Portal de Periódicos da CAPES

Emergency Physicians Accurately Interpret Video Capsule Endoscopy Findings in Suspected Upper Gastrointestinal Hemorrhage: A Video Survey

2013; Wiley; Volume: 20; Issue: 7 Linguagem: Inglês

10.1111/acem.12165

1553-2712

Andrew C. Meltzer, Carrie Pinchbeck, Sarah Burnett, Rasha Buhumaid, Payal Shah, Ru Ding, David E. Fleischer, Ian M. Gralnek,

Esophageal and GI Pathology

Acute upper gastrointestinal (GI) hemorrhage is a common emergency department (ED) presentation whose severity ranges from benign to life-threatening and the best tool to risk stratify the disease is an upper endoscopy, either by scope or by capsule, a procedure performed almost exclusively by gastroenterologists. Unfortunately, on-call gastroenterology specialists are often unavailable, and emergency physicians (EPs) currently lack an alternative method to endoscopically visualize a suspected acute upper GI hemorrhage. Recent reports have shown that video capsule endoscopy is well tolerated by ED patients and has similar sensitivity and specificity to endoscopy for upper GI hemorrhage. The study objective was to determine if EPs can detect upper GI bleeding on capsule endoscopy after a brief training session. A survey study was designed to demonstrate video examples of capsule endoscopy to EPs and determine if they could detect upper GI bleeding after a brief training session. All videos were generated from a prior ED-based study on patients with suspected acute upper GI hemorrhage. The training session consisted of less than 10 minutes of background information and capsule endoscopy video examples. EPs were recruited at the American College of Emergency Physicians Scientific Assembly in Denver, Colorado, from October 8, 2012, to October 10, 2012. Inclusion criteria included being an ED resident or attending physician and the exclusion criteria included any formal endoscopy training. The authors analyzed the agreement between the EPs and expert adjudicated capsule endoscopy readings for each capsule endoscopy video. For the outcome categories of blood (fresh or coffee grounds type) or no blood detected, the sensitivity and specificity were calculated. A total of 126 EPs were enrolled. Compared to expert gastroenterology-adjudicated interpretation, the sensitivity to detect blood was 0.94 (95% confidence interval [CI] = 0.91 to 0.96) and specificity was 0.87 (95% CI = 0.80 to 0.92). After brief training, EPs can accurately interpret video capsule endoscopy findings of presence of gross blood or no blood with high sensitivity and specificity. Interpretación Certera por los Urgenciólogos de los Hallazgos de la Videocápsula Endoscópica en la Sospecha de Hemorragia Digestiva Alta: Una Videoencuesta La hemorragia digestiva alta (HDA) es una patología frecuente en el servicio de urgencias (SU) cuya gravedad va desde la benignidad hasta el riesgo vital. La mejor herramienta para estratificar el riesgo de la enfermedad es una endoscopia oral, bien por endoscopio o por cápsula, un procedimiento realizado casi exclusivamente por médicos especialistas en digestivo. Desafortunadamente, los especialistas de digestivo de guardia no están a menudo disponibles, y los urgenciólogos en la actualidad no disponen de un método alternativo para visualizar endoscópicamente una sospecha de HDA aguda. Trabajos recientes han mostrado que la videocápsula endoscópica es bien tolerada por los pacientes del SU y tiene sensibilidad y especificidad similares a la endoscopia para la detección de la HDA. El objetivo de este estudio fue determinar si los urgenciólogos pueden detectar la HDA con una cápsula endoscópica tras una breve sesión formativa. Se diseñó una encuesta para mostrar ejemplos de cápsula endoscópica a los urgenciólogos y determinar si podrían detectar la HDA tras una breve sesión de formación. Todos los videos se generaron de un estudio previo en el SU con pacientes con sospecha de HDA aguda. La sesión formativa consistió en menos de 10 minutos de información sobre la historia clínica y ejemplos de videocápsulas endoscópicas. Los urgenciólogos se reclutaron en la reunión científica del American College of Emergency Physicians en Denver, Colorado, celebrada del 8/10/2012 al 10/10/2012. Los criterios de inclusión fueron ser residente o médico del SU, y el criterio de exclusión fue cualquier formación reglada en endoscopia. Se analizó la concordancia entre los urgenciólogos y el experto en cuanto a las lecturas adjudicadas de la cápsula endoscópica para cada videocápsula endoscópica. Se calculó la sensibilidad y la especificidad para las categorías de resultados de detección de sangrado (fresco o tipo de poso de café) o no sangrado. Se incluyeron 126 urgenciólogos. En comparación con la interpretación del experto en digestivo, la sensibilidad para detectar la sangre fue de un 94% (IC 95% = 0,91 a 0,96) y la especificidad fue de un 87% (IC 95% = 0,80 a 0,92). Tras una breve formación, los urgenciólogos pueden interpretar de forma certera en la videocápsula endoscópica la presencia o no de sangre de forma grosera con alta sensibilidad y especificidad. Acute upper gastrointestinal (GI) hemorrhage is a common presentation in the emergency department (ED) and may represent a potentially life-threatening condition.1 In 2007, GI hemorrhage from any source was the primary diagnosis for 570,000 visits to EDs and 292,000 patients were admitted to the hospital.2 The severity of upper GI bleeding ranges from mild and benign to life-threatening, and the best tool to risk stratify the disease is an upper endoscopy, a procedure performed almost exclusively by gastroenterologists. Unfortunately, on-call gastroenterologists may not be readily available to perform upper endoscopy,3 nasogastric tubes are uncomfortable for the patient,4 nasogastric tubes lack sensitivity and specificity for upper GI bleeding,5 and emergency physicians (EPs) lack an endoscopic method to accurately risk stratify acute upper GI hemorrhage. Without the ability to endoscopically risk stratify the severity of an upper GI bleeding event, approximately 85% of patients are admitted to the hospital for further evaluation, including upper endoscopy.6 There is a need for a new paradigm of risk stratification that can occur in the ED and without a gastroenterologist needing to be present at the bedside. The use of video capsule endoscopy (VCE) may represent such a new diagnostic paradigm for ED patients with acute upper GI hemorrhage.7-9 The potential advantages of VCE include the ability to be performed 24 hours a day, to be performed without conscious sedation,10 to be administered by any operator,11 to be well tolerated by patients,12 and to provide actionable results at the point of care.8 Finally, use of the VCE real-time viewer allows direct real-time visualization of capsule endoscopy findings at the patient bedside.13 Our goal was to determine if EPs with brief training in VCE can detect upper GI bleeding (fresh blood or coffee grounds versus no blood) on videos derived from patients with known upper GI bleeding. Emergency physicians were recruited at the American College of Emergency Physicians Scientific Assembly in Denver, Colorado, in the exhibitor hall from October 8, 2012, to October 10, 2012. Physicians were eligible at any level of training, including residents, fellows, or attendings who did not have prior formal endoscopy training. The study used a convenience sample of physicians who were attending the conference and whom we approached to participate. We estimate that greater than 70% of potential subjects with whom we personally interacted agreed to participate. Demographic data were collected at the time of subject enrollment. The EPs, as well as the study personnel conducting the study, were blinded as to the expert adjudication of each of the capsule videos. The subjects were compensated with a $10 gift card for their effort and time. George Washington University Institutional Review Board approved the study. This study was funded by Given Imaging, Yoqneam, Israel. The manufacturer made no claim to data or data analysis and did not review this manuscript prior to submission. The survey of EPs was designed on the basis of information from a prior study of four physicians (two EPs and two gastroenterologists), two interviews with individual physicians, and a review of the literature; the survey was pretested with the use of five trial surveys and adjusted for format, clarity, and length. The study was conducted and designed by investigators and authors listed. ACM, SB, and CP personally manned the booth to recruit subjects. The entire survey consisted of a 10-minute video that included a narrated introduction to the concept of capsule endoscopy, six multiple-choice questions designed to assess clinical need (see Table 1), five practice videos, and four test videos. Each physician was shown the same four videos in one of two randomly selected orders. Of the four test videos, two demonstrated fresh blood, one demonstrated coffee-ground blood, and one demonstrated no blood. Data were collected and managed using Sliderocket, a Web-based application designed to host video presentations and support data capture (http://www.sliderocket.com/). Videos were generated during a prior study that enrolled individuals presenting to the ED with hematemesis, coffee-ground emesis, and/or melena.9 Each video shown in the survey was scored prior to the study by three expert endoscopists blinded to each other's interpretation with 100% agreement. No patient identifiers were attached to the capsule endoscopy videos. The practice and test videos were each approximately 60 seconds in duration and were edited to demonstrate the oral pharynx, the esophagus, and the stomach; postpyloric images were not shown. The test videos were different from the practice videos. Each video was edited at the beginning to remove any identifiable facial features (the videos all began as the capsule entered the mouth) and then each video was clipped after the capsule was in the stomach for about 1 minute. Videos lasted about 1 minute total—a time we considered realistic for an EP to watch; anecdotally, the first 1 minute of the videos is usually sufficient to determine the presence of blood or bleeding in the esophagus or stomach. A prior study demonstrated that two untrained EPs who watched the full unedited videos demonstrated a 96% overall agreement with expert capsule endoscopists.9 There were two possible outcome categories for the videos: presence of fresh blood or coffee-ground blood or no blood present. Our analysis calculated the sensitivity and specificity of the EPs compared to expert endoscopist interpretation. We analyzed the agreement between the briefly trained EPs and the expert endoscopist ratings. For each outcome category, the sensitivity, specificity, and area under the received operating characteristic curve was calculated along with 95% confidence intervals (CIs). For calculation of the 95% CIs, SAS Proc GLMMIX (SAS Institute, Cary, NC) was used to adjust for EP as well as case/video random effects. Exploratory analyses examined and compared the above statistics for subgroups, such as the experience level of the EP and self-reported confidence level rating. Experience level was categorized as one of four categories: in residency or fellowship training, in practice less than 10 years, in practice between 11 and 20 years, and in practice greater than 20 years. Self-reported confidence level was rated using a 7-point Likert scale ("Rate from 7 to 1": 7 = extremely confident I could make a clinical assessment; 4 = moderately confident I could make a clinical assessment; 1 = not at all confident I could make a clinical assessment). Each of 126 EPs scored four capsule videos, for a total of 504 scores. We conducted a bivariate generalized linear fixed effects model to pool sensitivity and specificity of all physicians.8 We also examined the above statistics for subgroups by EP characteristics, including age, sex, board certification status, the experience level of the physician, and self-reported confidence level rating. In our study of 126 EPs, we demonstrated a sensitivity of 0.94 (95% CI = 0.91 to 0.96) and specificity of 87% (95% CI = 0.80 to 0.92) to detect blood in the esophagus or stomach after only a brief training session (Table 2). Study subjects were mostly younger physicians; 68% were less than 45 years old. In addition, 64% of study subjects were male, 72% were board-certified or board-eligible in emergency medicine, and 73% had already completed their residency or fellowship. Females and younger physicians demonstrated slightly higher but not statistically significant sensitivity and specificity to detect blood. Ninety percent of subjects reported a moderate or high level of confidence (greater than or equal to 4 of a possible 7) to interpret the capsule endoscopy videos. Very few physicians (6.4%) reported using clinical decision rules to make patient disposition decisions9 (Table 1). Approximately half of physicians (52.4%) used a nasogastric tube for evaluation, which was reported by the physician-subjects as having poor sensitivity and specificity (44.4%) and being painful for the patient (61.1%). In general, an increasing level of confidence corresponded to high sensitivity and specificity (Table 3). The investigation of acute upper GI hemorrhage is a diagnostic challenge for EPs because, with existing methodology, the ED physician is unable to directly endoscopically visualize the upper GI tract and estimate GI bleeding severity. Early endoscopy (defined as within 24 hours of patient presentation) is known to improve clinical outcomes of patients with suspected acute upper GI hemorrhage and allow safe discharge for patients with minimal disease.1, 14, 15 However, the use of early endoscopy is limited by the need for a gastroenterology specialist at the bedside, the need for procedural conscious sedation, and the invasive nature of the procedure. To our knowledge, the use of VCE in the ED for detection of acute upper GI hemorrhage has not been used outside of a research setting. The underlying premise of this research is that VCE, used within the ED, may improve our ability to diagnose, make safe disposition and treat patients with suspected acute upper GI hemorrhage compared to current paradigm of care. If ED physicians can detect bleeding with VCE after brief training, the current diagnostic paradigm for patients with suspected acute upper GI hemorrhage may change. Prior studies have shown that the presence of fresh or coffee-ground blood on capsule endoscopy had a sensitivity of 0.92 and a specificity of 0.78 compared to subsequent traditional endoscopy to detect a high-risk lesion.8, 9 For patients with high-risk lesions, early identification of acute upper GI hemorrhage has the potential to expedite hospital admission and subsequent clinical management. For patients with low-risk lesions, the use of VCE may reduce hospitalizations and costly inpatient endoscopy. We have not examined the ability of capsule endoscopy to quantify the amount of blood in the stomach. Stomach blood volume is currently not a marker of severity in established risk-stratification schemes, which use other measures to estimate volume loss such as laboratory and vital sign parameters. Although the up-front unit cost of capsule endoscopy is high, the potential for cost savings through its use may be significant in terms of prevention of hospital admission in selected low-risk patients, facilitation of a more appropriate inpatient admission site (intensive care unit [ICU] vs. non-ICU setting) for those patients requiring hospital admission, and more targeted use of expensive pharmacologic agents (intravenous proton pump inhibitors and splanchnic vasoactive agents). Capsule endoscopy should not replace traditional tube-based endoscopy, because it lacks the therapeutic capabilities of esophagogastroduodenoscopy (EGD) and video capsules cannot be directed or positioned. However, the detection of fresh or coffee-ground blood is an important endpoint because presumably all of these patients will need an inpatient EGD. Multiple prior studies have shown that both novice physicians and nonphysician extenders can interpret video capsule endoscopy with high sensitivity compared to a trained physician.11, 16 Capsule endoscopy is unique among endoscopic techniques because it never requires conscious sedation and the endoscopist does not need to be present at the time the procedure is performed. In addition, the risk of capsule retention, the worst potential complication, is estimated to be rare in this specific patient population and independent of the reader. We suspect that VCE will be able to match some of the diagnostic abilities of EGD to identify fresh blood/active bleeding. In addition, we suspect that VCE may allow more patients to be safely discharged from the ED without hospitalization. Prior studies have shown that if an EGD is performed in the ED, between 30 and 46% of ED patients with upper GI tract hemorrhage can be safely discharged.17-20 Previous studies have shown that the VCE is well tolerated in the ED and has similar test characteristics to traditional endoscopy.7-9 In general, EPs demonstrated high sensitivity and specificity to detect blood compared to expert gastroenterologist interpretation. Younger physicians, women physicians, and physicians who were more confident in their self-reported interpretative abilities performed slightly better. We suspect that the observed sex difference may have been due to a few male subjects who were color blind. A similar sex difference has been described in the ability of some male physicians to detect fecal occult blood, which has a blue detector color.21 One subject mentioned to us that he was color blind after completing the survey. Survey participants were attendants at the American College of Emergency Physician Scientific Assembly and may not be representative of EPs in general. The subject sample was collected from physicians who happened to be walking by the study booth. Moreover, the setting of the survey in a booth at a busy exhibitor hall may have made concentration difficult for some subjects; thus we may be underestimating the observed test characteristics in this study. A third limitation is due to our inability to exactly measure the response rate. We estimate that more than 70% of the physicians that we asked to participate agreed to participate, but some physicians walked up to the booth based on the booth's sign or from word of mouth. We are unable to calculate a definitive response rate of everyone who saw the sign or heard about the study from word of mouth. The final study limitation is the fact that the test videos shown may not represent all conceivable videos generated from video capsule endoscopy. After brief training, emergency physicians appear to be able to accurately interpret video capsule endoscopy for the endpoints of identifying gross blood or no blood with high sensitivity and specificity compared to expert gastroenterologist interpretation. The authors acknowledge Robert Shesser, MD.