A randomised controlled trial of Outpatient versus inpatient Polyp Treatment (OPT) for abnormal uterine bleeding

Definition of a uterine polyp

Uterine polyps are focal endometrial outgrowths that can occur anywhere within the uterine cavity. They contain a variable amount of glands, stroma and blood vessels, the relative amounts of which influence their visual appearance at hysteroscopy. Polyps may be soft and cystic or firm and fibrous; they may be pedunculated or sessile, single or multiple, and vary in size from small – with minimal uterine cavity distortion – to large, filling the whole cavity.

Aetiology of uterine polyps

Uterine polyps are composed of either functional and/or basal endometrium. They are typically a mixture of dense fibrous tissue (stroma), large and thick-walled vascular channels, and elongated glandular spaces of varying shapes1,2 which protrude into the uterine cavity. The underlying cause of uterine polyp formation remains unclear but is believed to be multifactorial. 3 They are thought to originate as focal areas of stromal and glandular overgrowth. This hyperplasia of the endometrium may occur within an area of mucosa that contains more hormonal receptors, thus making it more sensitive to oestrogenic stimulation. 4,5 The mechanism of uterine polyp development may differ according to menopausal status. In premenopausal women, a decrease in oestrogen and progesterone receptors within polyp stromal cells, although not the glandular component, has been found, which may make polyps less sensitive to cyclic hormonal changes. 6–8 Others have pointed to increased cell longevity, as a result of inhibition of apoptosis7,9 and biomarkers for altered gene expression have been documented within uterine polyps. 10,11

Diagnosis of uterine polyps

For many years the traditional method of diagnosing endometrial polyps was dilatation of the cervix and curettage of the endometrium [dilatation and curettage (D&C)]. However, this approach requires general anaesthesia, is associated with morbidity arising from uterine trauma, and has been shown to have limited diagnostic accuracy, because detecting and removing focal pathologies is notoriously problematic without direct visualisation. 12–15 The advent of convenient imaging of the uterine cavity using ultrasound or endoscopic technologies has largely replaced the D&C for endometrial evaluation; 16 when histological tissue samples are indicated, they can be obtained at the same time utilising blind, miniature outpatient biopsy devices17,18 or, in the case of hysteroscopy, guided biopsy. 19,20

Transvaginal ultrasound scan (TVS) is the least invasive outpatient test to evaluate the endometrium, as it avoids potentially painful instrumentation of the uterine cavity; the ultrasound probe simply sits within the top of the vagina. Uterine polyps typically appear as a hyperechoic lesion with or without cystic spaces, usually with regular contours and surrounded by a thin hyperechoic halo, or the polyp may appear as a non-specific endometrial thickening or focal mass within the endometrial cavity. TVS is often used as the first-line test for evaluating bleeding complaints because of its convenience and acceptability. 21 However, compared with saline infusion sonography (SIS), through which fluid is instilled to expand the uterine lumen and delineate the walls of the uterine cavity, or outpatient hysteroscopy (OPH), which directly visualises the inside of the uterus, TVS has poorer accuracy for diagnosing uterine polyps. 22,23 This is primarily because the sonographic findings are not specific to polyps, and other endometrial abnormalities – such as submucosal fibroids or endometrial irregularity – may have the same features, with the findings often subtle and easily overlooked especially for smaller polyps. 24 The development of ancillary technologies such as three-dimensional imaging and the use of colour or power Doppler may in time help improve the diagnostic accuracy of TVS for detecting focal pathologies within the uterus but evidence to date is lacking. 25,26

The gold standard investigative tools for diagnosing uterine polyps are SIS or OPH. Both tests have good accuracy for both detecting or excluding the presence of polyps within the uterus. 22,27–29 Although SIS is popular in some parts of Europe and North America, OPH is more widely practised in the UK and worldwide. This is because the skills of hysteroscopy are more widely attained by practising gynaecologists and it is safely practised30,31 but also because OPH allows convenient simultaneous removal of polyp (‘see and treat’),32 may be more cost-effective16 and appears to be preferred by women. 33,34 In contrast with OPH, SIS has the advantage of allowing assessment and visualisation of other pelvic structures, and visualisation of potential myometrial and adnexal abnormalities. 35

Prevalence and epidemiology of uterine polyps

The prevalence of uterine polyps in a general adult female population without abnormal uterine bleeding (AUB) is generally estimated to be around 10%. 32 Case series of asymptomatic women are generally small and estimates of prevalence imprecise; after TVS, uterine polyps were detected incidentally in 12% of premenopausal women36 and in 6–11% of infertile women without AUB. 37,38 In asymptomatic postmenopausal women, prevalences of 13%36 and 16%39 have been reported following investigation with ultrasound and hysteroscopy, respectively. However, they are found with increasing frequency in women undergoing investigation for problems with AUB (see Abnormal uterine bleeding and uterine polyps). In addition to AUB, risk factors for uterine polyp development are thought to include obesity, late menopause and the use of the partial oestrogen agonist tamoxifen. 36,40–42 The role of hormone replacement therapy (HRT) on polyp formation is unclear, with some studies supporting an association36,40 and others not. 43,44

Natural history

The natural history of endometrial polyps is not fully understood. 46 A proportion of small polyps may regress naturally without treatment, but the majority persist; in one series of 45- to 50-year-old asymptomatic women, 27% of 31 polyps regressed spontaneously during a 1-year follow-up. Polyps that regressed tended to be smaller,47 in keeping with an earlier case series. 46

Oncogenesis

The pathogenesis and oncogenic potential of uterine polyps are unclear. However, the vast majority of uterine polyps are benign, and endometrial cancer originating within the polyp is a rare occurrence. Case series of varied populations report a cancer prevalence of approximately 0.5–3%. 48–55 Asymptomatic women are estimated to have a 4- to 10-fold reduced risk of malignancy compared with those women with AUB. 55,56 Outside of AUB, other risk factors for malignancy within uterine polyps appear to include increasing age, postmenopausal status, obesity, diabetes53–55 and an increased polyp diameter. 55,56 The use of tamoxifen appears to increase the risk of atypical hyperplasia and malignancy within uterine polyps. 41,57

Atypical endometrial hyperplasia (EH) is considered to be a premalignant condition58 and the prevalence within polyps ranges between 1% and 3%. 53–56 Non-atypical hyperplasia has been reported to be found in up to 13% of polyps,53 but the oncogenic potential of this condition is generally low. A systematic review of observational series evaluated the prevalence of premalignant and malignant disease within uterine polyps. It reported malignant tissue changes within endometrial polyps in 0–12.9% of included studies, and hyperplastic change in 0.2–23.8% of polyps. They found postmenopausal symptomatic women to have the highest risk of premalignant and malignant tissue changes. 59 A more recent review60 similarly found symptomatic women with AUB bleeding to be at higher risk of premalignancy or malignancy within a uterine polyp; they reported the prevalence of endometrial neoplasia within polyps in women with symptomatic bleeding as 4.2% (195/4697) compared with 2.2% (85/3941) for those without bleeding [relative risk (RR) 1.97; 95% confidence interval (CI) 1.24 to 3.14]. Among symptomatic postmenopausal women with endometrial polyps, 4.5% (88 of 1968) had a malignant polyp in comparison with 1.5% (25/1654) of asymptomatic postmenopausal women (RR 3.36, 95% CI 1.45 to 7.80). The risk of premalignancy or malignancy within a uterine polyp was higher in symptomatic postmenopausal women (5.4%, 214/3946) compared with 1.7% (68/3997) in reproductive-aged women (RR 3.86, 95% CI 2.92 to 5.11).

Abnormal uterine bleeding

Abnormal uterine bleeding affects women of both reproductive (premenopausal women) and postreproductive (postmenopausal women) age. AUB is one of the four most common reasons for consulting a general practitioner (GP), and accounts for 70% of all referrals to hospital gynaecology clinics,61 making this complaint one of the commonest problems in gynaecology. A large proportion of health-care resources in both primary care and hospital settings are used up in managing this condition. 62

In premenopausal women, AUB manifests itself primarily as heavy menstrual bleeding (HMB), which has been defined as ‘excessive menstrual blood loss which interferes with the woman’s physical, emotional, social and material quality of life, and which can occur alone or in combination with other symptoms’. 63 HMB affects one in five women of reproductive age, with 5% of women aged 30–49 years consulting their GP each year because of the condition. 64 The overall prevalence of HMB in England and Wales has been estimated at 1.5 million women. 65 Uterine bleeding may be unscheduled, however, occurring outside of the expected time of the menstrual period. This irregular ‘breakthrough’ bleeding is known as intermenstrual bleeding (IMB) and is also common; a recent survey within primary care found the 2-year cumulative incidence of IMB to be 24% (95% CI 21% to 27%). 66

Postmenopausal bleeding (PMB) is also a common clinical problem in both general practice and secondary care (hospital settings). Women are most likely to present with PMB in the sixth decade of life, for which consultation rates in primary care are 14.3 per 1000 of the population. 67 PMB causes significant alarm and anxiety to women, who recognise vaginal bleeding after their periods have ceased as abnormal. Rapid referral to secondary care for investigation is indicated because between 5% and 10% of women with PMB will have endometrial cancer. 68 Postmenopausal women taking HRT may also develop problematic genital tract bleeding. Women taking sequential HRT regimens may present with either heavy scheduled bleeding or unscheduled, erratic bleeding, whereas women taking continuous, combined ‘no bleed’ HRT preparations present with unexpected bleeding, which is, by definition, unscheduled and abnormal. Similarly, women with breast cancer who are taking the partial oestrogen agonist tamoxifen may present with unscheduled bleeding. 69 Table 1 summarises the types of AUB.

Abnormal uterine bleeding and uterine polyps

The majority of women with symptomatic polyps present with AUB as described in the preceding section. With the advent of high-resolution pelvic ultrasound and hysteroscopic diagnosis, it has become clear that uterine polyps are highly prevalent during investigation of abnormal bleeding. The reported prevalence of endometrial polyps in general is considered to be between 20% and 30%,43,70,71 the variation reflecting the criteria used to define a polyp, the diagnostic test used, and the type of population studied. Although the prevalence of uterine polyps may be increased after the menopause,36 polyps are found commonly to affect both pre- and postmenopausal women across all age groups. 46 In recognition of the frequency in which uterine polyps are discovered in women of reproductive age, the International Federation of Gynecology and Obstetrics (FIGO) has recently accepted a new classification system for causes of AUB in the reproductive years, based on the acronym ‘PALM-COEIN’ with the ‘P’ denoting a ‘polyp’, i.e. describing AUB associated with the presence of uterine polyps. 72

The improved diagnostic accuracy, has led to the increased use of surgical intervention for the removal of polyps (‘polypectomy’), a procedure that is universally practised to resolve symptoms and to obtain tissue for histological examination. 45

Expectant management

The observation that polyps are an incidental finding in around 5–15% of women,36–39 the majority of polyps are benign60,75 and some may naturally regress46,47 has led some to question whether removal of uterine polyps is necessary,76 and indeed removal may subject women to unnecessary morbidity and wastage of scarce health service resources. Two RCTs have addressed this issue, randomising women with AUB and uterine polyps to expectant management or surgical removal. 75,76 One trial76 failed to recruit women with PMB because neither doctors nor patients were in equipoise and so were unwilling to participate. This finding is consistent with postmenopausal women having a preference for hysteroscopic diagnosis and treatment when an abnormality is found. 73 The other RCT randomised 150 women with uterine polyps, of which 60% had AUB symptoms. Overall, no reduction in periodic blood loss was demonstrated at 6 months’ follow-up, but IMB symptoms were significantly improved. 75 The findings from this study are limited, given that it was restricted to premenopausal women, the sample size was small (only 60% of the population included were symptomatic), the presenting complaints were heterogeneous and the study length of follow-up was short.

Medical management

Medical management is widely adopted for the treatment of menstrual complaints and includes the use of hormonal contraceptives. Although some of these women may have undiagnosed uterine polyps, evidence for the use of medical therapy is lacking and not recommended. 63 Gonadotrophin-releasing hormone analogues (‘GnRH-a’s) have been used prior to hysteroscopic resection of focal pathologies in premenopausal women77 but the costs and menopausal side effects are difficult to justify for the removal of uterine polyps. This is because polyps are successfully removed in the majority of cases without the need for adjunctive medical preparation, in contrast with submucous fibroids (SMFs). One small series evaluated different HRT regimens to see whether some have a reduced propensity to polyp formation. 78 The use of levonorgestrel-releasing intrauterine system (LNG-IUS) in women taking tamoxifen has been reported to reduce the incidence of endometrial polyps. 79

Surgical management: polypectomy

A UK national survey45 and two subsequent Dutch surveys80,81 confirmed that the vast majority of gynaecologists advocated surgical removal of polyps from the uterus after diagnosis with 854 of 918 (93%),45 455 of 553 (83%)80 and 411 of 585 (91%) respondents performing polypectomy. 81 In the UK, the predominant method for removal of uterine polyps was by blind avulsion or curettage, after hysteroscopic location of the focal lesion under general anaesthesia,45 whereas in the Netherlands removal under direct hysteroscopic vision under general or regional anaesthesia80,81 was the favoured approach.

Systematic Review performed during the Outpatient versus inpatient Polyp Treatment Trial

For this report we have undertaken a systematic review of the effectiveness of uterine polypectomy, building on our previously published systematic review,87 using updated methodological advances in search strategies, quality assessment and statistical analysis. 102,103 The objective of the review was to systematically review the literature to evaluate the effectiveness of uterine polypectomy for the treatment of AUB. Our secondary aims were to establish if the type of AUB, setting or technique influenced outcome.

Search strategy

We performed searches on the general bibliographic databases MEDLINE (1950–2013), EMBASE (1980–2013) and Cumulative Index to Nursing and Allied Health Literature (CINAHL) (1981–2013). Based on published advice, our search term combination for electronic databases was medical subject headings (MeSH) for polyps combined with word variants for endometrium (endometri* OR uter*) and surgical polypectomy (surgery OR curettage OR hysteroscopy OR polypectomy). Furthermore, all of the bibliographies of relevant studies were hand-searched to identify articles that were not captured by the electronic searches.

Study selection

Two reviewers independently selected articles in a two-stage process. First, abstracts obtained by either the electronic database searches or bibliography inspections were reviewed, and articles that could possibly fulfil the following criteria were selected for full text review.

Inclusion criteria

• Population Women with intrauterine polyps and AUB.

• Intervention Uterine polypectomy.

• Outcome Relief of AUB symptoms.*

[*Measured in general terms, e.g. objective, semiobjective or subjective measures of change in AUB; normalisation of bleeding patterns; satisfaction with AUB outcome; change in quality-of-life scores from baseline.]

Once articles were selected both reviewers used specially designed data abstraction forms to collect data on the main outcome measure that was relief of AUB symptoms. Secondary outcomes included technical feasibility, complications and polyp histology. Differences in article and information selection were solved by deliberation. Where a consensus could not be found a third reviewer (TJC) made the final judgement. No language restrictions were applied and translation available where necessary.

The strength of agreement between reviewers taking into account the play of chance was computed using kappa statistic (agreement is considered good if > 0.6 and very good if > 0.8).

Type of study included

All relevant randomised controlled studies were included. Owing to the small number of RCTs, non-randomised studies including both prospective and retrospective observational studies were also included.

Study quality assessment

The 2007 STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist was used to assess the quality of the observational studies,104 whereas the Cochrane risk of bias tool checklist was used to assess the quality of the randomised controlled studies. 102 Two reviewers independently scrutinised the articles against each element of the relevant checklist.

Synthesis of results

Originally, in the absence of heterogeneity, data pooling and meta-analysis was planned. However, owing to a lack of controlled studies this could be performed for only inpatient treatment compared with outpatient treatment. All other extracted data were tabulated to allow qualitative analysis.

Results of search

From the electronic search we obtained 1122 citations and a further two from searching reference lists of relevant articles. At this stage 1075 citations were excluded, based on a review of the abstracts and titles. An attempt was made to retrieve the remaining 49 articles for further scrutiny. One article could not be retrieved either online or via The British Library. Review of these articles showed that 32 did not meet the selection criteria. The characteristics of the excluded articles105 are described in Figure 1. There was a high level of agreement between reviewers for which articles should be retrieved for further scrutiny (kappa agreement = 0.92; p ≤ 0.001).

FIGURE 1.

Study selection process.

Included studies

The 17 studies23,75,91,97,101,105–114 (including TJ Clark, Birmingham Women’s Hospital, 2013, unpublished) that met our inclusion criteria enrolled a total of 1829 patients between 1989 and 2009. The population size ranged from 8 to 311, with only five studies having a population size of > 100. 12,16,21,22,29

Study quality and design

There were two randomised controlled studies: one comparing inpatient treatment with outpatient treatment (TJ Clark, unpublished), whereas the second compared polyp removal with observation for 6 months. 75 Of the remaining 15 observational studies, only two were controlled: one compared inpatient treatment with outpatient treatment97 and the second compared hysteroscopic morcellation to electrical resection. 105 The remaining 13 articles were uncontrolled observational studies. Only six of the studies were prospective, although the majority of the studies23,75,97,113,114 (plus TJ Clark, unpublished) collected the data consecutively (Figures 2 and 3; Table 2).

FIGURE 2.

Quality of observational studies using the STROBE guidelines. 104

FIGURE 3.

Study quality of the RCTs using the Cochrane risk of bias tool. 102

Participant characteristics

One study looked exclusively at women suffering from PMB,113 seven studies looked at only women who were premenopausal23,75,106,107,109,111,112 and the remaining nine studies91,97,101,105,108,110,114,115 (plus TJ Clark, unpublished) examined mixed populations of women with AUB.

Interventions

When the operative technique was described, polypectomy was performed under direct vision (hysteroscopically) with the exception of two studies (Clark et al. 97 and TJ Clark, unpublished) that had inpatient arms in which blind avulsion was used. 97 There were a variety of techniques described for polyp removal under direct vision including: scissors, polyp forceps, morcellator devices and a variety of bipolar instruments (Table 3).

Outcomes

There were large differences in the time of follow-up, ranging from 2 months110 to > 9 years91 (Table 4). Only two studies23,97 used a validated tool for measuring efficacy of polypectomy: both studies used a visual analogue scale (VAS). 23,97 The majority of studies defined the primary outcome as an improvement in symptoms of AUB as perceived by the patient. All of the studies23,75,91,97,101,105–115 (plus TJ Clark, unpublished) reported an improvement in symptoms from 60% to 100%. The study looking exclusively at postmenopausal patients presented a survival analysis curve. 113 The seven studies23,75,106,107,109,111,112 looking at premenopausal patients reported improvements in 60–100% of participants. The remaining nine studies91,97,101,105,108,110,114,115 (plus TJ Clark, unpublished) looking at mixed populations reported 65–100% symptomatic improvements. Two of these studies broke down recurrent AUB symptoms for those that were pre- and postmenopausal at treatment. 101,105 One of the studies101 found no significant difference in AUB at 1 year, although the numbers were small (27/34 premenopausal patients vs. 12/12 postmenopausal patients; p = 0.2). Although a larger study105 found that premenopausal women were more likely to have recurrence of symptoms [hazard ratio (HR) 2.42, 95% CI 1.42 to 4.11]. Another study109 broke down symptom outcome by types of AUB in premenopausal women; no differences in outcome were observed for those women complaining of HMB or IMB, although the study population was small (HR 1.29, 95% CI 0.61 to 2.73 and HR 0.42, 95% CI 0.09 to 1.76, respectively).

The two studies (Clark et al. 97 plus TJ Clark, unpublished) comparing inpatient to outpatient treatment reported no difference in symptom improvement, although the study sizes were small: 11 of 12 (92%) after outpatient treatment compared with 13 of 14 (93%) after inpatient treatment,97 and 18 of 22 (82%) after outpatient treatment compared with 17 of 26 (65%) after inpatient uterine polypectomy (RR 1.25, 95% CI 0.88 to 1.83). In both studies the outpatient polypectomies were performed under direct vision, whereas inpatient polypectomy was performed ‘blindly’.

There was one other controlled observational study105 and that compared mechanical polyp resection (using a morcellator) compared with electrical resection. Overall, 36 of 172 patients (21%) undergoing electrical resection and 21 of 139 patients (15%) undergoing intrauterine morcellation (IUM) reported recurrence of AUB.

The randomised controlled study comparing resection of polyps with observation for 6 months reported no difference in periodic blood loss using the pictorial blood assessment chart (PBAC) but did report a significant decrease in recurrence of gynaecological symptoms (e.g. IMB and vaginal discharge) in those women having polypectomy [7/75 patients (9.3%) vs. 28/75 control patients (37.3%); p < 0.001]. 75

Current practice: surgical method and setting

Outpatient evaluation of the uterus for common gynaecological complaints is now commonplace and has superseded traditional inpatient admission for D&C under general anaesthesia. 21,116 Polyps are highly prevalent and being increasingly identified within the uterus in women presenting with AUB, whether it be HMB, IMB or PMB. 66,70,117–119 Recent technological advances in endoscopy have resulted in improved high-definition optical imaging and digital data capture. Moreover, hysteroscopy systems have become miniaturised, such that they are increasingly portable and easy to insert into the uterine cavity without the need for wide cervical dilatation and blind uterine exploration. Alongside these developments, miniature ancillary instrumentation has been developed, allowing precise targeting of focal pathologies, such as uterine polyps, using a variety of mechanical and electrosurgical equipment32,85–88,96–100 (see Surgical management: polypectomy). These developments have facilitated the concept of ‘see and treat’ hysteroscopy in a convenient outpatient setting. The approach obviates the need for general anaesthesia. Hysteroscopic intervention is not simply restricted to diagnosis, but when pathology amenable to treatment is identified then simultaneous treatment is carried out. 21,32,85,86,88

Despite these technological advances and the apparent safety, convenience and feasibility of outpatient intervention, as well as the high prevalence of uterine polyps associated with AUB, outpatient surgical removal of uterine polyps remains infrequently practised. As described above (see Surgical management: polypectomy), surgical polyp removal is universally practised in the UK45 and elsewhere, although the most prevalent technique differs between the UK and the Netherlands, two countries for which national surveys of practice pertaining to polyp treatment have been conducted. 45,80,81 In the UK the default method for removal of uterine polyps was by blind avulsion or curettage after hysteroscopic location of the focal lesion under general anaesthesia,45 whereas in the Netherlands it was removal under direct hysteroscopic vision under general or regional anaesthesia. 80,81

In 2001 in the UK, removal of polyps under direct vision using hysteroscopic techniques was generally restricted to those gynaecologists with an interest in endoscopic surgery, with such practice being more common in members of endoscopic societies. 45 Although outpatient removal of uterine polyps is possible using blind curettage and/or avulsion, techniques developed for surgery under general anaesthesia, such approaches are associated with increased uterine trauma and discomfort. 12–15,32,82–84 Thus, to set up an outpatient service, the ability to perform hysteroscopic polyp removal under direct vision is a prerequisite. Furthermore, the concomitant removal of focal uterine lesions under direct vision may allow for more complete removal and hence better symptomatic outcomes, in addition to the potential advantages to women and their doctors in terms of increased efficiency, convenience and choice.

Notwithstanding the observed differences in hysteroscopic and blind polyp treatment between the UK and the Netherlands, common to both countries was a propensity among gynaecologists to conduct uterine polypectomy as an inpatient under general anaesthesia. In the UK survey, 19% of respondents performed outpatient uterine polypectomy, but only half of these did so routinely. 45 In the Netherlands, 27% of respondents performed polypectomy in an outpatient setting, but those gynaecologists working in teaching hospitals were twice as likely to undertake such procedures than their colleagues who were practising within non-teaching institutions (39% vs. 19% respectively; p < 0.001). 80 It seems, therefore, that outpatient polypectomy is restricted to those gynaecologists, often working within teaching hospitals, with specific interests or skills in endoscopy. Given the high disease burden associated with uterine polyps,74 the ubiquity of polypectomy and the potential morbidity and resource use associated with inpatient surgery, this difference in practice is unsustainable. It is possible that practice has changed since these surveys were conducted a decade ago, but a subsequent follow-up Dutch survey published last year81 found no such change, confirming again that the vast majority of gynaecologists advocating surgical removal of polyps from the uterus did so using general anaesthesia. Those performing outpatient procedures were again twice as likely to be based within teaching hospitals (43% vs. 19%, respectively; p < 0.001). 81 Within the UK it is also likely that practice is much as it was a decade ago, and this contention is supported by the small proportion of UK centres approached to participate in the Outpatient versus inpatient Polyp Treatment (OPT) Trial who were eligible; the main reason for ineligibility being an absence of a therapeutic OPH service (TJ Clark, Birmingham Women’s Hospital, 2013, personal communication).

The reasons for variation in practice and persistence with inpatient surgical treatment under general anaesthesia are likely to be multiple. The absence of an established outpatient diagnostic hysteroscopy service; no access to, or unfamiliarity with, equipment; lack of necessary surgical skills; and financial considerations restricting the development of new services are likely to play their part to a varying degree. As with any new health technologies, however, evidence of effectiveness and cost-effectiveness is necessary once safety and feasibility of the intervention has been established. The lack of effectiveness and cost-effectiveness data is likely to be a key factor driving the current status quo.

Need for a large simple trial of outpatient uterine polypectomy compared with inpatient uterine polypectomy in abnormal uterine bleeding

Two systematic reviews59,87 (see Systematic review performed during the Outpatient versus inpatient Polyp Treatment Trial) provide evidence which suggests that uterine polypectomy is a safe and technically successful procedure for the treatment of AUB and results in an improvement in AUB symptoms. However, the quality of existing research is poor, introducing a substantial potential for bias, and no RCTs compare their impact upon patient symptoms and other patient-centred outcomes. On this latter point, qualitative data addressing patient factors are also fundamental considerations, such as patient preferences for choice of treatment setting as well as their tolerance and acceptability of interventional hysteroscopic procedures in an outpatient setting without general anaesthesia. OPH is known to be associated with discomfort and can induce anxiety. 32,120 Thus, practitioners and patient fears over inducing pain may also influence the continued use of inpatient admission to hospital and general anaesthesia. Moreover, the relative cost-effectiveness of outpatient polyp treatment (OPT) compared with traditional inpatient approaches remains unclear.

The limitations placed upon intrauterine surgery in the conscious outpatient, which include pain tolerance and problems with access or manipulation of miniature hysteroscopic equipment, may translate into reduced feasibility and poorer clinical outcomes. However, even if this were proven, the advantages to women of outpatient intervention in terms of safety, convenience and efficiency may outweigh any inferiority in clinical effectiveness. In addition to these considerations, the inflated cost of miniaturised technologically advanced equipment required for most outpatient procedures combined with the potential for poorer clinical outcomes may offset the efficiency of outpatient polypectomy, even when it is performed immediately following diagnosis at OPH – the ‘see and treat’ approach. Thus, outpatient polypectomy may remain an attractive and preferred treatment option, even if it were not much worse than, or ‘non-inferior to’, standard inpatient treatment.

In light of these considerations, the traditional RCT objective of establishing superiority of a new treatment, in this case outpatient polypectomy, was thought to be inappropriate. Establishing that outpatient treatment is not unacceptably worse than standard inpatient approaches requires a non-inferiority trial.

Thus further research in the form of an adequately powered RCT between treatment settings (outpatient vs. inpatient), stratified by type (i.e. pattern) of AUB, is required to assess the therapeutic role, patient acceptability, effectiveness and cost-effectiveness of uterine polypectomy in AUB both in the short and longer term. The need for such a trial was evident from publications25,32,87,121,122 and supported by UK consultant gynaecologists. Our national survey in 2001 indicated that 268 of 854 (31%) of gynaecologists performing uterine polypectomy were supportive of a trial comparing inpatient with outpatient uterine polypectomy. 45 This implies that the newer outpatient approach had been introduced in some centres without definite evidence but opinion regarding its use is not yet solidified (i.e. collective equipoise) making the need for a trial even more urgent. Furthermore, a recent practice guideline on the diagnosis and management of endometrial polyps produced by the American Association of Gynecologic Laparoscopists (AAGL) highlighted the ‘paucity of high-quality data in the subject area of endometrial polyps given the common occurrence of this pathology. The following considerations are proposed for future research: 1. Randomized trials of women with abnormal uterine bleeding to evaluate the clinical outcome of polypectomy and 2. Cost comparisons of different methods for hysteroscopic removal of polyps, including outpatient and outpatient locations’. 25

Objectives

• To help us to optimise the trial design for a robust large scale, multicentre study with adequate statistical power to evaluate OPT reliably.

• To demonstrate the acceptability of randomisation.

• To establish standardised operating procedures for trial management, and piloted questionnaires and consent forms patients.

• To inform the sample size of a larger study.

Methods

All women with AUB who are referred for a diagnostic OPH between January and July 2000 at the Birmingham Women’s Hospital were approached for consent to participate in this pilot trial. AUB was defined according to four categories: (1) PMB; (2) unscheduled bleeding while on HRT or tamoxifen for > 6 months; (3) IMB in women > 40 years old; and (4) excessive menstrual bleeding refractory to medical therapy. At OPH, those women with a benign uterine polyp [endometrial polyp or pedunculated (grade 0) fibroid] as described by Clark et al. ,32 were randomised to immediate outpatient polypectomy under local anaesthesia or delayed (within 3 months of randomisation) inpatient uterine polypectomy under general anaesthesia as a day case. Women were excluded if hysteroscopic features suggested a malignant lesion or when additional pelvic pathology necessitated hysterectomy. Inclusion and exclusion criteria, in full, are shown below.

Inclusion criteria

• AUB requiring diagnostic hysteroscopy.

• Finding of a benign polyp on diagnostic hysteroscopy.

• No hysteroscopic features suspicious of malignancy.

• Need for polypectomy or myomectomy.

Exclusion criteria

• Hysteroscopic features suggesting malignant lesion.

• Additional pathology necessitating hysterectomy.

Interventions

Outpatient uterine polypectomy was performed under direct hysteroscopic vision using Versapoint^® spring tip bipolar electrodes (Gynecare, Ethicon, Somerville, NJ, USA) as previously described. 32,97 Inpatient uterine polypectomy was performed under general anaesthesia by traditional D&C, blind avulsion with or without prior localising hysteroscopy or under direct vision using an operative hysteroscope. In most instances, wide dilatation of the cervical canal was required to accommodate the larger diameter inpatient instruments within the uterus.

Outcomes

Patient completed outcomes, administered by postal questionnaire, were collected at baseline then at 3, 6 and 12 months post surgery. These consisted of:

• Bleeding response: a 100-mm VAS to measure bleeding in the last month. The scale was anchored at each end by ‘none’ and ‘continuous’. For those women who expected bleeding (premenopausal and those on sequential HRT) we considered a 50% reduction from baseline in bleeding score as a ‘success’. For those women for whom no bleeding was expected (postmenopausal) we considered a score of 0–3 mm a ‘success’.

• Disease-specific health-related quality of life (HRQL) was evaluated using the disease-specific Menorrhagia Multi-attribute Assessment Scale (MMAS). 123 Scores range from 0 (worst) to 100 (best).

• Generic HRQL using the generic EuroQol European Quality of Life-5 Dimensions (EQ-5D) instrument. 124 Scores range from –0.59 (worst) to 1.0 (best). Health thermometer scores range from 0 (worst) to 100 (best).

• Sexual satisfaction and functioning using the Index of Sexual Satisfaction (ISS). 125 Scores range from 0 (best) to 100 (worst).

Randomisation and blinding

Sixty women were randomised. The University of Birmingham Clinical Trials Unit provided third-party randomisation. For the randomisation sequence, variable block size in balanced groups was used to avoid any possibility of foreknowledge. Surgeons and patients could not be blinded in this study (to prevent performance bias and measurement bias) owing to the nature of the comparison, but randomisation was concealed such that the surgeons would be kept unaware of the patient’s group allocation until they had completed the diagnostic OPH, thereby preventing selection bias. In addition, the outcome assessments were conducted by self-administered questionnaires, avoiding possible bias arising from the influence of clinicians’ knowledge of patient’s group allocation at medical follow-up.

Statistical considerations

No formal sample size calculation was made. As many women as possible were randomised over the 6-month period; we considered this length of time to be long enough to achieve our stated objectives.

Summary statistics were produced to describe the demographics of the women randomised. No formal hypothesis testing was attempted on the outcome data, as with this size of sample we could not expect any statistically significant differences; this was not our objective here. Summary statistics for patient completed outcome measures at each time point are presented alongside estimates of uncertainty around point estimates: 95% two-sided CIs for RRs and mean differences between groups. Estimates of differences were adjusted for baseline using analysis of covariance. Analysis was by intention to treat (ITT).

Summary of results

In total, 60 women were randomised to either inpatient or outpatient uterine polypectomy (Figure 4). The questionnaire response rate at 6 months was 54 of 60 (90%), although not all of the components of the questionnaire were always completed.

FIGURE 4.

Pilot OPT Trial profile.

The baseline characteristics of the women randomised are shown in Table 5. Some slight imbalance was seen in some of the parameters that would be expected in a study of this size.

Bleeding response

Completed responses were available for 45 of 60 (75%) of the women randomised at 6 months. At this time point, successful bleeding response followed outpatient uterine polypectomy in 19/22 (86%) women, compared with 16/23 (69%) after inpatient uterine polypectomy at 6 months (Table 6). Similar proportions were seen at 1 year. Detailed VAS bleeding scores for those women who were expecting bleeding are detailed in Table 7.

Quality of life

Results for the disease-specific (MMAS) and generic quality of life (EQ-5D) scores are given in Tables 8 and 9. Scores generally appeared to improve from baseline. Sexual satisfaction score are given in Table 10; no obvious increasing or decreasing trend was noted here. Low response rates were noted for the MMAS and sexual satisfaction questionnaires.

Impact on planning for a definitive trial

Although AUB may affect the sexual domain of life quality, the response to the ISS was low. This probably reflected the sensitive nature of the questions presented to women of all ages, some of whom may have felt them intrusive or irrelevant. Moreover, the cumulative burden of completing various outcome questionnaires may have been a disadvantage. In view of the low response rate and the concern that potentially inclusion of the ISS may compromise return of the primary outcome in the definitive trial, a decision was made to no longer include the ISS as one of the secondary outcome measures. The response rate to the condition-specific MMAS was also low. This was mainly because it is a condition-specific questionnaire designed for use in women with HMB. However, in the pilot study the MMAS was included within the self-completed outcome booklet that was sent to all women and was not restricted to the minority of women who actually presented with HMB. Consequently, the MMAS was considered irrelevant and ignored by the majority of participants. The use of disease-specific questionnaires, such as the MMAS, has been encouraged as an important and more sensitive patient-reported outcome in trials. 126,127 In light of this we planned to retain the MMAS in the multicentre trial but make it clear within the patient-reported outcome questionnaires that the MMAS should be completed by only those women with regular periods or hormone-induced withdrawal bleeds, even if their presenting complaint is unscheduled bleeding rather than HMB.

Participants found that understanding the concept of the VAS, as applied to amount of bleeding, was difficult. This reflected a lack of familiarity in providing responses on a continuous scale, especially as most women with AUB included in the pilot did not have HMB, so that the nature of the bleeding, not the amount of bleeding, was the primary concern. It was clear therefore that if we were to maintain the relevance and generalisability of our work in a future, larger-scale trial we would have to include all types of AUB found in association with uterine polyps (i.e. HMB, IMB and PMB) and measurement of the primary outcome – successful alleviation of AUB symptoms – needed to be revised.

This pilot RCT appeared to show that outpatient treatment was comparable to inpatient treatment in terms of symptomatic relief (86% vs. 69% respectively). However, given the difficulties with VAS score responses in this population as previously detailed we exercised caution in directly applying these proportions to inform the sample size of the definitive study. Furthermore, it should be noted that they were derived from a small population in a single specialist ambulatory tertiary referral centre. Uncontrolled series included in the original systematic review87 showed symptomatic relief in up to 100% of women treated as inpatients. The only controlled series included in the review reported an 80% vs. 90% satisfaction with treatment outcome in favour of inpatient treatment. Thus, all this evidence was used in combination to inform the sample size calculation for the proposed multicentre RCT.

Clinical factors and prognosis

As outlined above (see Abnormal uterine bleeding), AUB comprises a number of complaints in women of both pre- and postreproductive age. Research to date is limited with regard to the impact of polypectomy upon the type of bleeding complaint: HMB, IMB or PMB. In addition, AUB and polyp formation may be affected by the use of exogenous sex steroid hormones found in HRT or the hormone partial agonist tamoxifen,36,40,41 commonly used in the treatment of breast cancer. Symptomatic outcomes may also be influenced by the location of the polyp within the uterus and its morphology. Fundally sited lesions can be harder to remove because the attachment to the uterine wall is less accessible. 32 Firmer fibrous lesions can be harder to cut through with mechanical instruments and retrieve from the uterine cavity via the narrow cervical canal. In contrast, softer glandular polyps may be more mobile and difficult to stabilise and detach but, because of their compressible consistency, they can be easier to extract from the uterine cavity. These variables were therefore identified a priori as potential prognostic factors.

Design implications for the substantive study

Although outpatient treatment of uterine polyps necessitates in most cases the use of innovative miniature endoscopic equipment, it is the treatment setting rather than the intervention itself that is novel. The limitations of outpatient polypectomy discussed previously could potentially result in reduced feasibility and poorer clinical outcomes. However, even if this were proven, the advantages to women of outpatient intervention in terms of safety, convenience and efficiency may outweigh any inferiority in clinical effectiveness. In addition to convenience, avoidance of inpatient stay and facilities may translate into substantially reduced costs to health services. Thus, outpatient polypectomy may remain an attractive and preferred treatment option, even if it were not much worse than, or ‘non-inferior to’, standard inpatient treatment.

In light of these considerations, the traditional RCT objective of establishing superiority of a new treatment, in this case outpatient polypectomy, was thought to be inappropriate. We therefore framed our primary objective in terms of establishing that outpatient treatment was not unacceptably worse than standard inpatient approaches. Therefore, we designed a non-inferiority trial. Based upon our own experience and consultation with potential collaborating centres, we believed that outpatient would be the treatment of choice even if 25% fewer women (in relative terms) had alleviated symptoms at 6 months. Thus, the margin of non-inferiority was set at 0.75.

The pilot study did not formally collect data on the number of women declining randomisation on the basis of a preference for immediate outpatient treatment under local anaesthesia or delayed inpatient uterine polypectomy under general anaesthesia. However, experience suggested a substantial proportion expressed a strong preference for one or the other setting. Women choosing their treatment setting may tend to report more favourable outcomes, being empowered by having been part of the decision process, or conversely may have high expectations of success and be disappointed with a less-than-complete resolution of their symptoms. Given the potential for interaction between choice, or conversely lack of choice if randomised, and outcome, we planned a parallel patient preference study using the same outcome measures as the RCT.

Objectives of the Outpatient polyp treatment study

In undertaking the OPT study, we aimed to:

1. test the hypothesis that in women with AUB associated with benign uterine polyp(s), OPT achieved as good, or no more than 25% worse (in relative terms), alleviation of bleeding symptoms at 6 months, compared with standard inpatient treatment (principal objective)

2. test the hypothesis that response to uterine polyp treatment differed according to the pattern of AUB and menopausal status by three secondary analyses:

   1. premenopausal women compared with postmenopausal women

   2. IMB compared with excessive menstruation

   3. postmenopausal women on HRT compared with those not on HRT

3. explore the variation in the effectiveness of outpatient polyp treatment compared with standard inpatient polyp treatment at different periods of follow-up (12 and 24 months)

4. assess patient acceptability and impact on HRQL

5. perform an economic evaluation for cost-effectiveness.

Screening and consent prior to outpatient hysteroscopy

All women with AUB seen in hospital outpatient clinics and undergoing a diagnostic OPH were considered for the trial. The trial was introduced to them in the outpatient clinic and a comprehensive, evidence-based patient information sheet was provided, either at the first clinic visit or with the appointment letter for the hysteroscopy. Participant information sheets and consent forms were provided to each centre in English and other languages as appropriate to their local community.

Before the procedure, the women were given the chance to discuss the risks and benefits of uterine polypectomy in the outpatient setting using local anaesthesia and in the inpatient setting under general anaesthesia, the process of randomisation and the follow-up requirements with the consultant gynaecologist and/or gynaecology nurse. It was carefully explained that the final decision about eligibility would be taken during the hysteroscopic examination and would be dependent on the findings; therefore, consent was required before the procedure. Women were informed that the process of randomisation would prolong the diagnostic procedure time by up to 2 minutes. Women were also made aware that if the allocation was outpatient polypectomy then the procedure would be undertaken immediately in most instances and treatment would take an additional 10–15 minutes on average, whereas if the allocation was inpatient polypectomy, the diagnostic hysteroscope would be removed and she would be given another appointment for the inpatient procedure within 8 weeks. Women were also informed that only about one in four women will have a uterine polyp and therefore be eligible for the OPT Trial. Each woman appreciated that if a polyp was not found, appropriate treatment would be offered but she would not be recruited into the trial. In this case, the consent form, baseline questionnaire and randomisation form would be destroyed.

In centres at which a ‘see and treat’ outpatient approach was not used (i.e. outpatient as well as inpatient polypectomies were scheduled for a later date), discussion about participation into the OPT Trial and randomisation could take place after the diagnostic hysteroscopy in eligible women known to have uterine polyp.

Determining eligibility

All women with AUB who provided written informed consent to participate in the OPT study and satisfied the eligibility criteria were included. Those women consenting to the RCT were to be randomised following the diagnostic OPH procedure. Those consenting to participate in the preference study were treated according to their preference. The practitioner inspected the uterine cavity, according to their standard hysteroscopic protocol, to determine the presence of uterine polyp(s), absence of any excluding pathology and technical feasibility for outpatient polypectomy. For the purpose of the OPT Trial a uterine polyp was defined at diagnostic hysteroscopy as:

A discrete outgrowth of endometrium, attached by a pedicle, which moves with the flow of the distension medium.[32] Polyps may be pedunculated or sessile, single or multiple and vary in size (the variable amount of glands, stroma and blood vessels that constitute the polyp will influence their macroscopic appearance [i.e. glandulocystic polyps or firmer, more fibrous polyps (indistinguishable in some instances from grade 0 submucous fibroids)].

The following inclusion/exclusion criteria were applied to assess eligibility.

Randomisation

If the woman was found to be eligible for the OPT Trial, the gynaecologist or member of his/her team obtained a randomised allocation during the hysteroscopic examination. Randomisation notepads were provided to investigators and used to collate the necessary information prior to randomisation. Participants were entered and randomised into the trial in a 1 : 1 ratio via a short telephone call to the Birmingham Clinical Trials Unit, or by logging into a secure web-based randomisation system. To avoid any possibility of foreknowledge, the randomiser needed to provide the name and date of birth of the participant, and confirm the eligibility and stratification criteria, whereupon a randomised allocation was provided and a trial number allocated.

Women with strong preference for treatment setting

It was anticipated that some women may be prepared to participate in the OPT study but not wish to be randomised between surgical treatments because they held clear preference for immediate outpatient treatment under local anaesthesia or conversely delayed inpatient uterine polypectomy under general anaesthesia. These women were invited to participate in a preference study and entered the same way as those randomised except that they received the treatment for which they had a preference. These women were followed up for 6 months.

Stratification of randomisation

A minimisation procedure using a computer-based algorithm was used to avoid chance imbalances in important variables. The variables chosen were:

• premenopausal compared with postmenopausal women

• postmenopausal women on HRT compared with those not on HRT

• whether the predominant abnormal bleeding complaint is (excessive) HMB or IMB/unscheduled bleeding

• history of using tamoxifen (current or previous user vs. those who have never used)

• location of uterine polyp (fundal vs. non-fundal)

• type of uterine polyp (endometrial ‘glandulocystic’ vs. fibrous).

For analysis purposes (see Chapter 3), because of the small numbers of the included women taking HRT, or with a history of use of tamoxifen, we narrowed these variables down to a shorter list of three:

• type of bleeding PMB/HMB/IMB (for definitions, see Table 1)

• location of uterine polyp (fundal vs. non-fundal)

• type of uterine polyp (endometrial ‘glandulocystic’ vs. fibrous).

Surgical procedures

A named investigator(s), with suitable training and experience in both outpatient and inpatient uterine polypectomy, performed all surgical procedures within participating centres.

Withdrawal from the Outpatient versus inpatient Polyp Treatment Trial

All women who consented to the randomised OPT Trial, or to the preference study, were followed up and asked to complete postal questionnaires, regardless of actual treatment received.

If a woman specifically requested a treatment setting after randomisation then her choices were respected but did not necessitate withdrawal from the trial, nor did failure of the outpatient procedure or subsequent inpatient treatment. In both circumstances, it was sensitively explained to women that follow-up information was still very important and, unless they wished to withdraw completely from the trial, they would be followed up. The OPT Trial office was notified of any withdrawal of consent for further follow-up.

Serious and unexpected adverse events

There may be mortality and morbidity associated with either polypectomy procedure, therefore all serious adverse events (SAEs) were to be reported by fax to the OPT Trial office as soon as possible. This report was to be followed within 2 days by a completed SAE form. For the purposes of this study, SAEs were defined as those that are fatal, life-threatening, disabling or requiring prolonged hospitalisation, and had resulted from the hysteroscopy, the polypectomy procedure, the anaesthetic or postoperative recovery, for example deep-vein thrombosis and hospital-acquired infections.

Other management at discretion of local doctors

Apart from the trial treatments allocated at randomisation, all other aspects of patient management were left entirely at the discretion of the local investigators.

Format

Patient-orientated outcomes were collected using a postal questionnaire, which included a combination of disease-specific and generic measurement instruments, tailored according to the initial symptom at presentation. The postal questionnaires were sent from the OPT Trial office with postage-paid envelopes 2 weeks before the due date. Reminders were sent to patients if the questionnaire was not returned within 1 week of the due date and attempts were made to contact the patient by phone if the questionnaire was not returned by 2 weeks after the due date.

Timing of assessments

The primary time point was 6 months post treatment. In addition, assessments took place at baseline (i.e. time of recruitment and randomisation to OPT Trial), 12 and 24 months post treatment.

Primary clinical outcome measure: treatment success

The woman’s own assessment of bleeding symptoms, using a dichotomous outcome measure, was used to establish if the treatment had been successful. This outcome was to be considered in a non-inferiority framework (see Statistical analysis, below). The question used for this measure was dependent on whether the patient was pre- or postmenopausal, the predominant complaint at randomisation and the type of HRT they were using, if any. This is because successful alleviation of AUB varies according to the type of bleeding associated with the presence of a uterine polyp(s). Eradication of bleeding is not the treatment goal following polypectomy in women with HMB or scheduled withdrawal bleeds on sequential HRT, but rather a reduction in bleeding to acceptable levels. In contrast, persistent unscheduled bleeding or PMB is always considered abnormal. Thus for these categories of AUB, complete cessation of symptoms is the clinical objective as opposed to a reduction in quantity. Further details can be seen in Figure 6. In all cases a ‘yes’ response was defined as a success (outcome measure ‘1’ in the outcome measure boxes, see Figure 6).

FIGURE 6.

Outpatient polyp treatment pathways for outcome measures. LMP, last menstrual period.

Secondary clinical outcome measures

Patient-completed measures:

• MMAS123 A multiattribute utility, designed to measure the impact of HMB upon HRQL. It has been evaluated for its reliability and face validity (condition-specific instrument). Summary scores range from 0 (severely affected) to 100 (not affected). A modified version of the form was used for women whose symptoms did not specifically relate to the questions on the form (group C in Figure 6). Our objective here was to use the responses from this group to explore and develop the use of a modified questionnaire for those patients for whom bleeding is not expected; results are not presented in this report.

• EuroQol EQ-5D-3L124 EQ-5D-3L (European Quality of Life-5 Dimensions, three-level version) is a standardised instrument for use as a generic measure of health outcome. Applicable to a wide range of health conditions and treatments, it provides a simple descriptive profile and a single index value for health status. Scores range from –0.59 (health state worse than death) to 1.0 (perfect health state).

• EuroQol health thermometer124 A VAS score ranging from 0 (worst health state imaginable) to 100 (perfect health state). Scores range from –0.59 (health state worse than death) to 1.0 (perfect health state).

• Likert bleeding scale All patients were asked how their bleeding had responded to treatment using a Likert scale with four response options: ‘much better’, ‘little better’, ‘same’, ‘worse’.

• VAS to measure monthly bleeding quantity and duration It is now well established that objective measures of blood loss are not particularly relevant to women’s subjective perception of bleeding symptoms. 126 For those patients with HMB (group A in Figure 6), our pilot work97,128 has demonstrated that improvements in VAS scores correlate very well with improvements in categorical and condition-specific quality-of-life measures. The reliability of VAS has been established in the assessment of chronic gynaecological conditions, such as pain,129 and change in individual VAS scores were considered to have sufficient psychometric strengths to be used in the research of AUB involving large group comparisons. 130 Scores range from ‘0’ (no days bleeding in the last month) to ‘100’ (bleeding every day in the last month), and similarly ‘0’ (no bleeding in the last month) to ‘100’ (heaviest imaginable bleeding in the last month).

• Perioperative measures Type, location and number of uterine polyps; size of largest polyp; the need for dilatation of the cervix; the use of a vaginal speculum; the technique used to remove the polyp(s) [hysteroscopic (direct vision) or non-hysteroscopic (blind) or combination]; the instruments used to remove the polyp(s) [hysteroscopic (electrosurgery or mechanical or combination) or non-hysteroscopic instrumentation]; the technique used to retrieve the polyp(s) [hysteroscopic or non-hysteroscopic; removal success, defined as complete removal and retrieval of polyp(s) from the uterine cavity]; the time taken to complete the polypectomy; the time in the outpatient room or operating theatre; details of any operative or postoperative complications, including vasovagal reactions (a liberal clinical diagnosis to define vasovagal episodes was used here, i.e. any woman who feels faint requiring her to lie supine or with her head down); and details of any further treatment prescribed for bleeding.

• Procedure acceptability Questions assessing the patients’ experiences of the procedures focused on pain [VAS; range ‘0’ (no pain) to ‘100’ (worst imaginable pain)] during the procedure (outpatient polypectomy only), at 1 hour after the procedure and on discharge from hospital; acceptability of procedure (totally, generally, fairly acceptable or unacceptable); embarrassment (extremely, moderately, little, none); and ‘yes/no’ answers to questions regarding whether to recommend the procedure to a friend, have the same treatment again and, upon reflection, whether the alternative treatment would have been preferred.

• Other measures during follow-up Additional medical treatments for bleeding; further polyp removal; gynaecological surgery; use of the following for reasons due to bleeding: outpatient clinic, hospital (day case or inpatient), GP (surgery or at home); days off work; visits to hospital for reasons unrelated to bleeding; visits to gynaecologist (not necessarily related to bleeding).

Statistical considerations

Statistical analysis

Independent Trial Steering Committee

The Trial Steering Committee (TSC) provided independent supervision for the trial, providing advice to the Chief and Co-Investigators and the Sponsor on all aspects of the trial and affording protection for patients by ensuring that the trial was conducted according to the Medical Research Council (MRC) Guidelines for Good Clinical Practice in Clinical Trials (www.mrc.ac.uk/documents/pdf/good-clinical-practice-in-clinical-trials/). The TSC consisted of the following independent members: Professor J Thornton (chairperson, Professor of Obstetrics and Gynaecology, University of Nottingham); Professor P O’Donovan (Professor of Medical Innovation and Gynaecology, University of Bradford); E Nicholls (Birmingham Women’s Hospital, Patient and Public Involvement representative); Professor J Deeks (University of Birmingham, Statistician); and Dr S Petrou (University of Oxford, Health Economist). Professor Deeks stepped down as a TSC member during the study follow-up period after taking up a role as Director as the Birmingham Clinical Trials Unit. He was not replaced on the Committee as the study was shortly due to finish.

Data Monitoring and Ethics Committee: determining when clear answers have emerged

The role of the DMEC was to advise the chairperson of the TSC if, in their view, it had become apparent during the period of recruitment that outpatient treatment was clearly inferior to standard inpatient treatment. Alternatively, new evidence might have emerged from other sources that outpatient polypectomy is definitely less effective than inpatient polypectomy. To protect against this, during the period of recruitment to the study, interim analyses of major end points were supplied, in strict confidence, to an independent DMEC, along with updates on results of other related studies, and any other analyses that the DMEC requested. Pragmatic stopping criteria were used; the DMEC were to advise the chairperson of the TSC if, in their view, any of the randomised comparisons in the trial provided both (a) ‘proof beyond reasonable doubt’ that outpatient treatment is so inferior from inpatient treatment that non-inferiority can never be demonstrated for all, or some, women, and (b) evidence that might reasonably be expected to influence the patient management of many clinicians who were already aware of the other main trial results. The TSC then decided whether to close or modify any part of the trial. Unless this happened, however, the Trial Management Group (TMG), TSC, the investigators and all of the central administrative staff (except the statisticians who supply the confidential analyses) remained unaware of the interim results.

The DMEC consisted of the following independent members: Professor M Lumsden (chairperson, Professor of Obstetrics and Gynaecology, University of Glasgow); Professor S Bhattacharya (Professor of Obstetrics and Gynaecology, University of Aberdeen); and Dr C Cummins (Paediatric Epidemiologist, University of Birmingham and Birmingham Children’s Foundation Trust). They met on three occasions and recommended continuing with the study with no change to the protocol.

Other outcomes

Similar numbers of women reported taking additional medical treatments for their bleeding during the 2 years of follow-up (Tables 20 and 21). At 6 months, 15% (31/209) of women were using the LNG-IUS in the inpatient group, compared with 9% (21/228) in the outpatient group. The corresponding figures at 1 and 2 years were 10% (29/219) compared with 8% (19/227) and 11% (21/195) compared with 11% (24/214).

In total over the 2-year follow-up period, 43/254 (17%) women in the outpatient group and 21/253 (8%) in the inpatient group had at least one further polyp removed (RR 2.0, 95% CI 1.2 to 3.3; p = 0.003) (see Figure 8; note that a small number of women had more than one further removal, which explains the slight discrepancy between total numbers). The total number of gynaecological operations other than polyp removal over the period of follow-up was also higher in the outpatient group [61/230 (27%) vs. 36/219 (16%), RR 1.6, 95% CI 1.1 to 2.3; p = 0.01] (Table 22). If hysteroscopies are ignored, this difference is still statistically significant [39/230 (17%) vs. 20/219 (9%), RR 1.8, 95% CI 1.1 to 3.1; p = 0.01]. All surgery types (hysterectomy, endometrial ablation and others) were more common in the outpatient group.

General practitioner and hospital use owing to bleeding appeared similar in both groups (Table 23).

Principal findings

The results of this trial demonstrate that outpatient polypectomy was non-inferior to inpatient polypectomy for the successful alleviation of AUB when compared with our prespecified non-inferiority margin of 25%. Polypectomy successfully treated AUB in 73% (166/228) of the women in the outpatient group, compared with 80% (168/211) in the inpatient group (equivalent to a NNTH of ‘6’ at the lower bound of the 95% CI) at 6 months, with no evidence of interaction of type of bleeding, location or type of polyp with treatment allocation. This treatment success was maintained at both 12 and 24 months, with no differences noted between outpatient and inpatient surgery at any time point. The duration and amount of bleeding were significantly reduced following both outpatient and inpatient treatment and no differences were identified according to treatment allocation. Similarly, a significant improvement in generic and disease-specific HRQL was seen following polypectomy in both treatment groups, which was maintained at 12 and 24 months. Rates of re-referral to a gynaecologist were similar between treatment groups, although the requirement for subsequent gynaecological operative intervention, in the form of endometrial ablation or hysterectomy, was twice as high following outpatient treatment over the 2 years’ follow-up.

Outpatient-based therapy, such as polypectomy, has been facilitated by technological advances in instrumentation, allowing focal pathologies to be removed under direct vision. However, the assumed convenience of avoiding hospital admission and anaesthesia may be diminished if the procedure is infeasible or acceptable in the majority of women. This trial estimated that outpatient polypectomy was successfully completed in four out of five women but the odds of failure to complete polyp removal was two and a half times more likely compared with traditional inpatient treatment. This equates to a NNTH of ‘9’; for every nine outpatient polypectomies performed, an additional one procedure will fail compared with inpatient treatment. Patient discomfort was the main factor predisposing to failure in the outpatient compared with the inpatient setting. Both outpatient and inpatient treatment times were short, i.e. 10 minutes on average. Pain scores were 45 out of 100 on average during outpatient polypectomy, levels which are similar to other commonly adopted intrauterine procedures, such as diagnostic hysteroscopy or endometrial biopsy (M Connor, Royal Hallamshire Hospital, Sheffield, 2013, personal communication). Within 1 hour of treatment, the mean pain scores had reduced to 28 out of 100 implying that pain intensity is moderated quickly. Postoperative pain at 1 hour and on discharge from hospital were significantly higher in the outpatient group than in the inpatient group but the small differences in absolute mean pain scores and their low intensity level suggest that these observations are unlikely to be of clinical significance. Outpatient polypectomy was also statistically less acceptable than inpatient treatment. However, only 2% (5/225) of women found outpatient-based treatment unacceptable and 90% (200/223) of women were prepared to have such treatment again or recommend the outpatient procedure. Nevertheless, with the benefit of hindsight, approximately one in five women in both treatment groups reported a desire to have had the alternative setting for treatment.

Outpatient treatment appeared to be safe; the most common operative complication being vasovagal fainting episodes that occurred in around 1 in 14 (7%, 17/241) women. The majority of polypectomies were conducted hysteroscopically, i.e. polyp detachment under direct vision, but this was done significantly more often in the outpatient group than in the inpatient treatment group (78%, 175/225 vs. 56%, 122/217, respectively). Most polyps (56%, 127/227) were retrieved under direct hysteroscopic vision in the outpatient setting, whereas the converse was true for inpatient therapy, which used traditional blind mechanical extraction in 66% (147/223) of cases. Intra-operative haemorrhage and trauma to the upper genital tract was restricted to a minority of women in the inpatient treatment group. However, four of these traumatic cases were uterine perforations, with one case requiring emergency laparotomy to repair bowel injuries and the other resulting in a pelvic haematoma and subsequent abscess requiring readmission and a period of hospitalisation. These two procedures were conducted blindly. Although traditional blind mechanical instrumentation of the uterus is generally safe, the serious morbidity and mortality that can arise from unrecognised intra-abdominal trauma using such techniques in an anaesthetised patient is well recognised. 12,13

Strengths and limitations of the study

The strengths of our trial include strict randomisation, its size, the multicentre design, the relatively low rates of loss to follow-up, and the tailoring of assessment of outcomes to the primary complaint. Extensive sensitivity analyses were carried out to assess the impact of missing responses and re-intervention on the primary outcome. Some limitations of our study should be noted. These include variation in the practice of outpatient polypectomy, with several techniques and forms of instrumentation being utilised, although the most common approach was bipolar electrosurgery, used in 54% (124/228) of cases. Furthermore, a ‘see and treat’ approach was adopted for outpatient polypectomy in 72% (174/242) of cases, whereas the others were undertaken at a subsequent appointment. PP analyses and a range of further sensitivity analyses did not change our conclusions supporting the robustness of our findings. Furthermore, most women were successfully treated regardless of the type of bleeding complaint with no evidence that relative effectiveness of outpatient compared with inpatient treatment varied by bleeding subtype. Although we could examine the relative efficacy of treatment according to menopausal status, as well as type of bleeding (HMB vs. IMB), we were unable to fulfil one of our original objectives – to compare women with PMB according to whether they were taking HRT. In our pilot study, women with PMB on HRT formed the largest single population subgroup. This contrasted sharply with recruitment to the OPT Trial, for which very few women on HRT and PMB presented, thereby precluding further analysis. This finding coincides with the decline in women taking HRT over recent years following the publication of large cohort studies showing no clear health benefits of long-term HRT use. 135

The overall effect of polypectomy on HMB did not appear to be influenced by the concomitant use of the LNG-IUS, which was fitted immediately post polypectomy in 14% (31/229) of outpatient treatments and 19% (43/222) of inpatient treatments. The type and location of uterine polyp did not influence relative treatment effectiveness.

The pragmatic design of the OPT RCT allowed for the inclusion of women presenting with all types of AUB complaints, enhancing the generalisability of our findings. Successful alleviation of AUB symptoms had to be defined differently according to one of the three categories of AUB: HMB, IMB and PMB. We used the woman’s own assessment of bleeding symptoms, using a dichotomous outcome measure, to establish if the treatment had been successful. We believe that the simplicity of this approach was understandable to women and clinically relevant, but the validity of our chosen outcome has not been tested. Even if women interpret the primary outcome questions differently, the internal validity (outpatient vs. inpatient) should not have been affected. It was reassuring to note that the secondary outcome Likert scale responses for ‘Compared with before your treatment, would you say your bleeding is? (much better to worse)’ were almost identical to the primary outcome measure. In addition, the primary bleeding outcome appeared to be related to an improvement in generic HRQL data measured using EQ-5D scores after treatment.

We also sanctioned the use of any described technique for effecting uterine polypectomy in either setting, but inevitably this led to an imbalance between groups according to treatment method. The outpatient approach to polypectomy was conducted under direct vision using varying technologies in the majority of women, whereas general anaesthesia permitted the use of larger, more traumatic instrumentation and adoption of ‘blind’ techniques in most women. Thus, the technique used may have affected the success of treatment rather than setting alone. However, this is unlikely because polypectomy is a simple procedure and the presumed mechanism of treatment efficacy upon AUB symptoms relates to their removal, however achieved. The study was not powered to determine whether a particular surgical method was more feasible, safe, effective or acceptable.

Implications for practice

Outpatient polyp treatment appears to be safe, feasible, acceptable and effective for the treatment of HMB, IMB and PMB. Outpatient uterine polypectomy should be made routinely available within the NHS in light of these favourable findings and the high disease burden associated with AUB and uterine polyps. Diagnostic OPH facilities are widely available within most NHS hospitals. Diagnostic OPH is familiar to many gynaecologists but additional training would be necessary to become competent in therapeutic procedures such as polypectomy. However, the additional costs and resources required for such training are uncertain. At present only the minority of NHS hospitals are able to provide operative OPH. The ambulatory setting should not be restricted to hospital-based practice, but the establishment of appropriately equipped and staffed community-based services should be explored.

The safety and convenience of outpatient treatment and the equivalent effectiveness in ameliorating AUB symptoms should be weighed against the increased likelihood of failure to complete the procedure, perioperative pain experience and reduced acceptability in the conscious outpatient compared with the anaesthetised inpatient. Women should be made aware of the risk of failure to completely remove the uterine polyp in one in every five outpatient treatments and, as a result, the possible necessity for subsequent inpatient treatment. They should be informed that outpatient intervention is associated with pain, albeit of generally moderate intensity and for a short duration of time. In addition, women should be cognisant of the trend towards reduced acceptability of the overall procedure compared with inpatient approaches, but that at least 8 out of 10 women find the outpatient procedure to be totally or generally acceptable. The thorough counselling of women in this way may present some challenges given that the majority of units effect immediate treatment following polyp diagnosis at hysteroscopy, the so-called ‘see and treat’ approach to management. Current practices for obtaining informed consent should be reviewed and consideration given to providing contemporary written material to women prior to their appointment and verbal information during the pre-hysteroscopy consultation.

Conclusion

We believe that evaluating all types of AUB within our trial has enhanced the generalisability of our findings. However, at present, outpatient uterine polypectomy, with or without the use of local anaesthetic, is not widely available within the NHS and many other health-care systems. Despite this caveat, ‘ambulatory’, ‘outpatient’ or ‘office’-based diagnostic and therapeutic interventions are becoming increasingly prevalent, driven by advances in technology and a desire to enhance recovery. Thus, this trial is timely, novel and relevant to contemporary clinical practice.

Our trial has demonstrated comparable clinical effectiveness of outpatient polypectomy in alleviating gynaecological bleeding complaints at 6, 12 and 24 months. However, despite the convenience and apparent safety of the outpatient treatment setting there were more technical failures reflecting the increased complexity of the procedures in a conscious outpatient compared with an anaesthetised inpatient. The clinical significance of the observations of increased postoperative pain and reduced patient acceptability associated with outpatient compared with inpatient treatment may be inconsequential, given the high overall levels of patient acceptability and relatively low postoperative pain intensity for outpatient polypectomy.

In light of the non-inferiority of outpatient polypectomy and its overall safety, feasibility and acceptability, therapeutic outpatient hysteroscopic services should be established, providing women with a choice of treatment setting. Women should be informed of the relative advantages and disadvantages of either treatment approach, so that they can make an informed decision.

Principal findings

Forty-four per cent of the 906 women agreeing to participate in the OPT study expressed a strong preference for a particular treatment setting and were unwilling to be randomised. Of these women, 81% (324/399) expressed a preference for outpatient treatment. The results of the preference study demonstrate that there was no difference between outpatient and inpatient polypectomy for the successful alleviation of AUB at 6 months. Over 80% of women reported successful symptomatic treatment regardless of the type of AUB. The duration and amount of bleeding were significantly reduced following both outpatient and inpatient treatment, and no differences were identified according to treatment preference. Similarly, a significant improvement in generic and disease-specific HRQL was seen following polypectomy in both treatment groups. Rates of re-referral to a gynaecologist were similar between treatment groups, as was the need for further medical or gynaecological operative intervention.

The preference study estimated that outpatient polypectomy was successfully completed in 90% (282/312) of women and did not demonstrate any increased likelihood of failure. However, this is likely to be due to unknown confounders and the small sample size because an increased likelihood of failure with outpatient polypectomy was seen in the RCT. Possible explanations for this are the selection of more motivated women and technically less challenging polyps to remove in the outpatient setting. It is also important to note that the power to detect a difference between groups is much smaller in the preference study than in the RCT because the groups are unequal; the inpatient group in the preference study being only 30% of the size of the corresponding group in the RCT. Both outpatient and inpatient treatment times were short, with a mean treatment time of 10 minutes in both groups. Perioperative pain scores for outpatient treatment were similar in the preference study to those observed in the RCT. As with the RCT, mean pain scores at 1 hour following the procedure and upon discharge had diminished but were higher in the outpatient polypectomy group than in the inpatient group, albeit differences were small at low levels of mean pain intensity. Consistent with observations from the RCT, 2% (7/299) of women found outpatient-based treatment to be unacceptable and there was a significant trend towards reduced acceptability with outpatient as opposed to inpatient therapy. However, no differences were observed between groups in the proportion of women who would recommend the procedure to a friend, undergo the same treatment again or have preferred an alternative treatment (again, we cannot rule out this being due to the small sample size).

Outpatient treatment appeared to be safe and in keeping with findings from the RCT; the most common operative complication was self-limiting vasovagal fainting episodes, which occurred in around 1 in 18 (6%, 17/302) women. The majority of polypectomies were conducted hysteroscopically, i.e. polyp detachment under direct vision, but, as observed in the randomised study, this was done significantly more often in the outpatient group than in the inpatient treatment preference group [82% (246/299) vs. 56% (36/64), respectively). Most polyps were retrieved under direct hysteroscopic vision in the outpatient setting, whereas the converse was true for inpatient therapy, which used traditional blind mechanical extraction in 69% (43/63) of cases.

Strengths and limitations of the study

The baseline characteristics of participants – type of bleeding, site, number and type of polyp – were similar to the RCT, apart from a slightly older population. The strengths of our preference study include its size, the multicentre design, the inclusion of participants who were ethnically representative of the UK population, the relatively low rates of loss to follow-up, and the tailoring of assessment of outcomes to the primary complaint. Selection bias inherent in the observational design precludes reliable comparison between treatment groups in contrast with the RCT; however, our analysis did make statistical adjustment for the obvious confounding factors (type of bleeding, site and type of polyp). In contrast with the RCT, failure of complete polyp removal was no different between treatment settings. The overall level of failure over both groups was lower. This finding may reflect the selection of more favourable, better motivated women and technically ‘less challenging’ polyps to surgically remove as an outpatient. There remained a trend towards reduced acceptability with outpatient treatment despite women exerting a preference for this setting, although other measures of acceptability were non-significant in contrast with the RCT, but this may reflect the smaller sample or other unknown confounders. The 4 : 1 preference for outpatient over inpatient resulted in a smaller inpatient cohort for comparison and, as a result, estimates in this group of the study are imprecise. Additional limitations of our preference study include variation in the practice of outpatient polypectomy and a small number of participants failing to get their chosen treatment.

Implications for practice

Outpatient polyp treatment appears to be safe, feasible, acceptable and effective for the treatment of HMB, IMB and PMB in women expressing a preference for such treatment. At present most NHS hospitals are unable to provide operative OPH. However, our study has shown that, where such services exist, and women are offered a choice, most recognise the potential benefits of outpatient treatment. Immediate ‘see and treat’ polypectomy was offered in 63% (196/312) of outpatient treatments. Thus, there is evidence on both clinical and patient preference grounds, to support prioritising the provision of ambulatory gynaecology services, such as outpatient hysteroscopic polypectomy within NHS hospital- and community-based environments. When providing women with a choice for treatment of this common condition, it is imperative that they should be informed that the experience of outpatient intervention could be associated with more pain and reduced acceptability than scheduled inpatient approaches to treatment under general anaesthesia. This information should be tempered by the finding that at least 8 out of 10 women find the outpatient procedure to be totally or generally acceptable. Although this preference study did not demonstrate any difference between technical treatment failures, this is likely to reflect the influence of selection bias and women should be informed of the increased risk of failure, as observed in the RCT, when making their decision as regards treatment setting.

Conclusions

This observational study has shown that the majority of women expressing a treatment preference choose to have outpatient treatment of uterine polyps associated with AUB symptoms. Although the data derived from this observational study are subject to selection bias, the clinical outcomes for all categories of AUB are in keeping with the higher-quality data obtained from the parallel RCT. The current situation of limited and eclectic provision of outpatient or ‘ambulatory’ gynaecological therapeutic gynaecological services for AUB is untenable and needs urgently addressing, based upon the findings of the OPT project. The preferences expressed by women in this study and the current patchy provision of such services imply that within most NHS hospitals women are being denied a choice of treatment setting. Consequently, many women are being subjected to the inconvenience and greater burden of inpatient hospital treatment, which if offered an alternative outpatient treatment option they could avoid.

Costs

All costs in the analysis are in UK pounds (£), based on 2011–12 values. 137 An a priori decision was made to use two separate methods to estimate the costs in this analysis. The first method was to use the published standard sources of costs for NHS procedures [NHS reference costs 2011–12138 and Personal Social Services Resource Unit (PSSRU) Costs 2012],137 which was preferred for the base-case analysis as it ensured that the results would be generalisable to all centres in the UK. The second method was to estimate the costs of inpatient and outpatient polypectomy by estimating the costs of the individual components of these procedures (bottom-up costing). These costs were used in a sensitivity analysis to ensure not only the generalisability of the results to centres outside the UK, but also to ensure that the reference costs used in the base case were representative of the ‘real world’ costs of these procedures. Given that there was significant heterogeneity in the types of procedures carried out and in the range of equipment used (see Appendix 16), we estimated resource use and costs for these procedures as carried out at the Birmingham Women’s Hospital (shown in Appendices 20–22). Health and Community Health Services (HCHS) pay and price indices were used to inflate costs, where appropriate. 137

Unit costs (Table 34) were attached to the cumulative resource use in each treatment group in order to calculate costs in both arms of the trial. Outpatient and inpatient polypectomy costs were estimated as per the NHS reference costs138 (Tables 35 and 36). All patients recruited into the trial were assessed in an outpatient clinic and underwent an OPH at the clinic where the uterine polyps were diagnosed. Although most of the patients randomised to an outpatient procedure were treated on their initial visit, 28% were scheduled to attend their treatment in a second clinic. All patients randomised to inpatient were pre-assessed in a nurse-led clinic to ensure that they were suitable to receive general anaesthesia before being scheduled for their day-case inpatient procedure.

Outcomes

Outcome data of interest within the trial were patient-reported effectiveness of the procedure and also QALY gains at 6 and 12 months. Within the OPT Trial, the woman’s own assessment of bleeding symptoms was used to establish if the treatment had been successful using a dichotomous outcome measure (see Chapter 2, Primary clinical outcome measure: treatment success). Those patients whose predominant complaint before the procedure was PMB or IMB were determined to have had a successful treatment if their bleeding stopped. In women with HMB, treatment was successful if the patient reported that their bleeding had returned to an acceptable level following the procedure.

All the patients in the trial were asked to complete EQ-5D-3L questionnaires at baseline, 6 months and 12 months, and the responses obtained were used as the basis for the CUA.

Assumptions

It was necessary to make the following pragmatic assumptions before the analysis could be carried out.

• All of the centres involved in the trial were assumed to have the same expertise and to have followed similar protocols in the management of these patients.

• Only the related events (REs) that occurred within 1 year of the procedure and were deemed relevant to the polypectomy were included in the analysis. Related events included immediate complications of the procedure, all hysterectomies (irrespective of the indication), endometrial ablation and hysteroscopy/endometrial biopsy procedures within this time frame. Costs relating to further polyp-related procedures, if any, were also estimated and stated separately to the REs.

• The costs of procurement of the different hysteroscope camera systems used by participating centres in this study [Versascope^® (Gynecare, Ethicon, Somerville, NJ, USA); Olympus 5-mm rigid hysteroscope (Olympus, Southend-on-Sea, UK) Storz Bettocchi hysteroscope (Karl Storz – Endoskope, Tuttlingen, Germany] were assumed to be the same (see Appendix 20).

• The costs obtained from the manufacturers of this equipment varied by around £1000.

Analysis

Cost-effectiveness analysis was carried out at 6 and 12 months, based on the outcomes expressed in natural units, which were estimated using responses obtained from patients to the question regarding effectiveness of the procedure. The results are expressed in terms of cost per additional patient whose symptoms improved with the procedure at 6 months and 12 months, respectively. A CUA was also carried out at 6 and 12 months and the results are expressed as additional cost incurred per QALY gained. Quality-of-life estimates were derived from EQ-5D responses provided by patients at baseline, 6 and 12 months by applying the standard UK tariff values. These estimates were then used to calculate total QALYs over 6 and 12 months for every individual in the study, using standard methods. 139

The presentation of results in QALYs allows comparison of the results with other available and recently published studies. 139 A range of one-way sensitivity analyses were carried out to explore the robustness of the base-case results to plausible variations during the uptake of these procedures in routine NHS use. In addition, a probabilistic sensitivity analysis (PSA) of the base case was carried out to enable the simultaneous exploration of the uncertainties in the cost and outcome data. The results of these analyses are presented in terms of incremental cost-effectiveness ratios (ICERs), which reflect the additional cost per additional outcome of interest of outpatient treatment compared with inpatient treatment. The results of the CUA are presented using cost-effectiveness acceptability curves (CEACs) to reflect sampling variation and uncertainties in the appropriate threshold cost-effectiveness value.

The analysis took the perspective of the NHS following the current recommendation from National Institute for Health and Care Excellence (NICE). 139 A wider societal perspective was also estimated using the patients self-reported out-of-pocket costs and is discussed within the sensitivity analysis within this report. As the time frame of this economic evaluation is only 1 year, discounting was not necessary. Also, given that there was no significant difference in the baseline EQ-5D scores (see Table 5), a baseline adjustment was not performed during the analysis.

Base-case analysis

We carried out two main analyses on the trial data. The base-case analysis was based on the ITT principle. A secondary analysis as per the treatment received by these patients irrespective of randomisation (the ‘PP’ analysis) was also carried out. Deterministic and PSAs were carried out for both of these methods.

Intention to treat

The base-case analysis estimated costs and outcomes as per ITT. In this method, patients are compared within treatment groups to which they were originally randomised irrespective of the treatment received. 140 This method of analysis allows the estimates to follow real-life scenarios in which patients may not always receive the planned treatment. Not using ITT analysis can often exaggerate the benefits of a given intervention, and it is widely acknowledged as the least biased of analysis options. 141

Per-protocol analysis

Within the OPT Trial, equal numbers of patients were randomised to either outpatient or inpatient polypectomy. However, 13% of those randomised to inpatient and 6% of those randomised to outpatient group received the ‘wrong’ treatment within the trial (i.e. outpatient treatment when randomised to inpatient and vice versa). Furthermore, 23 patients did not receive any treatment following randomisation (see Figure 8).

A PP analysis was carried out to look at the effect of treatment received on the outcome estimates. Therefore, in this analysis all patients who received outpatient treatment were compared with those who received inpatient polypectomy irrespective of the treatment to which they were randomised. Those who did not receive any treatment were disregarded for the purposes of this analysis.

Missing data

Within the trial, resource-use and outcome data could not be collected on all randomised patients (reasons outlined in the CONSORT flow chart; see Figure 8). Multiple imputation, a statistical technique that retains overall population variability and the relationship between observations, is considered useful only when > 10% data are missing. 142 Therefore, when > 10% of data were missing within the trial, these were treated as missing at random (MAR) and estimated using MCMC multiple imputation method. In this method we produced five simulated, complete versions of the data set using Stata 12 software. Each of these simulated data sets was analysed using standard methods and the results combined so as to produce estimates and CIs that incorporate missing data uncertainty. Given that the proportion of missing data was > 10% at most of the data analysis points (Tables 37 and 38), multiple imputation was carried out at all data points. The results obtained are summarised in Table 39.

Skewed data

As the majority of cost and outcome data are skewed, normal parametric methods cannot be used to calculate the differences in means and, subsequently, the ICER. Bootstrapping is a non-parametric approach that can be used to compare arithmetic means without making any assumptions regarding the sampling distributions. In this analysis, 3000 bootstrapping replications were undertaken in order to calculate the 95% CIs around the differences in mean costs and outcomes.

Sensitivity analysis

We carried out both deterministic and PSAs to explore data uncertainty during this economic evaluation.

Deterministic sensitivity analysis (DSA) was carried out to assess the uncertainty associated with input parameters for both analyses. This technique estimates the effect of changing a single parameter (i.e. either cost or effectiveness) on the overall ICER obtained. The point estimates used for all the other parameters remain unchanged. Within the current analysis, the following four options were considered.

Probabilistic sensitivity analysis

Probabilistic sensitivity analysis uses random numbers to generate multiple possible estimates of cost and outcomes from within the probability distributions (i.e. the Monte Carlo principle). A total of 5000 simulations were carried out using the Monte Carlo principle. The distribution around each parameter was specified by its baseline estimate and a bound (upper or lower) of the estimated 95% CI. The advantage of this method is that all parameter uncertainties can be incorporated simultaneously into the analysis. The result of this analysis allowed the generation of a CEAC that demonstrates the probability of an intervention being cost-effective at different willingness-to-pay thresholds.

Deterministic sensitivity analysis

Holding the outcome data constant, DSAs were carried out by changing the cost data. As shown in Table 41, inpatient polypectomy remained more expensive than outpatient treatment in all of the scenarios considered. The ICERs for DSA-1 to DSA-3 are similar to those obtained by the base-case analysis, and somewhat lower when the new tariffs are applied for outpatient treatment (DSA-4). The DSA results obtained by PP analyses (see Appendix 26) are also similar to those obtained for the base case. Therefore, we can surmise that the results obtained by this analysis are not sensitive to plausible changes in the costs of these procedures.

Probabilistic sensitivity analysis

Probabilistic sensitivity analysis showed that the likelihood of outpatient polypectomy being effective (when EQ-5D estimates are used as outcome of interest) is similar to that of inpatient treatment as per the ITT analysis (Figures 21 and 22) clearly demonstrate that although inpatient treatment is definitely more expensive than the outpatient treatment, the difference in effectiveness is more uncertain. The mean cost differences between the groups as per the PSA were £661 and £658 at 6 and 12 months, respectively (the inpatient costs are higher). The corresponding values for QALY difference were 0.0002 and 0.0005, respectively, at 6 and 12 months (the inpatient QALY gain is higher). This suggests that the effects of the treatment on costs and QALYs beyond 6 months, as expected, are minimal.

FIGURE 21.

Probabilistic sensitivity analysis: outpatient vs. inpatient treatment – ITT analysis at 6 months.

FIGURE 22.

Probabilistic sensitivity analysis: outpatient vs. inpatient treatment – ITT analysis at 12 months.

The CEACs shown in Figure 23 demonstrate that given that both treatments are equally effective, the cheaper treatment will be the preferred procedure at lower willingness-to-pay (WTP) thresholds. Indeed, inpatient and outpatient therapy start to become equally cost-effective only at a WTP threshold of £90,000. A similar picture was seen when PSA was carried out using the PP data. The results of this analysis are shown in Appendices 27 and 28. The CEAC derived from the PP analysis is shown in Appendix 29.

FIGURE 23.

Cost effectiveness acceptability curve outpatient vs. inpatient treatment: ITT analysis.

Figures 21 and 22 show PSA, which simultaneously represents uncertainty in cost and QALY values. The x- and y-axes represent the incremental effectiveness and cost of outpatient treatment compared with inpatient treatment respectively. These figures represent the results for ITT analysis at 6 and 12 months. Most of the values fall in the lower-right and lower-left quadrants, demonstrating that inpatient therapy is more expensive than outpatient treatment. The equal distribution between the left and right quadrants suggests that there is considerable uncertainty regarding the effectiveness of one treatment over the other (based on QALY values). In other words, the effectiveness of both treatments is similar.

Figure 23 shows a CEAC illustrating the uncertainty around the cost-effectiveness estimates by demonstrating the likelihood of an intervention being cost-effective at a given cost threshold compared with the proposed alternative. In this case, given that the likelihood of effectiveness of both treatments is roughly the same, the cheaper treatment is considered most cost-effective at baseline.

Principal findings

We found that inpatient polypectomy was more expensive than outpatient treatment and marginally more effective, resulting in slightly higher point estimates of self-reported effectiveness and QALY values at 6 and 12 months. The differences in costs and outcomes between these procedures were fairly constant at these time points, reflecting that the treatment has very little long term (i.e. beyond 6 months) implications on health and resource use. The ICERs obtained by cost-effectiveness and CUA were very high, reflecting the equivalence in effectiveness between these procedures. Although the mean estimates of outcomes appear to favour inpatient treatment, it is important to note that there was considerable uncertainty around these point estimates. This was explored further using PSA, clearly demonstrating that although outpatient therapy is definitely cheaper than inpatient treatment, the effectiveness estimates are uncertain, with the likelihood of effectiveness being roughly equal in both groups at 6 and 12 months.

A range of cost variations, which were considered plausible during the implementation of these treatment pathways within the NHS, were considered. However, these did not make a difference to the conclusions from the base-case analysis. It is notable that the bottom-up costs estimated for outpatient and inpatient procedure during this analysis were quite close to the tariffs from the NHS reference costs138 (see Appendices 21 and 22).

Although this was a non-inferiority trial and there was equivalence in effectiveness while costs were different between the groups (the so-called ‘weak dominance’ situation as postulated by Drummond et al. 147), a cost-minimisation analysis was not considered to be appropriate, as this would not provide adequate information regarding the uncertainty in the estimates. This is better estimated by carrying out a full cost-effectiveness and PSAs. 148

Strengths and limitations of the study

The OPT study is the largest randomised prospective study to estimate the costs and effectiveness of inpatient treatment compared with outpatient treatment for uterine polyps in women with AUB. Economic data in this study were collected prospectively alongside clinical outcome data, enabling accurate estimation of the costs and outcomes. In addition, robust techniques were used to account for data uncertainty, missing data and the skewness of data. Although the main analysis is mainly relevant to the UK, the sensitivity analyses enhance the generalisability of the findings of the study.

It was assumed for the purposes of this analysis that all of the participating UK centres had similar expertise with regards to inpatient and outpatient treatments, but it is possible that centres where outpatient therapy has been offered on the NHS for many years perform the procedure better than those with relatively inexperienced clinicians. In the interests of simplicity and clarity, the economic analysis did not explore the differences in costs and outcomes in subgroups of patients considered within the trial (e.g. premenopausal vs. postmenopausal, different age groups, etc.). The clinical effectiveness results for these groups were roughly similar and it was, therefore, felt that no further value could be added by extending the analysis. It should be noted that training costs were not included in this analysis. However, diagnostic hysteroscopy is now routinely performed in the outpatient setting and given this familiarity with the technique, performing an outpatient polypectomy may need relatively little further training. In addition, it is envisaged that all current clinical trainees (and future consultants) may have acquired the skills necessary for this procedure during their training.

This is the first study to estimate prospectively and compare the costs and outcomes of inpatient treatment with outpatient treatment for this condition. 147 A previous analysis was carried out in this area and a CEA was performed but this was a single-centre retrospective audit that included only a small number of patients. 147

Implications for practice

Outpatient polypectomy is a cost-effective procedure and should be recommended as best use of limited NHS resources in patients with AUB due to uterine polyps.

Conclusion

Outpatient polyp treatment was less expensive than traditional inpatient treatment and similarly effective, resulting in slightly lower self-reported effectiveness and QALY values at 6 and 12 months. The differences in costs and outcomes between these procedures were fairly constant at these time points, reflecting that the treatment has very few longer-term implications on health and resource use. The ICERs for inpatient compared with outpatient treatment obtained by cost-effectiveness and CUAs were very high, reflecting the equivalence in effectiveness between these procedures. Sensitivity analyses clearly demonstrated that although outpatient therapy was definitely cheaper than inpatient treatment, there was uncertainty around the effectiveness estimates implying the effectiveness of the two alternatives was broadly similar. Thus, outpatient polypectomy appears to be more cost-effective relative to the current inpatient approaches for polypectomy at current acceptable WTP thresholds for the NHS.

Methodology

The research aim was addressed by undertaking qualitative interviews with a sample of women in each group of the trial in order to collect more in-depth data on both the process and the views of women about their experience than could be collected by quantitative questionnaires. Interviewing allowed women to introduce areas of discussion that were important to them at a time when the experience was still recent but when they had also had the opportunity for reflection.

Sample

The initial intention was to interview a 10% sample of women from the inpatient and outpatient groups of the study, which would give a sample size of 24 patients from each group, although the basic rule of data saturation would have been applied. That is recruitment would have stopped for each group when no new data were being generated on the issues identified as important to many of the women.

This sampling strategy changed when it was realised that a significant number of women were stating a treatment preference, and also that some women who were randomised or stated a preference for outpatient treatment were not treated immediately. It was felt by the team that these factors could affect notions of acceptability, and so recruitment was stratified to include all of these groups. This resulted in greater outpatient than inpatient numbers. Preference patient recruitment was also weighted towards outpatient as 324 of 399 in this cohort chose to be treated as an outpatient.

The final sample (41 women) is detailed in Table 42.

Women were recruited to the qualitative study during the random allocation process of the trial. They were informed that they had been allocated for interview and consent was taken by the research nurse at the centre. They were then telephoned by one of the qualitative research team to arrange a suitable time for interview, and to answer any questions. Women were recruited from most of the trial sites.

Method

To ascertain issues of importance to women before the main study took place a focus group was conducted with women who had undergone outpatient hysteroscopy in the previous 2 years. Although six women were due to take part in the focus group, only three attended, but it still proved to be a useful exercise, as important issues were raised around pain levels and information received about the procedure, which were included in the interview schedules.

Semistructured interviews were conducted by phone to include a wide range of research sites from across the country in the sample. In a study comparing telephone with face-to-face interviews it was found that telephone interviews tended to be of shorter duration because less detail was given in answers by participants. 154 However, it was acknowledged by the author that there may be good reasons for using telephone interviews, and in the current study the multicentred nature of the trial lead to the adoption of this approach. Interviews took place around 1 week following treatment and took between 15 and 30 minutes. All were recorded with permission. The time interval was chosen in order that women would be able to reflect on the experience with minimum retention bias and to reduce the chance of being influenced in their responses by gratitude to doctors and other hospital staff. Slightly different schedules were developed for randomised and preference patients with questions focusing on the motivation to take part in the research or to choose their treatment method, and their experience of it. Women were also encouraged to raise issues of concern to them that were not asked by the researchers. Interview tapes were transcribed verbatim. To ensure content and face validity, the interview schedule was reviewed after the first three interviews. 155

Data analysis

Data were analysed using thematic analysis, which is used to identify, analyse and report patterns within the data. 156 A staged approach to analysis was undertaken. In the first instance, transcripts were read and data were analysed independently by each researcher, with the subsequent generation of codes and initial themes. The process for carrying out this task was decided individually but involved intensive reading of each transcript, looking for consistency, contradictions and recurring themes. Next the two researchers met to compare codes and themes, and to decide on final themes which reflected the data. Third, an independent qualitative researcher sampled the transcripts and agreed the themes as consistent with the data. In an additional phase to the thematic analysis model used, we also explored inconsistencies within the data. Achieving consensus within the team constituted a form of verification. As the qualitative literature on the subject of hysteroscopy was sparse, verification through embedding this research within it was not possible.

Characteristics of sample

Women who were recruited to the study suffered from PMB (n = 22), IMB (n = 10) or HMB (n = 9). Women who experienced PMB tended to seek medical advice immediately and were referred to a gynaecologist by their GP without initial investigations or treatment. Those whose main complaint was IMB or HMB were more likely to adopt a ‘wait and see’ approach until symptoms persisted or worsened, and some of these women were sent for scans or treated with medication to reduce bleeding before being referred by their GP.

Participation in a randomised trial

The women who agreed to be randomised all had similar reasons for taking part and these are consistent with previous research. 152 Altruistic reasons around helping other women were common, although a few women also personalised this with comments such as ‘my daughter may need this in the future’ (1388*). Other women gave more abstract reasons to do with research more generally. One randomised inpatient said:

I tend to think that any research for anything is a good thing, and we can only go forward with doing the research.

1116

Two women had concerns about both treatment options, so were happy to have the decision taken for them.

I wasn’t too happy to be put to sleep because basically I’m always sick when I wake up. But then if you look at it the other way I wouldn’t be too happy if they hurt me. So they took the choice from me, so that’s good.

1340

Having been randomised the majority of women declared themselves happy with the group to which they had been assigned. A few women did not really understand the randomisation process, and one thought that the computer was deciding on a treatment, based on clinical factors.

The preference patients had more individual reasons for choosing one treatment option over the other. Most women choosing outpatient treatment wanted it over and done with in one hospital visit, and even although for one of these women the procedure could not be completed she still thought it was the right choice. She commented:

[The doctor] tried her best and it just wasn’t to be. It wasn’t anyone’s fault, it’s just the way it is.

5134

Of the other women, two had a fear of anaesthetics, one had a pre-existing medical condition, one had children to make arrangements for, and one did not want to take time off work. Very few women in the study chose inpatient treatment and the four interviewed for the qualitative study all spoke of a previous bad experience of hysteroscopy or other procedures under local anaesthetic, or embarrassment at being in stirrups, which made them want a general anaesthetic.

In both groups of the study a number of women told of how they had consulted friends and family before attending the clinic, or the nurse in the clinic, in order to make a decision about the procedure, but this had to be weighed up against their personal feelings. For example one woman randomised to outpatient treatment said:

I had a lot of reservations about the procedure because people had told me how painful it was, but I don’t react very well to anaesthetics so for me I was glad it went the way it did.

1019

(*Patient trial no.)

Experience of the procedure

Women were asked about their care during the procedure and whether/how their dignity was maintained. This was an area of importance for women, particularly those in the outpatient group, and, as reported above, it was a factor in the decision of one woman to opt for inpatient treatment.

Most women in the outpatient groups of the trial, whether randomised or preference, were pleased to get the procedure over in one hospital visit, and some expressed surprise that it could all be done without them having to come back. However, those outpatients who had to return for treatment were also satisfied with the process, as there is a general expectation of health services that an initial appointment is followed by a separate appointment for treatment, so they viewed it as normal practice.

Only four women in the outpatient groups reported feeling embarrassed at the physical position they were in and some likened it to childbirth in terms of being exposed in front of strangers. However, most women recognised that there was no alternative, and were resigned to some embarrassment. As one woman commented:

I’ve had four children. It is more undignified than anything else. It’s a bit embarrassing, but it’s got to be done.

1002

The majority of women felt that the staff did as much as possible to maintain their dignity during the procedure, and appreciated having one nurse at their head to talk them through it or to take their mind off it, which may explain the low numbers of women feeling embarrassed during the procedure. There was almost unanimous praise for the sensitive way that the nurses and doctors treated them, putting them at ease and explaining what was happening, and it was implied or directly stated that this influenced their overall satisfaction. A few women felt that the number of people in the room, particularly the presence of staff not directly involved in caring for them, such as medical students or other doctors, added to their embarrassment, and one woman felt that the staff were uncaring.

Most women feel that they were given sufficient time to recover following the procedure, although a small minority felt pressured to leave. Postoperative recovery was generally quick in this group, unless a lot of pain was experienced, which will be discussed below.

For inpatients having a general anaesthetic, dignity was not such an issue, even although they knew that they would be exposed in theatre. Two women felt that they were kept waiting for too long for the operation, as they had been starved, but otherwise all of the women were satisfied with their care. Postoperatively, women in this group reported more fatigue, and took longer to recover as they had had a general anaesthetic, but none reported this as a major drawback. One woman was concerned that no doctor came to the ward to discharge her but, again, this may be to do with patient expectations of how health services operate rather than an actual need to see a doctor. One woman felt that she would have benefitted from staying in hospital overnight (she had a small child) but all of the other inpatients were pleased to go home soon after the procedure.

Pain

Women were all asked about any pain that they experienced during and after the procedure and this was the common factor for all within both groups of the trial, although there was quite a diversity of responses. Pain can be divided into that experienced during the procedure, immediately after and over the following few days.

Obviously pain during the procedure was experienced only by women in the outpatient group. It is noted though that for a small group of patients in the preference group the request for a general anaesthetic was because they reported a very painful previous experience during the OPH. Reports from women in the outpatient group ranged from ‘mild discomfort’ to ‘excruciating pain’ but was reported as difficult to cope with by nearly one-third of all participants treated as outpatients. The preference patients reported less pain than the randomised ones, using descriptions such as ‘uncomfortable’, ‘bearable’ and ‘better than expected’. As one woman described it:

I geared myself up for it being really painful, you know extremely painful cramps, but I haven’t had any of that, touch wood. Actually it was a really good process.

5279

This group also reported little postoperative pain, describing it like wind or period pain, although one woman did get a sharp pain after 2 days, which her GP attributed to thrush.

Randomised patients on the other hand described a wide range of pain experiences both during and after the procedure. Most women who were randomised to an outpatient procedure expected it to be painful, and that was the reality in all but two of the cases. They spoke of cramping pains, a few said that it hurt more than they were expecting and one described it as ‘excruciating’ but for the most part women were quite stoical. They commented that ‘it had to be done’ or ‘it was over in 10 minutes’ and spoke of the trade-off between an element of pain and a quick and convenient procedure. One woman rationalised the amount of pain by her own attitude, stating that she was always very tense during internal examinations and could not relax during the procedure, which had to be abandoned.

I can’t say that’s anything to do with what they were doing, I think a lot was due to me not relaxing properly.

1350

A minority of women felt little or no discomfort during the procedure.

There was a fair bit of oohing and aahing but I am not sure you would call it painful.

1002

I was expecting it to be uncomfortable and it wasn’t. I was almost waiting for something to happen when he said ‘that’s it’.

1029

After the procedure, some women found that the pain disappeared fairly quickly, usually within 2 hours, but others reported that it lasted for a few hours or in four cases even a couple of days, and one woman felt it was worse after the procedure than during it. The main description of pain at this time was that it was crampy, or like a period pain, and was generally short lived. Some women, but not all, had been warned that they may feel like this following the procedure.

Those randomised to inpatient treatment reported postoperative fatigue rather than pain, and tended to experience less postoperative pain than outpatients.

In summary, most women felt some pain either during the procedure (outpatients) and/or in the hours or days following. This was generally a crampy pain that was described as bearable, but for some women it was more uncomfortable. For women randomised to outpatient treatment this was felt to be worth the convenience of a quick procedure but some participants expressed the opinion that the level of pain experienced might be difficult for all women to cope with. Preference patients tended to report less pain overall than those who were randomised.

Information needs

All women should have received the trial patient information sheet before attending the clinic. Some women denied having this and others admitted to not reading it, which affected the level of knowledge of the two treatments. However some of these reported that clinic staff gave detailed information to allow them to make an informed decision on whether to take part.

I think that every point was explained clearly and when I asked questions they gave me the clear answer and support and so I was happy with everything.

1105

Despite receiving information, not all women understood the randomisation process. Three ‘preference’ outpatients thought that they were taking part in the trial, and two said that they did not take part in the trial because they did not understand it.

Most of those women who did agree to be randomised knew what this involved and what the trial groups were. One, however, thought that the computer made the treatment decision on the basis of clinical factors; another that the consultant decided; and there were a few other misconceptions. One woman who was randomised to inpatient treatment did not know that outpatient treatment was carried out without anaesthetic, to which she would not have agreed, as she found the diagnostic hysteroscopy so painful.

Most women were happy with the information that they received preoperatively but there were two suggestions for additional information. One woman said that it should be made clear that after treatment patients would not be able to drive and so should bring someone with them; another that you should be given some idea of how long you would be in the hospital, because of car parking, care responsibilities, etc. Two women who were starved for long periods before inpatient treatment felt that the length of time for starving should be more closely aligned to the operation time.

Three women wanted more specific information regarding pain relief for those having outpatient treatment. This amounted to recommending specific painkillers for the type of pain likely to be experienced (e.g. ibuprofen) and a time to take it for it to be most effective during the procedure.

Women who had inpatient procedures were more likely to feel that they had received adequate information on discharge than those who were outpatients, possibly because they received general discharge information. The type of information that preference and randomised outpatients would like to have had focused on two issues: what to expect in terms of recovery and when they could resume normal activities. Women did not know whether postoperative pain or bleeding was normal and, if so, how long it should last before they should seek medical help. They were also unsure whether to consult the hospital or their GP and felt that a contact for problems could be provided.

Resuming normal activities revolved around issues, such as going back to work, driving and resuming sexual activity.

Most women assumed that the polyp was benign and that once it was taken out that was the end of the matter, but three wondered about the histology and what would happen to the result. Two women reported having had previous gynaecological cancer and were understandably more anxious about receiving a result from the histology but women were generally unaware whether it would be sent to them or not.

Differences between inpatient and outpatient experience

The main difference in experience was that outpatients reported some pain and embarrassment during the procedure, and inpatients had some level of fatigue from the anaesthetic. Only the inpatients reported returning to work before they were really ready but this was an issue for only two patients. Inpatients also had different ideas over the level of medical care they should receive, expecting a doctor to discharge them and follow-up appointments. As discussed above this may be because of expectations of the health care system and being admitted to hospital.

I understand how busy the consultants are but it would have been nice if the consultant or one of the team had come down prior to us going home.

5258

Two inpatients commented on the length of the day, as they were required to be admitted early morning for an afternoon operation. Conversely, outpatients often commented on how quickly they were seen and the procedure carried out, and were pleasantly surprised.

None of the inpatients commented on having to attend an appointment for a hysteroscopy and return for removal of the polyps, but two women who opted for inpatient treatment did cite a painful hysteroscopy as a rationale for their choice.

Women in the outpatient groups balanced the pain of the procedure with the convenience of the ‘one-stop shop’, and often set that within the context of their lives. So, for example, one woman described herself as a workaholic who could not stand the idea of taking sick leave. One visit to the clinic and no anaesthetic allowed her to go back to work the next day. Another woman commented that she did not want to explain details of her health to her employer, and a short appointment allowed her to maintain her privacy. Many women in the outpatient groups felt that the pain and embarrassment of the procedure was compensated by not having to return to the hospital or take time off work and the financial cost that these would incur. Avoiding a general anaesthetic was also a positive aspect, particularly for those who had had previous poor experience of this or who had a complicating medical condition, such as one woman who had epilepsy.

When comparing treatments, another woman randomised to outpatient commented:

[with inpatient treatment] I have to make a lot of arrangements for my daughter and everything so I was quite happy with this.

1098

Three women thought that if they had been asked to go home after the hysteroscopy and return at a later date they may not have done, which they thought was an advantage of a ‘one-stop clinic’. One likened it to the dentist telling you that you need a tooth extracting.

You know that if you go home you won’t come back.

1029

Differences between randomised and preference patients

The rationale for choosing to enter the trial varied, with randomised patients expressing altruistic reasons around helping others or generalised views of the benefits of research, whereas preference patients tended to have more individualised reasons for their choice.

Minority experiences

Two women, both allocated to outpatient treatment, had a negative experience that they found to be very traumatic and which is worth reporting, even although they were a small minority. The first woman had undergone removal of polyps in the past under general anaesthetic but was happy to take part in the trial. She expected some pain during the procedure, as friends had told her it ‘wasn’t very nice’ (1329). However, she described what she went through as horrendous, and something she would never go through again.

It got to the point I think my body went into shock. I was shaking, you know, I couldn’t keep my legs still. They were trying to take my mind off it asking what I considered stupid questions at the time. Obviously they were doing a job and it got to a point where I just couldn’t speak I was in so much pain. I think one of the nurses realised and said are you able to cope and the doctor just said ‘oh, she’ll be alright’.

1329

Later in the interview, she expressed the view that the outpatient treatment should not be offered at all:

It’s torture, it’s barbaric and it shouldn’t be allowed to be done.

1329

This woman’s procedure was completed. However, in the case of the second woman who reported extreme pain and shaking, saying that she was in shock, the procedure was terminated in favour of a later inpatient procedure. This patient also reports that she was in pain and bleeding heavily for several days and would never agree to a similar procedure being done without anaesthetic again. She commented that she had not realised that it would be a one-stop procedure, even although she had agreed to be randomised. She thought that after the hysteroscopy the procedure would be over and she would return for an anaesthetic, as she was in pain from the start. Like the woman above, she describes being in shock and shaking badly due to the pain, but, unlike the first woman, she commented that the clinic staff tried to support and relax her.

Satisfaction with the procedure

There was a high level of satisfaction from the women interviewed, from both groups of the study and from randomised and preference patients.

Women who had outpatient treatment largely felt that the procedure was talked through and they knew what was happening, for most a nurse was with them either explaining or talking to take their mind off it. Women appreciated the presence of a nurse throughout the procedure who was not involved in their clinical care to hold their hand and reassure them. This was a very positive aspect of the study and greatly appreciated by patients. Only one patient (discussed in the minority experiences above) felt that she was not treated with care and empathy.

All appreciated the fact that it was over quickly and they could get back to normal almost immediately.

The fact that it takes 5 minutes and it can be done there and then seems to make perfect sense really.

1029

Although some women did find it painful, this was expected and did not alter their overall satisfaction. One woman randomised to outpatient treatment summed up the views of the majority:

I suppose it was painful but it was more like they explained it a crampy period type pain for the first few hours after, but it wasn’t too bad at all, not like I was expecting . . . I was quite pleased the way it all went and I did go back to work the next day, so I don’t think you can hope for much more than that really.

1019

Women who were admitted for general anaesthetic were also satisfied with their treatment, apart from the long wait experienced by two women. Those who had a lot of pain during the hysteroscopy generally felt pleased to be randomised to inpatient treatment, and it was the reason for two preference patients’ decisions. As with the outpatient group, these women felt that the care they received was good.

Principal findings

The most obvious finding of the study was that women expressed satisfaction with their treatment, whatever their preference for inpatient or outpatient treatment, and whether or not they received it. For most of the women their symptoms were an inconvenience rather than a major disruption to their lives, something that they wanted dealing with quickly, and a ‘see and treat’ intervention could be seen as a proportionate response to the problem. The value was that it fitted into their lives, and is consistent with the immediacy and convenience of other aspects of modern living, such as internet banking and 24-hour shopping. A number of women spoke of the advantage of outpatient treatment as being discharged quickly and getting back to normal. However, women who were admitted in a two-stage procedure, whether for outpatient or inpatient treatment, also expressed satisfaction, and for seven women (four preference and three randomised) the idea of invasive treatment while awake was not acceptable, and therefore the option of general anaesthetic is still needed, even although this would seem to be a minority choice. However, this overall finding of satisfaction with the processes and procedures must be weighed against the levels of pain and discomfort experienced by a significant number of the women taking part.

In contrast with findings by Gupta et al. 120 in this study few women expressed feeling anxious preoperatively, although many spoke of expecting some pain during the procedure. This may be because this study was retrospective, whereas Gupta et al. 120 administered the State–Trait Anxiety Inventory pre procedure. It is also possible that as medical procedures are increasingly being carried out on an outpatient basis there is more familiarity and acceptance of them by patients, and indeed some women had previous experience of this for polyps and other conditions.

One finding that is hard to explain is that ‘preference’ patients reported less pain than ‘randomised’ patients. This may be an artefact of a small sample size, or preference patients wanting to justify their choice. There may also be a difference between a population that agrees to be randomised in research and that which does not, and it is important to obtain data from both, particularly in research such as this for which there was a large preference group.

Strengths and limitations of the study

No previous study has used a qualitative methodology to explore women’s experience of OPH, and to compare inpatient and outpatient experience. Undertaking a qualitative study within the OPT Trial has provided an additional perspective on being in a medical trial and has confirmed the overall satisfaction reported in the rating scales. It has also allowed for an expansion of the two minority experiences that women reported. The advantage of the approach was that we were able to explore issues that women raised of concern to them rather than the researcher-defined issues, thereby supplementing quantitative quality-of-life data with more in-depth exploration providing contextualised views. In addition, timing the semistructured interviews one week following treatment provided more perspective to women on their overall treatment.

The main limitations of the study were using telephone interviews, and not being able to obtain respondent validation – issues that may be linked. The disadvantages of telephone interviews have been discussed above, but in a multicentre study the choices are to restrict interviewing to a group of study sites in close proximity or to use the geographical range and to interview by telephone. We made the decision to try to recruit women from a range of centres from across the country, as we considered this to be more reflective of the whole study. Finally, as no previous qualitative study has been conducted on women’s experience of OPH, it was not possible to verify our findings by embedding them within existing literature.

Implications for practice

1. In view of the level of pain and distress experienced by some women, the development of ‘see and treat’ clinics and the option of an anaesthetic should be available if this is patient choice.

2. Patients need information about taking pain relief before the procedure, and the most appropriate analgesia to take. Top-up analgesia should be available during the procedure if necessary.

3. A dedicated rest area could be made available for patients who find the procedure painful and distressing.

4. The presence of a support nurse who is not part of the team carrying out the procedure was unanimously appreciated and should be considered good practice.

5. An information leaflet needs to contain information about what to expect from the procedure (quotes from previous patients are useful), pain relief, who will receive biopsy results, what to expect post procedure in terms of pain and bleeding, and returning to normal activities, work, resuming sexual activity) and who to contact if complications occur.

6. Further qualitative research should be conducted into the experience of undergoing OPH and similar gynaecological procedures.

Conclusion

The value of a qualitative study within a RCT is that it provides an additional type of data, and allows participants the opportunity to raise issues of importance to them, rather than those set by the research team. This research compared two treatment settings and procedures for a condition, uterine polyps, which causes inconvenience rather than major disruption to life. As such for most women a service that offers a timely and sensitive treatment, with a quick return to normal life is the optimum choice. For the majority of women in this study this was provided by ‘see and treat’, but for a minority this was not an acceptable option.