{"paper_id":"05dc6d17-bcb0-41cf-b7e2-ceb2f1fbe8ad","body_text":"Trial participants are frequently excluded based on their symptoms rather \nthan their condition: A systematic review of Cochrane reviews and their \ncomponent trials \nKatie Stockinga*, Andrew Watsonb, Jamie J Kirkhama, Jack Wilkinsona, Andy Vaila \na Centre for Biostatistics, Faculty of Biology and Health Sciences, University of Manchester, \nManchester, UK \nb Department of Obstetrics and Gynaecology, Tameside & Glossop Acute Services NHS Trust, \nAshton-Under-Lyne, UK \n* Correspondence address. Centre for Biostatistics, Division of Population Health, Health Services \nResearch and Primary Care, University of Manchester, Rm 1.307 Jean McFarlane Building, \nUniversity Place, Oxford Road, Manchester, M13 9PL, UK. E-mail: katie.stocking@manchester.ac.uk  \n \nAbstract  \nObjectives: Identify strategies used in the design of recent randomised controlled trials \n(RCTs) and their associated Cochrane reviews where patients with the same gynaecological \ncondition present with different symptoms.  \n \nStudy Design and Setting: We searched the Cochrane library (February 2022) for reviews in \npolycystic ovarian syndrome (PCOS) and endometriosis. Reviews were included if the \nintervention was intended to treat all condition-specific symptoms. We restricted to trials \npublished since 2012 to consider ‘current’ approaches. For each trial we recorded the number \nof potentially eligible participants excluded as a direct result of the chosen strategy.  For each \nreview we recorded the numbers of RCTs and participants excluded unnecessarily. \nResults There were 89 distinct PCOS trials in 13 reviews, and 13 Endometriosis trials in 11 \nreviews. Most trials restricted their eligibility to participants with specific symptoms (55% \nPCOS, 46% endometriosis). The second most common strategy was to measure and analyse \nclinical outcomes that were not relevant to all participants (38% PCOS, 31% endometriosis). \nReviews excluded 27% of trials based just on outcome data. \nConclusions: Current gynaecological research is inefficient. Most trials either exclude \npatients who could benefit from treatment or measure outcomes not relevant to all \nparticipants. \nRegistration: PROSPERO (CRD42022334776) \nKeywords: Research waste; Randomised Clinical Trial; Cochrane Review; Gynaecology; \nOutcome selection. \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\nWhat is new? \nKey findings \n Over a quarter of Cochrane reviews included in this review excluded trials based on \nthe outcomes reported. \n Typically, recent randomised controlled trials in Polycystic Ovarian Syndrome and \nEndometriosis trials either exclude patients who could potentially benefit from the \ntreatment given, or measure outcomes of no relevance to some participants. \nWhat this adds to what is known? \n Strategies developed that are employed in the design and measurement of outcomes in \ngynaecological trials. \n There are multiple sources of waste in the current gynaecological research landscape. The \npopulation of patients available is under-utilised by excluding patients based on the outcomes \nmeasured, or alternatively, researchers are measuring outcomes in patients who do not \nexperience the associated symptom(s). \nWhat is the implication and what should change now? \n Gynaecological patients experience heterogeneity in their symptoms and therefore it is crucial \nto employ appropriate outcome measures in order to reduce research waste. Cochrane \nReviews should include all trials which report outcomes that are relevant to the population \nof interest if the intervention under investigation is deemed to plausibly treat the associated \nsymptom(s).  \n \n1. Introduction \nThe principal role of a randomised controlled trial (RCT) is to evaluate whether a medical \nintervention is safe and effective. In order for this to happen, it is imperative that researchers measure \noutcomes which are both appropriate and relevant to the population of interest. Although randomised \ncontrolled trials remain the gold standard tool for treatment evaluation, many, through poor design, \ncontribute to the overwhelming problem of waste in research [1-4]. In the Lancet collection of papers \non waste in medical research, it was estimated that $240 billion of annual research expenditure is \nwasted [3, 5-8]. It is indeed true that much work is being done to reduce this figure, however, there is \nstill much room for improvement [9, 10] . Inefficient studies that fail to address questions that matter \nto both patients and stakeholders emphasise the importance that we need to do less, but better, \nresearch [11].   \n \nOften, in the case of gynaecological conditions such as Polycystic Ovarian Syndrome (PCOS) and \nEndometriosis, patients require different things from their care, at different stages of their lifetime \n[12-15]. Not all patients with the same diagnosis will experience all of the associated complications, \nand, although their most bothersome symptom might differ, it is common to receive the same \ntreatment. If we take PCOS, for example, the Rotterdam criteria are the most widely used \nclassification for diagnosis and proposes that PCOS is present if the patient has at least two of the \nthree characteristics: oligo- and/or anovulation, clinical and/or biochemical hyperandrogenism, and \npolycystic ovaries on ultrasound [16]. These criteria for the diagnosis of PCOS in itself has \nconsequences, as by definition, not all patients with PCOS have all the possible manifestations of the \ndisorder and therefore do not experience the same symptoms and health risk factors [12]. For \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\nendometriosis, patients may have symptoms dominated by pelvic pain, infertility, or both. In the post-\nreproductive era the reduction in quality-of-life from menstrual dysfunction may predominate. In the \nlikely scenario that potential trial participants have no, or little, symptoms in common, it presents the \nproblem of how researchers select relevant, patient-important outcomes and design RCTs that are not \nwasteful.  \nOne of the most notable inclusions to the movement to reduce waste in research is the development of \ncore outcome sets (COS). The Core Outcome Measures in Effectiveness Trials (COMET) Initiative \nencourages the application of agreed standardised sets of outcomes. These outcome sets represent the \nminimum that should be measured and reported in clinical trials in specific clinical areas [17-19]. \nRecently developed core outcome sets in both Polycystic Ovarian Syndrome (PCOS) and \nendometriosis undoubtedly violate the concept of COS, as authors in both instances concluded that \nnot all outcomes could be reported in all trials [20, 21]. This is demonstrative of the fact that in the \nfield of gynaecology, it is not one-size-fits-all.  \n \nThis prompts the question of how we should design trials so as to incorporate this heterogeneity, as \nwell as the implications for systematic reviews and meta-analysis. Due to the multifactorial nature of \na patient’s symptoms, the design and recruitment of gynaecological trials is challenging and is \napproached in different ways. We conducted a systematic review to investigate how diverse \nsymptoms and patient populations are currently handled in gynaecological trials, in which the \nintervention could plausibly be used to treat all symptoms related to the diagnosis. We aimed to \nidentify the methodological strategies applied within randomised controlled trials in Cochrane \nReviews, where an intervention is hypothesised to have potential benefit for patients with the index \ncondition.  \n \n2. Methods \nThe study design was a systematic review with descriptive statistics. Our overall approach was to \nidentify systematic reviews in the conditions of interest and to examine the characteristics and \nmethodological practice of their included and excluded trials. Protocol registration was with the \nProspective Register of Systematic Reviews (registration number: CRD42022334776). Full details of \nmethods are given there but summarised below. \nIn February 2022 searches were undertaken to identify systematic reviews contained in the Cochrane \nLibrary on interventions for PCOS or endometriosis. We considered trials list under both ‘included’ \nand ‘excluded’ categories. \n2.1 Study inclusion criteria \nDecisions regarding eligibility were made by discussion with AW, a consultant in gynaecology. \nCochrane intervention reviews and RCTs were eligible for inclusion from 2012 onwards to give an \noverview of current practice. Cochrane reviews were included only if the intervention under \ninvestigation was intended to treat the underlying condition, and would therefore plausibly treat all \ncondition-specific symptoms. For example, in vitro fertilisation would not be an eligible treatment, as \nit would be used only for fertility outcomes.   \n2.2 Data extraction \nTwo reviewers (KS and AW) screened all titles and abstracts against the inclusion criteria. Any \ndisagreements were resolved through discussion. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\nOutcomes categories were pre-specified (appendix), along with whether each outcome was primary, \nsecondary, or unspecified.  This information was also extracted for each of the trials, along with the \nsetting, intervention type, funding source, design, size (number of participants randomised), number \nof participants excluded for symptom/outcome-related reasons and the number of participants \ncontributing to the primary outcome.  \nSeven different strategies were anticipated for the RCTs, as demonstrated in Figure 1. We developed \nthese strategies as part of an iterative process. As a pilot, we developed some potential strategies we \nhad seen during our time as researchers, assessed several trials and determined if the strategy used by \neach trial was different to those we anticipated. We then discussed and re-evaluated until no new \nstrategies were found. We note that this may not be an exhaustive list, and we considered whether any \nadditional strategies not included in our list were used. Each trial was categorised according to the \nstrategy used, in the order listed, i.e. if the trial did not fit the criteria to be the first strategy, it was \nconsidered for the next, and so on.   \nFor ease of context and reference, we named the strategies and will refer to them as such throughout: \n Participant-specific Outcome Strategy - patient-specific primary outcome was chosen, for \nexample patient satisfaction or success.  \n Composite Outcome Strategy - composite outcome, for example the resumption of \nmenstruation or hirsutism improvement.  \n Universal Eligibility Strategy - measured outcomes regardless of their relevance to the \npatients in the study, e.g. a study which did not restrict to patients experiencing amenorrhea (a \nlack of menstruation) but measured resumption of menstrual period as an outcome in every \npatient.  \n Grouped Universal Eligibility Strategy would be trials designed the same way as the \nUniversal Eligibility Strategy, but the statistical analysis would be restricted to those who \nexperienced amenorrhea at baseline, as a subset analysis.  \n Restricted Eligibility Strategy was allocated where a trial measured only clinical outcomes \nbased on the symptoms of all patients in the study, i.e. a fertility trial, interested only in \nfertility-based outcomes, where the inclusion criteria is patients experiencing subfertility. \n Downstream Eligibility strategy was studies measuring only quality-of-life outcomes. \n Upstream Eligibility strategy was studies reporting no clinical outcomes, only biomarker \noutcomes. \nFor each trial, where available, we recorded the number of potentially eligible participants excluded \nas a direct result of the chosen strategy relative to the achieved sample size.  That is, how much larger \ncould the recruitment have been without an outcome-defined restriction on eligibility criteria. Where \navailable, the number of potentially eligible participants were taken from the CONSORT flow chart. \nThis was taken to be the number of participants that were found to be ineligible for outcome-related \nreasons pre-randomisation. Similarly, for each Cochrane review we recorded the numbers of \nidentified RCTs excluded from consideration for reasons such as not reporting review-specific \noutcomes. To determine which trials were excluded from each Cochrane review and the reason for \nexclusion, we used the ‘Characteristics of excluded studies’ section of the review. We selected all \nexcluded trials for assessment where the reason for exclusion was related to the population of patients, \noutcome reporting or selection.  \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\n \nFigure 1 - Strategies for randomised controlled trials in gynaecology. \n2.3 Study quality assessment and data analysis \nWe accepted the published risk of bias assessment for the included studies. For the studies that were \nexcluded from the Cochrane review, and therefore have no associated risk of bias assessment, these \nwere independently assessed using Cochrane’s risk of bias tool as a surrogate for their quality.  \nDescriptive analyses were undertaken. All data synthesis was exploratory and included the calculation \nof a mean number of exclusions.  \nWe aimed to analyse whether calendar year of publication and study-level factors (e.g. sample size, \nnature of intervention, study quality assessed by risk of bias) were associated with inefficiency, using \nFisher’s Exact test as an exploratory analysis. For this, date of publication was divided into pre-2017 \nPatient-specific \noutcome\n• Patient-specific outcome chosen, i.e. measurement of \noutcomes that are identified as most important to each patient, \nor asking for satisfaction at the end of the study.\nComposite \noutcome\n• Composite outcome e.g. resumption of menstruation or \nhirsutism improvement.\nUniversal & \nGrouped \nUniversal \nEligibility\n• Universal- Measurement of one or more clinical outcomes that \nare not relevant to all participants with analysis overall.\n• Grouped - Measurement of one or more clinical outcomes that \nare not relevant to all participants, analysis only of subsets.\nRestricted \nEligibility\n• Measurement of clinical outcomes based on symptoms of all \npatients in study (e.g. by restricting eligibility criteria)\nDownstream & \nUpstream\n• Downstream- Outcome(s) measured is 'downstream' (such as \nquality of life) \n• Upstream- Outcome(s) measured is 'upstream' (such as at \nbiomarker level)\n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\nand 2017 or after; nature of intervention was medical, surgical or other; risk of bias was judged as \neither in the primary analysis or not if selection bias was deemed low using Cochrane Gynaecology \nand Fertility’s guidance [22] and strategy was compared as Universal Eligibility, Restricted Eligibility \nor Other strategy. \n \n3. Results \n3.1 Study selection and characteristics \nFor PCOS, there were 31 Cochrane reviews screened at the title and abstract stage, of which 13, \ncontaining 239 trials (included those that were excluded from the associated Cochrane review), met \nthe inclusion criteria. There were 136 trials which were excluded from this review for reasons relating \nto design and access, information can be found in the PRISMA (Figure A1). Duplication occurred in \ntwo ways, the first, where the same trial appeared in multiple Cochrane reviews. We removed these \nduplicate trials, including only the first time, chronologically, that the trial appears in a review (2 trials \nappeared three times, 7 appeared twice). The second form of duplication occurred on 3 instances \nwhere publications had re-analysed original trial data, reporting different outcomes. In this case, all \ndata were collected and considered as one trial. Therefore, a total of 89 trials in 13 Cochrane reviews \ncontributed to the findings of this review. \nFor endometriosis, 32 Cochrane reviews were screened at the title and abstract stage with 11, \ncontaining 19 trials, meeting the inclusion criteria. Six trials were excluded as they were abstract only \n(n=3), inaccessible (n=1) or had no locatable publication (n=2), therefore, a total of 13 trials in 11 \nCochrane reviews were included in our systematic review (Figure A2). The trial characteristics are \nsummarised in Table 1. Most commonly, PCOS patients were recruited from obstetrics and \ngynaecology clinics (40%) and fertility clinics (35%), with very few research teams recruiting from \nthe community (4%). The majority of endometriosis patients were recruited from obstetrics and \ngynaecology clinics (85%). \nTable 1 - Characteristics of RCTs identified in PCOS and endometriosis \nDemographics PCOS  \n(n=89) \nEndo  \n(n=13) \nIncluded in Cochrane Review  \nYes  \nNo  \n \n62 (70%) \n27 (30%) \n \n12 (92%) \n1 (8%) \nNumber Randomised \nMedian (IQR) \nMinimum \nMaximum \n \n87 (50, 122) \n15 \n1000 \n \n130 (60, 159) \n40 \n360 \nType of Trial \nParallel  \nCrossover  \nFactorial  \n \n87 (98%) \n1 (1%) \n1 (1%) \n \n13 (100%) \n- \n- \nIntervention Type \nMedical \nSurgical \nLifestyle \nAlternative \nMedical & Surgical \nMedical & Lifestyle \nMedical & Alternative \nLifestyle & Alternative \n \n54 (58%) \n8 (9%) \n7 (8%) \n10 (11%) \n5 (5%) \n1 (1%) \n5 (5%) \n7 (3%) \n \n4 (31%) \n5 (38%) \n- \n- \n4 (31%) \n- \n- \n- \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\n \n \n \n \n \n \n \n \n*Other = Acupuncture, Morphology, Nutrition, Physiopathology, Medical and Surgical sciences, Urology. \n3.2 Trial strategies  \nOnly 7 trials included in this review reported the number of participants excluded pre-randomisation, \nfor outcome-based reasons. However, these 7 trials excluded a total of 990 participants (median 16, \nIQR: 4-229, minimum: 3, maximum: 704). To give context to this, the total sample size accrued by \nthese 7 trials was 2744 (median 172, IQR: 46-750, minimum: 45, maximum: 1000).  \nThe most common strategy used by researchers in PCOS and endometriosis was Restricted Eligibility \n(55% and 46%, respectively). In PCOS, we found that over half of the trials that used Restricted \nEligibility focused solely on fertility-based outcomes (59%), demonstrated in Table 2. For \nendometriosis there was more variation in the outcomes reported, with a third (33%) of Restricted \nEligibility studies reporting pain only outcomes and a third (33%) reporting fertility only outcomes. \nUniversal Eligibility was the second most used, with 38% of PCOS trials using this approach. Of \nthese trials, the majority measured combinations of outcomes, most commonly choosing to measure \nmultiple clinical outcome combinations (41%) or fertility and other clinical outcomes (29%). There \nwere 31% of endometriosis trials that used this strategy. Similarly to the PCOS trials, they measured \ncombinations of outcomes: fertility and pain (50%) and patient important plus other outcomes (50%) \nNo PCOS trials and only one endometriosis trial that employed the Universal Eligibility strategy \nmade note of the numbers of patients experiencing the primary outcome at baseline. Therefore, we \nwere unable to calculate the ratio of patients not experiencing the primary outcome of interest in \nrelation to the sample size randomised. \nGrouped Universal Eligibility was used in 15% of identified endometriosis trials, and only 1% in \nPCOS trials. The Upstream Outcome strategy was used in 8% of endometriosis trials, and 6% of \nPCOS trials. There were no instances where the Patient-specific Outcome, Composite Outcome or \nDownstream Outcome strategies were used. \nThere was no significant difference in the strategies used in pre-2017 compared to 2017 and after in \nPCOS (p=0.76) or endometriosis (p=0.63). Interventions tested were also not significantly different \nbetween trial strategies in PCOS (p=0.05) or endometriosis (p=0.27). When comparing the studies \nwhich could plausibly have been included in a primary analysis, using the RoB assessment, only 5 out \nof the 102 possible PCOS and endometriosis trials were classified as low risk. \n3.3 Review-level findings \nOf the 27 trials excluded from the Cochrane reviews in PCOS, most were excluded as the authors of \nthe review were interested in fertility outcomes: 74% (n=20) because they were non-fertility studies, \nFunding Source \nNon-Commercial  \nCommercial  \nMixed  \nNone  \nNo Info  \n \n28 (31%) \n1 (1%) \n3 (3%) \n21 (24%) \n36 (40%) \n \n7 (54%) \n2 (15%) \n1 (8%) \n- \n3 (23%) \nMulticentre  \nYes  \nNo  \n \n6 (7%) \n83 (93%) \n \n4 (31%) \n9 (69%) \nSetting \nFertility clinic  \nObstetrics and gynaecology  \nCommunity  \nOther  \n \n31 (35%) \n36 (40%) \n5 (6%) \n13 (15%) \n \n0 \n11 (85%) \n0 \n2 (15%) \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\n15% (n=4) were PCOS with or without infertility. The remaining trials were excluded for having no \noutcomes of interest (n=2, 7%) and fertility outcomes only (n=1, 4%). These 27 trials had a median \nnumber of 60 people randomised, (IQR: 45-88, minimum: 26, maximum: 233). Similarly, the trial \nexcluded from the endometriosis reviews was on the basis of not reporting the review’s outcome of \ninterest. Overall, the exclusion of trials on the basis of the outcomes they reported totalled 27% of \navailable RCTs (28/102). \n \nTable 2 - Proportion of outcome combinations in each strategy. \n \nOutcome \nCombination \nStrategy \nUniversal \nEligibility \nGrouped Universal \nEligibility \nRestricted \nEligibility \nUpstream  \nOutcome \nPCOS \n(n=34) \nEndo \n(n=4) \nPCOS \n(n=1) \nEndo \n(n=2) \nPCOS \n(n=49) \nEndo \n(n=6) \nPCOS \n(n=5) \nEndo \n(n=1) \nFertility only 0 0 0 0 29 \n(59%) \n2 \n(33%) \n0 0 \nPain only 0 0 0 0 0 2 \n(33%) \n0 0 \nFertility + \nQoL + Other \nclinical \noutcome(s) \n6 \n(18%) \n0 0 0 0 0 0 0 \nFertility + \nOther clinical \noutcome(s)* \n10 \n(29%) \n2  \n(50%) \n1 \n(100%) \n0 13 \n(26%) \n0 0 0 \nQoL + Other \nclinical \noutcome(s) \n4 \n(12%) \n0 0 2 \n(100%) \n \n0 1 \n(17%) \n0 0 \nClinical \noutcome(s) + \nTreatment \nsatisfaction \n0 2 \n(50%) \n0 0 0 1 \n(17%) \n0 0 \nOther clinical \noutcome \ncombinations \n14 \n(41%) \n0 0 0 7 \n(14%) \n0 0 0 \nBiomarkers \nonly \n0 0 0 0 0 0 5 \n(100%) \n1  \n(100%) \nPercentage given is within condition. *fertility + pain for endometriosis \n \n4. Discussion \n \nWe conducted our systematic review to examine the current strategies used for the design of and \nrecruitment to randomised controlled trials, by researchers. We focused on PCOS and endometriosis \nas exemplar conditions due to the heterogeneity of their symptoms, however, we anticipate the \nfindings to be applicable to gynaecological research in general, and in other conditions where patients \nexperience varying symptoms. \n \nSystematic reviews are considered the gold standard of evidence for decision-making, used to collate, \ncritique and summarise evidence. However, we found that the Cochrane reviews included in this \nreview, excluded many trials, based on the outcomes they reported. Whilst at first glance, choosing a \nresearch question, and selecting trials that report outcomes relating to that research question does not \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\nappear to be erroneous, there are issues with this. Our review considered only Cochrane reviews of \ninterventions which could, plausibly, be used regardless of symptoms. Where authors exclude trials \nbased on their outcomes, they are reducing the number of patients with data available to add to the \nevidence base. Dwan and colleagues have previously discussed the prevalence and impact of \nexcluding studies from reviews on the basis of the relevance of outcome data, with further works \nobserving that doing so leads to potentially biasing the conclusions of systematic reviews, along with \nwaste in production and reporting of research [23, 24]. Therefore, it would be advantageous for \nreviews to consider trials regardless of reported outcomes, and instead consider a multitude of \nsymptoms and outcomes that could plausibly be treated with the intervention under investigation. This \ncould in turn reduce the number of reviews needed, and hence waste from duplication of effort.  \n \nIn order to tackle the demonstrated waste in gynaecological trials we need to focus first on improving \nthe quality of the research questions, ensuring they matter to patients and clinicians. Three of the \nstrategies we anticipated: Participant-specific Outcome, Composite Outcome, and Downstream \nOutcome were not being utilised and require consideration. It is likely that these strategies are not \ncurrently perceived as attractive to use as those that are more common. Participant-specific outcomes, \nsuch as measuring a patient’s most bothersome symptom or allowing a patient to set their own \npersonalised goals, are relatively novel ideas, and although they have previously been used in \ngynaecological trials, further consideration is needed for their statistical advantages [25, 26]. \nComposite outcomes allow research to address more than one aspect of a patient’s health status, but \ntheir use is widely debated, with interpretation difficult [27-30]. Similarly, although a well-established \ninstrument for providing evidence of an individual’s physical, emotional and social health, \ninterpretation of treatment effect can also be difficult when using a quality-of-life measurement. Cox \net al suggests that “in practice the main difficulty is likely to be in separating the real treatment-by-\nindividual interaction from noise” [31].  \nIn the current research landscape, we are typically seeing two main strategies employed in \ngynaecological trials, which we refer to as Universal Eligibility and Restricted Eligibility. In the first, \noutcomes measured are of relevance to everybody in the trial. This means patients are excluded due to \nstringent inclusion criteria and are unable to be involved in a trial of treatment that may be of clinical \nbenefit to them. Secondly and conversely, some trials include participants which are not experiencing \nthe symptom at baseline, resulting in the inability to provide useful information, as they cannot \ncontribute to the outcome of interest. For example, researchers record BMI as an outcome for patients \nnot overweight at baseline. In April 2022, the Royal College of Obstetricians and Gynaecology \n(RCOG) reported that gynaecological care waiting lists in England had grown the most substantially \ncompared to any other medical specialty, seeing an over 60% increase on pre-pandemic levels [32]. \nThere is no shortage of people seeking and requiring gynaecological treatment. Both of the strategies \nthis review identified as most prevalent in gynaecological research do not utilise the potential \npopulation and are inefficient, often leading to research that lacks in impact and results in waste.  \n \n5. Conclusions \nAs a minimum, participants whose symptoms could potentially benefit from any specific treatment \nshould have a chance to receive said treatment. This research identified multiple sources of waste in \nthe current gynaecological research landscape. We have shown that the population of patients \navailable is often under-utilised by excluding patients based on the outcomes measured, or \nalternatively, researchers are measuring outcomes in patients who do not experience the associated \nsymptom(s). \nGynaecological patients experience heterogeneity in their symptoms and therefore it is crucial to \nemploy patient-specific outcome measures. Not only would this reduce research waste from a budget \nand resource perspective, it would also, most importantly, alleviate patient burden. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\nAcknowledgements \nThe authors thank the Public Patient Involvement group that have given their time to help support this \nresearch and shape the strategies developed. \nAuthors’ contributions \nKatie Stocking: Conceptualization; Data curation; Formal analysis; Methodology; Writing – Original \ndraft preparation. Andrew Watson: Conceptualization; Data curation; Validation; Writing – \nReviewing and editing. Jamie J Kirkham: Conceptualization; Methodology; Supervision; Writing – \nReviewing and editing. Jack Wilkinson: Methodology; Supervision; Writing – Reviewing and editing \nAndy Vail: Conceptualization; Methodology; Supervision; Writing – Reviewing and editing. \nFunding \nKatie Stocking, Doctoral Research Fellow (NIHR301756), is funded by the National Institute for \nHealth and Care Research (NIHR) for this research project. The views expressed in this publication \nare those of the author(s) and not necessarily those of the NIHR, NHS or the UK Department of \nHealth and Social Care. \n \n \nReferences \n[1] Chalmers I, Glasziou P. Avoidable waste in the production and reporting of research evidence. \nLancet. 2009;374:86-9. \n[2] Ioannidis JP. Clinical trials: what a waste. BMJ. 2014;349:g7089. \n[3] Ioannidis JP, Greenland S, Hlatky MA, Khoury MJ, Macleod MR, Moher D, et al. Increasing value \nand reducing waste in research design, conduct, and analysis. Lancet. 2014;383:166-75. \n[4] Yordanov Y, Dechartres A, Atal I, Tran VT, Boutron I, Crequit P, et al. Avoidable waste of research \nrelated to outcome planning and reporting in clinical trials. BMC Med. 2018;16:87. \n[5] Al-Shahi Salman R, Beller E, Kagan J, Hemminki E, Phillips RS, Savulescu J, et al. Increasing value \nand reducing waste in biomedical research regulation and management. Lancet. 2014;383:176-85. \n[6] Chalmers I, Bracken MB, Djulbegovic B, Garattini S, Grant J, Gulmezoglu AM, et al. How to \nincrease value and reduce waste when research priorities are set. Lancet. 2014;383:156-65. \n[7] Chan AW, Song F, Vickers A, Jefferson T, Dickersin K, Gotzsche PC, et al. Increasing value and \nreducing waste: addressing inaccessible research. Lancet. 2014;383:257-66. \n[8] Glasziou P, Altman DG, Bossuyt P, Boutron I, Clarke M, Julious S, et al. Reducing waste from \nincomplete or unusable reports of biomedical research. Lancet. 2014;383:267-76. \n[9] Gillies K, Chalmers I, Glasziou P, Elbourne D, Elliott J, Treweek S. Reducing research waste by \npromoting informed responses to invitations to participate in clinical trials. Trials. 2019;20:613. \n[10] Minogue V, Cooke M, Donskoy AL, Vicary P, Wells B. Patient and public involvement in reducing \nhealth and care research waste. Res Involv Engagem. 2018;4:5. \n[11] Altman DG. The scandal of poor medical research. BMJ. 1994;308:283-4. \n[12] Escobar-Morreale HF. Polycystic ovary syndrome: definition, aetiology, diagnosis and treatment. \nNat Rev Endocrinol. 2018;14:270-84. \n[13] Hart R, Hickey M, Franks S. Definitions, prevalence and symptoms of polycystic ovaries and \npolycystic ovary syndrome. Best Pract Res Clin Obstet Gynaecol. 2004;18:671-83. \n[14] Geukens EI, Apers S, Meuleman C, D'Hooghe TM, Dancet EAF. Patient-centeredness and \nendometriosis: Definition, measurement, and current status. Best Pract Res Clin Obstet Gynaecol. \n2018;50:11-7. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint \n\n[15] Johnson NP, Hummelshoj L, Adamson GD, Keckstein J, Taylor HS, Abrao MS, et al. World \nEndometriosis Society consensus on the classification of endometriosis. Hum Reprod. 2017;32:315-\n24. \n[16] Rotterdam EA-SPcwg. Revised 2003 consensus on diagnostic criteria and long-term health risks \nrelated to polycystic ovary syndrome (PCOS). Hum Reprod. 2004;19:41-7. \n[17] Gargon E, Williamson PR, Altman DG, Blazeby JM, Tunis S, Clarke M. The COMET Initiative \ndatabase: progress and activities update (2015). Trials. 2017;18:54. \n[18] Williamson PR, Altman DG, Bagley H, Barnes KL, Blazeby JM, Brookes ST, et al. The COMET \nHandbook: version 1.0. Trials. 2017;18:280. \n[19] Williamson PR, Altman DG, Blazeby JM, Clarke M, Devane D, Gargon E, et al. Developing core \noutcome sets for clinical trials: issues to consider. Trials. 2012;13:132. \n[20] Al Wattar BH, Teede H, Garad R, Franks S, Balen A, Bhide P, et al. Harmonising research \noutcomes for polycystic ovary syndrome: an international multi-stakeholder core outcome set. Hum \nReprod. 2020;35:404-12. \n[21] Duffy J, Hirsch M, Vercoe M, Abbott J, Barker C, Collura B, et al. A core outcome set for future \nendometriosis research: an international consensus development study. BJOG. 2020;127:967-74. \n[22] CGF. Cochrane Gynaecology and Fertility - Guidance for Authors \nhttps://cgf.cochrane.org/sites/cgf.cochrane.org/files/public/uploads/cgf_guidance_for_review_upd\nates_and_new_reviews_rob_1_final_v2.3.pdf; 18th January 2023 \n[23] Dwan KM, Williamson PR, Kirkham JJ. Do systematic reviews still exclude studies with \"no \nrelevant outcome data\"? BMJ. 2017;358:j3919. \n[24] Kirkham JJ, Altman DG, Chan AW, Gamble C, Dwan KM, Williamson PR. Outcome reporting bias \nin trials: a methodological approach for assessment and adjustment in systematic reviews. BMJ. \n2018;362:k3802. \n[25] Ettinger B, Hait H, Reape KZ, Shu H. Measuring symptom relief in studies of vaginal and vulvar \natrophy: the most bothersome symptom approach. Menopause. 2008;15:885-9. \n[26] Tanmahasamut P, Jirasawas T, Laiwejpithaya S, Areeswate C, Dangrat C, Silprasit K. Effect of \nestradiol vaginal gel on vaginal atrophy in postmenopausal women: A randomized double-blind \ncontrolled trial. J Obstet Gynaecol Res. 2020;46:1425-35. \n[27] Cordoba G, Schwartz L, Woloshin S, Bae H, Gotzsche PC. Definition, reporting, and \ninterpretation of composite outcomes in clinical trials: systematic review. BMJ. 2010;341:c3920. \n[28] Freemantle N, Calvert M, Wood J, Eastaugh J, Griffin C. Composite outcomes in randomized \ntrials: greater precision but with greater uncertainty? JAMA. 2003;289:2554-9. \n[29] Goldberg R, Gore JM, Barton B, Gurwitz J. Individual and composite study endpoints: separating \nthe wheat from the chaff. Am J Med. 2014;127:379-84. \n[30] Ross S. Composite outcomes in randomized clinical trials: arguments for and against. Am J \nObstet Gynecol. 2007;196:119 e1-6. \n[31] Cox DR, Fitzpatrick R, Fletcher AE, Gore SM, Spiegelhalter DJ, Jones DR. Quality-of-Life \nAssessment: Can We Keep It Simple? Journal of the Royal Statistical Society Series A (Statistics in \nSociety). 1992;155:353-93. \n[32] RCOG. More than half a million women face prolonged waits for gynaecology care - Royal \nCollege of Obstetricians & Gynaecologists https://www.rcog.org.uk/news/more-than-half-a-million-\nwomen-face-prolonged-waits-for-gynaecology-care/ 18th January 2023 \n \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprintthis version posted April 21, 2023. ; https://doi.org/10.1101/2023.04.21.23288922doi: medRxiv preprint","source_license":"CC0","license_restricted":false}