内镜黏膜下剥离术培训体系建设分析

目的研究如何挑选合适的内镜医师接受内镜黏膜下剥离术(ESD)培训。方法纳入2017年1月至2018年6月在南京鼓楼医院消化内镜中心进行ESD培训的3期共41名学员,收集相关资料,包括姓名,年龄,性别,培训前独立操作胃镜例数,独立操作肠镜例数,肠镜操作手法,是否使用窄带光成像(NBI)、放大内镜(ME),是否单独开展内镜黏膜切除术(EMR)、ESD、内镜超声检查(EUS)、经内镜逆行胰胆管造影术(ERCP)等指标。安排每位学员完成4次离体猪食管模型ESD,记录剥离标本面积、操作时间,计算平均操作速度评价学员学习指标,并记录术中穿孔发生情况。对影响ESD学员操作速度的相关因素进行线性回归分析。结果41名学员中男26名、女15名,年龄(36.07±4.44)岁,离体动物实验切除标本大小(4.67±1,61)cm2,操作时间(24.54±5.97)min,操作速度(0.19±0.05)cm2/mino单因素线性回归分析提示,学员培训前独立操作胃镜例数(n>5000例,P=0.0018),独立单人操作肠镜例数(n>3000例,P=0.0001),上消化道早癌发现例数(n>30例,P=0.0003),下消化道早癌发现例数(n>10例,P=0.0197)及培训前独立开展ME(P=0.0478)、EMR(P=0.0196)、ESD(P=0.0003)情况与学员离体猪食管ESD操作速度密切相关;但只开展NBI(P=0.5329)、ERCP(P=0.5007)、EUS(P=0.7668)与学员操作速度无明显相关性。ESD学员操作速度与穿孔率呈负相关(P<0.0001)。多元线性回归分析提示培训前独立开展EMR(P=0.029)、ESD(P=0.034)与ESD学员操作速度密切相关。结论ESD培训学员培训前独立胃镜操作例数>5000例,独立单人肠镜操作例数>3000例,上消化道早癌发现例数>30例,下消化道早癌发现例数>10例,以及培训前独立开展ME、EMR、ESD,对提高ESD培训效果有明显帮助。本研究为我们今后挑选合适的内镜学员进行ESD培训提供了理论依据。     

进阶式教学模式在超声内镜培训中的初步应用

目的 超声内镜已经成为诊治胆胰系统疾病及消化道疾病不可或缺的手段之一,但这项操作技术仍具有相当的难度。目前,我国超声内镜专家数量远远低于医疗需求,有关培训课程匮乏,培训形式不同,相关研究更为缺乏。针对超声内镜学习规律,结合进阶式教学的理念,我们提出了超声内镜进阶式培训模式,旨在研究该方法与传统长程培训方法相比,学员的超声内镜能力是否提升更为明显。方法 我们将超声内镜学习分为3个阶段,6个层次,分层次阶段学习。纳入无超声内镜基础但熟练掌握胃肠镜的10名学员,总的学习时间为1年,随机分两组,一组接受传统的长期培训,另一组接受进阶式培训模式。对比两组学员对超声内镜知识及技术的掌握程度。结果 进阶式教学模式在最终考核中超声内镜能力明显大于传统教学组,但因两组人数均较少,还无法进行统计学之间的比较。此外,进阶式教学模式的学员离岗时间明显缩短,在国内各医院相对人手缺乏的情况下更具有优势。结论 进阶式教学模式教学优势显著,应大力推广。     

实时监控系统在胶囊内镜检查干预过程中的应用价值观察

目的探讨胶囊内镜实时监控系统在胶囊内镜检查干预过程中的应用价值及其对全小肠检查完成率的影响。方法对2007年3月-2008年8月问因疑似小肠疾病而行胶囊内镜检查的120例患者应用实时监控系统进行视频追踪,并对幽门通过时间延迟者使用胃镜及圈套器将胶囊内镜直接送入十二指肠以减少其在食管和胃内的滞留时间;与2002年5月-2007年2月问以传统检查方式接受胶囊内镜检查的329例患者进行胶囊内镜幽门通过时间、小肠转运时间以及全小肠检查完成率的比较。结果传统检查组与实时监控组比较,胶囊内镜幽门通过中位数时间（30min对34min,P=0．787）及小肠转运时间[（275±100）min对（289±96）min,P=0．256）]差异均无统计学意义。实时监控组中胶囊内镜全小肠检查完成率显著高于传统检查组（84．2％对73．6％,P=0．019）。结论实时监控系统可及时发现检查过程中的异常现象,并指导内镜医生准确地实施临床干预,在其监测下进行干预可提高全小肠检查完成率。     

胃镜下将胶囊内镜经幽门送入小肠两种方式的比较研究

[目的]探讨胃镜下单用圈套器与胃镜前端安置透明帽联合使用圈套器两种方式将胶囊内镜经幽门送入小肠的操作技巧以及对胶囊内镜检查的影响。[方法]回顾66例胃镜下送胶囊内镜通过幽门的操作资料,分为单用圈套器组(27例)与安置透明帽后使用圈套器组(39例),比较2组操作时间、操作过程中胶囊内镜从圈套器掉落的次数、全小肠检查完成率(CRE)、小肠通过时间(SBTT)的差异。[结果]单用圈套器组、透明帽联合圈套器组,平均操作时间分别为(29.67±8.92)min、(13.88±6.01)min,2组比较t=5.708,P0.05;胶囊内镜掉落次数分别为(5.42±1.83)次、(2.35±1.27)次,2组比较t=5.328,P0.05;CRE分别为77.8%、82.1%,2组比较χ~2=0.184,P0.05;SBTT分别为(232.00±79.08)min、(289.12±45.84)min,2组比较t=2.347,P0.05。[结论]胃镜前端安置透明帽后使用圈套器较单用圈套器能更快、更稳定地摄取胶囊内镜将其送入小肠,能争取更多的小肠检查时间,但不能提高全小肠检查完成率。     

借助胃镜完成小儿胶囊内镜检查

胶囊内镜作为一种新的诊断技术,无论对于成人还是儿童,已成为检查小肠疾病的一种有效方法。胶囊内镜大小为26 mm×11 mm,对于小儿,其相对较大的体积可能造成患儿吞咽困难。我们完成了2例患儿经胃镜辅助下实施的胶囊内镜检查,该2例均因年龄因素不能自主吞咽胶囊内镜。     

无痛胃镜肠镜与胶囊内镜检查同日进行的可行性探讨

[目的]探讨无痛胃镜肠镜与胶囊内镜检查同日进行的可行性。[方法]接受胶囊内镜检查的患者中,32例为同日接受无痛胃镜肠镜检查后、接着行胶囊内镜检查的患者(同日组),81例为先行电子胃肠镜检查后、择日只接受胶囊内镜检查的患者(择日组)。观察2组胶囊的运行时间,病变检测率,不良反应。[结果]胃内运行的平均时间:同日组为(30.75±35.97)min,择日组(31.81±32.13)min,2组比较差异无统计学意义(P0.05);小肠内运行的平均时间:同日组(294.97±122.55)min,择日组(267.87±117.82)min,2组比较差异无统计学意义(P0.05)。2组均未发生无不良反应,无一例发生胶囊滞留。[结论]胶囊内镜检查操作安全、无创伤,有较高的诊断价值,与无痛胃镜、肠镜同日执行是可行的,并可避免多次肠道准备带来的不适。     

促胃肠动力药在胶囊内镜检查中的作用研究

目的探讨胶囊内镜检查前口服促胃肠动力药对胶囊内镜检查中胃肠转运时间、检查完成率及诊断率的影响。方法将2004年9月至2005年9月间因疑似小肠疾病的患者60例纳入本研究，随机分为2组，每组30例。服药组胶囊内镜检查前口服促胃肠动力药莫沙比利10mg，对照组则不服药物。记录胶囊内镜检查中的胃肠转运时间，比较两组全小肠检查完成率及诊断阳性率。结果服药组平均胃排空时间为29min（2～133min），短于对照组的54min（2～275min）（P=0．035）；两组小肠通过时间（248min vs．281min）差异无统计学意义（P=0．3492）；服药组完成全小肠检查率（93．3％）高于对照组（66．7％）（P=0．021）；服药组胶囊内镜诊断率（73．3％）与对照组（50％）相比差异无统计学意义（P=0．11）。结论胶囊内镜检查前口服促胃肠动力药能缩短胃排空时间，提高全小肠检查完成率。     

消化内镜技师培训方法探讨

目的探讨消化内镜技师三级培训方法,以指导技师的临床教学工作。方法根据学员工作年资和基础,分初、中、高三种级别进行培训,各级别按理论培训、实践技能培训、考核三阶段实施。结果初级培训学员共54名,优秀率100％;中级培训学员共27名,优秀10名,良好15名,差2名;优秀率37％;高级培训学员共3名,优秀3名,优秀率100％。结论中级技师培训是消化内镜技师培训的重点和难点。     

胶囊内镜教育与培训研究进展

胶囊内镜(capsule endoscopy,CE)因具备无痛无创、安全有效、无严重并发症等特点,现已成为临床上消化道疾病重要的检查手段,弥补了传统消化内镜的不足。随着CE技术的不断发展,患者对于CE检查需求度的不断提升,按专业化标准培训CE专业人员已成为当务之急。本文就CE培训现状、CE学习曲线以及CE和磁控胶囊胃镜培训模式等四个方面进行综述,为探索CE培训教育体系和模式提供参考。     

新型过氧乙酸溶液消毒消化内镜的效果评价

目的 探讨新型过氧乙酸溶液应用于消化内镜消毒的临床价值.方法 选择消化内镜40条分成两组,每组胃镜和肠镜各10条,观察组接受新型过氧乙酸溶液消毒10 min,对照组接受常规2％戊二醛溶液消毒10 min,比较两组消毒合格率,评价新型过氧乙酸溶液消毒的有效性.有效性获得证实后,再选择消化内镜80条分成四组,每组胃镜和肠镜各10条,分别接受新型过氧乙酸溶液消毒2 min、3 min、4 min和5 min,比较消毒合格率,探讨新型过氧乙酸溶液消毒的高效性.结果 对照组胃镜消毒合格率为100％ (10/10),肠镜消毒合格率为90％(9/10),消毒后胃镜和肠镜均能检出菌落;观察组胃镜和肠镜消毒合格率均为100％,且消毒后胃镜和肠镜均不再检出菌落,明显优于对照组.新型过氧乙酸溶液消毒3 min组、4 min组和5 min组胃镜消毒合格率均为100％(10/10),明显高于2 min组的30％ (3/10) (P ＜0.05),但3 min组仍可检出菌落;2 min组肠镜消毒均不合格,4 min组和5 min组肠镜消毒合格率均为100％,明显高于3 min的80％ (8/10) (P ＜0.05),但4 min组仍可检出菌落.结论 新型过氧乙酸溶液可用于临床消化内镜的消毒,较2％戊二醛溶液更符合消化内镜高水平消毒的要求.     

Validation of computer simulation training for esophagogastroduodenoscopy: Pilot study

BACKGROUND: Little is known regarding the value of esophagogastroduodenoscopy (EGD) simulators in education. The purpose of the present paper was to validate the use of computer simulation in novice EGD training. METHODS: In phase 1, expert endoscopists evaluated various aspects of simulation fidelity as compared to live endoscopy. Additionally, computer-recorded performance metrics were assessed by comparing the recorded scores from users of three different experience levels. In phase 2, the transfer of simulation-acquired skills to the clinical setting was assessed in a two-group, randomized pilot study. The setting was a large gastroenterology (GI) Fellowship training program; in phase 1, 21 subjects (seven expert, intermediate and novice endoscopist), made up the three experience groups. In phase 2, eight novice GI fellows were involved in the two-group, randomized portion of the study examining the transfer of simulation skills to the clinical setting. During the initial validation phase, each of the 21 subjects completed two standardized EDG scenarios on a computer simulator and their performance scores were recorded for seven parameters. Following this, staff participants completed a questionnaire evaluating various aspects of the simulator's fidelity. Finally, four novice GI fellows were randomly assigned to receive 6 h of simulator-augmented training (SAT group) in EGD prior to beginning 1 month of patient-based EGD training. The remaining fellows experienced 1 month of patient-based training alone (PBT group). Results of the seven measured performance parameters were compared between three groups of varying experience using a Wilcoxon ranked sum test. The staffs' simulator fidelity survey used a 7-point Likert scale (1, very unrealistic; 4, neutral; 7, very realistic) for each of the parameters examined. During the second phase of this study, supervising staff rated both SAT and PBT fellows' patient-based performance daily. Scoring in each skill was completed using a 7-point Likert scale (1, strongly disagree; 4, neutral; 7, strongly agree). Median scores were compared between groups using the Wilcoxon ranked sum test. RESULTS: Staff evaluations of fidelity found that only two of the parameters examined (anatomy and scope maneuverability) had a significant degree of realism. The remaining areas were felt to be limited in their fidelity. Of the computer-recorded performance scores, only the novice group could be reliably identified from the other two experience groups. In the clinical application phase, the median Patient Discomfort ratings were superior in the PBT group (6; interquartile range [IQR], 5-6) as compared to the SAT group (5; IQR, 4-6; P = 0.015). PBT fellows' ratings were also superior in Sedation, Patient Discomfort, Independence and Competence during various phases of the evaluation. At no point were SAT fellows rated higher than the PBT group in any of the parameters examined. CONCLUSION: This EGD simulator has limitations to the degree of fidelity and can differentiate only novice endoscopists from other levels of experience. Finally, skills learned during EGD simulation training do not appear to translate well into patient-based endoscopy skills. These findings suggest against a key element of validity for the use of this computer simulator in novice EGD training.     

Training with video imaging improves the initial intubation success rates of paramedic trainees in an operating room setting

STUDY OBJECTIVE: Video imaging of intubation as seen by the laryngoscopist has not been a part of traditional instruction methods, and its potential impact on novice intubation success rates has not been evaluated. METHODS: We prospectively tracked the success rates of novice intubators in paramedic classes who were required to watch a 26-minute instructional videotape made with a direct laryngoscopy imaging system (video group). We compared the prospectively obtained intubation success rate of the video group against retrospectively collected data from prior classes of paramedic students (traditional group) in the same training program. All classes received the same didactic airway instruction, same mannequin practice time, same paramedic textbook, and were trained in the same operating room with the same teaching staff. RESULTS: The traditional group (n=113, total attempts 783) had a mean individual intubation success rate of 46.7% (95% confidence interval 42.2% to 51.3%). The video group (n=36, total attempts 102) had a mean individual intubation success rate of 88.1% (95% confidence interval 79.6% to 96.5%). The difference in mean intubation success rates between the 2 groups was 41.4% (95% confidence interval 31.1% to 50.7%, P <.0001). The 2 groups did not differ in respect to age, male sex, or level of education. CONCLUSION: An instructional videotape made with the direct laryngoscopy video system significantly improved the initial success rates of novice intubators in an operating room setting.     

Assessment of a bronchoscopy simulator.

The study objective was to validate a flexible bronchoscopy simulator by determining if it could differentiate between expert and novice bronchoscopists. A subsequent evaluation phase was then done to determine whether use of the simulator would improve the rate of bronchoscopy skill acquisition for new pulmonary fellows. A multicenter prospective cohort study was performed using a bronchoscopy simulator. Three cohorts were evaluated based on the number of bronchoscopies previously performed: "experts" (> 500, n = 9), "intermediates" (25 to 500, n = 8), and "novices" (none, n = 11). Each participant performed two simulated cases with performance measures being recorded by the simulator. Performance measures that distinguished between groups were then used to evaluate the learning curve for new fellows training on the simulator. A randomized-controlled trial was then conducted comparing the quality of bronchoscopy performance for new pulmonary fellows who were trained either with conventional methods or with the simulator. Expert bronchoscopists performed better on the simulator than intermediates who performed better than novices in terms of procedure time, percentage of segments visualized, time in red-out, and wall collisions. Training of new fellows demonstrated that after performing 20 bronchoscopic simulations, the skill level acquired with the simulator significantly improved in terms of speed, percentage of segments visualized, time in red-out, and collisions. Fellows trained on the simulator performed better than fellows trained using conventional methods during their first actual bronchoscopies as assessed by procedure time (815 versus 1,168 s, p = 0.001), a bronchoscopy nurse's subjective quality assessment score (7.7 +/- 0.3 versus 3.7 +/- 2.5, p = 0.05), and by a quantitative bronchoscopy quality score (percentage of segments correctly identified/procedure time, 0.119 +/- 0.015 versus 0.046 +/- 034, p = 0.03). In conclusion, the bronchoscopy simulator was able to accurately assess bronchoscopy experience level. Training new fellows on the bronchoscopy simulator leads to more rapid acquisition of bronchoscopy expertise compared with conventional training methods. This technology has the potential to facilitate bronchoscopy training and to improve objective evaluations of bronchoscopy skills.     

Establishing a true assessment of endoscopic competence in ERCP during training and beyond: a single-operator learning curve for deep biliary cannulation in patients with native papillary anatomy

BACKGROUND: Deep cannulation of the common bile duct (CBD) in patients with native papillary anatomy can be used as a marker of competence at ERCP. OBJECTIVE: The primary aim of this study was to analyze a single-operator learning curve for supervised ERCPs in patients with native papillary anatomy and to assess the development of endoscopic competence, defined as the ability to deeply cannulate the CBD in the setting > or =80% of the time. Posttraining outcomes were evaluated as proof of training. DESIGN: A retrospective review: 1097 ERCP procedures were analyzed, 697 were performed during ERCP training (July 2002-July 2003), 400 were performed after training as an independent operator, 499 and 303 procedures for training and posttraining periods, respectively, were performed with the intent of deep cannulation of CBD in patients with native papillary anatomy. Procedures were chronologically grouped into subsets. Success rates were plotted against time. SETTING: Single center. MAIN OUTCOME MEASUREMENTS: Rate of successful deep biliary cannulation. RESULTS: The successful cannulation rate increased from 43% at the beginning of training to > or =80% after 350 to 400 supervised procedures. The success rate continued to improve posttraining with an aggregated success rate of >96% for the next 300 procedures performed as an independent operator. LIMITATIONS: Single operator. CONCLUSIONS: Achievement of a satisfactory success rate for deep biliary cannulation in patients with native papillary anatomy should be tracked by ERCP trainers and trainees. The consistent achievement of > or =80% success at deep biliary cannulation in such patients should become a standard for ERCP training programs to produce skilled and competent therapeutic biliary endoscopists.     

The complementary Erlangen active simulator for interventional endoscopy training is superior to solely clinical education in endoscopic hemostasis--the French training project: a prospective trial

BACKGROUND: The Erlangen Active Simulator for Interventional Endoscopy (EASIE) using ex-vivo porcine organs was introduced in 1997. The present study should analyze whether repeated EASIE simulator training in endoscopic hemostasis led to superior performance compared with a traditionally educated group. The results were compared with a similar project in New York. METHODS: Thirty-five French GI fellows were enrolled. Baseline skills evaluation was performed in four disciplines (manual skills, injection/coagulation, clip application and variceal ligation) using the compactEASIE-simulator equipped with an upper gastrointestinal organ package for bleeding simulation. The same, translated evaluation forms (from the prior New York project) were used. Subsequently, fellows were randomized into group A (n=17, only clinical education) and group B (n=18, additional three simulator trainings). Group B was trained the next day and after 4 and 7 months by experts of the French Society of Gastrointestinal Endoscopy. Both groups performed routine and emergency endoscopies at their home hospitals during the study period. Both groups were re-evaluated blindly after 9 months. RESULTS: The learning curve for group B showed a significant improvement in all disciplines (P<0.004) whereas group A improved significantly in only two of four disciplines at blinded final evaluation (manual skills P=0.02, injection/coagulation P=0.013). The direct comparison of groups B and A at blinded final evaluation showed significantly superior ratings for group B in all disciplines (P<0.006) and significantly shorter performance times in two disciplines (P=0.016 each). The comparison with the similar 'New York project' revealed that preexisting differences in skills were adjusted by the training. CONCLUSION: Complementary trainings (three workshops in 7 months) in endoscopic hemostasis using the compactEASIE improved skills compared with a solely clinical education. The results of the 'New York project' were confirmed and benefits were independent from the medical educational system.     

Computer-based virtual reality colonoscopy simulation improves patient-based colonoscopy performance

BACKGROUND:

Colonoscopy simulators that enable one to perform computer-based virtual colonoscopy now exist. However, data regarding the effectiveness of this virtual training are limited.

OBJECTIVE:

To determine whether virtual reality simulator training translates into improved patient-based colonoscopy performance.

METHODS:

The present study was a prospective controlled trial involving 18 residents between postgraduate years 2 and 4 with no previous colonoscopy experience. These residents were assigned to receive 16 h of virtual reality simulator training or no training. Both groups were evaluated on their first five patient-based colonoscopies. The primary outcome was the number of proctor ‘assists’ required per colonoscopy. Secondary outcomes included insertion time, depth of insertion, cecal intubation rate, proctor- and nurse-rated competence, and patient-rated pain.

RESULTS:

The simulator group required significantly fewer proctor assists than the control group (1.94 versus 3.43; P≤0.001), inserted the colonoscope further unassisted (43 cm versus 24 cm; P=0.003) and there was a trend to intubate the cecum more often (26% versus 10%; P=0.06). The simulator group received higher ratings of competence from both the proctors (2.28 versus 1.88 of 5; P=0.02) and the endoscopy nurses (2.56 versus 2.05 of 5; P=0.001). There were no significant differences in proctor-, nurse- or patient-rated pain, or attention to discomfort.

CONCLUSIONS:

Computer-based colonoscopy simulation in the initial stages of training improved novice trainees’ patient-based colonoscopy performance.     

Impact of skill and experience of the endoscopist on the outcome of endoscopic sphincterotomy techniques.

BACKGROUND: Our aim was to assess the influence of the skill and experience of the endoscopist on the success and risk of endoscopic sphincterotomy techniques. METHODS: The outcome of all endoscopic sphincterotomies (n = 1335) carried out between 1988 and 1995 were retrospectively analyzed with respect to the endoscopist performing the procedure. Endoscopists were differentiated according to whether they had previous experience with endoscopic sphincterotomy techniques (n > 100) and the frequency of endoscopic sphincterotomy during the study period (>40, 26 to 40, 10 to 25, <10 per year). RESULTS: Indications for endoscopic sphincterotomy techniques and technical execution had only a minor influence on the results of endoscopic sphincterotomy and were comparable for the individual endoscopists. The overall success rate of endoscopic sphincterotomy was 94.4% and did not significantly differ among the endoscopists. The overall complication rate of endoscopic sphincterotomy was 7.3%. Endoscopists learning endoscopic sphincterotomy techniques with a case frequency of less than 10 procedures per year had a consistently high complication rate (10.5%). Those learning endoscopic sphincterotomy techniques with a case frequency of more than 25 procedures per year had an above-average complication rate for their first 40 endoscopic sphincterotomy procedures and a significant decrease in complication rate as the number of procedures increased. The complication rate for experienced endoscopists was 7.7%. There were distinct and, in one case, significant differences in complication rates between individual endoscopists (11.5% vs. 4.8%, p = 0.01). However, when corrected for multiple testing, there were no significant differences at the p < 0. 05 level. The endoscopic sphincterotomy frequency of the endoscopist was the only significant risk factor for complications. Endoscopists with a frequency of more than 40 procedures per year had a significantly lower complication rate (5.6%) than endoscopists with a lower case frequency (9.3%, p < 0.05). CONCLUSIONS: A low endoscopic sphincterotomy frequency is, even for endoscopists with previous experience with the procedure, a risk factor for complications after endoscopic sphincterotomy. The learning of endoscopic sphincterotomy techniques requires a minimum of 40 procedures, but also after 100 procedures a further decrease of the complication rate can be expected.     

仿真结肠镜运用系统反馈独立训练与教师指导训练的比较研究

目的本研究拟开展仿真内镜不同训练模式对结肠镜技能建立的作用研究,以期发现最优化的训练模式。方法将学员分为2组,A组采用自学法,B组教师指导法。训练完成后根据成绩判断哪种训练方法最适合结肠镜受训学员。结果两组总体评分无显著性差异,各环节比较包括安全性、准确度、残气量、肠袢形成及操作时间方面均无显著性差异。结论本研究表明应推荐自学法为仿真结肠镜的训练方法。     

Interobserver and intraobserver variation in endoscopic assessment of GERD using the "Los Angeles" classification.

BACKGROUND: A new endoscopic classification of gastroesophageal reflux disease (GERD) has been proposed, and the term mucosal break has been introduced to describe mucosal damage. This new classification was evaluated by endoscopists with different levels of experience. METHODS: Fifty endoscopic photographs for each of 20 randomly selected patients with GERD were assessed retrospectively by three groups of seven endoscopists classified by experience: group 1 (100 to 500 procedures), group 2 (500 to 3000), and group 3 (more than 3000). The new classification was modified by adding grade 0 to describe healed mucosal breaks, so that there were five grades. All photographs were assessed twice at an interval of more than 1 week, and kappa statistics were used to determine observer variation. RESULTS: Interobserver variation within group 3 (kappa = 0.39, n = 21) and between groups 3 and 2 (kappa = 0.36, n = 49) was significantly different (p < 0.01) from that between groups 3 and 1 (kappa = 0.26, n = 49). Intraobserver variation in group 1 (kappa = 0.39, n = 7) was significantly different (p < 0.01) from that in group 2 (kappa = 0.51, n = 7) and group 3 (kappa = 0.54, n = 7). CONCLUSIONS: Observer variation depends on level of endoscopic experience. Only experienced endoscopists should use the new classification for grading of GERD.