Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

Journal of Crohn's and Colitis (2012) 6, 692–697

Lewis Score: A useful clinical tool for patients withsuspected Crohn's Disease submitted tocapsule endoscopyBruno Rosa⁎, Maria João Moreira, Ana Rebelo, José Cotter

Gastroenterology Department, Alto Ave Hospital Center, Guimarães, Portugal

Received 17 August 2011; received in revised form 6 November 2011; accepted 3 December 2011

⁎ Corresponding author at: GastroenTel.: +351 253540330; fax: +351 2534

E-mail addresses: bruno.joel.rosa@jcotter@mail.telepac.pt (J. Cotter).

1873-9946/$ - see front matter © 2011doi:10.1016/j.crohns.2011.12.002

KEYWORDSCapsule endoscopy;Crohn's Disease;Lewis Score

Abstract

Background/aims: The Lewis Score (LS) can assess inflammatory activity on small bowel capsuleendoscopy (SBCE). We aimed to evaluate the LS usefulness in the setting of suspected Crohn'sDisease (CD).

Methods: Retrospective single-center study including 56 patients undergoing SBCE for sus-pected CD. Patients were divided into three groups, according to clinical presentation: Group1 (28 patients): suspected CD not supported by the International Conference on Capsule Endos-copy (ICCE) criteria; Group 2 (19 patients): suspected CD based on two ICCE criteria; Group 3 (9patients): patients fulfilling three or more criteria. Inflammatory activity was assessed with theLS. The diagnosis of CD required a minimum follow-up of 6 months after SBCE, basing on clinicalevaluation, endoscopic, histological, radiological, and/or biochemical investigations.Results: SBCE detected significant inflammatory activity (LS≥135) in 23 patients (41.1%), being5 patients from Group 1 (17.8%), 11 from Group 2 (57.9%) and 7 from Group 3 (77.8%) (pb0.05).CD was diagnosed in 23 patients (41.1%): six patients from Group 1 (21.4%), 10 from Group 2(52.6%) and 7 from Group 3 (77.8%) (pb0.05). CD was diagnosed in 82.6% of patients with signif-icant inflammatory activity on CE (LS≥135), but in only 12.1% of those having a LSb135(pb0.05). The LS Positive Predictive Value, Negative Predictive Value, Sensitivity and Specificitywere 82.6%, 87.9%, 82.6% and 87.9%, respectively.Conclusions: The LS may be a valuable diagnostic tool in the setting of suspected CD. Patients notfulfilling the ICCE criteria have lower LS and fewer are diagnosed with CD during follow-up.© 2011 European Crohn's and Colitis Organisation. Published by Elsevier B.V. All rights reserved.

terology Department, Alto Ave21308.gmail.com (B. Rosa), mj.more

European Crohn's and Colitis

Hospital Center, Guimarães, Rua dos Cutileiros, 4835–044, Guimarães.

ira@netcabo.pt (M.J. Moreira), airebelo_@hotmail.com (A. Rebelo),

Organisation. Published by Elsevier B.V. All rights reserved.

Table 1 Patients' demographics, clinical characteristics,endoscopy and imaging prior to SBCE.

Group 1 Group 2 Group 3 Overall

Number ofpatients (%)

28 (50%) 19 (34%) 9 (16%) 56(100%)

SexFemale 16 12 6 34 (61%)

693The Lewis Score for suspected small bowel Crohn's Disease

Introduction

Small bowel capsule endoscopy (SBCE) is currently the mostsensitive diagnostic technique to noninvasively detect earlysmall bowel mucosal disease.1–3 One of the widely acceptedindications for SBCE is the setting of suspected Crohn'sDisease (CD). However, the lack of specificity of endoscopicfindings, suboptimal interobserver agreement and lack oftissue sampling may increase the risk of inaccurate diagno-ses. The Lewis Score (LS),4 which has been integrated intothe latest software from the PillCam® (Given®, RAPID Read-er®), has been developed with the purpose of increasing theobjectivity and maximizing interobserver agreement, whenassessing SBCE inflammatory activity. It uses a standardterminology for the description of endoscopic lesions, thecapsule endoscopy structured terminology (CEST),5–6 andgrades the inflammatory activity through a rank of severity,with the premise that the final numerical score reflects thephysician's global assessment of disease.

In this study, we aimed to evaluate if the assessment andgrading of the severity of inflammation on SBCE with the LSmay be useful as a diagnostic tool for patients withsuspected CD.

Materials and methods

We conducted a retrospective single-center study, includingfifty-six consecutive patients (61% female, mean age 36±14.7 years [16–74]) with suspected CD submitted to SBCE.Exclusion criteria were medication with aspirin or nonsteroi-dal anti-inflammatory drugs (NSAIDs) in the last month,presence of endoscopic inflammatory lesions in the colonos-copy and/or upper endoscopy, patients with obstructivesymptoms in the last 6 months, and patients with less than6 months of follow-up after CE.

The International Conference on Capsule Endoscopy(ICCE) recommended that patients with suspected CDpresenting with suggestive symptoms (A) plus either extrain-testinal manifestations (B), inflammatory markers (C), orabnormal imaging studies (D) should be selected to undergoSBCE7 — Fig. 1. Based on this algorithm, we defined threegroups of patients, according to their clinical presentation:Group 1 — patients that did not fulfill the ICCE criteria;Group 2 — patients with suspected CD based on conjugationof two criteria; and Group 3 — patients with three or morecriteria, thus having the highest pretest probability of CD.Twenty-eight patients were included in Group 1 (50%), 19in Group 2 (34%) and 9 in Group 3 (16%). All patients had

Figure 1 ICCE criteria for suspected Crohn's Disease.7

been submitted to colonoscopy prior to SBCE, with retro-grade ileoscopy being performed on 47 patients (83.9%).Forty-one patients (73.2%) were submitted to upper endos-copy prior to SBCE. Twenty-six patients (46.4%) had beensubmitted to small bowel imaging with small bowel follow-through or computerized tomography (CT) scan; nine ofthose patients (34.6%) had radiological features suggestiveof CD, although no evidence of stricturing lesions. Patients'demographics, as well as clinical characteristics (ICCE cri-teria) and endoscopic, imaging features previous to SBCEare summarized in Table 1.

All patients followed a clear liquid diet for 24 h plus 12 hfasting prior to CE (PillCam® SB, Given® Imaging Ltd. Yoq-neam, Israel). No oral purge was administered. We usedRAPID Reader® 5 to review CE images, and systematicallyassessed inflammatory activity for each patient with deter-mination of the respective LS, using the automatic calcula-tor included in the RAPID® software. The LS allowed theranking of inflammatory activity into three levels: 1) no dis-ease or clinically insignificant disease (LSb135); 2) mild dis-ease (135≤LS≤790); and 3) moderate or severe disease(LSN790).4

The diagnosis of CD during the follow-up period (minimumsix months) was established basing on a combination of clin-ical evaluation, endoscopic, histological, radiological, and/or biochemical investigations.

Statistical analyses were performed using SPSS version 16.0(SPSS Inc., Chicago, Illinois). Comparisons of categorical datawere performed using the chi-squared test. For numeric vari-ables we used one-way analysis of variance (ANOVA).

Results

Patients were followed-up for a mean period of 21.2±12.9 months [6–58].

Twenty-six patients (46.4%) underwent small bowel imagingprevious to SBCE, either with small bowel follow-through or CTscan. Nine of those patients (34.6%) had some radiologic

Male 12 7 3 22 (39%)Age 35.1±

13.842.3±16.2

25.7±6.8 36.0±14.7

Criteria(ICCE)

A (27)C+D (1)

A+C (16)A+B (2)A+D (1)

A+C+D (7)A+B+C (1)A+B+C+D (1)

Prior upperendoscopy

19(67.8%)

13(68.4%)

9 (100%) 41(73.2%)

Priorileoscopy

24(85.7%)

14(73.7%)

9 (100%) 47(83.9%)

Prior SBimaging

10(35.7%)

8 (42.1%) 8 (88.9%) 26(46.4%)

694 B. Rosa et al.

features already suggesting CD; in this subset of patients, 6(66.7%) revealed significant inflammation on SBCE (SL≥135)and were diagnosed with CD during follow-up, with histologicalconfirmation in 50%.

The mean Lewis Score (LS) of patients in Groups 1, 2 and 3was 179 [0–4164], 674 [0–3922] and 1410 [0–5364], respec-tively (pb0.05). Significant inflammatory activity (LS≥135)was detected in 5 (17.8%), 11 (57.9%) and 7 (77.8%) patients(pb0.05), and it was moderate or severe (LSN790) in 1(3.6%), 3 (15.8%) and 3 (33.3%), respectively (p=0.55). Over-all, significant inflammatory activity was detected in 23 pa-tients (41.1%), and CD was confirmed during the follow-up in19 of those patients (Positive Predictive Value, PPV=82.6%).On the other hand, in 33 patients (58.9%), SBCE revealed nodisease or clinically insignificant disease (LSb135), and in 29of those patients, the diagnosis of CD was excluded duringthe follow-up (Negative Predictive Value, NPV=87.9%).SBCE had identified significant inflammatory activity(LS≥135) in 19 of the 23 patients who were diagnosed withCD during follow-up (Sensitivity=82.6%). Conversely, itrevealed no disease or clinically insignificant disease(LSb135) in 29 of the 33 patients who were not diagnosedwith CD during the follow-up period (Specificity=87.9%) —Table 2.

The diagnosis of CD was established in 82.6% of the pa-tients with significant inflammatory activity on SBCE(LS≥135), but in only 12.1% of those who had a LSb135(pb0.05). Overall, the diagnosis of CD was established dur-ing follow-up in 23 patients (41.1%), being 6 patients fromGroup 1 (21.4%), 10 patients from Group 2 (52.6%) and 7 pa-tients from Group 3 (77.8%) (pb0.05). In 13 of those patients(56.5%), the diagnosis was histologically confirmed duringthe follow-up (6 patients underwent colonoscopy with retro-grade ileoscopy and biopsies from the terminal ileum, 4 pa-tients underwent single-balloon enteroscopy and 3 patientswere submitted to surgery), while in the other 10 patients(43.5%) the diagnosis was established by the referring

Table 2 SBCE findings and follow-up: summary of results.

Group 1 Group 2

Number of patients (%) 28 (50%) 19 (34%)Lewis Score 179 [0–4164] 674 [0–39Significant inflammatory activity(LS≥135)

5 (17.8%) 11 (57.9%)

Moderate or severe inflammatoryactivity (LSN790)

1 (3.6%) 3 (15.8%)

SBCE transit time 290.1±101.3[77–525]

290.6±107[54–480]

Incomplete SBCE 1 (3.6%) 3 (15.8%)Capsule retention 0 2 (10.5%)Confirmed diagnosis of CD(follow-up)

6 (21.4%) 10 (52.6%)

Positive histology for CD(follow-up)

5 (17.9%) 3 (15.8%)

Follow-up (months) 18.7±11.4 [7–52] 23.1±14.7SBCESensitivityNegative predictive valueSpecificityPositive predictive value

physician during the follow-up, based on a combination ofclinical evaluation, response to therapy, as well as addition-al imaging and/or biochemical data — Table 2. If a positivehistology had been required to firm the diagnosis of CD dur-ing follow-up, 13 patients would have met that criterion,being 5 patients from Group 1 (17.9%), 3 patients fromGroup 2 (15.8%) and 5 patients from Group 3 (55.6%)(pb0.05). SBCE features would be different in this setting,with higher Sensitivity and NPV, and lower Specificity andPPV, as follows: Sensitivity=84.6% (11 patients withLS≥135 among those 13 patients with histologically con-firmed CD during follow-up); NPV=93.4% (31 patients with-out histological confirmation of CD out of 33 with LSb135);Specificity=72.1% (31 patients with LSb135 out of the 43 pa-tients in which CD was not histologically confirmed); andPPV=47.8% (11 patients with histologically confirmed CDout of 23 with LS≥135).

Overall, SBCE transit time was 287.6±101.2 min[54–525], with no significant differences between groups.In our series, the capsule did not reach the cecum withinthe battery lifespan in 9 patients (16.1%); this proportionwas significantly different between groups (3.6% in Group 1versus 15.8% in Group 2 and 55.6% in Group 3, pb0.05). Cap-sule retention occurred in four patients (7.1%) – 2 patients inGroup 2 and 2 patients in Group 3 – Table 2.

The Table 3 summarizes the demographic and clinicalcharacteristics of the 23 patients with confirmed CD duringthe follow-up.

Discussion

Up to 33% of CD patients have small bowel disease out of thereach of standard endoscopy, making the diagnosis morechallenging. In this setting, SBCE may contribute to an earli-er diagnosis of CD, or otherwise play a key role at excludingthe diagnosis when no significant lesions are found.8-10

Group 3 Statistics Overall

9 (16%) 56 (100%)22] 1410 [0–5364] pb0.05 545 [0–5364]

7 (77.8%) pb0.05 23 (41.1%)

3 (33.3%) p=0.055 7 (12.5%)

.8 258.8±93.6[187–394]

p=0.843 287.6±101.2[54–525]

5 (55.6%) pb0.05 9 (16.1%)2 (22.2%) ns. 4 (7.1%)7 (77.8%) pb0.05 23 (41.1%)

5 (55.6%) pb0.05 13 (23.2%)

[6–58] 24.9±13.2 [6–45] p=0.339 21.2±12.9 [6–58]

82.6%87.9%87.9%82.6%

Table 3 Demographic and clinical characteristics of the 23patients with confirmed CD during the follow-up.

Age (years) (mean±SD) 36.3±14.8Gender (female) n (%) 16 (69.6%)Lewis Score (LS) n (%)

b135 4 (17.4%)135≤LSb790 13 (56.5%)≥790 6 (26.1%)

Diagnosis of Crohn'sDisease (CD)

ICCE criteria at presentation

Group 1: 6 patients(21.4%)

A (6)

Group 2: 10 patients(52.6%)

A+C (9); A+B (1)

Group 3: 7 patients(77.8%)

A+C+D (5); A+B+C (1); A+B+C+D(1)

SBCE transit time (min)(mean±SD)

293.7±11.4

Incomplete SBCE n (%) 6 (26.1%)Capsule retention n (%) 3 (13.0%)Follow-up (months)(mean±SD)

19.5±11.8

695The Lewis Score for suspected small bowel Crohn's Disease

Indeed, a recently published OMED-ECCO consensus statedthat SBCE may identify lesions compatible with CD in somepatients in whom conventional endoscopic and radiographicimaging modalities have been nondiagnostic, and that a nor-mal SBCE has a high negative predictive value for activesmall bowel CD.10-12 However, the lack of specificity ofSBCE endoscopic findings, suboptimal interobserver agree-ment and lack of tissue sampling may increase the risk of in-accurate diagnoses, highlighting the importance of adoptinga common language to describe endoscopic lesions, as wellas uniform criteria to grade and classify the severity of theinflammation. Several indices have been developed in an ef-fort to distinguish significant endoscopic findings from theinnocent lesions found in the general population; while nei-ther system has gained general acceptance, the LS hasbeen integrated into the latest software from the PillCam®(Given®, RAPID Reader®), making it more accessible.4 Thisscoring system has been refined to quantify mucosal damagebased on villous appearance (edema), ulcer, and stenosis,

A. Villous oedema B. Ulcer

Figure 2 A. Villous edem

while minor mucosal breaks, erythema, nodularity or villousatrophy are not scored, because of their lower clinical rele-vance or lower interobserver agreement — Fig. 2. Ascoreb135 corresponds to normal or clinically insignificantmucosal inflammatory change, a score between 135 and790 is mild disease, and a score≥790 corresponds to moder-ate or severe disease. When compared to the study by Mowet al.,13 which considered the presence of more than threeulcers to presume a diagnosis of CD, or one other study byVoderholzer et al.,14 where it was the detection of morethan ten aphthoid or erosive lesions, the LS may theoretical-ly increase the diagnostic yield of SBCE, as it includes pa-tients with villous edema and/or stenosis with noassociated mucosal ulceration. However, a caveat must beremarked: although the LS can quantitatively assess thenumber and severity of mucosal damage, it grades inflam-mation despite of its etiology, which means that it cannotdistinguish among different diagnostic entities, and thus itmust be interpreted with caution in each individual patient.

This study shows that the diagnosis of CD was establishedmuch more frequently in patients with significant inflamma-tory activity on SBCE (LS≥135). In our opinion, the routineuse of the LS to characterize and grade small bowel inflam-matory activity, when interpreted in the right clinical con-text, may be of great value in the setting of suspected CD,to assist clinicians in the interpretation of SBCE lesions andto guide decisions towards a more early and accuratediagnosis.

Similarly to other studies reported in the literature,15 ourstudy is limited to the fact that there is no gold standard forthe diagnosis of small bowel CD. The diagnosis of CD wasassessed during the follow-up with a combination of clinicalevaluation, endoscopic, histological, radiological, and/orbiochemical investigations. In almost half of the patients(43.5%, n=10), the diagnosis was assumed by the referringphysician without histological confirmation, based on a com-bination of clinical evaluation, response to therapy, as wellas additional imaging and/or biochemical data during theperiod of follow-up. We think that this embracing and com-prehensive follow-up is of paramount importance, as thereis actually no single gold standard for the diagnosis of CD;as a matter of fact, it is interesting to remark that if wehad considered only the 13 patients with histological confir-mation during follow-up, CD would only have been assumedin 5 patients from Group 1 (17.9%) instead of 6 (21.4%), 3

C. Stenosis

a. B. Ulcer. C. Stenosis.

696 B. Rosa et al.

patients from Group 2 (15.8%) instead of 10 (52.6%) and 5 pa-tients from Group 3 (55.6%) instead of 7 (77.8%); besides,the features for SBCE would be different, with higher Sensi-tivity (84.6% vs. 82.6%) and NPV (93.4% vs. 87.9%), and withlower Specificity (72.1% vs. 87.9%) and PPV (47.8% vs.82.6%).

A limitation of our study is represented by its retrospec-tive design, and the fact that the characteristics of thestudy population were somewhat heterogeneous. As patientswere divided into three groups according to clinical presen-tation, the number of patients in each subgroup was rela-tively small, and the number of patients was not equallydistributed among groups, as the majority of those patients(50%) referred to Group 1. Nevertheless, this study designallowed us to evaluate, for each group individually, therate of significant small bowel inflammatory activity onSBCE and its influence on subsequent management,highlighting the paramount importance of a careful selectionof suspected CD patients to be submitted to SBCE. As a mat-ter of fact, while the concept of suspected CD is still a mat-ter of debate, with implications in the selection of thepatients, the recognition that isolated abdominal pain or di-arrhea should not constitute an indication for CE is being in-creasingly recognized.16–17 As our study included patientsthat did not fulfill the ICCE criteria (Group 1), it allowed usto conclude that those patients had the lowest rate of signif-icant small bowel inflammatory activity, with lower LS, andlower percentage of patients being diagnosed with CD duringfollow-up.

In our series, ileoscopy was not performed prior to SBCEin 9 patients (16.1%). The fact that during follow-up the di-agnosis of CD was confirmed based on ileoscopy in 6 patients(26% of patients with CD), highlights the importance of per-forming routine retrograde ileoscopy during colonoscopy inthe setting of suspected CD. We also emphasize that, inour study, the diagnosis of CD was established in 16 patientsout of the 47 who had a prior negative ileoscopy (34%), andSBCE showed significant inflammatory activity in 13 ofthose patients. In fact, other studies had previously shownthat in a high percentage of patients, ileal involvement inCD may be outside the range of the ileoscopy, focusing theimportance of SBCE in this setting.18

The rate of incomplete enteroscopies was 16.1%, similarto the rates reported in other studies considering allindications.19–20 This proportion was significantly higher inpatients selected according to the ICCE criteria (3.6% inGroup 1 versus 15.8% in Group 2 and 55.6% in Group 3), theretention rate for diagnosed CD being 3/23 (13.0%). Inmost of the cases, this probably occurred due to a delay ofcapsule progression next to segments with inflammatory ac-tivity, such as edema or inflammatory stenoses. Two ofthose patients spontaneously passed the capsule within afew weeks, while in the other patient the capsule had tobe surgically retrieved, due to obstructive symptoms and un-successful endoscopic retrieval with antegrade single-balloon enteroscopy. Overall, the capsule retention ratewas higher (7%) when compared to other studies in the liter-ature describing retention rates around 1.4% in the setting ofsuspected CD.8,21 Although there are currently no firm rec-ommendations to perform imaging small bowel studiesand/or patency capsule in the feature of suspected CD, allof the patients in our series who happened to retain the

capsule had been previously submitted to imaging studieswho did not suggest the presence of any stenosis. This con-forms to other studies which demonstrate the low reliabilityof clinical and imaging studies in predicting the existence ofstenoses.21–23 In our institution, we do not have availableAgile® patency system, which may be a useful tool to use be-fore CE in patients at risk for capsule retention.24

To summarize, it is our strong belief that this study sup-ports the routine use of the LS to characterize and gradethe small bowel inflammatory activity on SBCE. When inter-preted in adequate clinical context, it may be of great valuein the setting of suspected CD.

Conflict of interest

None of the authors have any potential conflict of interest todeclare.

Acknowledgments

BR carried out the studies and data analyses and drafted themanuscript. MJM conceived the study, participated in its de-sign and coordination and helped to draft the manuscript. ARparticipated in the design of the study and performed thestatistical analysis. JC critically revised the manuscript andfinally approved the version to be submitted. All authorsread and approved the final manuscript.

References

1. Leighton JA, Legnani P, Seidman EG. Role of capsule endoscopyin inflammatory bowel disease: where we are and where we aregoing. Inflamm Bowel Dis 2007;13(3):331–7.

2. Marmo R, Rotondano G, Piscopo R, et al. Meta-analysis: capsuleenteroscopy vs. conventional modalities in diagnosis of smallbowel diseases. Aliment Pharmacol Ther 2005;22:595–604.

3. Triester SL, Leighton JA, Leontiadis GI, et al. A meta-analysis ofthe yield of capsule endoscopy compared to other diagnosticmodalities in patients with non-stricturing small bowel Crohn'sdisease. Am J Gastroenterol 2006;101:954–64.

4. Gralnek IM, Defranchis R, Seldman E, et al. Development of acapsule endoscopy scoring index for small bowel mucosalinflammatory change. Aliment Pharmacol Ther 2008;27(2):146–54.

5. Korman LY. Standard terminology for capsule endoscopy.Gastrointest Endosc Clin N Am 2004;14:33–41.

6. Korman LY, Delvaux M, Gay G, et al. Capsule endoscopystructured terminology (CEST): proposal of a standardized andstructured terminology for reporting capsule endoscopyprocedures. Endoscopy 2005;37:951–9.

7. Mergener K, Ponchon T, Gralnek I, et al. Literature review andrecommendations for clinical application of small-bowelcapsule endoscopy, based on a panel discussion by internationalexperts. Consensus statements for small-bowel capsule endos-copy. Endoscopy 2007;39:895–909.

8. Chermesh I, Eliakim R. Capsule endoscopy in Crohn's disease —indications and reservations. J Crohns Colitis 2008;2:107–13.

9. Figueiredo P, Almeida N, Lopes S, et al. Small-bowel capsuleendoscopy in patients with suspected Crohn's disease — diagnos-tic value and complications. Diagn Ther Endosc 2010; 2010:ID101284. Epub

10. Bourreille A, Ignjatovic A, Aabakken L, Loftus EV, Eliakim R,Pennazio M, et al. Role of small-bowel endoscopy in the

697The Lewis Score for suspected small bowel Crohn's Disease

management of patients with inflammatory bowel disease: aninternational OMED-ECCO consensus. Endoscopy 2009;7(41):618–37.

11. Eliakim R. Video capsule endoscopy of the small bowel. CurrOpin Gastroenterol 2010;26(2):129–33.

12. Swaminath A, Legnani P, Kornbluth A. Video capsule endoscopyin inflammatory bowel disease: past, present, and future redux.Inflamm Bowel Dis 2010;16:1254–62.

13. Mow WS, Lo SK, Targan SR, et al. Initial experience with wire-less capsule enteroscopy in the diagnosis and management of in-flammatory bowel disease. Clin Gastroenterol Hepatol2004;2(1):31–40.

14. Voderholzer WA, Beinhoelzl J, Rogalla P, et al. Small bowelinvolvement in Crohn's disease: a prospective comparison ofwireless capsule endoscopy and computed tomographyenteroclysis. Gut 2005;54(3):369–73.

15. Yousfi MM, De Petris G, Leghton JA, et al. Diaphragm diseaseafter use of nonsteroidal anti-inflammatory agents: first reportof diagnosis with capsule endoscopy. J Clin Gastroenterol2004;38:686–91.

16. Bardan E, Nadler M, Chowers Y, Fidder H, Bar-Meir S. Capsuleendoscopy for the evaluation of patients with chronic abdominalpain. Endoscopy 2003;35(8):688–9.

17. Fry LC, Carey EJ, Shiff AD, et al. The yield of capsule endoscopyin patients with abdominal pain or diarrhea. Endoscopy2006;38(5):498–502.

18. Oshitani N, Yukawa T, Yamagami H, et al. Evaluation of deepsmall bowel involvement by double-balloon enteroscopy inCrohn's disease. Am J Gastroenterol 2006;101(7):1484–9.

19. Westerhof J, Weersma RK, Koornstra JJ. Risk factors forincomplete small-bowel capsule endoscopy. GastrointestEndosc 2009;69(1):74–80.

20. Lewis BS. Expanding role of capsule endoscopy in inflammatorybowel disease. World J Gastroenterol 2008;14(26):4137–41.

21. BCheifetz AS, Lewis BS. Capsule endoscopy retention: is it acomplication? J Clin Gastroenterol 2006;40(8):688–91.

22. Y.Li F, Gurudu SR, De Petris G, et al. Retention of the capsuleendoscope: a single-center experience of 1000 capsule endoscopyprocedures. Gastrointest Endosc 2008;68(1):174–80.

23. Z.Cheifetz AS, Kornbluth AA, Legnani P, et al. The risk of retentionof the capsule endoscope in patients with known or suspectedCrohn's disease. Am J Gastroenterol 2006;101(10):2218–22.

24. Herrerias JM, Leighton JA, Costamagna G, et al. Agile patencysystem eliminates risk of capsule retention in patients withknown intestinal strictures who undergo capsule endoscopy.Gastrointest Endosc 2008;67(6):902–9.