- Email: [email protected]
Performance of given suspected blood indicator
THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2003 by Am. Coll. of Gastroenterology Published by Elsevier Inc.
Vol. 98, No. 12, 2003 ISSN 0002-9270/03/$30.00 doi:10.1016/j.amjgastroenterol.2003.07.006
Performance of Given Suspected Blood Indicator Suthat Liangpunsakul, M.D., Lori Mays, R.N., and Douglas K. Rex, M.D., F.A.C.G. Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana
OBJECTIVES: Given Imaging (Yoqneam, Israel) has developed a suspected blood indicator (SBI) function for its wireless capsule endoscopy (WCE) software. The SBI detects blood and marks appropriate images for interrogation. The sensitivity and accuracy of SBI and its role in examining WCE studies is unknown. Our aim was to evaluate the accuracy of this new software in detecting small bowel lesions. METHODS: WCE reports of all patients referred for WCE in October and November, 2002, were reviewed. The images from each patient were reviewed by experienced gastroenterologists at 15 frames/s. The findings detected by gastroenterologists were compared to those recognized by SBI. Arteriovenous malformations (AVMs), ulcers, erosions, and sites of active bleeding were considered significant lesions. RESULTS: A total of 24 patients (16 women and eight men, mean age 59 yr) were studied during this period. The indications for the study were iron deficiency anemia in 18 patients and abdominal pain in six patients. A total of 109 lesions were identified by gastroenterologists (47 AVMs, 18 active bleeding ulcers, 18 ulcers without bleeding, and 26 erosions). Active bleeding seen by gastroenterologists was secondary to AVMs (five cases), jejunal ulcers (11 cases, all in the same patient), and gastric ulcers (two cases). A total of 31 potential areas of blood were identified by SBI. When compared to those findings recognized by gastroenterologists, 28 lesions were correctly identified (six AVMs, 13 active bleeding ulcers, (three AVMs and 10 jejunal ulcers), seven nonbleeding ulcers, and two nonbleeding erosions. The overall sensitivity, positive predictive value, and accuracy of SBI to detect significant small bowel lesions were 25.7%, 90%, and 34.8%, respectively. If only actively bleeding lesions in the small bowel were considered, SBI had sensitivity, positive predictive value, and accuracy of 81.2%, 81.3%, and 83.3%, respectively. CONCLUSIONS: SBI has good sensitivity and positive predictive value for actively bleeding lesions in the small bowel. Complete review of the study by the physician is still needed. SBI should be considered as a complementary and rapid screening tool for gastroenterologists to identify actively bleeding lesions. (Am J Gastroenterol 2003;98: 2676⫺2678. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION Wireless capsule endoscopy (Given Imaging, Yoqneam, Israel) has advanced examination of the small intestine (1–3). Recently, a suspected blood indicator (SBI) function was developed as part of the Given software to provide easy screening for potential sites of active bleeding. The software recognizes color changes suggestive of blood and marks them for the individual reading. The accuracy of SBI and its role in examining WCE studies are unknown. The objective of this study was to evaluate the accuracy of this new software to detect small intestinal lesions.
MATERIALS AND METHODS All patients undergoing WCE during October and November, 2002, at our institution were reviewed for this study. Permission to conduct the review was granted by the Institutional Review Board at the Indiana University and Purdue University (Indianapolis, IN). All patients fasted overnight before swallowing the Given M2A video capsule (Given Imaging). Laxative preparation was not used. The images from each patient were reviewed by five experienced gastroenterologists at 15 frames/s. All detected abnormalities were highlighted in thumbnails for subsequent analysis. The findings detected by gastroenterologists were compared to those recognized by SBI. All thumbnail endoscopy images and reports were subsequently reviewed by two of the authors (S.L., D.R.) to confirm the findings that were read initially by other gastroenterologists. No discrepancies involving the interpretation of the presence or absence of Table 1. Significant Lesions Identified During Wireless Capsule Endoscopy in 24 Patients Lesion
Number (%)
AVM Active bleeding Gastric ulcers Jejunal ulcers* Jejunal AVMs Ulcers without bleeding Erosions Total
47 (43.1) 18 (16.5) 2 (1.8) 11 (10) 5 (4.7) 18 (16.5) 26 (23.9) 109 (100)
AVM ⫽ arteriovenous malformation. * All lesions were in the same patient.
AJG – December, 2003
Performance of Given Suspected Blood Indicator
Figure 1. Suspected blood indicator (SBI) function as part of the WCE software (left). The SBI detects blood and marks (in red line) appropriate images for interrogation. SBI correctly identified blood in the jejunum adjacent to an ulcer.
blood were encountered. Arteriovenous malformations (AVMs), ulcers, erosions, and sites of active bleeding were considered significant lesions.
RESULTS A total of 24 patients (16 women and eight men, mean age 59 yr) were studied during this period. The indications for the study were iron deficiency anemia in 18 patients and abdominal pain in six. Gastroenterologists identified 109 lesions (Table 1). Active bleeding seen by the gastroenterologists was secondary to AVMs (five lesions), jejunal ulcers (11 lesions, all in the same patient), and gastric ulcers (two lesions). A total of 31 potential bleeding areas were identified by SBI (Figs. 1 and 2). When compared to findings recognized by GE, 28 lesions were correctly identified (six AVMs, 13 active bleeding ulcers, seven nonbleeding ulcers, and two nonbleeding erosions). Of the 13 active bleeding lesions, three were AVMs and 10 were jejunal ulcers. The overall sensitivity, specificity, positive predictive value, and negative predictive value of SBI to detect all lesions were 25.7%, 85%, 90%, and 17.3%, respectively
2677
Figure 2. Another jejunal ulcer with active bleeding correctly identified by SBI.
(Table 2). If only actively bleeding small bowel lesions are considered, SBI had sensitivity and positive predictive value of 81.2%. There were three actively bleeding lesions that were not detected by SBI (two jejunal AVMs and one jejunal ulcer). For nonbleeding AVMs, the sensitivity and positive predictive value of SBI were 12.8% and 66.7%. The sensitivity of SBI to detect nonbleeding ulcers was poor (38.9%). The capabilities of SBI to detect significant small bowel lesions are summarized in Table 2.
DISCUSSION The SBI function developed by Given Imaging is designed to highlight actively bleeding areas for rapid review by the physician. We found that SBI has good sensitivity and positive predictive value for bleeding lesions. The sensitivity of SBI to detect nonbleeding AVMs as well as ulcers without bleeding (a function for which it was not designed) was poor. A total of 41 AVMs were identified by gastroenterologists but were missed by SBI. Our study suggests that the SBI function can be used for quick screen-
Table 2. Diagnostic Accuracy of SBI to Detect Significant Small Bowel Lesions Capability of SBI to Detect Significant GI Lesions (%) Lesions
Sensitivity (95% CI)
Specificity (95% CI)
PPV (95% CI)
NPV (95% CI)
Accuracy (95% CI)
Overall lesions AVMs Overall active bleeding Active small bowel bleeding Ulcers without bleeding Erosions
25.7 (17.5–33.9) 12.7 (3.2–22.3) 72.2 (51.5–92.9) 81.2 (62.1–100) 38.9 (16.3–61.4) 7.69 (2.55–17.9)
85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100)
90.3 (79.9–100) 66.7 (35.8–97.4) 81.3 (62.1–100) 81.3 (62.1–100) 70 (41.6–98.4) 40 (2.9–82.9)
14.2 (9–24.8) 29.3 (17.6–41.0) 77.2 (59.7–94.7) 85 (69.3–100) 60.7 (42.6–78.8) 41.5 (26.3–56.5)
34.8 (26.6–43.1) 34.3 (22.9–45.7) 78.9 (65.9–91.9) 83.3 (71.1–95.5) 63.2 (47.8–78.5) 41.3 (27.1–55.5)
NPV ⫽ negative predictive value; PPV ⫽ positive predictive value.
2678
Liangpunsakul et al.
ing of the small bowel for active bleeding. This could be useful if the interpreter does not have time when a study becomes available to perform a complete diagnostic reading. Alternatively, the SBI function might simply be used as adjunct to interpretation of the study. Clearly, the current version of SBI does not substitute for detailed review of the entire study. In summary, SBI has good sensitivity particularly in the small bowel for actively bleeding lesions. Complete review of the WCE study by the physician is still needed. SBI should be considered as a complementary and rapid screening tool for physicians to identify actively bleeding lesions.
AJG – Vol. 98, No. 12, 2003
Reprint requests and correspondence: Douglas K. Rex, M.D., F.A.C.G., Professor of Medicine, Division of Gastroenterology and Hepatology, Department of Medicine, 550 N. University Boulevard, UH 4100, Indianapolis, IN 46202. Received Mar. 14, 2003; accepted July 21, 2003.
REFERENCES 1. Iddan G, Meron G, Glukhovsky A, Swain P. Wireless capsule endoscopy. Nature 2000;40:725–9. 2. Gong F, Swain P, Mills T. Wireless endoscopy. Gastrointest Endosc 2000;51:725–9. 3. ASGE Technology Assessment Committee. Wireless capsule endoscopy. Gastrointest Endosc 2002;56:621–24.
Vol. 98, No. 12, 2003 ISSN 0002-9270/03/$30.00 doi:10.1016/j.amjgastroenterol.2003.07.006
Performance of Given Suspected Blood Indicator Suthat Liangpunsakul, M.D., Lori Mays, R.N., and Douglas K. Rex, M.D., F.A.C.G. Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana
OBJECTIVES: Given Imaging (Yoqneam, Israel) has developed a suspected blood indicator (SBI) function for its wireless capsule endoscopy (WCE) software. The SBI detects blood and marks appropriate images for interrogation. The sensitivity and accuracy of SBI and its role in examining WCE studies is unknown. Our aim was to evaluate the accuracy of this new software in detecting small bowel lesions. METHODS: WCE reports of all patients referred for WCE in October and November, 2002, were reviewed. The images from each patient were reviewed by experienced gastroenterologists at 15 frames/s. The findings detected by gastroenterologists were compared to those recognized by SBI. Arteriovenous malformations (AVMs), ulcers, erosions, and sites of active bleeding were considered significant lesions. RESULTS: A total of 24 patients (16 women and eight men, mean age 59 yr) were studied during this period. The indications for the study were iron deficiency anemia in 18 patients and abdominal pain in six patients. A total of 109 lesions were identified by gastroenterologists (47 AVMs, 18 active bleeding ulcers, 18 ulcers without bleeding, and 26 erosions). Active bleeding seen by gastroenterologists was secondary to AVMs (five cases), jejunal ulcers (11 cases, all in the same patient), and gastric ulcers (two cases). A total of 31 potential areas of blood were identified by SBI. When compared to those findings recognized by gastroenterologists, 28 lesions were correctly identified (six AVMs, 13 active bleeding ulcers, (three AVMs and 10 jejunal ulcers), seven nonbleeding ulcers, and two nonbleeding erosions. The overall sensitivity, positive predictive value, and accuracy of SBI to detect significant small bowel lesions were 25.7%, 90%, and 34.8%, respectively. If only actively bleeding lesions in the small bowel were considered, SBI had sensitivity, positive predictive value, and accuracy of 81.2%, 81.3%, and 83.3%, respectively. CONCLUSIONS: SBI has good sensitivity and positive predictive value for actively bleeding lesions in the small bowel. Complete review of the study by the physician is still needed. SBI should be considered as a complementary and rapid screening tool for gastroenterologists to identify actively bleeding lesions. (Am J Gastroenterol 2003;98: 2676⫺2678. © 2003 by Am. Coll. of Gastroenterology)
INTRODUCTION Wireless capsule endoscopy (Given Imaging, Yoqneam, Israel) has advanced examination of the small intestine (1–3). Recently, a suspected blood indicator (SBI) function was developed as part of the Given software to provide easy screening for potential sites of active bleeding. The software recognizes color changes suggestive of blood and marks them for the individual reading. The accuracy of SBI and its role in examining WCE studies are unknown. The objective of this study was to evaluate the accuracy of this new software to detect small intestinal lesions.
MATERIALS AND METHODS All patients undergoing WCE during October and November, 2002, at our institution were reviewed for this study. Permission to conduct the review was granted by the Institutional Review Board at the Indiana University and Purdue University (Indianapolis, IN). All patients fasted overnight before swallowing the Given M2A video capsule (Given Imaging). Laxative preparation was not used. The images from each patient were reviewed by five experienced gastroenterologists at 15 frames/s. All detected abnormalities were highlighted in thumbnails for subsequent analysis. The findings detected by gastroenterologists were compared to those recognized by SBI. All thumbnail endoscopy images and reports were subsequently reviewed by two of the authors (S.L., D.R.) to confirm the findings that were read initially by other gastroenterologists. No discrepancies involving the interpretation of the presence or absence of Table 1. Significant Lesions Identified During Wireless Capsule Endoscopy in 24 Patients Lesion
Number (%)
AVM Active bleeding Gastric ulcers Jejunal ulcers* Jejunal AVMs Ulcers without bleeding Erosions Total
47 (43.1) 18 (16.5) 2 (1.8) 11 (10) 5 (4.7) 18 (16.5) 26 (23.9) 109 (100)
AVM ⫽ arteriovenous malformation. * All lesions were in the same patient.
AJG – December, 2003
Performance of Given Suspected Blood Indicator
Figure 1. Suspected blood indicator (SBI) function as part of the WCE software (left). The SBI detects blood and marks (in red line) appropriate images for interrogation. SBI correctly identified blood in the jejunum adjacent to an ulcer.
blood were encountered. Arteriovenous malformations (AVMs), ulcers, erosions, and sites of active bleeding were considered significant lesions.
RESULTS A total of 24 patients (16 women and eight men, mean age 59 yr) were studied during this period. The indications for the study were iron deficiency anemia in 18 patients and abdominal pain in six. Gastroenterologists identified 109 lesions (Table 1). Active bleeding seen by the gastroenterologists was secondary to AVMs (five lesions), jejunal ulcers (11 lesions, all in the same patient), and gastric ulcers (two lesions). A total of 31 potential bleeding areas were identified by SBI (Figs. 1 and 2). When compared to findings recognized by GE, 28 lesions were correctly identified (six AVMs, 13 active bleeding ulcers, seven nonbleeding ulcers, and two nonbleeding erosions). Of the 13 active bleeding lesions, three were AVMs and 10 were jejunal ulcers. The overall sensitivity, specificity, positive predictive value, and negative predictive value of SBI to detect all lesions were 25.7%, 85%, 90%, and 17.3%, respectively
2677
Figure 2. Another jejunal ulcer with active bleeding correctly identified by SBI.
(Table 2). If only actively bleeding small bowel lesions are considered, SBI had sensitivity and positive predictive value of 81.2%. There were three actively bleeding lesions that were not detected by SBI (two jejunal AVMs and one jejunal ulcer). For nonbleeding AVMs, the sensitivity and positive predictive value of SBI were 12.8% and 66.7%. The sensitivity of SBI to detect nonbleeding ulcers was poor (38.9%). The capabilities of SBI to detect significant small bowel lesions are summarized in Table 2.
DISCUSSION The SBI function developed by Given Imaging is designed to highlight actively bleeding areas for rapid review by the physician. We found that SBI has good sensitivity and positive predictive value for bleeding lesions. The sensitivity of SBI to detect nonbleeding AVMs as well as ulcers without bleeding (a function for which it was not designed) was poor. A total of 41 AVMs were identified by gastroenterologists but were missed by SBI. Our study suggests that the SBI function can be used for quick screen-
Table 2. Diagnostic Accuracy of SBI to Detect Significant Small Bowel Lesions Capability of SBI to Detect Significant GI Lesions (%) Lesions
Sensitivity (95% CI)
Specificity (95% CI)
PPV (95% CI)
NPV (95% CI)
Accuracy (95% CI)
Overall lesions AVMs Overall active bleeding Active small bowel bleeding Ulcers without bleeding Erosions
25.7 (17.5–33.9) 12.7 (3.2–22.3) 72.2 (51.5–92.9) 81.2 (62.1–100) 38.9 (16.3–61.4) 7.69 (2.55–17.9)
85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100) 85 (69.3–100)
90.3 (79.9–100) 66.7 (35.8–97.4) 81.3 (62.1–100) 81.3 (62.1–100) 70 (41.6–98.4) 40 (2.9–82.9)
14.2 (9–24.8) 29.3 (17.6–41.0) 77.2 (59.7–94.7) 85 (69.3–100) 60.7 (42.6–78.8) 41.5 (26.3–56.5)
34.8 (26.6–43.1) 34.3 (22.9–45.7) 78.9 (65.9–91.9) 83.3 (71.1–95.5) 63.2 (47.8–78.5) 41.3 (27.1–55.5)
NPV ⫽ negative predictive value; PPV ⫽ positive predictive value.
2678
Liangpunsakul et al.
ing of the small bowel for active bleeding. This could be useful if the interpreter does not have time when a study becomes available to perform a complete diagnostic reading. Alternatively, the SBI function might simply be used as adjunct to interpretation of the study. Clearly, the current version of SBI does not substitute for detailed review of the entire study. In summary, SBI has good sensitivity particularly in the small bowel for actively bleeding lesions. Complete review of the WCE study by the physician is still needed. SBI should be considered as a complementary and rapid screening tool for physicians to identify actively bleeding lesions.
AJG – Vol. 98, No. 12, 2003
Reprint requests and correspondence: Douglas K. Rex, M.D., F.A.C.G., Professor of Medicine, Division of Gastroenterology and Hepatology, Department of Medicine, 550 N. University Boulevard, UH 4100, Indianapolis, IN 46202. Received Mar. 14, 2003; accepted July 21, 2003.
REFERENCES 1. Iddan G, Meron G, Glukhovsky A, Swain P. Wireless capsule endoscopy. Nature 2000;40:725–9. 2. Gong F, Swain P, Mills T. Wireless endoscopy. Gastrointest Endosc 2000;51:725–9. 3. ASGE Technology Assessment Committee. Wireless capsule endoscopy. Gastrointest Endosc 2002;56:621–24.