery with TCRE

Secondary:

Pictorial blood assessment chart (PBAC) score; uterine fluid absorption; adverse events

NotesReasons for withdrawals not specified, but minimal loss to follow-up. Authors contacted to clarify data but information not available, and measures of variation for some outcomes were imputed
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskOpaque sealed envelopes
Blinding of participants and personnel (performance bias)
All outcomes
Low riskSurgeon blind to interventions for endometrial thinning. Participants blind to treatment with danazol or placebo (not with GnRHa, as this was given by injection)
Blinding of outcome assessment (detection bias)
All outcomes
Low riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskReasons not given for withdrawals, but minimal loss to follow-up
Selective reporting (reporting bias)Unclear riskFull information not provided on some of the prespecified outcomes
Other biasUnclear riskNo table provided to assess whether baseline variables in randomly assigned groups were similarly distributed. Source of funding not reported

Fraser 1996

Methods

A two-centre (Sydney, Melbourne) open randomised study

Randomly assigned in a 1:1 ratio with use of a computer-generated randomisation scheme

Power calculation for sample size performed—300 women per group required to show a difference in menstrual blood loss with a 95% confidence interval and a power of 90%. Authors rationalised that 25 women per group would be sufficient, based on previous studies

No intention-to-treat analysis

Pharmaceutical company stated as source of funding

Participants

60 women

Inclusion criteria: younger than 50 years of age, 21- to 42-day menstrual cycle, menorrhagia (diagnosed clinically), not responding to conventional medical treatment

Exclusion criteria: submucous fibroids, adenomyosis, hormonal agent used within two months, danazol or goserelin used within six months

Rollerball ablation used for surgery

Interventions

1. GnRH analogue—two goserelin implants (3.6 mg) four weeks apart, the first on day 2 of the menstrual cycle (30 women)

2. Danazol 200 mg orally twice daily for two months (30 women)

Outcomes

1. Endometrial thickness at surgery (subjective visual assessment by surgeon)

2. Uterine cavity length (before therapy and three months postoperatively)

3. Duration of operation

4. Difficulty of operation (subjective assessment by surgeon as simple, intermediate, difficult)

5. Distension medium absorption

6. Intraoperative blood loss (from haemoglobin in irrigation fluid using a modified alkaline-haematin method)

7. Menstrual blood loss at one, two, three and six months after surgery (using alkaline-haematin method)

8. Dysmenorrhoea for six months after surgery (by participant questionnaire)

9. Side effects during treatment

Notes

Three women withdrew from goserelin group and one woman from danazol group for reasons unrelated to the study medication. These women were excluded from data analysis

Dysmenorrhoea was graded using an ordinal scale as none, mild, moderate and severe

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated random number table
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen study with participants undergoing different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskMinimal dropouts and reasons given—unlikely to affect assessment of estimates
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskInsufficient data provided about baseline factors. Drugs for study provided by drug manufacturer

Garry 1996

Methods

A single-centre open randomised trial

Method of randomisation not described

Power calculation for sample size performed—at least 60 women per group required to detect a difference of 1.2 mm between goserelin and danazol at 90%

No intention-to-treat analysis

Pharmaceutical company stated as source of funding

Participants

160 women

Inclusion criteria: premenopausal, over 20 years old, dysfunctional uterine bleeding requiring hysterectomy

Exclusion criteria: pregnancy, breast feeding, hormonal agents in the previous month, danazol or GnRH analogues in the previous six months

Nd:YAG laser used for surgery

Interventions

1. GnRH analogue—two goserelin implants (3.6 mg) at eight and four weeks before surgery (38 women)

2. Danazol 200 mg four times a day for eight weeks (33 women)

3. GnRH analogue—one goserelin implant (3.6 mg) at four weeks before surgery (38 women)

4. Danazol 200 mg four times per day for four weeks (36 women)

Outcomes

1. Endometrial thickness before and after pretreatment (by ultrasound, histology and subjective measurement by surgeon)

2. Uterine cavity length (before and after therapy)

3. Duration of operation

4. Distension medium absorption

5. Menstrual loss (assessment by clinician at six months)

6. Participant satisfaction (assessment by clinician at six months)

7. Side effects during treatment

Notes

Six women withdrew after randomisation. 15 women did not complete the study

Postoperative menstrual loss assessed on ordinal scale as amenorrhoea, hypomenorrhoea, normal flow or no improvement

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen-label study in which participants received different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNo reasons given for dropouts and not balanced across randomly assigned groups
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskFunding by pharmaceutical company

Jack 2005

Methods

Single-centre open randomised trial

Computer-generated balanced random number blocks in 1:1 ratio used

Power calculation for sample size: at least 90 participants per group required to have 80% power to detect a 20% difference in satisfaction and a 10% difference in acceptability

Modified intention-to-treat analysis for most outcomes (no inclusion of 13 dropouts after randomisation because women did not fit inclusion criteria of the trial)

Funding by the Scottish Executive Health Department

Participants

210 women (13 of these excluded/dropped out after randomisation)

Inclusion criteria: complaint of excessive menstrual; agreeable to microwave endometrial ablation under local anaesthesia; normal endometrial pathology; family completed; uterine size ≤ 12 weeks

Exclusion criteria: not reported

Surgery by microwave

Location: Aberdeen, UK

Interventions

1. Either Danazol 200 mg twice a day for four to five weeks or GnRHa (Goserelin) 3.6 mg subcutaneous injection five weeks before surgery

2. No preoperative treatment—surgery in the postmenstrual phase of the cycle

Outcomes

Primary:

1. Acceptability of treatment (at two weeks)

2. Satisfaction rate (at 12 months)

3. Participant satisfaction rate (at five years)

Secondary:

1. Duration of operation

2. Endometrial thickness

3. SF-12 QOL scores

4. Menstrual blood loss at 12 months and five years

5. Side effects

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated balanced random number blocks
Allocation concealment (selection bias)Low riskSealed opaque sequentially numbered randomisation envelopes at a separate site—telephoned through to a distant site
Blinding of participants and personnel (performance bias)
All outcomes
High riskParticipants had different preoperative protocols
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot mentioned whether outcome assessment was blinded
Incomplete outcome data (attrition bias)
All outcomes
Low riskReasons carefully specified for exclusions before treatment and dropouts after treatment—unlikely to affect assessments of estimates
Selective reporting (reporting bias)Low riskAll prespecified outcomes reported
Other biasLow riskGroups balanced at baseline and no evidence of other potential bias

Kriplani 2001

Methods

A single-centre randomised trial

Participants were reported as blinded, but lack of a placebo control group may have unmasked the participants

Method of randomisation by computer0generated randomisation table

Power calculation for sample size not reported by power calculations reported after data analysis

Intention-to-treat analysis not reported, but no dropouts were reported

Source of funding not reported

Participants

50 women

Inclusion criteria: history of heavy menstrual bleeding confirmed by pictorial chart assessment, but details not given; no further desire for childbearing; uterus < 12 weeks gravid size; negative cervical cytology; benign endometrial histology

Exclusion criteria: bleeding disorder

Surgery by resectoscope

Location: New Delhi, India

Source: not described

Interventions

1. DMPA (depot medroxy progesterone acetate—150 mg six weeks before surgery + another dose after surgery)

2. No preoperative or postoperative treatment—immediate surgery in the postmenstrual phase

Outcomes

1. Duration of surgery (minutes)

2. Fluid deficit (mL)

3. Postoperative amenorrhoea (up to four years of follow-up)

4. Requirement for further surgery (up to four years)

5. Endometrial thickness (mm)

NotesNo reported dropouts, but number of eligible women not reported
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated randomisation table
Allocation concealment (selection bias)Unclear riskNot described
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskStated as single-blind but not clear who was blinded
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo dropouts reported
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskNot clear whether groups were balanced at baseline

Kriplani 2002

Methods

Single-centre open randomised trial

Computer-generated randomisation method

No power calculation for sample size reported

Reasons for withdrawals not given, but minimal at one year and not likely to affect estimates. Estimates at six years only for a subgroup of all randomly assigned participants

Intention-to-treat analysis not reported

Source of funding not reported

Participants

132 women

Inclusion criteria: excessive menstrual blood loss (defined) confirmed by PBAC (value not given); no bleeding disorders; negative cervical cytology and benign endometrial histology; uterus ≤ 12 weeks in size; no further desire for childbearing

Exclusion criteria: not reported

Resectoscope used for surgery

Interventions

1. Danazol 400 to 600 mg/d for four to six weeks before surgery

2. No preoperative treatment—surgery in the postmenstrual phase

Duration: six years (subgroup of participants only)

Outcomes

1. Endometrial thickness (mm)

2. Fluid deficit (mL)

3. Duration of operation

4. Requirement for repeat surgery

5. Menstrual blood loss at one and six years

6. Dysmenorrhoea

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer generated
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskParticipants had different pretreatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskFor immediate postoperative outcomes, minimal loss to follow-up; for assessment of menstrual symptoms and additional surgery at six years, approximately half of each randomly assigned group lost to follow-up, so these assessments were biased
Selective reporting (reporting bias)Unclear riskNo published protocol identified and no adverse events reported
Other biasLow riskGroups balanced at baseline and no evidence of other potential bias

Lissak 1999

Methods

A single-centre open randomised trial

Randomly assigned using a random number table

Pilot study—no power calculation for sample size and intention-to-treat analysis not reported

No source of funding stated

Participants

30 women

Inclusion criteria: 30 to 55 years of age

Menorrhagia confirmed using pictorial chart

Failed therapy with oral contraceptive agents

Exclusion criteria: Active or chronic pelvic inflammatory disease, undiagnosed vaginal bleeding, previous caesarean section

Thermal balloon used for ablation

Interventions

1, GnRH analogue—decapeptyl depot 3.75 mg one month before surgery

2. No pretreatment

Outcomes

1. Uterine cavity length

2. Duration of operation

3. Menstrual loss at six months

4. Mean duration of menstrual flow (days) at six months

5. Participant satisfaction

Notes

Duration of operation not included in this review, as with balloon ablation, this is independent of any pretreatment

Unequal numbers in randomly assigned groups—13 in one group and 17 in the other

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskTable of random numbers
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskParticipants underwent different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo dropouts reported
Selective reporting (reporting bias)Unclear riskNo prior published protocol
Other biasUnclear riskUnequal numbers in randomly assigned groups

Rai 2000

Methods

Single-centre open randomised trial

Randomly assigned using sealed opaque envelopes

No power calculation for sample size reported. No withdrawals, so analysis by intention to treat

No source of funding stated

Participants

100 women

Inclusion criteria: not stated

Exclusion criteria: not stated

Resectoscope used for surgery

Interventions

1. GnRH analogue—naferelin nasal spray twice daily for eight weeks (25 women)

2. Danazol—200 mg three times per day for eight weeks (25 women)

3. Progestogen—medroxyprogesterone acetate 10 mg three times daily for eight weeks (25 women)

4. No preoperative therapy—surgery immediately after menstruation ceased (25 women)

Outcomes

1. Endometrial thickness (objective assessment by pathologist)

2. Endometrial histology after therapy

3. Menstrual loss at 12 months (assessment by patient)

4. Participant satisfaction at 12 months

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskMethod not reported
Allocation concealment (selection bias)Low riskNumbered sealed opaque envelopes
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskParticipants not blinded, but surgery performed by one blinded surgeon and endometrial outcomes assessed by blinded histopathologist
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskAs above
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo withdrawals after randomisation
Selective reporting (reporting bias)Unclear riskNo prior published protocol
Other biasUnclear riskNot clear whether groups were balanced at baseline, as no table of values reported

Romer 1996

Methods

A single-centre open randomised trial

Method of randomisation not described

No power calculation for sample size reported, and intention-to-treat analysis not reported

No source of funding stated

Participants

40 women

Inclusion criteria: clinically diagnosed menorrhagia

Exclusion criteria: cavity length over 10 cm, submucous fibroids, malignant pathology on biopsy

Rollerball ablation used for surgery

Interventions

1. GnRH analogue—decapeptyl depot four weeks preoperatively (10 women)

2. Danazol—200 mg three times per day for four weeks (10 women)

3. Progestogen—orgametril (lyestrenol) 10 mg per day for four weeks (10 women)

4. No preoperative therapy—surgery on day 5 to 9 of cycle (10 women)

Outcomes

1. Endometrial thickness (subjective assessment by surgeon)

2. Endometrial histology after therapy

3. Menstrual loss at six months (assessment by clinician)

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskParticipants underwent different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskParticipants underwent different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo dropouts reported
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskNot clear whether groups balanced at baseline

Shawki 2002

Methods

A multi-centre (n = 3) open randomised parallel-group trial

Method of randomisation not described

No power calculation for sample size and no reported dropouts, so analysis by intention to treat

No source of funding reported

Participants

131 women

Inclusion criteria: abnormal uterine bleeding refractory to medical management at a level to justify hysterectomy; no longer desiring to have children; uterine size < 12 weeks; participant refusal of continued medication attempts

Exclusion criteria: neoplasia

Endometrial destruction by resectoscope or by rollerball, according to surgeon preference

Interventions

1. GnRHa 3.6 mg sc for one month (23 women)

2. GnRHa 3.6 mg sc for three months (26 women)

3. Danazol 400 to 600 mg per day for three months (26 women)

4. Medroxyprogesterone acetate (MPA) before surgery (27 women)

5. Dilatation & curettage immediately before the procedure (39 women)

Outcomes

1. Menstrual blood loss at 12 months

2. Duration of operation

3. Side effects

Duration: one year

NotesNumbers unbalanced between randomly assigned groups
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskParticipants had different pretreatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskParticipants had different pretreatment protocols
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo details of attrition
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskNo indication whether groups balanced at baseline. Numbers in randomly assigned groups very different

Sorensen 1997

Methods

A single-centre open randomised trial

Randomised by computer into three groups

No power calculation for sample size reported

No intention-to-treat analysis

Pharmaceutical company stated as source of funding

Participants

60 women

Inclusion criteria: clinically diagnosed menorrhagia

Exclusion criteria: premalignant or malignant endometrial histology on biopsy, fibroids larger than 5 cm in diameter

Resectoscope used for surgery

Interventions

1. GnRH analogue—one goserelin implant (3.6 mg) four to six weeks before surgery

2. GnRH analogue—two goserelin implants (3.6 mg) one four to six weeks before surgery and one on the day of surgery

3. No preoperative therapy—timing of surgery within cycle not reported

Outcomes

1. Weight of endometrial strips removed

2. Duration of operation

3. Distension medium absorption

4. Menstrual loss at six and 12 months (as assessed by participant questionnaire, measurement of the number of sanitary towels used and number of days of menstrual loss)

5. Dysmenorrhoea at 12 months (as assessed by participant questionnaire)

6. Patient satisfaction (as assessed by patient questionnaire)

NotesStandard deviation for distension medium absorption estimated from mean and range
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer randomisation
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen study in which participants received different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskOpen study in which participants received different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
Low riskMinimal dropouts and balanced between randomly assigned groups—unlikely to affect assessment of estimates
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskFunding by pharmaceutical company

Sowter 1997

Methods

A single-centre open randomised trial

Method of randomisation not described

No power calculation for sample size reported

Analysis not by intention to treat

No source of funding stated

Participants

77 women

Inclusion criteria: clinically diagnosed menorrhagia, younger than age 50

Exclusion criteria: not described

Resectoscope used for surgery

Interventions

1. GnRH analogue—two goserelin implants (3.6 mg) eight and four weeks preoperatively (35 women)

2. No preoperative therapy—surgery performed in early proliferative phase of menstrual cycle (40 women)

Outcomes

1. Menstrual loss at 12 to 24 months (as assessed by participant questionnaire)

2. Dysmenorrhoea at 12 to 24 months (as assessed by participant questionnaire)

3. Participant satisfaction at 12 to 24 months (as assessed by participant questionnaire)

Notes 
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen study in which participants received different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskOpen study in which participants received different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
High riskWomen not analysed in their randomly assigned groups, and assessment of estimates could be biased
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskBaseline characteristics compared according to treatment received, not according to randomisation schedule

Sutton 1994

Methods

A single-centre open randomised trial

Randomly assigned in a 1:1 ratio

Randomisation method not described

No power calculation for sample size reported

No intention-to-treat analysis

No source of funding stated

Participants

120 women

Inclusion criteria: premenopausal, 25 to 50 years of age with clinically diagnosed menorrhagia; uterus less than 12 weeks of gestation in size, cavity length less than 12 cm

Exclusion criteria: endometrial hyperplasia present on biopsy; hormonal agents used in previous two months, or danazol or goserelin used in previous six months

Resectoscope used for surgery

Interventions

1. GnRH analogue—two goserelin implants at eight and four weeks before surgery (55 women)

2. Danazol—200 mg orally four times per day for eight weeks before surgery (55 women)

Outcomes

1. Endometrial thickness (by ultrasound, histology and subjective visual assessment by surgeon)

2. Duration of operation

3. Difficulty of operation (assessment by surgeon)

4. Menstrual loss at 12 weeks (as assessed by clinician)

5. Participant satisfaction at 12 weeks (as assessed by clinician)

Notes

10 women withdrew from study before taking medication

Seven danazol participants withdrew from study because of side effects

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskRandomisation method not described
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
High riskOpen study with participants undergoing different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskOpen study in which participants received different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
High riskWithdrawals unbalanced between groups and related to effects of treatment received
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasUnclear riskNot clear whether groups balanced at baseline

Taskin 1996

Methods

A single-centre (Turkey) double-blind randomised trial

Randomly assigned in a 1:1 ratio using a table of random numbers—concealment of allocation not described

No power calculation for sample size reported

Intention-to-treat analysis not reported

No source of funding described

Participants

13 women

Inclusion criteria: dysfunctional uterine bleeding that did not respond to medical therapy; consent to undergo hysteroscopy

Exclusion criteria: abnormal smear; suspected uterine malignancy; submucosal fibroids or masses distorting uterine lining; adnexal pathology; "large" (not defined) uterine volume

All women underwent routine cervical smear; endometrial sampling; transvaginal and transabdominal ultrasound examinations

Rollerball used for surgery

Interventions

1. GnRH analogue—depot goserelin (3.75 mg) on day 21 of cycle, with a second dose given four weeks later (seven women)

2. Placebo—two saline injections four weeks apart (six women)

Surgery was performed 10 to 14 days after the last injection under general anaesthesia

Outcomes

Primary:

Degree of fluid absorption (mL)

Other:

Operating time (minutes)

NotesNo reported dropouts
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskTable of random numbers
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
Low riskStated as double-blind with placebo control
Blinding of outcome assessment (detection bias)
All outcomes
Low riskStated as double-blind with placebo control
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo reported dropouts
Selective reporting (reporting bias)Unclear riskNo prior published protocol identified
Other biasLow riskGroups appear balanced at baseline; no source of funding reported

Taskin 1998

Methods

A single-centre parallel-group randomised controlled trial

Randomisation on 1:1 ratio using a table of random numbers—concealment of allocation not reported

No power calculation for sample size reported

Intention-to-treat analysis not reported

No source of funding described

Participants

17 women

Inclusion criteria: diagnosis of dysfunctional uterine bleeding that did not respond to conservative treatment

Exclusion criteria: abnormal cervical smear; suspected uterine malignancy; submucosal fibroids or masses distorting the endometrial lining; adnexal pathology

All women had preoperative cervical smears, endometrial sampling and transvaginal and transabdominal ultrasound examinations

Surgery by rollerball

Interventions

1. GnRH analogue (not specified) 3.75 mg on day 21 of the menstrual cycle and four weeks later, with surgery scheduled 10 to 14 days after the second dose

2. Placebo (saline injections) under same regimen approximately two months before surgery

Outcomes

Primary:

Amount of irrigant and irrigant deficit

Secondary:

Operation time

NotesAuthors contacted for clarification of data, but information not available
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskTable of random numbers
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
Low riskInvestigators stated as blinded to allocation; not explicitly stated whether participants were blinded, although placebo treatment makes this likely. Outcomes unlikely to be affected by participants' knowledge of treatment
Blinding of outcome assessment (detection bias)
All outcomes
Low riskInvestigators stated as blinded to allocation; not explicitly stated whether participants were blinded, although placebo treatment makes this likely. Outcomes unlikely to be affected by participants' knowledge of treatment
Incomplete outcome data (attrition bias)
All outcomes
Low riskNo reported dropouts after randomisation
Selective reporting (reporting bias)Unclear riskResults of all outcomes measured not clearly reported
Other biasLow riskGroups appeared similar at baseline. Funding source not reported

Vercellini 1996

Methods

Multi-centre randomised trial (three centres in Italy)

Centralised randomisation in a 1:1 ratio by telephone using computer-generated randomisation scheme

Power calculation for sample size—80% power to detect a difference in fluid absorption of 300 mL; at level of 5% 30 women required per group

No intention-to-treat analysis

Pharmaceutical company stated as source of funding

Participants

71 women

Inclusion criteria: aged 40 or older with menorrhagia diagnosed clinically and with pictorial blood loss assessment chart, uterus less than eight weeks of gestation in size

Exclusion criteria: pregnant, breast feeding, recently used hormonal agents or inhibitors of menstruation, fibroids larger than 3 cm in diameter

Interventions

1. GnRH analogue—goserelin implant (3.6 mg) eight and four weeks before surgery (35 women)

2. No preoperative therapy—surgery in early proliferative phase of menstrual cycle (36 women)

Outcomes

1. Endometrial thickness (subjective visual assessment by surgeon)

2. Duration of operation

3. Distension medium absorption

4. Time to return to work or to resume domestic activities

NotesEight women withdrew from the study before surgery for reasons unrelated to their therapy. Two additional women withdrew after surgery
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskComputer-generated randomisation list
Allocation concealment (selection bias)Low riskCentralised allocation by telephone
Blinding of participants and personnel (performance bias)
All outcomes
High riskWomen had different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskWomen had different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskWithdrawals unbalanced between groups
Selective reporting (reporting bias)Unclear riskNo prior protocol identified
Other biasLow riskNo other potential source of bias identified

Vilos 2010

  1. a

    D & C: Dilatation & curettage (operation in which the cervix is dilated and the endometrial lining of the womb scraped).

    PBAC: Pictorial blood assessment chart (for women to chart their blood loss via symbols).

    sc: subcutaneous (injection under the skin).

Methods

A single-centre parallel-group randomised controlled trial

Randomisation method not described

No power calculation for sample size reported

No intention-to-treat analysis

Source of funding not reported

Participants

105 women

Inclusion criteria: participant’s Higham menstrual score > 150 one to two months before surgery

Exclusion criteria: Genital tract malignancy; unresolved endometrial hyperplasia; anatomic or pathologic uterine anomalies; history of previous classical caesarean section or transmural myomectomy; intrauterine pregnancy; acute genital and/or urinary tract infection; women wishing to preserve their fertility; women expecting amenorrhoea as an outcome; IUCD in place; failed previous endometrial ablation

Interventions

1. GnRH analogue—goserelin (3.6 mg sc) four weeks before surgery

2. Suction curettage immediately before surgery

Surgery by ThermaChoice II balloon endometrial ablation

Outcomes

1. Reduction in Higham score

2. Combined amenorrhoea/hypomenorrhoea

3. Eumenorrhoea

4. Success of treatment (normal bleeding or less)

5. Dysmenorrhoea

6. Premenstrual syndrome symptoms

7. Participant satisfaction rates

8. Further surgery (any or hysterectomy)

9. Adverse events

10. Health-related quality of life

All outcomes assessed at 12 months

Notes

Five women withdrew after randomization but before treatment

Two women withdrawn at time of surgery as unable to be treated with balloon endometrial ablation for reasons not associated with treatment

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskMethod of randomization not explained
Allocation concealment (selection bias)Low riskConcealment was achieved by placing the randomization code into an opaque envelope, and treatment allocation was revealed after entry criteria had been met and informed consent obtained
Blinding of participants and personnel (performance bias)
All outcomes
High riskNo mention of blinding of participants or personnel; however, would be difficult to blind between suction curettage and goserelin as different treatment protocols
Blinding of outcome assessment (detection bias)
All outcomes
High riskNo mention of blinding of outcome assessors to interventions and allocations, but would be difficult to blind given different treatment protocols
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskAttrition rate > 10%, but no mention in article about number of participants required to gain adequate power
Selective reporting (reporting bias)High riskA lot of outcomes looked, at but not all reported or reported adequately
Other biasUnclear riskBaseline characteristics of both groups similar, except for uterine balloon volume (50% larger in suction curettage group)

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Alford 1996This comparative study was excluded because randomisation was not used. Drug regimens were changed sequentially as authors "gained more experience of their effectiveness". The effects of leuprolide acetate (a GnRH analogue) (24 participants), danazol (six participants), tamoxifen (four participants) and medroxyprogesterone acetate (Depo-Provera) (six participants) were compared. Endometrial atrophy at surgery and postoperative menstrual symptoms were evaluated. Leuprolide acetate and danazol appeared to be superior to tamoxifen and medroxyprogesterone acetate, but the duration of follow-up time was also shorter in these first two groups
Cicenelli 2007This randomised study was excluded because the participant group consisted of women with endometrial polyps who were undergoing hysteroscopic surgery. It does not meet the inclusion criteria for participants
Cooper 1996This randomised trial used cervical resistance to dilatation as its only treatment outcome measure within a larger randomised trial comparing hysterectomy with hysteroscopic treatments for menorrhagia. The effects of goserelin (60 participants) in women allocated to endometrial resection or laser ablation were compared with no treatment (20 participants) in women allocated to hysterectomy. Goserelin increased cervical resistance. This study also evaluated in a double-blind randomised manner the effect of misoprostol in reducing GnRH-induced cervical resistance to dilatation (64 women randomly assigned to misoprostol or placebo). Misoprostol had no beneficial effect. The study was excluded because no other intraoperative or postoperative outcomes were assessed, and the study treatment groups did not meet inclusion criteria
Cooper 2001This was a prospective double-arm controlled observational study assessing endometrial pretreatment with GnRHa versus immediate ablation with the 3D bipolar NovaSure system. No indication suggested that the groups had been allocated randomly
Florio 2010This study was excluded, as it looked at hysteroscopic surgery in women with endometrial polyps rather than at endometrial ablation techniques. It does not meet the inclusion criteria for participants
Florio 2012This study was excluded, as it looked at hysteroscopic surgery rather than endometrial ablation techniques
Gabbanini 2009This study was excluded, as it looked at hysteroscopic surgery in women with intrauterine pathology (polyps, submucous myoma, septate uterus) rather than endometrial ablation techniques. It does not meet the inclusion criteria for participants
Mavrelos 2010This randomised study was excluded because it looked at the use of preoperative GnRH analogues before hysteroscopic treatment of women with submucous fibroids. It does not meet the inclusion criteria for participants
Muzii 2010This randomised study was excluded because it looked at the efficacy of GnRH analogue treatment before hysteroscopic resection of submucous myomas in participants with abnormal uterine bleeding. It does not meet the inclusion criteria for participants
Perino 1993A prospective comparative study of 193 participants undergoing hysteroscopic surgery for uterine septa, submucous fibroids and abnormal uterine bleeding. Within this study were 56 women who were undergoing endometrial resection for abnormal uterine bleeding with benign pathology. 36 of these women received two injections of leuprolide acetate depot (3.75 mg) at 28-day intervals. Operating time, intraoperative blood loss, volume of distension medium used and number of subsequent treatment failures at 12 months after surgery were evaluated. The study was excluded because no randomisation technique was used to determine treatment allocation
Rich 1995A randomised study of 48 women comparing danazol, norethisterone, cyproterone acetate and medroxyprogesterone acetate taken for four weeks before endometrial ablation. The study was excluded, as it did not meet inclusion criteria for outcome measures
Serden 1992A comparison of preoperative GnRH analogue (leuprolide acetate) (56 women), danazol (26 women), progestogens (medroxyprogesterone acetate or norethindrone acetate) (nine women) and no therapy (46 women). GnRH analogues appeared to result in the highest amenorrhoea rate, but this group also had the shortest duration of follow-up. This study was excluded because although participants were randomly allocated to different treatments, the numbers in each group and the duration of follow-up were very different, suggesting that true randomisation was not used
Triolo 2006This randomised study was excluded because the participant group consisted of women with endo-uterine pathologies (polyps, submucous myoma, septate uterus) who were undergoing hysteroscopic surgery. It does not meet the inclusion criteria for participants
Trivedi 1999A comparison of pretreatment with danazol (28 women), goserelin (four women), prior suction curettage (three women), progestogens (four women) and no endometrial preparation (625 women) and their effects on postoperative symptoms and participant satisfaction. This study was not randomised
Vercellini 1994A comparison of goserelin depot (3.6 mg) (33 participants) and surgery in proliferative phase of the cycle (22 participants). The study compared duration of operation, distention medium absorption, difficulty of surgery, duration of postoperative bleeding, time to return to work and time to resume normal domestic activities. Goserelin appeared to reduce operating time and distension medium absorption. The duration of postoperative bleeding and the time to return to work were unaffected. The study was excluded because participant choice rather than randomisation was used to determine treatment allocation
Zapico 2005This study was excluded because it was not randomised, and it does not meet the inclusion criteria for study design

Characteristics of studies awaiting assessment [ordered by study ID]

Gannon 1994

MethodsAbstract not available
Participants 
Interventions 
Outcomes 
Notes 

Mayonda 1994

MethodsOpen randomised parallel-group study
Participants48 women. Inclusion criteria: menorrhagia, endometrial ablation planned as treatment
Interventions(1) Danazol; (2) cyproterone acetate; (3) norethisterone; (4) medroxyprogesterone acetate for four weeks before ablation
Outcomes(1) Endometrial thickness by ultrasonography; (2) side effects
NotesAbstract presented at the Victor Bonney Society Meeting in 1994

McDonald 1994

MethodsParallel-group open-label randomised controlled trial
Participants42 women. No inclusion or exclusion criteria reported in abstract
Interventions

(1) goserelin implant (3.6 mg) one month before surgery; (2) no preoperative treatment

Surgery with intramenstrual resection

OutcomesEase of surgery, fluid loss, amenorrhoea rates, further surgery
NotesNo reply received from author

Nagele 1995

MethodsParallel-group placebo-controlled randomised trial
Participants

49 women

No inclusion criteria reported

Exclusion criteria: irregular menses, uterus > 12 weeks in size, submucous fibroids > 5 cm; recent gynaecological treatment, serious illness

Interventions(1) Danazol 200 mg three times daily for six weeks before surgery; (2) placebo
OutcomesAdverse events, menstrual loss and other unspecified outcomes
NotesAbstract presented at 27th British Congress of Obstetrics and Gynaecology

Watermeyer 2005

MethodsOpen randomised parallel-group study
Participants

40 women

Inclusion criteria: validated dysfunctional uterine bleeding

Interventions(1) GnRHa agonist; (2) timed treatment after end of a period
OutcomesMenstrual loss; amenorrhoea rates; participant satisfaction
NotesAuthor could not be reached by email. Abstract presented at the Royal College of Obstetricians and Gynaecologists 6th International Scientific Meeting, September 2005

Ancillary