A Randomized Trial Comparing Wireless Capsule Endoscopy With Push Enteroscopy for the Detection of Small-Bowel Lesions

Transcription

1 GASTROENTEROLOGY 2000;119: RAPID COMMUNICATIONS A Randomized Trial Comparing Wireless Capsule Endoscopy With Push Enteroscopy for the Detection of Small-Bowel Lesions MARK APPLEYARD,* ZVI FIREMAN, ARKADY GLUKHOVSKY, HAROLD JACOB, REUVEN SHREIVER, SRINATHAN KADIRKAMANATHAN,* ALEXANDRA LAVY, SCHLOMO LEWKOWICZ, EYTAN SCAPA, # RONA SHOFTI,** PAUL SWAIN,* and ASSAF ZARETSKY** *Department of Gastroenterology, Royal London Hospital, London, England; Given Imaging Ltd., Yoqneam, Israel; Gastroenterology Institute, Hillel Yoffe Medical Center, Hadera, Israel; Department of Diagnostic Radiology, Rambam Medical Center, Haifa, Israel; B nai Zion Medical Center, Haifa, Israel; # Instititue of Gastroenterology, Liver Diseases and Nutrition, Assaf Harofe Medical Center, Zerifin, Israel; and **Laboratory Animal Unit, Faculty of Medicine, Technion Institute of Technology, Haifa, Israel Background & Aims: Wireless capsule endoscopy is a new, painless method of imaging the entire small bowel. It has not been compared with push enteroscopy. We compared the sensitivity, specificity, and safety of capsule and push enteroscopy in detecting small-bowel lesions. Methods: Nine to 13 radiopaque, colored beads (3 6 mm diameter) were sewn in random order inside 9 canine small bowels, half within the first meter, and confirmed on x-ray. After recovery, the number, order, and color of beads were assessed in 23 capsule enteroscopies and 9 push enteroscopies in a random order. The surgeons, push enteroscopists, capsule video interpreters, and pathologist were blinded to the others findings. Results: The capsules identified more beads than push enteroscopy (median, 6 [range, 2 9] vs. 3 [range, 2 6 beads]; P < 0.001). The sensitivity of the capsule was 64% compared with 37% for push enteroscopy. The specificity was 92% for capsule enteroscopy and 97% for push enteroscopy. The capsules identified significantly more beads beyond the reach of the push enteroscope (median, 4 [range, 2 7] vs. 0; P < ). Hair, ingested plastic, ulceration, submucosal swelling, and worms were clearly identified by the capsule. The capsules passed safely through the animals with no significant histologic findings. Conclusions: Wireless capsule endoscopy detected more abnormalities in the small bowel than push enteroscopy. The small bowel is the most difficult part of the bowel to examine because of its remoteness from the mouth and anus. Conventional endoscopic techniques for examining the small bowel are also limited by the length of the small intestine ( m), 1 its free intraperitoneal location constrained by mesenteric attachments and multiple complex looped configuration. As a result, successful endoscopic examination of the small bowel is far less developed than in all other parts of the gastrointestinal (GI) tract. A barium small-bowel series, the most commonly used investigation, is not able to demonstrate flat lesions such as angiodysplasias, one of the most common pathologic lesions found, 2,3 and is not very sensitive for raised lesions. 4 The need for endoscopic examination of the small bowel is now well established, 5 particularly for obscure GI bleeding, 6 8 but also in the diagnosis, screening, and treatment of small-bowel tumors and polyposis syndromes 9 as well as inflammatory diseases of the small bowel. 10 Of patients presenting with obscure GI bleeding in the presence of normal upper GI endoscopy and colonoscopy findings, a small-bowel source can be identified in up to 40% of cases with push or sonde enteroscopy. In one series, in 26% of examinations, lesions were detected only by sonde enteroscopy, confirming distal small-bowel disease. 11 Push enteroscopy requires a skilled endoscopist; the procedure requires between 15 and 45 minutes in the endoscopy suite and is uncomfortable and often painful, usually requiring sedation and analgesia. In addition, the instrument can only examine between 80 and 120 cm beyond the ligament of Treitz, 12 and occasional complications may occur, usually relating to the use of an overtube to facilitate deep intubation of Abbreviations used in this paper: CMOS, complimentary metal oxide silicon; GI, gastrointestinal; IQR, interquartile range; LED, light-emitting diode by the American Gastroenterological Association /00/$10.00 doi: /gast

2 1432 APPLEYARD ET AL. GASTROENTEROLOGY Vol. 119, No. 6 the small bowel. Improved instrument design and therapeutic capability have led to an increasing use of push enteroscopy. Sonde endoscopy, in theory, has the potential to examine the entire small bowel. A long scope ( mm) is inserted transnasally into the stomach, pushed through the pylorus with a gastroscope passed through the mouth, 13 and carried by peristalsis of a balloon inflated at the tip. The procedure time, commonly 6 8 hours, is uncomfortable and often painful; does not allow tip deflection, biopsies, or marking of the bowel; but can sometimes allow inspection of the entire small bowel. Among patients referred for sonde examination, in up to 10% the endoscope will not be able to be passed; 10% will have complications, mainly epistaxis, but including perforation; and up to 75% of the time the distal ileum is not reached with only 50% 80% of the entire smallbowel mucosa visualized. 14 Sonde-type enteroscopy is currently rarely performed. Despite recent improvements in performance and wider availability of push enteroscopes, it remains difficult to decide when to resort to laparoscopy or operative enteroscopy in patients with difficult recurrent bleeding of uncertain origin. There is room for improvement in imaging of the small intestine. 15 The recent development of a wireless video-capsule system 16,17 propelled by peristalsis, 16 not requiring air insufflation, 18 has allowed for the first time painless imaging of the entire small bowel. The aim of this study was to compare the performance, sensitivity, specificity, and safety of capsule and push enteroscopy in detecting abnormalities in the small bowel in a specifically developed animal model. The study was also designed to assess the ability of the capsule endoscope to detect lesions beyond the reach of a push enteroscope. Materials and Methods Animals had beads sewn into the small bowel at laparotomy under anesthetic by the surgical team. The beads were inserted in a predetermined randomized order, and the number, site, and color of the bead were recorded. Half were inserted within a meter of the pylorus, within the potential range of the push enteroscope. A week later, push enteroscopy and 1 3 capsule enteroscopies were performed in a random order under sedation by the endoscopy team. Macroscopic damage was assessed at postmortem. The histologic appearances of the small intestine were interpreted by a pathologist. The images transmitted by the capsule were assessed by 2 experienced endoscopists. The different teams were blinded to each others findings. The protocol for this study was approved by the Haifa University Animal Study Review Board, Haifa, Israel. Power Calculation The main performance objective of the study was to prove the ability of the capsule to detect beads in areas not reached by the enteroscope. The sample size calculation could lead to an extremely small sample group because finding a single bead in a single animal in the distal small bowel would support the claim. However, to gain more information on the capsule performance, the number of beads per animal and number of animals were set to meet the following 2 requirements: (1) the probability of detecting at least 2 beads more than the push enteroscope in each animal was 90% and (2) the probability that at least 5 animals had at least 2 more beads detected than by push enteroscopy was 90%. The level of significance chosen to be tested was P Assuming the probability that the capsule can detect a single bead in a single animal is equal to 60%, then, to meet the first requirement, the minimal number of beads per animals would be 7. The required number of animals to guarantee requirement 2 (assuming requirement 1 is fulfilled) is 8. It was thus recommended to set the minimum sample size to 8 animals, each with at least 7 beads implanted. The numbers were also chosen to allow a calculation of the sensitivity and specificity of the capsule in comparison to push enteroscopy. The numbers were also set to be large enough to give useful information on safety, technical limitations of the prototype capsule, and the image reception. Randomization and Blinding of the Study Using sealed envelopes, randomization was applied to the order and color of the beads sewn into the small bowel. The order of push and capsule enteroscopy was similarly randomized. The surgeons, push enteroscopists, interpreters of the video-capsule images, and pathologist were blinded to each others findings. Capsule Videoendoscope The capsules (Given M2A video capsule; Given Imaging Ltd., Yoqneam, Israel) measured mm, had a battery life of approximately 6 hours, and were propelled by peristalsis. The capsules contained a complimentary metal oxide silicon (CMOS) chip camera, transmitter, light-emitting diode (LED) illumination, and silver oxide batteries (Figure 1). They transmitted continuous video images at 2 frames per second during their passage through the GI tract, and the images were recorded using a solid-state recorder and aerial system that was applied to the skin of the abdomen. The recorder was plugged into the workstation in which the images were processed and then viewed on a computer monitor using a specifically designed software package, which allowed the images to be watched at various speeds with pausing and reversing capability. Some unprocessed images were available in real-time. Push-Type Videoenteroscope A push-type videoenteroscope (240 cm in length, VS3 3430; Pentax, Tokyo, Japan) was used.

3 December 2000 COMPARISON OF CAPSULE AND PUSH ENTEROSCOPY 1433 Capsule Delivery System A suction device was designed to deliver the capsule into the duodenum using the enteroscope. It was back-loaded through the biopsy channel of the endoscope, and the capsule was ejected hydraulically by forcing water through the suction tubing once the endoscope and capsule were in the duodenum. The capsule within the suction device had to be positioned precisely at the pyloric orifice and advanced simultaneously with a wave of gastric peristalsis, which sometimes placed the capsule rapidly past the first 2 beads despite immediate release of the capsule (Figure 2). Animals Nine mongrel dogs, 4 male and 5 female, weighing kg were used in the endoscopic study. Each dog was uniquely identified by attaching to its neck-chain a small metallic disk with an identification number. This identification number was used to identify the animal on all study material. A control animal (dog 10) had an active capsule inserted into the stomach, which was later retrieved after it had passed through the GI tract. This animal underwent no surgery, anesthesia, or enteroscopic procedure. No images were recorded, and histology was reviewed according to the protocol outlined. Anesthetic and Preparation The animals were anesthetized with isoflurane (Forane; Abbot Laboratories, Maidenhead, England). During laparotomy, artificial small-bowel lesions were created by sewing between 9 and 13 radiopaque, colored beads of different shapes and sizes (3 6 mm) in a random order into the small bowel by veterinary surgeons. White, red, green, yellow, and black beads, some with markings, were used. The beads were discoid, oval, and multifaceted in shape. The beads were sewn in via small enterotomies made in the small bowel, secured with 2 separate full-thickness sutures, with approximately half of the beads sewn within 1mofthepylorus (assuming enteroscopy has the capability examining this length of small bowel in the human). The animals recovered for 7 days after surgery before enteroscopy. Enteroscopy Push enteroscopy and between 1 and 3 capsule enteroscopies were performed in a randomized order with ketamine hydrochloride (Ketalar; Parke Davis, Eastleigh, UK) and diazepam (Valium; Roche, Welwyn Garden City, UK) anesthetic. The push enteroscopies were performed, videorecorded, Figure 1. (A) A schematic diagram of the capsule endoscope. 1, optical dome; 2, short focal length lens; 3, white LEDs; 4, CMOS imager; 5, battery; 6, transmitter; 7, antenna. (B) A photograph of the capsule endoscope. Figure 2. X-ray of the (A) enteroscope and (B) delivery device inserting the capsule (C) into the duodenum (D) past the first bead. (E) Radiopaque markers indicating bead location.

4 1434 APPLEYARD ET AL. GASTROENTEROLOGY Vol. 119, No. 6 Table 1. Characteristics of Study Animals Animal Gender Weight (kg) implanted x-rayed 1 F F M F M M F M F F and interpreted initially in real-time by 2 experienced enteroscopists who reported on any implanted threads and beads or other lesions seen, with the aim of reaching as far into the small bowel as possible. The videos of the push enteroscopies were then subsequently assessed by 2 independent enteroscopists to ensure consistency with the interpretation of the capsule enteroscopies. The range of the enteroscope was defined by the most distal bead detected and by measurements on the enteroscope once the stomach loop had been reduced and the scope was starting to fall back after reaching its most distal point. The capsule enteroscopes were inserted into the duodenum using the insertion device. Images were captured using aerials applied to the anterior abdominal wall and downloaded for processing and subsequent interpretation. X-ray screening recorded the number of beads present after each examination. The capsules were retrieved for examination and testing after they had passed through the GI tract. Capsule Image Interpretation Two independent gastroenterologists with experience of enteroscopy viewed all of the capsule videos, reporting again on threads and beads as well as other lesions. Variation in bead coloration and shape helped the video viewers check that bead detections were not duplicated. The beads were included in the analysis only if identified by both reviewers. The enteroscopists and capsule video viewers were blinded to the findings of each other, the x-ray data, and the surgical data. Transit times through the small bowel were calculated by measuring the time taken for the capsule to travel from the first sighting of a bead in the small bowel to the last sighting. Development of the Model During feasibility studies, a porcine model was initially used, but gastric emptying was very slow. Despite attempts at endoscopic delivery of the capsule into the small bowel with enteroscopes or after formation of a gastrostomy or partial gastrectomy, small-bowel motility was severely impaired under anesthetic and the capsules did not move. Smallbowel motility did not seem to be affected as much by anesthetic agents in the dog; however, the hypertonic pylorus in the dog delayed the capsule gastric emptying time to beyond 10 hours. Because the capsules had a running time of 6 hours in early studies, the capsule only transmitted images from the esophagus and stomach. To overcome this problem, the capsules were delivered into the small bowel endoscopically with the aid of a specifically designed suction device described above. Histologic Assessment At postmortem immediately after the last enteroscopic examination, 5 6-cm-long segments of the small bowel were obtained for histologic analysis from various locations along the small bowel. The bowel was inspected immediately for macroscopic signs of mucosal damage. The first 2 3 bowel segments were selected from the proximal small bowel within reach of the enteroscope, and another 3 4 segments were taken distally where only the capsule had passed. The specimens were fixed in formalin and sent to an independent veterinary pathology laboratory for histologic processing (PathoVet Laboratory, Kfar Bilu, Israel). One transverse slice section, approximately 4 5 mm thick, was taken from the center of each submitted tissue specimen. These slices were separately placed in previously marked cassettes for routine paraffin embedding, followed by microtome sectioning performed the next day ( hours later). The microtome sections were 4 m thick and were placed on glass slides, routinely stained with H&E, and examined by a single independent pathologist again blinded to any of the previous findings. Definitions Sensitivity was defined as the number of beads detected, divided by the number of beads present on x-ray, multiplied by 100. Specificity was defined as the agreement on the presence of each bead, divided by the total number of beads seen by either of the 2 endoscopists, multiplied by 100. Statistical Methods 2 test, with Yates correction when appropriate, and Mann Whitney U test were used to assess the results. A P value of 0.05 was considered statistically significant. Results Characteristics of Study Animals Ten dogs were used with a male-to-female ratio of 4:6. Their weights ranged from 24 to 30 kg. The number of beads implanted was 9 13 (total, 99), but this was Figure 3. Capsule endoscope images of (A) normal small-bowel mucosa, (B) a multifaceted green bead, (C) a round white bead with black markings, (D) a submucosal polypoidal bulge in the jejunum, (E) plastic in the cecum, (F )anascaris worm with its GI tract visible, (G) black hair, and (H ) 2 small superficial ulcers in the jejunum.

5 December 2000 COMPARISON OF CAPSULE AND PUSH ENTEROSCOPY 1435 reduced to 9 12 (total, 89) after 7 days when the animals were x-rayed (Table 1) after the initial enteroscopy. This was presumably because some of the sutures holding the beads had cut through, and the beads had been passed. In dog 1, the sutures holding the beads onto the smallbowel mucosa were not full thickness such that the x-ray confirmed only 9 of the 13 inserted beads after the first capsule enteroscopy. After push enteroscopy, at which some further beads were detached by the procedure, only 4 beads remained at x-ray. As a result, further capsule enteroscopies were not performed. Subsequently, the beads were sewn in using full-thickness sutures and separated double ties. Dog 4 only had 4 beads remaining of 9 inserted after the second capsule enteroscopy, so further capsules were not inserted. The third capsule examination in dog 7 was prevented by the absence of peristalsis and by large amounts of food content obscuring the view. The only intervention in dog 10 was the delivery of an active capsule into the stomach, which was later retrieved intact. Capsule and Push Enteroscopy Findings The capsules transmitted clear continuous video images for up to 6 hours (Figure 3A C). All capsules

6 1436 APPLEYARD ET AL. GASTROENTEROLOGY Vol. 119, No. 6 Table 2. Number of Detected by Push and Capsule Enteroscopies Animal Push enteroscopy Insertion depth (cm) detected Capsule 1 Capsule endoscopy, beads detected Capsule 2 Capsule (4) a (control) a During the second capsule enteroscopy in dog 4, only 4 beads were present on x-ray. were retrieved intact, and careful examination showed that they were undamaged, the dome window remained clear, and there had been no leakage from the encapsulated batteries. The median number of beads detected at push enteroscopy was 3 with a range of 2 6 (interquartile range [IQR], 3 4) with small-bowel insertion depths of cm beyond the pylorus. Significantly more beads were detected by both video interpreters during capsule enteroscopy with a median of 6 (range, 2 9 [IQR, 6 7]; P 0.001; Table 2). During the 9 push enteroscopic examinations, 33 of 89 implanted beads were identified. The capsule identified 143 from the possible 225 beads during the 23 video-capsule examinations (P , 2 test with Yates correction; Table 3). The overall sensitivity of the push enteroscope in detecting implanted beads within the entire small bowel was 37% compared with 64% for the capsule. The first 1 2 beads in the proximal duodenum were frequently missed in part because of the capsules being delivered beyond them, but predominantly because of the illumination from the enteroscope flooding the capsule imaging system producing a whitened uninterpretable image. By comparing push and capsule enteroscopy videos, it was calculated that on average, 1.5 beads were missed per examination. If these had been seen, the sensitivity of the capsule would increase to 79%. The range of the push enteroscope as determined by the most distal bead seen, taking into account the known order and color of beads, allowed assessment of the sensitivity of both types of enteroscopy to be calculated within this range. The push enteroscope had a sensitivity of 94% in identifying the beads within its range compared with an overall sensitivity of 53% for the capsule within the same range (Table 4). The sensitivity for the capsule in this range is probably underestimated, given the likelihood of the first 1 or 2 duodenal beads being missed and the loss of some beads from dog 4 during the second capsule enteroscopy. However, the capsule identified 96 beads beyond the reach of the push enteroscope, with a median of 4 beads per examination (range, 2 7 [IQR, 3 5]; P ). The specificity for the identification of beads using the push enteroscope, as judged by the agreement between 2 video interpreters against the total number of beads reported by either of the video interpreters, was 33 of 34 (97%). Food and hairs adherent to the beads sometimes made capsule identification difficult even with push enteroscopy, in which it was of course possible to manipulate the endoscope and wash the area of interest to improve identification. The specificity of the capsule was 92% (143/156). The yellow and white beads were more difficult to spot, especially when partly covered with adherent hair or food. On average, a 4-hour video could be viewed using the specifically designed software in roughly 20 minutes. Unexpected pathology was diagnosed using the capsule endoscope, supporting the expectation that this form of endoscopy might show pathology in clinical trials in humans. Several Ascaris worms were seen as well as 2 acute ulcers, a submucosal polypoidal bulge in the small bowel, and some ingested plastic tubing in the cecum. Many of these abnormalities were identified in areas beyond the reach of the push enteroscope. Just under half of the worms as well as half of the hairs were identified independently by the push enteroscopists as well as the video-capsule interpreters. The resolution of the capsule images allowed easy identification of villi, threads, hairs, and their color and the GI tract within an Ascaris worm (Figure 3D H ). Good views of the cecum were obtained even though no attempt was made to clear the colon. The results as presented include only beads seen by both viewers of the capsule videos. However, there was generally good agreement between the video-capsule interpreters on the presence of beads or surgical thread in isolation. The overall agreement for beads was 92.2% Table 3. Total Numbers of Seen Compared With Total Implanted detected missed Total no. of beads Sensitivity (%) Push enteroscopy Capsule enteroscopy

7 December 2000 COMPARISON OF CAPSULE AND PUSH ENTEROSCOPY 1437 Table 4. Number of Detected by Push Enteroscopy and by the Capsule Within the Push Enteroscope Range Animal in push enteroscope range Push enteroscope detected/sensitivity of the capsule enteroscope Sensitivity (5) a 4 100% 2/50% / / % 3/100% 3/100% 1/33.3% % 2/66.7% 3/100% 2/66.7% % 1/33% 0/0% / % 2/50% 2/50% 2/50% % 2/66.7% 1/33.3% 1/33.3% % 2/33% 3/50% / % 5/83.3% 5/83.3% 3/50% % 0/0% 1/33.3% 1/33.3% a It is likely that 1 of the 5 beads sewn within the push enteroscope range in dog 1 was no longer present because, according to x-ray, 4 were lost from the original 13 beads sewn in. and 80.2% for thread. In addition, the other pathologic lesions seen, including 2 acute ulcers, a submucosal polypoidal bulge, hairs, and the intestine inside Ascaris worms, were identified by both video interpreters. The median small-bowel transit time for the capsule was 28 minutes (range, minutes). Macroscopically, there was no evidence of mucosal damage. Between 6 and 8 sections of small bowel were examined by an independent histopathologist for any sign of mucosal damage. A small number of minor changes were noted in the 70 sections examined. Occasional or focal epithelial attenuation was reported in 4 sections, 3 of which were in the push enteroscope range. Mild mucosal edema was reported in 2 sections, again within the range of the push enteroscope. The histopathologist concluded that the sections examined appeared to be those of normal dogs, with neither overt inflammation nor damage to the mucosa. The very few areas of inflammation that were observed within the serosa were interpreted as the result of surgical handling. Discussion The wireless capsule endoscope was repeatedly able to identify beads sewn beyond the reach of the push enteroscope and also identified unexpected pathology. All capsules were retrieved intact with no sign of damage. The capsule endoscope successfully transmitted clear, color video images of the dog stomach, small bowel, and some of the colon for up to 6 hours. The transparent dome window overlying the videocamera lens remained clear throughout the procedure, with any debris collecting on the dome being wiped off by peristalsis as the dome sweeps past the bowel wall. The images of the small bowel obtained were of good quality with villi easily seen. The capsule was also able to transmit clear images of objects such as hair, with a diameter of 300 m, and tell their color. Overall, the video images were considered by the experienced endoscopists to be slightly inferior to the best new video push enteroscopes, but better than those seen with sonde-type enteroscopes. This study has demonstrated that endoscopy of the small bowel is possible without air insufflation. The capsule outer casing is water-tight and designed to be bite resistant. As a result, the batteries contained within the capsule are never in contact with the bowel. The batteries used were of silver oxide type. The toxicity of accidental ingestion of such batteries has been reported in a large pediatric series of 2382 cases. 19 No adverse effects were noted unless the battery became impacted in the esophagus. Lithium- or mercury-containing batteries were more dangerous. The capsules caused no demonstrable damage macroscopically or microscopically to the small bowel. The small number of minor histologic changes that were seen tended to be in the push enteroscope range rather than parts of the small bowel only visited by the capsule. The capsule endoscopic images are new, and the oscillating nature of the video as the capsule is propelled forward and sometimes backward by peristalsis can be difficult to interpret initially; therefore, although the general agreement between the video viewers was good, the interpretations were not identical. We believe that endoscopists used to seeing similar video images would have little difficulty in rapidly getting used to this new endoscopic video format. More than half of the dogs studied were long haired, and clumps of ingested hair caught on beads and threads. This occasionally made definite bead or thread identification difficult, which would not be an issue in the human small bowel. The animals were anesthetized during the capsule examinations, which could have affected small-bowel motility, which in turn may have an effect on the performance of

8 1438 APPLEYARD ET AL. GASTROENTEROLOGY Vol. 119, No. 6 the capsule. However, the median dog small-bowel transit time of 28 minutes suggests reasonably well-preserved small-bowel motility. The overall sensitivity for picking up lesions randomly sewn into the full length of the small bowel was 64% compared with 37% for push enteroscopy. The sensitivity of the capsule within the range of the push enteroscope was 53% vs. 94% for push enteroscopy, confirming the high sensitivity of push enteroscopy. Most of the beads missed by the capsule in the push enteroscope range were the first 1 or 2 in the proximal duodenum where images were not captured as a result of being flooded with light from the combined illumination of the capsule and enteroscope and some capsules being delivered distal to these beads. The sensitivity of the capsule for this range would have improved significantly, had these proximal beads been identified. Human trials comparing push enteroscopy and capsule enteroscopy are needed to assess the true sensitivity of a swallowed capsule in diagnosing small-bowel lesions in the human small bowel. This study has provided the first images of pathology using a new type of endoscopy. It has shown the ability of the system to identify lesions in parts of the small bowel that could not be reached by push enteroscopy. The prospect of painless endoscopy is attractive. We have tested a new form of wireless endoscopy and compared it with push enteroscopy. Video capsule enteroscopy performed well when compared with push enteroscopy. This new form of endoscopy may have a role in the investigation of many small-bowel conditions. References 1. Underhill BM. Intestinal length in man. BMJ 1955;2: Dodda G, Trotman BW. Gastrointestinal angiodysplasia. J Assoc Acad Minor Phys 1997;8: Foutch PG. Angiodysplasia of the gastrointestinal tract. Am J Gastroenterol 1993;88: Nolan DJ, Traill ZC. The current role of the barium examination of the small intestine. Clin Radiol 1997;52: Swain CP. The role of enteroscopy in clinical practice. Gastrointest Endosc Clin North Am 1999;9: Lewis BS, Waye JD. Chronic gastrointestinal bleeding of obscure origin: role of small bowel enteroscopy. Gastroenterology 1988; 94: Morris AJ, Wasson LA, MacKenzie JF. Small bowel enteroscopy in undiagnosed gastrointestinal blood loss. Gut 1992;33: Foutch PG, Sawyer R, Sanowski RA. Push-enteroscopy for diagnosis of patients with gastrointestinal bleeding of obscure origin. Gastrointest Endosc 1990;36: Rossini FP, Risio M, Pennazio M. Small bowel tumors and polyposis syndromes. Gastrointest Endosc Clin North Am 1999;9: Gay GJ, Delmotte JS. Enteroscopy in small intestinal inflammatory diseases. Gastrointest Endosc Clin North Am 1999;9: Berner JS, Mauer K, Lewis BS. Push and sonde enteroscopy for the diagnosis of obscure gastrointestinal bleeding. Am J Gastroenterol 1994;89: Lewis BS. The history of enteroscopy. Gastrointest Endosc Clin North Am 1999;9: Lewis BS, Waye JD. Total small bowel enteroscopy. Gastrointest Endosc 1987;33: Seensalu R. The sonde exam. Gastrointest Endosc Clin North Am 1999;9: Mosse CA, Swain CP. Technical advances and experimental devices for enteroscopy. Gastrointest Endosc Clin North Am 1999; 9: Iddan G, Meron G, Glukhovsky A, Swain P. Wireless capsule endoscopy. Nature 2000;405: Gong F, Swain P, Mills T. Wireless endoscopy. Gastrointest Endosc 2000;51: Appleyard MN, Gong F, Mills T, Mosse CA, Swain CP. Endoscopy without air insufflation (abstr). Gastrointest Endosc 2000;51: AB Litovitz T, Schmitz BF. Ingestion of cylindrical and button batteries: an analysis of 2382 cases. Pediatrics 1992;89: Received September 13, Accepted October 12, Address requests for reprints to: Paul Swain, M.D., Department of Gastroenterology, Royal London Hospital, Whitechapel Road, Whitechapel, London E11BB, England. Fax: (44)