Levonorgestrel‐releasing intrauterine device (LNG‐IUD) for symptomatic endometriosis following surgery
Abstract
Background
Endometriosis is a condition characterised by the presence of ectopic deposits of endometrial‐like tissue outside the uterus, usually in the pelvis. The impact of laparoscopic treatment on overall pain is uncertain and a significant proportion of women will require further surgery. Therefore, adjuvant medical therapies following surgery, such as the levonorgestrel‐releasing intrauterine device (LNG‐IUD), have been considered to reduce recurrence of symptoms.
Objectives
To determine the effectiveness and safety of post‐operative LNG‐IUD in women with symptomatic endometriosis.
Search methods
We searched the following databases from inception to January 2021: The Specialised Register of the Cochrane Gynaecology and Fertility Group, CENTRAL (which now includes records from two trial registries), MEDLINE, Embase, PsycINFO, LILACS and Epistemonikos. We handsearched citation lists of relevant publications, review articles, abstracts of scientific meetings and included studies. We contacted experts in the field for information about any additional studies.
Selection criteria
We included randomised controlled trials (RCTs) comparing women undergoing surgical treatment of endometriosis with uterine preservation who were assigned to LNG‐IUD insertion, versus control conditions including expectant management, post‐operative insertion of placebo (inert intrauterine device), or other medical treatment such as gonadotrophin‐releasing hormone agonist (GnRH‐a) drugs.
Data collection and analysis
Two review authors independently selected studies for inclusion, and extracted data to allow for an intention‐to‐treat analysis. For dichotomous data, we calculated the risk ratio (RR) and 95% confidence interval (CI) using the Mantel‐Haenszel fixed‐effect method. For continuous data, we calculated the mean difference (MD) and 95% CI using the inverse variance fixed‐effect method.
Main results
Four RCTs were included, with a total of 157 women. Two studies are ongoing. The GRADE certainty of evidence was very low to low. The certainty of evidence was graded down primarily for serious risk of bias and imprecision.
LNG‐IUD versus expectant management
Overall pain: No studies reported on the primary outcome of overall pain.
Dysmenorrhoea: We are uncertain whether LNG‐IUD improves dysmenorrhoea at 12 months. Data on this outcome were reported on by two RCTs; meta‐analysis was not possible (RCT 1: delta of median visual analogue scale (VAS) 81 versus 50, P = 0.006, n = 55; RCT 2: fall in VAS by 50 (35 to 65) versus 30 (25 to 40), P = 0.021, n = 40; low‐certainty evidence).
Quality of life: We are uncertain whether LNG‐IUD improves quality of life at 12 months. One trial demonstrated a change in total quality of life score with postoperative LNG‐IUD from baseline (mean 61.2 (standard deviation (SD) 14.8) to 12 months (mean 70.3 (SD 16.2) compared to expectant management (baseline 55.1 (SD 17.0) to 57.0 (SD 33.2) at 12 months) (n = 55, P = 0.014, very low‐certainty evidence).
Patient satisfaction: Two studies found higher rates of satisfaction with LNG‐IUD compared to expectant management; however, combining the studies in meta‐analysis was not possible (n = 95, very low‐certainty evidence). One study found 75% (15/20) of those given post‐operative LNG‐IUD were "satisfied" or "very satisfied", compared to 50% (10/20) of those in the expectant management group (RR 1.5, 95% CI 0.90‐2.49, 1 RCT, n=40, very low‐certainty evidence). The second study found that fewer were "very satisfied" in the expectant management group when compared to LNG, but there were no data to include in a meta‐analysis.
Adverse events: One study found a significantly higher proportion of women reporting melasma (n = 55, P = 0.015, very low‐certainty evidence) and bloating (n = 55, P = 0.021, very low‐certainty evidence) following post‐operative LNG‐IUD. There were no differences in other reported adverse events, such as weight gain, acne, and headaches.
LNG‐IUD versus GnRH‐a
Overall pain: No studies reported on the primary outcome of overall pain.
Chronic pelvic pain: We are uncertain whether LNG‐IUD improves chronic pelvic pain at 12 months when compared to GnRH‐a (VAS pain scale) (MD ‐2.0, 95% CI ‐20.2 to 16.2, 1 RCT, n = 40, very low‐certainty evidence).
Dysmenorrhoea: We are uncertain whether LNG‐IUD improves dysmenorrhoea at six months when compared to GnRH‐a (measured as a reduction in VAS pain score) (MD 1.70, 95%.CI ‐0.14 to 3.54, 1 RCT, n = 18, very low‐certainty evidence).
Adverse events: One study suggested that vasomotor symptoms were the most common adverse events reported with patients receiving GnRH‐a, and irregular bleeding in those receiving LNG‐IUD (n = 40, very low‐certainty evidence)
Authors' conclusions
Post‐operative LNG‐IUD is widely used to reduce endometriosis‐related pain and to improve operative outcomes. This review demonstrates that there is no high‐quality evidence to support this practice. This review highlights the need for further studies with large sample sizes to assess the effectiveness of post‐operative adjuvant hormonal IUD on the core endometriosis outcomes (overall pain, most troublesome symptom, and quality of life).
PICOs
Plain language summary
Use of a levonorgestrel‐releasing intrauterine device (LNG‐IUD) for reducing pain in women who have had surgery for endometriosis
What is endometriosis?
Endometriosis is a chronic pain syndrome associated with the presence of endometrial (womb‐like) tissue outside the womb, usually in the pelvis, that can lead to infertility and pelvic pain (pain below the belly button).
How is endometriosis treated?
Endometriosis is usually managed with hormonal medications, surgery, or a combination of both. The progestogen levonorgestrel is one such hormonal medication that is believed to stop the growth of endometrial tissue outside the womb.
What is the aim of the review?
The aim of this review was to assess whether the use of a LNG‐IUD was beneficial for managing associated painful symptoms and improving the quality of life and patient satisfaction in women who have recently had surgery for endometriosis.
What did the review find?
At this stage, there is not enough evidence to support the use of LNG‐IUD after surgery to reduce pain caused by endometriosis. Although there was some evidence of a benefit in reducing painful periods and improving quality of life and satisfaction when using LNG‐IUD after surgery, the certainty of evidence was low to very low, due to the small numbers of studies and study participants, as well as flaws in study design. This suggests that further studies are required before a recommendation can be made for its use.
Authors' conclusions
Summary of findings
| Postoperative use of LNG‐IUD compared to no postoperative treatment for symptomatic endometriosis following surgery | ||||||
| Patient or population: Patients with symptomatic endometriosis following surgery | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Expectant management | Postoperative use of LNG‐IUD | |||||
| Overall pain at 12 months | No studies reported on this outcome. | |||||
| Chronic pelvic pain at 12 months | No studies reported on this outcome. | |||||
| Dysmenorrhoea at 12 months | Tanmahasamut 2012 found a median reduction of 50 mm (0‐78) (pain scores on a 100mm VAS) P < .001 Vercellini 2003 found a median reduction of 30 mm (25‐40) (pain scores on a 100mm VAS) P = 0.021 | Tanmahasamut 2012 found a median reduction of 81 mm (51.5 ‐87.5) (pain scores on a 100mm VAS) P < .001. Vercellini 2003 found a median reduction of 50 mm (35–65) (pain scores on a 100mm VAS) P = 0.021 | ‐ | 95 (two studies) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD reduces dysmenorrhoea in women with symptomatic endometriosis. |
| Improvement of the most troublesome symptom at 12 months | No studies reported on this outcome. | |||||
| Quality of life at 12 months | Measured using Thai version SF‐12 scoring. Baseline score 56.1 (SD 16.5) to 57.0 (SD 33.2) at 12 months . | Measured using Thai version SF‐12 scoring. Baseline 61.3 (SD 16.4) to 70.3 (SD 16.2) at 12 months. | ‐ | 55 participants (1 study) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD improves quality of life in women with symptomatic endometriosis. |
| Patient satisfaction at 12 months | Measured using the Likert scale. Vercellini 2003 found 50% (10/20) of those in the expectant management group were satisfied or very satisfied (RR 1.5, 95% CI 0.90‐2.49). Tanmahasamut 2012 found that fewer were very satisfied in the expectant management group when compared to LNG‐IUD (RR 0.64, 95% CI 0.33–1.24; P = .184 as reported by trial authors). | Measured using the Likert scale. | ‐ | 95 participants (2 studies) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD improves patient satisfaction in women with symptomatic endometriosis. |
| Adverse events | There was a significant increase in women reporting bloating (P = .021) and melasma (P = .15) in those who received post‐operative LNG‐IUD compared to those receiving no post‐operative treatment. | ‐ | 55 participants (1 study) | ⊕⊝⊝⊝ | We are uncertain of whether post‐operative LNG‐IUD increases adverse events following surgery for symptomatic endometriosis. | |
| The risk in the intervention group (and its 95% confidence interval (CI)) is based on theassumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). §Follow‐up duration was 12 months in all 3 trials except one (Gomes 2007), which was 6 months. CI: Confidence interval; LNG‐IUD: levonorgestrel intrauterine device; RR: Risk ratio; SD: Standard deviation; VAS: Visual analogue scale | ||||||
| GRADE Working Group grades of evidence | ||||||
| a we downgraded twice for very serious imprecision (wide confidence intervals, small sample size) b we downgraded once for serious risk of bias (high risk of performance bias) | ||||||
| Outcome | Illustrative comparative risks* (95% CI) | Relative effect (95% CI) | No of participants (Studies) | Certainty of the evidence
| Comments
| |
|---|---|---|---|---|---|---|
|
| Assumed risk | Corresponding risk | ||||
|
| No postoperative treatment | Postoperative use of LNG‐IUD | ||||
| Overall pain at 12 months | No studies reported on this outcome | |||||
| Chronic pelvic pain at 12 months | The mean chronic pelvic pain score (VAS) in the control group was 37
| MD 2 lower (20.23 lower to 16.23 higher) | ‐ | 40 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain of the difference in efficacy between LNG‐IUD and GnRH‐a with regard to chronic pelvic pain. The evidence was downgraded due to the high risk of bias and the small sample size of the one trial included. |
| Dysmenorrhoea at 12 months | The mean score (reduction in VAS) in the control group was 0.4
| MD 1.7 higher (0.14 lower to 3.54 higher) | ‐ | 18 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain of the difference in efficacy between LNG‐IUD and GnRH‐a with regard to dysmenorrhoea. The evidence was downgraded due to the high risk of bias and the small sample size of the one trial included. |
| Improvement of the most troublesome symptom at 12 months | No studies reported on this outcome | |||||
| Quality of life at 12 months | No studies reported on this outcome | |||||
| Patient satisfaction at 12 months | No studies reported on this outcome | |||||
| Adverse events | The most common side effects were irregular | The most common side | ‐ | 40 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain on whether there is a difference in adverse events in women treated with post‐operative LNG‐IUD compared with GnRH‐a. |
| The risk in the intervention group (and its 95% confidence interval (CI)) is based on theassumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). §Follow‐up duration was 12 months in both trials. | ||||||
| GRADE Working Group grades of evidence | ||||||
| a we downgraded twice for very serious imprecision (wide confidence interval, data from one small trial) b we downgraded once for serious risk of bias (unclear risk of selection bias; high risk of performance bias) | ||||||
Background
Description of the condition
Endometriosis is a common gynaecological disease affecting 10% of the female population (Eskenazi 1997). It is defined as the presence of ectopic deposits of endometrial‐like tissue found outside the uterus in women of reproductive age (Johnson 2017). These deposits usually, but not exclusively, occur in the pelvis and induce a chronic inflammatory reaction which can lead to infertility and pelvic pain (Ballard 2008; Kennedy 2005). Although the exact pathogenesis is uncertain, the disease is thought to be oestrogen‐driven and due to retrograde menstruation (migration of endometrial deposits into the peritoneal cavity during menstruation) (Henderson 1991; Sampson 1927; Tal 2019).
Pelvic pain is the primary complaint of women suffering from endometriosis and presents in numerous forms (Bellelis 2010; Hirsch 2016). Dysmenorrhoea is by far the most frequent symptom and form of pain reported by women with endometriosis (Bellelis 2010; Vercellini 1996). However, depending on the location of the lesions, women may also experience deep dyspareunia, and dyschezia (Seracchioli 2008; Thomassin 2004).
Laparoscopy with histological sampling is the gold standard for diagnosing endometriosis (Dunselman 2014). A clinical staging system has been developed to allow researchers and clinicians during laparoscopy to describe the extent of the disease (ASRM 2006). However, grading systems have been criticised for their inability to provide clinically relevant disease and treatment prognostication (Johnson 2017). Other modalities such as transvaginal ultrasound can aid in the diagnosis of endometriosis, however, it is less accurate for uterosacral, vaginal, and rectovaginal septum involvement (Bazot 2004).
Various treatment options exist for endometriosis, including ovarian suppression therapy, surgical treatment, or a combination of these strategies. Surgical treatment of endometriosis aims to remove visible areas of endometriosis and restore the anatomy by dividing adhesions. Surgically, there are several techniques by which endometriosis can be removed or destroyed, each with potential advantages, disadvantages, and differences in efficacy (Bafort 2020; Duepree 2002; Fedele 2004; Ford 2004; Leonardi 2020). The role of surgery differs depending upon the site of the endometriotic deposits, the extent of disease, and reproductive status of the women (Bafort 2020; NICE 2004; Nicklin 1999). Surgery may be as conservative as laparoscopic cauterisation of endometriotic deposits or as radical as hysterectomy with bilateral oophorectomy and resection of portions of the bowel.
It was previously proposed that laparoscopic treatment may improve pelvic pain associated with endometriosis, but this assertion has recently been brought into question (Bafort 2020).
Certainly, a proportion of women continue to experience pain following laparoscopic surgery or the pain recurs within one to two years after surgery (Crosignani 1996; Guo 2009; Sutton 1994; Vercellini 2000). The lesion subtype does not appear to affect the rate of recurrence and the lesions tend to re‐present as the same subtype excised in the initial surgery (Nirgianakis 2020). It is estimated that up to 50% of women will require further surgery within the first five years due to the recurrence of pain (Shakiba 2008). Therefore, adjuvant medical therapies have been considered in addition to surgery to reduce surgical treatment failures and recurrence rates. Postoperative medical therapy should theoretically induce resorption of microscopic foci and lesions that could not be surgically removed, reduce the risk of iatrogenic dissemination of endometriotic cells, and improve pain relief (ASRM 2006; Vercellini 2000). Medical therapy aims to inhibit the growth of endometriotic implants by suppression of ovarian steroids and induction of a hypo‐estrogenic state (Vercellini 1998). Anti‐oestrogens and regimens that induce either medical menopause (such as gonadotrophin‐releasing hormone agonists (GnRH‐a)) or a pseudo‐pregnant state (e.g. continuous combined oral contraceptive or progestogens) are among the systemic medical therapeutic options used (Luciano 1988). Danazol was the main agent used in the 1980s, and GnRH‐a were the standard treatment in the 1990s (Vercellini 1998). In recent years, both of these have been superseded by progestogen, with or without oestrogen treatment, due to their superior side‐effect profile.
Description of the intervention
The levonorgestrel intrauterine device (LNG‐IUD), also known as the levonorgestrel intrauterine system (LNG‐IUS), provides a mechanism of local delivery of progestogen to the uterus and pelvis. Levonorgestrel is a 19‐nortestosterone that prevents decidualisation of the stroma inducing endometriotic atrophy (Vigano 2007; Beatty 2009).
How the intervention might work
Endometriosis is an oestrogen‐dependent condition; thus, treatment strategies often involve hormonal suppression. Locally administered levonorgestrel has a profound effect on the endometrium, which becomes atrophic and inactive, while ovulation is usually not suppressed (Crosignani 1997). Its effects are predominantly localised to the endometrium with the high concentrations of levonorgestrel inducing atrophy and pseudo‐decidualisation (Maruo 2001; Nilsson 1978; Silverberg 1986). The systemic levels following LNG‐IUD administration are less than those achieved with therapeutic oral or parenteral doses of progestogens (Du 1999; Fedele 2001; Luciano 1988; Moghissi 1999; Nilsson 1978) hence side effects should theoretically be less. When provided immediately or very soon after surgical removal or ablation of endometriotic patches, levonorgestrel is expected to suppress the regeneration of endometriosis (Tanmahasamut 2012).
Why it is important to do this review
A recent Cochrane Review (Chen 2020) found inconclusive evidence on the effect of timing (pre‐operative versus post‐operative) when using systemic hormonal therapies to improve endometriosis symptoms. As LNG‐IUD is a non‐systemic hormonal therapy, it was excluded from Chen 2020. Our Cochrane Review is an update (Abou‐Setta 2013) assessing whether there is evidence for the widely used practice of post‐operative LNG‐IUD to improve surgical outcomes and to prevent secondary recurrence.
Objectives
To determine if the LNG‐IUD improves pain symptoms, quality of life, and patient satisfaction when inserted postoperatively in women undergoing surgery for endometriosis, compared with expectant management, postoperative placebo (inert IUD), or alternative postoperative medical treatment (e.g. GnRH‐a).
Methods
Criteria for considering studies for this review
Types of studies
We included all randomised controlled trials (RCTs) with a parallel design. We excluded quasi‐randomised and cross‐over trials.
Types of participants
Women with any stage of endometriosis who had undergone any type of surgery for endometriosis that preserved their uterus (including diagnostic laparoscopy), with surgery no more than three months prior to randomisation, were eligible for inclusion. The diagnosis of endometriosis was clinical, based on laparoscopy or laparotomy findings, with or without histology. Studies were considered for inclusion if they incorporated a subgroup of patients who met the inclusion criteria and for whom data were available for analysis.
Types of interventions
Intervention
Post‐operative insertion of LNG‐IUD
Comparison
-
Post‐operative expectant management
-
Placebo (inert IUD)
-
Any other medical treatment (e.g. GnRH‐a)
Types of outcome measures
In 2020, a minimum set of core outcomes was developed to assess research reporting on endometriosis outcomes. This was generated using a formal consensus method involving endometriosis specialists, researchers, and endometriosis sufferers across the globe (Duffy 2020). The Cochrane Gynaecology and Fertility Group as supported the implementation of these core outcome set for endometriosis research. As such, we assessed these outcomes in this review.
Primary outcomes
1) Overall pain
Effectiveness of treatment: pain measured by validated pain scales, for example, visual analogue pain scale (VAS) scores, the McGill Pain Questionnaire (MPQ), a pain improvement rating scale, general pain experience, and a gynaecological pain questionnaire. All time‐points were measured, with a preference of outcomes reported at 12 months post‐randomisation.
Secondary outcomes
1) Chronic pelvic pain (CPP)
2) Dysmenorrhoea
3) Improvement of most troublesome symptom
4) Quality of life, as described by women
5) Patient satisfaction with treatment, as described by women
6) Adverse outcomes (e.g. treatment intolerance, side effects of intervention)
Outcome measures
Outcomes were reported using continuous or dichotomous values. If more than one measurement of the outcome was presented, we gave priority to the binary outcome. If binary outcomes were not reported, we selected the most frequently used validated scale across the included studies.
Effectiveness of treatment with regard to pain was measured by validated pain scales, for example, visual analogue scale (VAS) pain scores, the McGill Pain Questionnaire (MPQ), a pain improvement rating scale, general pain experience, and a gynaecological pain questionnaire. VAS pain scales involved a zero to100 mm horizontal line, with two descriptors, i.e. 'no pain' at the left end and 'intolerable pain' at the right end.
Psychological outcomes were indicated by scores such as depression scores (Hamilton Depression Rating Scale (HAM‐D) score, Hospital Anxiety Depression Scale (HADS) score), and mood scores.
Quality of life outcomes was indicated by, for example, the Medical Outcomes Study Short Form 36 (SF‐36), the Social Adjustment Survey (SAS‐WR), the Sickness Impact Profile (SIP), a general health questionnaire (GHQ), the revised Sabbatsberg Sexual Rating Scale (rSSRS) and EuroQOL‐5D (EQ‐5D).
Patient satisfaction outcomes were indicated with validated scales such as the Likert‐scale questionnaire, whereby participants would select from the options of 'very satisfied', 'satisfied', 'uncertain', 'dissatisfied', or 'very dissatisfied'.
Primary and secondary outcome time points
All time points for outcome measures were extracted. Studies were expected to assess outcomes beyond six months, to exclude a placebo effect. Priority was given to measures 12 months post‐randomisation. Where this was not available, the nearest time was chosen between six and 18 months, with a preference for outcomes reported later than 12 months.
Search methods for identification of studies
In consultation with the Cochrane Gynaecology and Fertility Group Information Specialist, we searched for all published and unpublished RCTs conducted to date, with no language restriction.
Electronic searches
We searched the following electronic databases for relevant trials:
-
The Cochrane Gynaecology and Fertility Group (CGF) Specialised Register of Controlled Trials, ProCite platform, searched on 12 January 2021 (Appendix 1);
-
CENTRAL via the Cochrane Register of Studies Online (CRSO), Web platform, searched on 12 January 2021 (Appendix 2). CENTRAL now contains records from CINAHL as well as from the World Health Organization's International Clinical Trials Registry Portal (ICTRP) and the ClinicalTrials.gov trials registry at the US National Institutes of Health;
-
MEDLINE (Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations), Ovid platform, searched from 1946 to 12 January 2021 (Appendix 3);
-
Embase, Ovid platform, searched from 1980 to 12 January 2021 (Appendix 4);
-
PsycINFO, Ovid platform, searched from 1806 to 12 January 2021 (Appendix 5).
The MEDLINE search was combined with the Cochrane highly sensitive search strategy for identifying randomised trials, which is described in the Cochrane Handbook (Version 5.1.0 chapter 6, 6.4.11).
We handsearched reference lists of relevant trials and systematic reviews retrieved by the search and contacted experts in the field to obtain additional trials. We also handsearched relevant journals and conference abstracts that are not covered in the CGF Specialised Register, in liaison with the Information Specialist.
Searching other resources
Other electronic sources of trials included:
-
LILACS (Latin American and Caribbean Health Science Information database), Web platform, searched on 18 January 2021;
-
Google Scholar, for recent trials not yet indexed in the major databases, Web platform, searched on 18 January 2021;
-
Epistemonikos database (www.epistemonikos.org), a multilingual database of health evidence, Web platform, searched on 18 January 2021.
Data collection and analysis
Data collection and analysis were conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021).
Selection of studies
All relevant trials identified by the search strategy were considered for inclusion in the review, irrespective of how outcome data were reported. Two review authors (TG and EG) independently screened titles and abstracts retrieved by the search using the Covidence platform (Covidence). We then retrieved the full texts of all potentially eligible studies and contacted authors for missing texts if necessary. Two review authors (TG and EG) independently examined full‐text articles for compliance with the inclusion criteria and selected eligible studies. Any missing information that was needed to make a decision about eligibility was sought from the principal investigators of the trials. Disagreements were initially discussed between the review authors (TG and EG), any disagreements that could not be resolved were discussed with the full team (TG, EG, YC and MW). A PRISMA flow chart (Liberati 2009) documents our selection process.
Data extraction and management
Two review authors (TG and EG) independently extracted data from eligible studies. Data extracted included study characteristics and outcome data. Where studies had multiple publications, we collated the reports of the same under a single study ID with multiple references. Any disagreements within the data extraction process were resolved by discussion between the two review authors (TG and EG). Any disagreements that could not be resolved were discussed with the other review authors (YC and MW).
Assessment of risk of bias in included studies
Two review authors (TG and EG) independently assessed the included studies for risk of bias, using the Cochrane 'Risk of bias' assessment tool (Higgins 2011). We assessed the risk of 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 reporting of outcomes); and other bias. Judgments were assigned as recommended in the Cochrane Handbook Section 8.5 (Higgins 2011). Disagreements were initially between TG and EG; any disagreements that could not be resolved were discussed with the other review authors (YC and MW).
Measures of treatment effect
For binary outcomes, data were extracted to allow for the calculation of the risk ratio (RR). For continuous outcomes, data were extracted to allow for the calculation of the mean difference (MD). If continuous outcome data were presented in differing formats (for example, severity of painful periods) the standardised mean difference (SMD) was calculated. Ordinal data (for example, quality of life scores) were treated as continuous data. We present the 95% confidence intervals (CI) for all outcomes.
Unit of analysis issues
The primary analysis was per woman randomised. Priority was given to measures reported 12 months post‐randomisation.
Dealing with missing data
To the extent possible, we analysed the data on an intention‐to‐treat basis (i.e. including all randomised participants in analysis, in the groups to which they were randomised). Attempts were made to obtain missing participant data from the original investigators. Where participant data were unobtainable, imputation was planned for the primary outcome only.
Missing summary and study design data were sought from the original investigators.
Assessment of heterogeneity
We assessed statistical heterogeneity by the measure of the I² statistic. An I² measurement greater than 50% was taken to indicate substantial heterogeneity (Higgins 2021). We also assessed homogeneity by visual inspection of the outcomes tables and by using the Chi2 test for heterogeneity with a 10% level of statistical significance; a P value of 0.1 was the cut‐off point for rejection of the null hypothesis of study homogeneity to limit type II errors.
Assessment of reporting biases
Reporting bias was assessed by comparing the study protocols, when available, and the methods sections to the results presented in the trial publications. If there were ten or more studies in an analysis, we planned to use a funnel plot to explore the possibility of small study effects (a tendency for estimates of the intervention effect to be more beneficial in smaller studies).
Data synthesis
We planned to perform a meta‐analysis on the results where at least two studies were suitable for inclusion. If a meta‐analysis was not possible due to an insufficient number of studies, we provided only a narrative description of the results. We pooled data from studies that were sufficiently similar to make meta‐analysis appropriate. We extracted and analysed data on an intention‐to‐treat basis. For binary outcomes, the overall RR (that is, the risk of having clinical symptoms) and 95% CI were calculated using a fixed‐effect model, as per the protocol. A fixed‐effect model was chosen at the protocol stage due to the assumption that the underlying effect size would be the same for all the trials in the analysis, and that there would likely be fewer than ten included studies, increasing imprecision in the random‐effects model. For continuous data, the overall MD and 95% CI were calculated using a fixed‐effect model. When only computed effect sizes were reported, they were combined using the generic inverse variance method. Statistical analysis was performed using Review Manager Web (RevMan Web 2021).
Subgroup analysis and investigation of heterogeneity
No subgroup analysis was planned. We planned to investigate heterogeneity where found, to determine the possible sources of heterogeneity (e.g. baseline severity grading, age, or LNG‐IUD dosage) to explain the observations. If the source was located, we planned to performed subgroup analyses to determine the effect of the heterogeneity on the outcome measures. Differences between subgroups would have been assessed using the formal Test for Subgroup Differences in Review Manager Web (RevMan Web 2021).
Sensitivity analysis
We planned to conduct sensitivity analyses for the primary outcome to determine whether the conclusions were robust to arbitrary decisions made regarding the eligibility and analysis. These analyses would have included consideration of whether the review conclusions would have differed if:
-
eligibility had been restricted to studies at low risk of bias (i.e. no high risk of selection bias);
-
a random‐effects model had been adopted;
-
alternative imputation strategies had been implemented (for example imputation of a mean rather than the last time‐point observation carried forward); or
-
the summary effect measure had been odds ratio rather than relative risk ratio.
Summary of findings and assessment of the certainty of the evidence
We followed Cochrane methods (Higgins 2021) to prepare 'Summary of findings' tables (summary of findings Table 1; summary of findings Table 2), using the GRADEpro GDT software (GRADEpro GDT). These tables evaluated the overall certainty of the body of evidence for each of the main review outcomes. We assessed the certainty of the evidence using GRADE criteria: risk of bias, consistency of effect, imprecision, indirectness, and publication bias. Judgments about evidence certainty (high, moderate, low, or very low) were made by two review authors (TG and EG) working independently, with disagreements resolved by discussion. Judgments were justified, documented, and incorporated into reporting of results for each outcome.
As per the review protocol, we extracted study data, formatted our comparisons in data tables, and prepared 'Summary of findings' tables before writing the results and conclusions of our review. The key considerations of our 'Summary of findings' tables included:
-
The main comparisons (LNG‐IUD versus expectant management or LNG‐IUD versus placebo IUD and LNG‐IUD versus any other medical treatment) were planned to appear at the front of the published review
-
We planned to present one table per comparison (LNG‐IUD versus expectant management or LNG‐IUD versus placebo IUD and LNG‐IUD versus any other medical treatment).
The outcomes presented are as follows.
-
Overall pain
-
Chronic pelvic pain
-
Dysmenorrhoea
-
Improvement of the most troublesome symptom
-
Quality of life
-
Patient satisfaction
-
Adverse effects
The same outcomes were presented for each comparison. The same comparisons and outcomes were reported in the abstract as in the 'Summary of findings' tables. All GRADE considerations were clearly described and were used to rank the certainty of evidence. For each assumed risk cited in the 'Summary of findings' tables we provided a source and rationale.
If a meta‐analysis was not possible, we planned to present the results narratively in a ‘Summary of findings’ table format.
Results
Description of studies
See Included studies and Characteristics of excluded studies.
Results of the search
Electronic searches and handsearches produced 531 citations, 80 of which were duplicates (Figure 1). After screening the titles and abstracts, 30 citations were considered to be potentially relevant to this review and were screened using the full‐text manuscripts. After the full‐text screening, 21 studies were excluded (10 duplicates, 11 with the wrong study design; see Figure 1). Two trials are still ongoing (Daoudom 2014; Lee 2017). We included four studies (Bayoglu 2011; Gomes 2007; Tanmahasamut 2012; Vercellini 2003). We placed three studies in Studies awaiting classification; Xu 2011 (awaiting translation from Chinese), Wang 2018, and Magos 2004 (awaiting full text).
Included studies
Study design and setting
The included studies all had a parallel design. One was undertaken in Brazil (Gomes 2007), one in Turkey (Bayoglu 2011), one in Thailand (Tanmahasamut 2012), and one (Vercellini 2003) in Italy.
Participants
The included studies involved 153 participants. The trials were all small with the number of women randomised ranging from 22 to 55. The age distribution of the participants in this review ranged from 18 to 45 and there were no significant differences in baseline demographics between groups.
Intervention
Two studies (Tanmahasamut 2012; Vercellini 2003) investigated post‐operative LNG‐IUD versus expectant management. Two studies (Bayoglu 2011; Gomes 2007) investigated post‐operative LNG‐IUD versus GnRH‐a.
Outcome measures
No studies reported on overall pain, the primary outcome.
Three of four studies (Gomes 2007; Tanmahasamut 2012; Vercellini 2003) reported on the secondary outcome, dysmenorrhoea. Gomes 2007 reported chronic pelvic pain which was cyclical, which the study authors confirmed is equivalent to dysmenorrhoea.
One study (Bayoglu 2011) reported on chronic pelvic pain.
One study reported on quality of life (Tanmahasamut 2012).
Three studies reported patient satisfaction (Bayoglu 2011; Tanmahasamut 2012; Vercellini 2003). However, one study (Bayoglu 2011) did not publish their results, and we were not able to obtain the missing data from the authors. We were thus unable to include data from this study in the analysis.
Two studies reported on adverse events (Bayoglu 2011; Tanmahasamut 2012).
No studies reported on the most troublesome symptom.
Apart from one study (Gomes 2007), which reported outcomes up to six months, all studies reported outcomes up to 12 months.
In studies reporting improvement in pain outcomes (Tanmahasamut 2012; Vercellini 2003), a VAS pain scale was used with the baseline score being taken prior to surgical or medical management. The VAS scales were graded from zero to 100 mm (a score of zero representing no pain, a score of 100 representing the worst pain).
Participant satisfaction was measured in two studies (Tanmahasamut 2012; Vercellini 2003) with a Likert‐scale questionnaire, whereby participants would select from the options of 'very satisfied', 'satisfied', 'uncertain', 'unsatisfied' or 'very dissatisfied'. One study (Bayoglu 2011) recorded patient satisfaction after 12 months; however, the method of collecting data and numerical results were not published.
Quality of life was measured in one study (Tanmahasamut 2012) using a Thai version Short Form‐56. This form provided data on physical and mental domains, as well as providing a total quality of life score. Unpublished data were retrieved from the authors to allow reporting on the difference in baseline to 12‐month scores.
Additional outcomes to those pre‐specified as our primary and secondary outcomes were reported by all four studies. Bayoglu 2011 reported on the total endometriosis severity profile (TESP) at baseline and 12 months following surgery, with or without LNG‐IUD. Gomes 2007 reported the change in American Society for Reproductive Medicine (ASRM) score at first look and second‐look laparoscopy, following six months of postoperative hormonal therapy (GnRH‐a or LNG‐IUD). Tanmahasamut 2012 commented on non‐cyclical pain and dyspareunia. Vercellini 2003 reported on dyspareunia and non‐menstrual pelvic pain at 12 months following surgery with (or without) postoperative LNG‐IUD.
Excluded studies
Thirty full‐text articles were assessed for eligibility; we excluded 21 from the review (Figure 1). Ten studies were excluded for having the wrong study design. Ten were excluded as they were found to be duplicates, after the initial exclusion of duplicates. Two studies (Daoudom 2014; Lee 2017) were not included in this review, as they are ongoing. These studies meet the inclusion criteria and should be considered for the update of this review.
Out of the 11 studies excluded for the incorrect study design, one study (Acien 2019) was excluded as the participants did not have surgically confirmed endometriosis and not all participants had surgical intervention as part of the trial. Two studies (Margatho 2018; Petta 2005) were excluded as participants had over three months between their surgeries and randomisation to LNG‐IUD. One study (Alhamdan 2010) was excluded as it included women with endometriosis and/or chronic pelvic pain. Despite attempts to contact the author, we were unable to obtain the data for women with only endometriosis and as the study was over 11 years old, the raw data was likely unobtainable. Two studies (Chen 2017; Seo 2018) were excluded as GnRH‐a were used in both the intervention and control arms, so the effectiveness of solely post‐operative LNG‐IUD could not be ascertained. Three studies (Lee 2018; Lockhat 2005; Taneja 2017) were excluded as they were not RCTs. One study (Yagamuti 2014) was excluded as it did not report any of the outcomes specified in our protocol. Finally, one study (Qiu 2017) was excluded as it was a letter to the editor (Chen 2017).
Risk of bias in included studies
Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
Randomisation in all included studies was performed using a computer‐generated randomisation sequence in a 1:1 ratio. Treatment allocation and allocation concealment were performed in accordance with the randomisation sequence using sealed envelopes with two of the trials (Tanmahasamut 2012; Vercellini 2003), describing them as sealed and opaque envelopes, whilst two trials (Bayoglu 2011; Gomes 2007) only reported that sealed envelopes were used without reporting any further details on opacity or sequential numbering of the envelopes.
Blinding
Blinding of participants
One trial was reported as an open‐label study (Vercellini 2003); one trial (Tanmahasamut 2012) attempted to blind participants, but there were no comments on whether dummy IUDs had been used, therefore it was possible that participants could have felt their IUD strings. Two trials did not report on blinding (Bayoglu 2011; Gomes 2007), but it is likely the participants were aware of their treatment as the use of dummy IUD or injections were not reported.
Blinding of outcome assessors
One trial was reported as an open‐label study (Vercellini 2003); one trial (Tanmahasamut 2012) was reported as double‐blind and two trials (Bayoglu 2011; Gomes 2007) did not comment on outcome assessor blinding.
Incomplete outcome data
In one trial (Vercellini 2003) displacement of the LNG‐IUD was observed in one woman in the LNG‐IUD group five months after insertion. In addition, one participant in each group was lost to follow‐up (at nine months in the LNG‐IUD group, and seven months in the control group). In the second trial (Bayoglu 2011), no drop‐outs were reported. In the third trial (Tanmahasamut 2012), four women (three in the control group and one in the LNG‐IUD group) were lost to follow‐up and one woman was removed from the control group due to protocol violation. Nevertheless, as all three trials used an intention‐to‐treat analysis, and the number of withdrawals was small, they were all considered to be at low risk of bias.
In the fourth trial (Gomes 2007), four patients were excluded (one in the LNG‐IUS group and three in the GnRH‐a group) because they declined second‐look laparoscopy. There was no evidence to suggest that the analysis corrected for bias, or that sensitivity analyses were performed showing that the missing data made little difference to the outcome; as such, the missing data could have influenced the true value, therefore was deemed at high risk of bias.
Selective reporting
Trial protocols were not available, but no evidence of selective reporting was evident from the trial reports.
Other potential sources of bias
No other potential source of bias was identified for these trials.
Effects of interventions
See: Summary of findings 1 Postoperative use of LNG‐IUD compared to expectant management for symptomatic endometriosis following surgery; Summary of findings 2 Postoperative use of LNG‐IUD compared to postoperative GnRH‐a for symptomatic endometriosis after surgery
LNG‐IUD versus expectant management
Primary outcome
Overall pain
No studies reported on the primary outcome, overall pain.
Secondary outcomes
Chronic pelvic pain
No studies reported on this outcome.
Dysmenorrhoea
Two trials (Tanmahasamut 2012; Vercellini 2003) comprising 95 participants assessed the effect of post‐operative LNG‐IUD on dysmenorrhoea; however, meta‐analysis was not possible. In both trials, participants were asked to report pain scores using a VAS scoring system. Due to the abnormal distribution of results, the studies were unable to report mean scores for meta‐analysis. As such, the median reduction scores were reported.
One study (Tanmahasamut 2012) found a median reduction of 81 mm with post‐operative LNG‐IUD compared to 50 mm in the expectant management group (P < 0.001). The second study (Vercellini 2003) found a reduction in VAS pain score by 50 mm (interquartile range (IQR) 35 mm to 65 mm) with post‐operative LNG‐IUD compared to 30 mm (IQR 25 mm to 40 mm) in the expectant group (P = .012).
Most troublesome symptom
No studies reported on this outcome.
Quality of life
One study of 55 participants (Tanmahasamut 2012) assessed quality of life (Table 1) using the Thai version SF‐36 form, and found an increase (P = 0.014) in the change in total quality of life mean score from baseline (61.2 (standard deviation (SD) 14.8) to 12 months (70.3 (SD 16.2)) with LNG‐IUD, compared to expectant management (from baseline 55.1 (SD 17.0) to 12 months 57.0 (SD 33.2)). There was a significant increase in the physical subscale score (P = .015) with post‐operative LNG‐IUD (baseline mean 56.8 (SD 17.5) to 68.0 (SD 16.1) at 12 months) compared to expectant management (baseline 55.1 (SD 17.0) to 54.9 (SD 32.1) at 12 months), but not the mental subscale score (P = .229).
| Quality of life | LNG‐IUD | Control | P value |
|---|---|---|---|
| Physical health at 0 months | 56.8 +/‐17.5 | 55.1 +/‐ 17.0 |
|
| Mental health at 0 months | 61.2 +/‐ 14.8 | 53.7 +/‐ 15.1 |
|
| Total score at 0 months | 61.3 +/‐ 16.4 | 56.1 +/‐ 16.5 |
|
| Physical health at 6 months | 63.4 +/‐ 15.3 | 56.1 +/‐ 29.6 |
|
| Mental health at 6 months | 65.6 +/‐ 13.2 | 52.5 +/‐ 28.4 |
|
| Total score at 6 months | 66.6 +/‐ 12.8 | 57.2 +/‐ 30.1 |
|
| Physical health at 12 months | 68.3 +/‐ 16.1 | 54.9 +/‐ 32.1 | .229 |
| Mental health at 12 months | 68.0 +/‐ 16.4 | 53.9 +/‐ 32.1 | 0.36 |
| Total score at 12 months | 70.3 +/‐ 16.2 | 57.0 +/‐ 33.2 | .014* |
Quality of life scores from Tanmahasamut 2012.
Statistical significance was performed for 12‐month results only.
* indicates statistical significance
Patient satisfaction
Two trials (Tanmahasamut 2012; Vercellini 2003) comprising 95 participants, assessed satisfaction, however combining the studies in meta‐analysis was not possible. We are uncertain of the benefits of LNG‐IUD on satisfaction. Although, both studies found that more women were satisfied or very satisfied with their treatment results in the LNG‐IUD group, the confidence intervals (CI) include the line of no effect.
Tanmahasamut 2012 found that the proportion of participants reporting that they were very satisfied was lower in the expectant management group when compared to LNG‐IUD, but there were no available data to include in the meta‐analysis. The trial authors reported that the CI included the line of no effect (RR 0.64, 95% CI 0.33 to 1.24; P = .184 as reported by trial authors, with no further information). Vercellini 2003 found that 75% (15/20) of those given post‐operative LNG‐IUD were satisfied or very satisfied compared with 50% (10/20) of those in the expectant management group; however, the CIs include the line of no effect (RR 1.5, 95% CI 0.90 to 2.49, 1 RCT, n=40, very low‐certainty evidence; Analysis 1.1). Care should be taken when interpreting these results given the small sample size, imprecision, and risk of bias due to insufficient participant blinding. The certainty of evidence has been downgraded to very low‐certainty evidence and the results should be interpreted as such.
Adverse events
One study of 55 participants (Tanmahasamut 2012) reported adverse events (Table 2). More women reported bloating (P = .021) and melasma (P = .015) in those who received post‐operative LNG‐IUD. There were no differences in other reported adverse events such as weight gain, acne, and headaches.
| Adverse event | LNG‐IUD group n =2 7 (%) | Expectant group n = 23 (%) | P‐value |
|---|---|---|---|
| Bloating | 10 (37.0) | 16 (69.6) | 0.021* |
| Acne | 16 (59.3) | 13 (56.5) | 0.849 |
| Oily skin | 20 (74.1) | 16 (69.6) | 0.730 |
| Melasma | 6 (22.2) | 0 (0) | 0.015* |
| Weight gain | 17 (62.9) | 13 (56.5) | 0.651 |
| Breast tenderness | 18 (66.7) | 9 (39.1) | 0.053 |
| Headache | 13 (48.1) | 17 (73.9) | 0.066 |
| Nausea | 11 (40.7) | 9 (39.1) | 0.910 |
| Leukorrhoea | 1 (3.7) | 3 (13.0) | 0.233 |
Adverse events from Tanmahasamut 2012
*Indicates statistical significance
LNG‐IUD versus GnRH‐a
Primary outcome
Overall pain
No studies reported on the primary outcome, overall pain.
Secondary outcomes
Chronic pelvic pain
We are uncertain whether LNG‐IUD improves chronic pelvic pain (VAS) at 12 months when compared to GnRH‐a (MD ‐2.0, 95% CI ‐20.2 to 16.2, 1 RCT, n = 40, very low‐certainty evidence; Analysis 2.1; Figure 4).
Chronic pelvic pain outcome ‐ LNG‐IUD Vs GnRH‐a
Dysmenorrhoea
One study reported on dysmenorrhoea measured as a reduction in VAS pain score (Gomes 2007). We are uncertain whether LNG‐IUD improves dysmenorrhoea at 6 months when compared to GnRH‐a. (MD 1.70, 95%.CI ‐0.14 to 3.54, 1 RCT, n = 22, very low‐certainty evidence; Analysis 2.2; Figure 5).
Most troublesome symptom
No studies reported on this outcome.
Quality of life
No studies reported on this outcome.
Patient satisfaction
One study recorded this outcome; however, in the published report, no data was reported for analysis. It was reported, however, that GnRH‐a resulted in a higher patient satisfaction score than LNG‐IUD.
Adverse events
One study of 40 participants (Bayoglu 2011) reported on adverse events. Vasomotor symptoms were the most common symptoms reported with patients receiving GnRH‐a, and irregular bleeding was most common in those receiving LNG‐IUD (Table 3).
| Adverse event | LNG‐IUD n = 20 (%) | GnRH n = 20 (%) |
|---|---|---|
| Irregular bleeding | 13 (65) | 0 (0) |
| One‐sided lower abdominal pain | 8 (40) | 0 (0) |
| Weight gain | 2 (10) | 1 (5) |
| Amenorrhoea | 0 (0) | 6 (30) |
| Vasomotor symptoms | 0 (0) | 10 (50) |
| Simple ovarian cysts | 11 (55) | 0 (0) |
Adverse events from Bayoglu 2011
Discussion
Summary of main results
No studies reported on the primary outcome of overall pain.
With respect to secondary outcomes, we are uncertain whether post‐operative LNG‐IUD improves dysmenorrhoea at 12 months, compared to expectant management, as data on this outcome were reported by two small RCTs providing low‐certainty evidence. Similarly, we are uncertain whether post‐operative LNG‐IUD improves quality of life (one small trial) or satisfaction (two small trials) at 12 months (very low‐certainty evidence).
Finally, compared to post‐operative GnRH‐a, we are uncertain whether post‐operative LNG‐IUD improves dysmenorrhoea or chronic pelvic pain.
Overall completeness and applicability of evidence
The uncertainty of this review is due to the small number of RCTs reporting on the endometriosis core outcomes (Duffy 2020); no trials reported on the primary outcome (overall pain), nor the secondary outcome, most troublesome symptom. Moreover, the studies that were included were small and lacked appropriate data for meta‐analysis. Furthermore, the included trials lacked proper reporting on all aspects of potential biases, thereby limiting the internal validity of the results. Additional large trials are needed, reporting the suggested endometriosis outcomes (overall pain, most troublesome symptom, and quality of life) in order to produce clinically relevant effect estimates.
Quality of the evidence
The available data came from four small trials that included 153 women in total. All four were at high risk of bias due to lack of blinding. One (Vercellini 2003) was an open‐label study; three (Bayoglu 2011; Gomes 2007; Tanmahasamut 2012) did not report dummy injections or IUDs, so it is possible the participants were aware of their allocated group.
Unfortunately, there were too few studies reporting the pre‐specified outcomes for most of the planned comparisons; as such, meta‐analysis was not possible. Using GRADE methods of assessment, as shown in summary of findings Table 1, the certainty of the evidence was graded as very low due to the inclusion of only two small trials, with a high risk of bias and imprecision.
Potential biases in the review process
Numerous steps were taken during the process of this review to reduce bias. Firstly, the search was developed and run by the Cochrane Gynaecology and Fertility group with no limitations in language or date. Secondly, screening and extraction were conducted independently by two review authors. Any conflicts that could not be resolved were discussed with the other review authors. Despite efforts to minimise bias, there were multiple outcomes assessed using evidence from a small number of trials, with a small sample size, which may have introduced bias and resulted in difficulties extrapolating clinically relevant conclusions. Furthermore, three studies remain in 'awaiting classification' as we were unable to translate the Chinese text for one of the studies, despite many attempts to recruit a translator. The second study is complete according to trial registries, but there are no published results. It was not clear whether the right population was included in the third study. These studies introduce the possibility of publication bias in this review.
Agreements and disagreements with other studies or reviews
In this updated review, one additional study was included (Gomes 2007), providing additional data on the comparison of postoperative LNG‐IUD vs GnRH‐a with regard to dysmenorrhoea and adverse events.
The current review findings contrast with those of the previous version (Abou‐Setta 2013). Although the findings of this review do demonstrate a possible benefit of post‐operative LNG‐IUD when compared with expectant management, there is a vast reduction in our level of certainty, and meta‐analysis was not possible. This is due to the difference in outcomes assessed, subsequent to the introduction of the core endometriosis outcome set (Duffy 2020). These outcomes were not routinely reported on in the included trials.
Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Chronic pelvic pain outcome ‐ LNG‐IUD Vs GnRH‐a
Comparison 1: Postoperative use of LNG‐IUD compared with expectant treatment in women with endometriosis, Outcome 1: Patient satisfaction
Comparison 2: Postoperative use of LNG‐IUD compared with GnRH‐a in women with endometriosis, Outcome 1: Chronic pelvic pain
Comparison 2: Postoperative use of LNG‐IUD compared with GnRH‐a in women with endometriosis, Outcome 2: Dysmenorrhoea
| Postoperative use of LNG‐IUD compared to no postoperative treatment for symptomatic endometriosis following surgery | ||||||
| Patient or population: Patients with symptomatic endometriosis following surgery | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Assumed risk | Corresponding risk | |||||
| Expectant management | Postoperative use of LNG‐IUD | |||||
| Overall pain at 12 months | No studies reported on this outcome. | |||||
| Chronic pelvic pain at 12 months | No studies reported on this outcome. | |||||
| Dysmenorrhoea at 12 months | Tanmahasamut 2012 found a median reduction of 50 mm (0‐78) (pain scores on a 100mm VAS) P < .001 Vercellini 2003 found a median reduction of 30 mm (25‐40) (pain scores on a 100mm VAS) P = 0.021 | Tanmahasamut 2012 found a median reduction of 81 mm (51.5 ‐87.5) (pain scores on a 100mm VAS) P < .001. Vercellini 2003 found a median reduction of 50 mm (35–65) (pain scores on a 100mm VAS) P = 0.021 | ‐ | 95 (two studies) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD reduces dysmenorrhoea in women with symptomatic endometriosis. |
| Improvement of the most troublesome symptom at 12 months | No studies reported on this outcome. | |||||
| Quality of life at 12 months | Measured using Thai version SF‐12 scoring. Baseline score 56.1 (SD 16.5) to 57.0 (SD 33.2) at 12 months . | Measured using Thai version SF‐12 scoring. Baseline 61.3 (SD 16.4) to 70.3 (SD 16.2) at 12 months. | ‐ | 55 participants (1 study) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD improves quality of life in women with symptomatic endometriosis. |
| Patient satisfaction at 12 months | Measured using the Likert scale. Vercellini 2003 found 50% (10/20) of those in the expectant management group were satisfied or very satisfied (RR 1.5, 95% CI 0.90‐2.49). Tanmahasamut 2012 found that fewer were very satisfied in the expectant management group when compared to LNG‐IUD (RR 0.64, 95% CI 0.33–1.24; P = .184 as reported by trial authors). | Measured using the Likert scale. | ‐ | 95 participants (2 studies) | ⊕⊝⊝⊝ | We are uncertain whether post‐operative LNG‐IUD improves patient satisfaction in women with symptomatic endometriosis. |
| Adverse events | There was a significant increase in women reporting bloating (P = .021) and melasma (P = .15) in those who received post‐operative LNG‐IUD compared to those receiving no post‐operative treatment. | ‐ | 55 participants (1 study) | ⊕⊝⊝⊝ | We are uncertain of whether post‐operative LNG‐IUD increases adverse events following surgery for symptomatic endometriosis. | |
| The risk in the intervention group (and its 95% confidence interval (CI)) is based on theassumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). §Follow‐up duration was 12 months in all 3 trials except one (Gomes 2007), which was 6 months. CI: Confidence interval; LNG‐IUD: levonorgestrel intrauterine device; RR: Risk ratio; SD: Standard deviation; VAS: Visual analogue scale | ||||||
| GRADE Working Group grades of evidence | ||||||
| a we downgraded twice for very serious imprecision (wide confidence intervals, small sample size) b we downgraded once for serious risk of bias (high risk of performance bias) | ||||||
| Outcome | Illustrative comparative risks* (95% CI) | Relative effect (95% CI) | No of participants (Studies) | Certainty of the evidence
| Comments
| |
|---|---|---|---|---|---|---|
|
| Assumed risk | Corresponding risk | ||||
|
| No postoperative treatment | Postoperative use of LNG‐IUD | ||||
| Overall pain at 12 months | No studies reported on this outcome | |||||
| Chronic pelvic pain at 12 months | The mean chronic pelvic pain score (VAS) in the control group was 37
| MD 2 lower (20.23 lower to 16.23 higher) | ‐ | 40 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain of the difference in efficacy between LNG‐IUD and GnRH‐a with regard to chronic pelvic pain. The evidence was downgraded due to the high risk of bias and the small sample size of the one trial included. |
| Dysmenorrhoea at 12 months | The mean score (reduction in VAS) in the control group was 0.4
| MD 1.7 higher (0.14 lower to 3.54 higher) | ‐ | 18 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain of the difference in efficacy between LNG‐IUD and GnRH‐a with regard to dysmenorrhoea. The evidence was downgraded due to the high risk of bias and the small sample size of the one trial included. |
| Improvement of the most troublesome symptom at 12 months | No studies reported on this outcome | |||||
| Quality of life at 12 months | No studies reported on this outcome | |||||
| Patient satisfaction at 12 months | No studies reported on this outcome | |||||
| Adverse events | The most common side effects were irregular | The most common side | ‐ | 40 participants (1 study) | ⊕⊝⊝⊝ VERY LOWa,b | We are uncertain on whether there is a difference in adverse events in women treated with post‐operative LNG‐IUD compared with GnRH‐a. |
| The risk in the intervention group (and its 95% confidence interval (CI)) is based on theassumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). §Follow‐up duration was 12 months in both trials. | ||||||
| GRADE Working Group grades of evidence | ||||||
| a we downgraded twice for very serious imprecision (wide confidence interval, data from one small trial) b we downgraded once for serious risk of bias (unclear risk of selection bias; high risk of performance bias) | ||||||
| Quality of life | LNG‐IUD | Control | P value |
|---|---|---|---|
| Physical health at 0 months | 56.8 +/‐17.5 | 55.1 +/‐ 17.0 |
|
| Mental health at 0 months | 61.2 +/‐ 14.8 | 53.7 +/‐ 15.1 |
|
| Total score at 0 months | 61.3 +/‐ 16.4 | 56.1 +/‐ 16.5 |
|
| Physical health at 6 months | 63.4 +/‐ 15.3 | 56.1 +/‐ 29.6 |
|
| Mental health at 6 months | 65.6 +/‐ 13.2 | 52.5 +/‐ 28.4 |
|
| Total score at 6 months | 66.6 +/‐ 12.8 | 57.2 +/‐ 30.1 |
|
| Physical health at 12 months | 68.3 +/‐ 16.1 | 54.9 +/‐ 32.1 | .229 |
| Mental health at 12 months | 68.0 +/‐ 16.4 | 53.9 +/‐ 32.1 | 0.36 |
| Total score at 12 months | 70.3 +/‐ 16.2 | 57.0 +/‐ 33.2 | .014* |
| Quality of life scores from Tanmahasamut 2012. Statistical significance was performed for 12‐month results only. * indicates statistical significance | |||
| Adverse event | LNG‐IUD group n =2 7 (%) | Expectant group n = 23 (%) | P‐value |
|---|---|---|---|
| Bloating | 10 (37.0) | 16 (69.6) | 0.021* |
| Acne | 16 (59.3) | 13 (56.5) | 0.849 |
| Oily skin | 20 (74.1) | 16 (69.6) | 0.730 |
| Melasma | 6 (22.2) | 0 (0) | 0.015* |
| Weight gain | 17 (62.9) | 13 (56.5) | 0.651 |
| Breast tenderness | 18 (66.7) | 9 (39.1) | 0.053 |
| Headache | 13 (48.1) | 17 (73.9) | 0.066 |
| Nausea | 11 (40.7) | 9 (39.1) | 0.910 |
| Leukorrhoea | 1 (3.7) | 3 (13.0) | 0.233 |
| Adverse events from Tanmahasamut 2012 *Indicates statistical significance | |||
| Adverse event | LNG‐IUD n = 20 (%) | GnRH n = 20 (%) |
|---|---|---|
| Irregular bleeding | 13 (65) | 0 (0) |
| One‐sided lower abdominal pain | 8 (40) | 0 (0) |
| Weight gain | 2 (10) | 1 (5) |
| Amenorrhoea | 0 (0) | 6 (30) |
| Vasomotor symptoms | 0 (0) | 10 (50) |
| Simple ovarian cysts | 11 (55) | 0 (0) |
| Adverse events from Bayoglu 2011 | ||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Patient satisfaction Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Chronic pelvic pain Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2.2 Dysmenorrhoea Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
