Performance measures for upper gastrointestinal endoscopy: A European Society of Gastrointestinal Endoscopy quality improvement initiative
Year: 2016
Type: review
Abstract: The European Society of Gastrointestinal Endoscopy (ESGE) and United European Gastroenterology (UEG) have identified quality of endoscopy as a major priority and we described our rationale for this in a first manuscript that also addressed the methodology of the quality initiative process.1 The identification of upper gastrointestinal (UGI) performance measures presents a considerable challenge, in contrast to the situation with colonoscopy for instance, where several performance measures (inspection time, adenoma detection rate, and interval cancers, among others) have been identified over the last decade.2, 3 Following the Quality in UGI Endoscopy meeting held in Lisbon in 2013, it was clear that there was a need to identify performance measures for the UGI tract, and that quality standards could be identified although there is a paucity of evidence. This lack of evidence helps however to identify research priorities for the development of clinical trials that will further validate and substantiate the implementation of performance measures. The aim therefore of the UGI working group was twofold: (a) to identify performance measures for UGI endoscopy; (b) to identify the evidence or absence of evidence that would develop the research priorities in this field. We used an innovative methodology to facilitate the quality initiative process, which combined a thorough search and standardized evaluation of the available evidence for each clinical question, followed by a Delphi process (http://is.njit.edu/pubs/delphibook/delphibook.pdf ) using an online platform.4, 5 This online platform permitted iterative rounds of modification and comment by all members of the UGI working group until agreement was reached on the performance measure. We now report these newly identified performance measures. We previously described the multistep process for the methodology to develop performance measures.1 Briefly, following the Lisbon meeting in 2013, a list of 56 possible performance measures was distributed to all of the working group members for comments, suggestions, and shortcomings in September 2014. Every participant was required to comment on all of the proposed performance measures during teleconferences that took place between 1 October–18 December 2014. All possible performance measures that were identified by this process were structured using the PICO framework (where P stands for population/patient; I for intervention/indicator; C for comparator/control; and O for outcome) to inform searches for available evidence to support the performance measures. This process resulted in 67 possible performance measures and 108 PICOs. Due to the timeframe for this first initiative and the wide range of pathology in the UGI tract, the working group had to prioritize general UGI endoscopy topics within the abundance of proposed performance measures and PICOs. As part of this prioritization, PICOs that were concerned with areas where guidelines were already available or under development were omitted. We also excluded PICOs that focused on: the assessment of effectiveness, or comparative effectiveness, of specific treatments (e.g. administration of proton pump inhibitors (PPIs) before endoscopy for acute bleeding, percentage of patients undergoing endoscopic resection in Barrett's esophagus with high grade dysplasia (HGD) before ablation); legal or local regulation (informed consent); histopathology (e.g. the need for confirmation/revision of a diagnosis of dysplasia by an independent pathologist); and service working group issues (e.g. adequate management of anticoagulants, sedation, etc.). The initial priority list was developed during a face-to-face meeting on 14 February 2015. In total, 44 PICOs were retained as the basis for literature searches. Several disease-specific performance measures were also developed (Barrett's esophagus, intestinal metaplasia in the stomach, and squamous cell cancer (SCC) in the esophagus). The PICOs, and the clinical statements derived from these which were organized into eight domains on the basis of their clinical applications, were adapted and/or excluded during the iterative rounds of comments and suggestions from the working group members during the Delphi process. The evolution and adaptation of the different PICOs and clinical statements during the Delphi process can be reviewed in the Supplementary Material. In total, working group members participated in four rounds of voting to agree on the performance measures in predefined domains and their respective thresholds, which are discussed below. The agreement that is given for the different statements refers to the last voting round in the Delphi process. A statement was accepted if at least 80% agreement was reached after a minimum of two voting rounds. The performance measures are displayed in boxes under the relevant domain. Each box describes the performance measure and the rationale behind its adoption, the agreement on acceptance during the modified Delphi process, and the grading of the available evidence, along with details of how the score should be measured and the desired threshold. During the Delphi process, the Quality Improvement Committee (QIC) chairs distinguished key performance measures from minor performance measures to assist service providers with decisions about the implementation of performance measures in their endoscopy services. Reasons to qualify a performance measure as minor included the measure being very disease specific (e.g. detection of neoplasia during surveillance of Barrett's esophagus or gastric intestinal metaplasia) or that its implementation might be relatively difficult and dependent on the availability of adequate software for auditing of the performance measures. The division and allocation of performance measures to key and minor performance measures was agreed by the UGI working group in an additional face-to-face meeting in April 2016. The number of cases that need to be audited to adequately assess if the threshold for a certain performance measure is reached can be calculated by estimating the 95% confidence intervals (CIs) for a predefined threshold and variable sample size (see Appendix 1). For reasons of practicality and feasibility when implementing an audit, the working group agreed that 100 procedures (or all, if <100 procedures had been performed) should be measured to assess the performance measure. Ideally this should be done at an individual procedure level but, as this requires robust and sophisticated software, we suggest that the assessment is first performed at a service level. If problems are detected at a service level, further analysis at an individual level is then required to identify possible targets for improvement. In the first round of development, the working group accepted 11 performance measures in total, after a total of four voting rounds in the modified Delphi process. The evidence quality (assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) criteria)6 for most of these performance measures is low; however, this does not indicate that a performance measure is not important. Taking into account both the feasibility of implementation and the possible impact on diagnostic quality and patient outcome, we identified six key and five minor performance measures (Figure 1 and Table 1 ). Nevertheless, all the performance measures were deemed valuable by the working group members and were obtained after a rigorous process, as described above. From a practical viewpoint, it may be desirable to implement the key performance measures first in those units that are not presently monitoring any performance measures. Once a culture of quality measurement is accepted and software is available, the minor performance measures may then further aid monitoring of the quality of UGI endoscopy. The domains and performance measures chosen by the working group. MAPS: management of precancerous conditions and lesions in the stomach; SCC: squamous cell carcinoma. All of the performance measures are described below, according to the domain to which they are attributed. The PICOs and statements that were used during the modified Delphi process to develop the performance measures can be found in the Supplementary Material. The statement numbers correspond to those used in Supplementary Material. Two studies were found that addressed these instructions and the duration of fasting prior to a scheduled UGI endoscopy.7, 8 In both studies, the authors mainly assessed the fasting time for liquids. The fasting time for solids was at least six hours prior to the endoscopy in both studies, with a good effect on visibility during endoscopy. In the study by Koeppe et al.,7 general discomfort was reported less frequently by patients who had a drink of water (200 ml) two hours before the procedure than in those who were fasting for solids and liquids for a full eight hours (18% vs 42%; p = 0.010). Even though the endoscopists subjectively observed more liquid in the stomach of the former group, no cases of aspiration were observed in the sample of 50 lightly sedated patients. De Silva et al.8 also reported lower discomfort scores when water ad libitum was allowed until one hour before the procedure (recorded volumes drunk were 200–410 ml) compared with no water being allowed during a six-hour pre-endoscopy fast (5.6 vs 9.7; p < 0.0001). No significant differences were found for complications and safety outcomes, apart from a significant difference in the volume of retained fluid in the gastric fundus, this being greater when water was drunk until one hour prior to the procedure, which was performed without sedation. Again no cases of aspiration were observed. The outcome "incomplete examination" was not reported in the retrieved studies. The outcome "good or normal visibility of gastric mucosa" could be used as an indirect outcome for the evaluation of incomplete examination. This outcome was consistently high after both two hours/one hour minimum of no fluids and nil by mouth for at least six hours (96% vs 98%7 and 93% vs 100%,8 respectively). Although no data from the two available studies directly assessed the duration of fasting for solids, it appears that an interval of at least six hours is safe and effective for UGI endoscopy in patients without any history or predisposing factors for delayed gastric emptying. For endoscopies that are planned to be performed in the afternoon, patient satisfaction may be increased if a small breakfast is allowed. Unlike with colonoscopy, we do not have a standardized scale to measure "gastric preparation" for UGI endoscopy. We advise that the contents of the stomach, such as food residues, blood, bile, or the presence of bubbles, should be reported, along with information on whether a waterjet system was used to improve mucosal visualization. Just recording this as a surrogate performance measure may however omit patients that are sent home again because they did not receive fasting instructions. Recording that proper instructions were given should therefore be done prior to the endoscopy itself and this could be included in the pre-assessment part of the endoscopy report (together with, for instance, the informed consent). This would mean that patients who show up for endoscopy having not received proper instructions and therefore have their endoscopy cancelled should be included in any audits of this performance measure. In contrast to colonoscopy, there is a paucity of data on the assessment of a complete procedure. As a definition, "reaching the duodenum" seems too trivial: it does not really reflect endoscopic competence as it is easy to achieve; there is no data comparable to cecal intubation rate in the colon that supports its relationship to better disease detection. During the discussions of the working group, there was a strong emphasis on trying to define this performance measure and searching the literature for an anatomical landmark or finding that might be related to disease detection. We formulated several PICOs to assess whether reaching any specific anatomical landmark yielded a better rate of diagnosis. One may speculate whether documentation that the major papilla has been visualized can serve as an auditable performance measure for completeness of the procedure in a patient referred for a complete UGI endoscopy. Analogous to cecal intubation, it cannot be achieved in all endoscopies and is less trivial than reaching the second portion of the duodenum. In the absence however of any data to support this, no consensus was reached on this statement (only 73% agreement) and the working group therefore formulated this as one of the research priorities (Table 2 ). We did however find one study that aimed to evaluate whether the length of time spent on UGI endoscopy improved the diagnostic yield. This was a retrospective cohort study by Teh et al.9 that aimed to determine the diagnostic yield for early neoplastic lesions in the stomach. The study included 837 symptomatic patients with no history of gastric cancer who underwent a first diagnostic endoscopy by one of 16 endoscopists. The mean examination time for the 224 examinations without any abnormal findings or biopsies taken was 6.6 min, which allowed the definition of a cut-off time of ≥7 min to distinguish between "slow" versus "fast" procedures. Afterwards, in a retrospective evaluation of the 837 endoscopies, they concluded that a "slow" endoscopist (who took on average at least seven minutes to perform a normal endoscopy) was twice as likely to detect high risk gastric lesions, defined as biopsy evidence of intestinal metaplasia, gastric atrophy, gastric dysplasia, or cancer (odds ratio (OR) 2.50, 95% CI 1.52–4.12) and three times as likely to detect a case of dysplasia or cancer (OR 3.42, 95% CI 1.25–10.38) than a "fast" endoscopist (who took fewer than seven minutes on average). A similar concept of measuring length of time for inspection, but in the specific context of Barrett's esophagus, has shown increased detection of dysplasia with an inspection time of one minute per cm of Barrett's esophagus.10 Only one study has evaluated the correlation between increased detection of gastric dysplasia or gastric cancer and other UGI endoscopic diagnoses.11 Park et al. retrospectively analyzed 54,889 records of patients who underwent a screening UGI endoscopy, performed by 66 experienced endoscopists, from 2006–2013 in a single center in Korea. Any diagnoses of reflux esophagitis, Barrett's esophagus, atrophic gastritis, intestinal metaplasia, erosion, ulceration, polyps, subepithelial lesions, xanthoma, angiodysplasia, or a diverticulum were recorded and the relevant records were re-evaluated with respect to increased detection of early gastric neoplasia. In multivariate analysis, the detection rates of gastric subepithelial lesions and gastric diverticula were independently associated with the detection rate of early gastric neoplasms. No data exist to support the view that photodocumentation of all normal anatomical landmarks and abnormal findings will improve diagnostic yield. However, photodocumentation should be considered a general quality improvement in comparison with previous reports that were made before dedicated reporting software was available. Most endoscopic systems now enable digital picture acquisition, therefore the working group strongly agreed on the inclusion of digital photography in reports. This measure is supported by endoscopic societies and experts suggest it might be an indirect quality indicator for careful inspection of the digestive lumen.12-14 The minimum number of pictures to be collected, combining relevance and applicability, in a normal endoscopic examination should be 10, namely: proximal esophagus, distal esophagus, Z line and diaphragm indentation, cardia and fundus in inversion, corpus in forward view including lesser curvature, corpus in retroflex view including greater curvature, angulus in partial inversion, antrum, duodenal bulb, and second part of duodenum. The working group suggested that it may be desirable to document more extensively in specific surveillance examinations, such as for Barrett's esophagus (e.g. one picture per cm of Barrett's esophagus),10 or where there are extensive gastric premalignant conditions (e.g. 21 pictures of the stomach).13 In addition, several validated classifications have been developed for specific pathologies. The working group agreed that the use of these classifications in conjunction with photodocumentation improves comparability and accurate information exchange among gastroenterologists, both in the clinical setting and for investigational purposes. In the UGI tract, this is especially true for the Los Angeles classification for the reporting of reflux esophagitis and the Prague classification for Barrett's esophagus.15-18 Implementation of this performance measure is inevitably dependent on the availability of image acquisition and software to incorporate images into the report. Because this performance measure simplifies and improves communication between different endoscopists, implementation of appropriate software should be prioritized by hospital policy makers. The working group recognizes that gastroenterologists performing procedures mainly in their own surgeries will often struggle to find a reasonable way to be reimbursed for the considerable cost of this software. The evidence for this performance measure is mainly derived from the study by Teh et al.8 By using a cut-off time of ≥7 min per endoscopy, from intubation to extubation, endoscopists performing above the cut-off (i.e. longer inspection times) detect two times as many high risk gastric lesions (intestinal metaplasia, gastric atrophy, gastric dysplasia, or cancer) and three times as many dysplastic lesions and gastric cancers. The study did not evaluate differing diagnostic yields in the esophagus or duodenum between endoscopists but it provides evidence for the stomach that is comparable to that for inspection times in the colon.18 The interval of three years in the statement from the Delphi process stems from the suggestion of the European consensus on "Management of precancerous conditions and lesions in the stomach (MAPS)" guideline.19, 20 The three-year interval was suggested among experts to be the best clinically applicable interval for endoscopic surveillance of extensive atrophy and/or extensive intestinal metaplasia. This three-year interval strategy has been shown more recently, in a European population between 50–75 years of age, to be cost-effective as a surveillance strategy.21 The quality of endoscopy is closely related to the quality of the report and the use of standardized terminology enables better communication between endoscopists and unequivocal discrimination of disease-related findings. The working group considered accurate reporting as one of the main topics for quality assurance. Besides reporting and documentation of anatomical landmarks, the correct use of available and validated terminology for specific diseases was considered to be a cardinal point for quality improvement. It enables the gathering of sound epidemiological data and is a prerequisite for auditing the quality of endoscopic reports. Although the literature searches for the PICOs did not render any evidence in terms of higher diagnostic yield or proven efficacy for better physician interaction and communication, the statements on implementing standardized terminology as a quality measure for accurate reporting all reached 100% agreement. The Paris classification is a uniform and relatively well established endoscopic classification for early neoplastic lesions,22, 23 with clinical in terms of the of the risk of and therefore for endoscopic however, is about the agreement of this the of this classification system has been in an study for assessment in the which a of and a mean agreement of This study that further research is to assess the of the Paris classification or to The Los Angeles classification for reflux disease was validated when it was in and that agreement for the assessment of mucosal areas of or and complications was Because of the availability of data and the that this classification is now used most the working group to implement the Los Angeles classification as the for endoscopic assessment of reflux the Prague classification is a relatively and which enables better communication between endoscopists. The score has been validated among and in two additional studies among and endoscopists, the of the Prague classification for the accurate of Barrett's esophagus and the length of the The on the diagnosis and management of UGI has strongly the uniform use of the as used in several studies the risk for and it is clinically that this classification is used in the endoscopy report in to the correct clinical management after endoscopy for UGI classification systems that should be are classification for the classification for grading of and classification for duodenal in Although there is less data available in terms of these systems are relatively to and have an clinical in terms of patient management and The working group accepted that although agreement was reached about the use of the standardized terminology in the modified Delphi process, its implementation may be not In in to data that will enable this performance measure to be there is a for an adequate reporting system that can a diagnosis (e.g. duodenal to the standardized terminology being used A prerequisite of such a reporting system is that it would to be at intervals and to be to individual endoscopists. from such reporting systems can to improve the performance for this for instance, the system can be adapted to a of the of the relevant classification system and that a report cannot be validated when a diagnosis has been the terminology is If such a system is in reports are and the of this by the endoscopist should be No studies were found that have directly addressed the comparison between measuring or not measuring the inspection One study to determine whether the inspection time in Barrett's esophagus with the detection of lesions and/or Barrett's or early was considered It was a analysis of data from a clinical of patients that the performance of for dysplasia detection during Barrett's esophagus The study found that greater of patients had an with inspection times min, min, min, ≥7 min, p < and a greater were found to have min, min, min, ≥7 min, p = The study suggested that an inspection time of 1 of Barrett's esophagus resulted in increased detection of neoplasia. Although this study had certain and did not reflect the of Barrett's esophagus dysplasia in a the working group members agreed to support this performance measure with a high of agreement. In contrast to the detection of colon polyps, where a basis seems to exist with to the measurement of inspection time during as a performance for UGI endoscopy there is a paucity of it seems reasonable to that a inspection of Barrett's esophagus may in better detection. The inspection time also the esophagus to improve proper and patient and the use of high definition endoscopy (i.e. high definition to high definition using a high definition this time, there are no data to support the use of any such as or but is there in when available. In the Barrett's and it was suggested that these should be used in experienced The implementation of this performance measure is again dependent on the availability and development of an reporting however, this is in it should be easy to One of the research priorities should be to whether there is a correlation between inspection time and increased neoplasia detection in Barrett's esophagus in a general studies addressed this clinical in patients with a history of and by endoscopy with Because this is a screening examination by a from a clinical point of view it to perform it in patients who have been previously with for their For the diagnosis of five of the studies in the rates of for early cancers, from of detected all eight studies increased yield for dysplasia from of The rates of lesions in this high risk group of patients were for dysplasia and for cancer after The in UGI endoscopy with of a by a or directly by the biopsy of the with the examination being two minutes In view of the that is a and relatively for which the available evidence is of the working group reached a high of agreement on the acceptance of this performance measure. The of biopsies of any followed by four biopsies taken 2 cm along the of the of the Barrett's all in different per level and per This is accepted in guidelines to be the for Barrett's esophagus The acceptance of this to several studies, which were mainly performed in an prior to that suggested better detection of neoplasia and a in In a retrospective cohort study including patients with cm Barrett's esophagus undergoing endoscopic patients received a biopsy and received a biopsy The detected more grade dysplasia vs p < and vs p = In the biopsy group, three patients of Barrett's esophagus compared with in the group. In with this et reported a cohort of patients for early Barrett's esophagus neoplasia and found that those without a prior diagnosis of dysplasia were more likely not to have the biopsy In the prior to an established endoscopic of early Barrett's esophagus neoplasia as to whether an better the presence of cancer in comparison to a less et the to four biopsies 1 cm for patients followed up after a diagnosis of and suggested that a 2 cm biopsy would of the cancers. In et found that a 2 cm interval for the biopsy was to detect cancer prior to using in experienced suggest that in the there may be a for to the but there are data to support Because of the acceptance of this in all the working group agreed the quality evidence, to the could serve as
Authors: Raf Bisschops, Miguel Areia, Emmanuel Coron, Daniela Dobru, Bernd Kaskas, Roman Kuvaev, Oliver Pech, Krish Ragunath, Bas Weusten, Pietro Familiari, Dirk Domagk, Roland Valori, Michal F Kaminski, Cristiano Spada, Michael Bretthauer, Cathy Bennett, Carlo Senore, Mário Dinis‐Ribeiro, Matthew D Rutter
Institutions: Universitair Ziekenhuis Leuven, Instituto Português de Oncologia de Coimbra Francisco Gentil, Universidade do Porto, Instituto Português de Oncologia Francisco Gentil, Centre for Health Technology and Services Research, Centre Hospitalier Universitaire de Nantes, Universitatea de Medicină, Farmacie, Științe și Tehnologie „George Emil Palade” din Târgu Mureș, University of Pittsburgh, Krankenhaus Barmherzige Brüder, Nottingham University Hospitals NHS Trust, NIHR Nottingham Digestive Diseases Biomedical Research Unit, Queen's Medical Centre, St. Antonius Ziekenhuis, Agostino Gemelli University Polyclinic, National Health Service, Gloucestershire Hospitals NHS Foundation Trust, Gloucestershire Royal Hospital, University of Oslo, The Maria Sklodowska-Curie National Research Institute of Oncology, Oslo University Hospital, Coventry University, Azienda Ospedaliera Citta' della Salute e della Scienza di Torino, Piedmont Reference Center for Epidemiology and Cancer Prevention, IPO Porto, Durham University, University Hospital of North Tees
Language: English
Subfield: Gastroenterology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2944570828
