comparison of high definition with standard white light endoscopy for detection of dysplastic...
TRANSCRIPT
ORIGINAL ARTICLE
Comparison of High Definition with Standard White LightEndoscopy for Detection of Dysplastic Lesions DuringSurveillance Colonoscopy in Patients with ColonicInflammatory Bowel DiseaseVenkataraman Subramanian, MD, DM, MRCP(UK),*,† Vidyasagar Ramappa, MD, MRCP(UK),†
Emmanouil Telakis, MD,† Jayan Mannath, MD, MRCP(UK),† Aida U. Jawhari, PhD, FRCP,†
Christopher J. Hawkey, DM, FRCP,† and Krish Ragunath, MPhil, FRCP†
Background: Dysplasia in colonic inflammatory bowel disease (IBD) is often multifocal and flat. High-definition (HD) colonoscopy improves
adenoma detection rates by improving the ability to detect subtle mucosal changes. The utility of HD colonoscopy in dysplasia detection in
patients with IBD has not been reported so far. We aimed to compare the yield of dysplastic lesions detected by standard definition (SD) white
light endoscopy with HD endoscopy.
Methods: A retrospective cohort study of patients with long-standing (>7 years) colonic IBD undergoing surveillance colonoscopy at Notting-
ham University Hospital was studied between September 2008 and February 2010. Details of diagnosis, duration of disease, and outcomes of the
colonoscopy were collected from the endoscopy database, electronic patient records, and patient notes.
Results: There were 160 colonoscopies (101 ulcerative colitis [UC] and 59 Crohn’s disease [CD]) in the SD group and 209 colonoscopies (147
UC and 62 CD) in the HD group. The groups were well matched for all demographic variables. Thirty-two dysplastic lesions (27 on targeted
biopsy) were detected in 24 patients in the HD group and 11 dysplastic lesions (six on targeted biopsy) were detected in eight patients the
SD group. The adjusted prevalence ratio of detecting any dysplastic lesion and dysplastic lesion on targeted biopsy was 2.21 (95% confidence
interval [CI] 1.09–4.45) and 2.99 (95% CI 1.16–7.79), respectively, for HD colonoscopy.
Conclusions: HD colonoscopy improves targeted detection of dysplastic lesions during surveillance colonoscopy of patients with colonic IBD
in routine clinical practice. Randomized controlled studies are required to confirm these findings.
(Inflamm Bowel Dis 2012;000:000–000)
Key Words: surveillance, UC, Crohn’s colitis, dysplasia, high definition
P atients with ulcerative colitis (UC) have an increased
risk for colorectal cancer (CRC) compared to the gen-
eral population.1 Cancer in UC occurs at a younger age
and increases with time, approaching 18% after 30 years of
disease.1 This increased risk has prompted both the North
American and UK gastroenterology societies to recommend
cancer prevention strategies.2,3 Random sampling through-
out the colon has been the mainstay of conventional sur-
veillance practice. Surveillance colonoscopy requires multi-
ple biopsies to be taken and processed, which is tedious,
expensive, and time-consuming. It has been estimated that
at least 33 biopsies are needed to achieve 90% confidence
to detect dysplasia if it is present.4 Surveillance colono-
scopy practices are not uniform and less than 10 biopsies
were noted to be taken based on gastroenterologists self-
reported practices for colonoscopic surveillance for UC.5
The focus of dysplasia in UC is flat and multifocal
and can be easily overlooked with conventional white light
endoscopy.6 There is growing evidence that the yield of
surveillance can be improved by addition of newer endo-
scopic methods that enhance the detection of subtle muco-
sal abnormalities like chromoendoscopy (CE) and auto-
fluorescence with narrow-band imaging (NBI).7 CE refers
to the topical application of dyes or pigments to improve
Received for publication April 8, 2012; Accepted April 9, 2012.
From the *Department of Gastroenterology and Leeds Institute of Molecular
Medicine, St James University Hospital, Leeds, UK, †Nottingham Digestive
Diseases Centre, Nottingham University Hospital, UK.
Venkataraman Subramanian was the recipient of an NIHR clinical
lecturership from the National Institute of Health Research (UK). Krish
Ragunath has received educational/grant support from Olympus (Keymed,
UK).
Reprints: Dr. Venkataraman Subramanian, Department of Gastroenterology, St
James University Hospital, Leeds LS9 7TF, UK (e-mail: venkat.subramanian@
leedsth.nhs.uk).
Copyright VC 2012 Crohn’s & Colitis Foundation of America, Inc.
DOI 10.1002/ibd.23002
Published online in Wiley Online Library (wileyonlinelibrary.
com).
Inflamm Bowel Dis 1
detection and delineation of surface abnormalities and is an
inexpensive adjunct to conventional endoscopy. It has been
shown to be useful in the detection of flat adenomas in the
sporadic setting as well as in patients with familial polypo-
sis syndromes.8,9 There is increasing evidence that most
dysplasias in UC are associated with visible mucosal
abnormalities.10,11 This has in turn led to increasing use of
endoscopic resection techniques to treat areas of raised,
visible dysplasia in patients with UC, without the need for
colectomy.12 There is therefore a need to improve detection
techniques during surveillance endoscopy. A recent meta-
analysis suggested that CE is superior to standard definition
(SD) white light colonoscopy for detection of dysplasia but
this technique has not been widely accepted in routine clin-
ical practice.13
High definition (HD) colonoscopy has also shown
promise in improving the detection of colorectal polyps
and adenomas.14 More recently HD with additional CE was
only shown to only marginally improve the adenoma detec-
tion rate compared to HD alone.15 With an increase in the
number of endoscopy units switching to HD endoscopes,
we sought to determine if the use of HD colonoscopes was
associated with increased detection of dysplastic lesions
both overall and targeted during surveillance colonoscopies
for patients with colonic inflammatory bowel disease (IBD)
in our institution.
MATERIALS AND METHODS
Data CollectionThe computerized clinical and endoscopic database of
the Nottingham University Hospitals served as the data source
for this retrospective cohort study. We reviewed the computer-
ized charts of consecutive patients with IBD undergoing colo-
noscopies between September 2008 and February 2010 at Not-
tingham University Hospital. Patient information was
collected regarding age, sex, diagnosis, extent of disease, use
of 5-aminosalicylates, use of immunomodulator medications
or biological drugs, number of biopsies taken during the colo-
noscopy, whether lesions were noted and if a dysplastic lesion
was confirmed on histopathology, and if colonoscopy was
done by a consultant, nurse endoscopist, or trainee and if an
HD scope and monitor was used. During colonoscopy infor-
mation was recorded about the location and size of any visible
abnormalities.
Only colonoscopies performed with CFH260DL (Olym-
pus, Keymed, UK) colonoscopies using an Olympus Excera
processor and an HD 1080i capable monitor (Olympus
OEV191H) were classed as high definition. All the other pro-
cedures were classed as standard definition and included
Olympus series scopes CF230L, CF240L, CF240 AL, CF240
DL, CF 260DL, and CF Q260DL. Bowel preparation was with
either two sachets of Picolax or four sachets of Kleen Prep
with two tablets of Senna.
During the time period of this study the British Society
of Gastroenterology recommended taking four quadrant biop-
sies every 10 cm2 and this was the protocol adopted by the
endoscopy unit at Nottingham. However, this was not
enforced and individual clinicians were allowed to practice
according to their clinical judgment. All endoscopists, how-
ever, took targeted biopsies of any lesions detected. Biopsy
materials were processed according to standard procedures
and read by an expert gastrointestinal pathologist. When there
was dysplasia or cancer the consensus opinion of two patholo-
gists was reported. Histology was classified according to the
Vienna criteria of gastrointestinal epithelial neoplasia ranging
from no intraepithelial neoplasia to invasive neoplasia.16
Patient Inclusion and Exclusion CriteriaPatients over the age of 18 years with a confirmed diag-
nosis of UC (at least left-sided) or Crohn’s colitis involving a
least one-third of the colon and disease duration of at least 7
years undergoing colonoscopy were included in the study.
Patients undergoing a planned therapeutic procedure or
referred from another center with a diagnosis of dysplasia/
cancer were excluded.
Statistical AnalysisNumeric variables are summarized with the sample
mean and standard deviation and categorical variables with
the number and percentage. The primary endpoints of interest
were the number of patients with any dysplastic lesion
detected and the number of patients with endoscopcially visi-
ble dysplastic lesions. The secondary endpoints were the num-
ber of high-grade dysplasia’s and cancers detected and the
number of patients with endoscopically visible flat dysplastic
lesions. The risk of association between the outcome variables
of interest and type of endoscopy (HD or SD) was estimated
using Poisson regression with robust standard errors rather
than the more commonly used logistic regression.17,18 The
choice of the Poisson model, which provides a prevalence
ratio (PR), was based on the fact that the outcomes were rela-
tively common. The Poisson model provides a more conserva-
tive estimate of the relative risk that is closer to its sample
value than when logistic regression is used in cross-sectional
studies. Potential confounders that were included in multiple
regression models were age, gender, duration of disease,
5-aminosalicylate use, extent of disease, exposure to immuno-
modulators or biologics, experience of colonoscopist, ade-
quacy of bowel preparation, and total number of biopsies
taken. A two-sided significance level was set at P < 0.05. The
statistical software SPSS v. 17 (Chicago, IL) and Stata v. 10
(College Station, TX) were used in all analyses.
RESULTSThe database search revealed 870 lower gastrointesti-
nal (GI) endoscopies that were done on 792 patients, of
which 98 were wrongly coded in the endoscopy database
Inflamm Bowel DisSubramanian et al
2
and did not have IBD, 188 were flexible sigmoidoscopes
not intended for surveillance, 106 did not meet the inclu-
sion criteria as they had less than 7 years of disease, 94
had limited colitis, two were referred with a diagnosis of
cancer high grade dysplasia (HGD) for further manage-
ment, and three were having planned therapeutic colonos-
copies for endoscopic resection of previously detected dys-
plastic lesions. In all, 369 colonoscopies done on 353
patients were included in the final analysis. Table 1 is a
summary of the patient demographics, disease characteris-
tics, and endoscopist characteristics for the 160 SD colo-
noscopies and 209 HD colonoscopies. The groups were
well matched for age, gender, diagnosis, extent of disease,
disease duration, and mean number of biopsies taken per
patient. There were significantly more patients on 5-amino-
salicylates in the HD group (89%) than the SD group
(81%). Significantly more colonoscopies in the HD
group (37%) were done by trainees compared to the SD
group (10%).
Eleven dysplastic lesions were detected in 8/160
colonoscopies in the SD group. Thirty-two dysplastic
lesions were detected in 24/209 colonoscopies in the HD
group. There were two patients with cancer and one patient
with HGD (noted on random biopsies) in the SD group.
There were five patients with cancer and two with HGD
(all picked up on targeted biopsies) in the HD group. Table
2 provides details of the number and characteristics of
lesions detected by the two modalities. Table 3 provides
the adjusted and unadjusted PR of picking up any dysplas-
tic lesion, HGD, or cancer, endoscopically visible dysplas-
tic lesions, and endoscopically visible flat dysplastic lesions
with HD colonoscopy compared to SD colonoscopy. Sig-
nificantly more dysplastic lesions were detected by HD
colonoscopy and there was significantly increased detection
of dysplastic lesions noted on targeted biopsies with HD
colonoscopy.
DISCUSSIONOur study shows that the use of HD colonoscopies is
associated with higher dysplasia detection rate in compari-
son to SD colonoscopy. This was most apparent for dys-
plastic lesions detected by targeted biopsies, especially on
the right side of the colon where 18 lesions (16 on targeted
biopsy) were detected by HD colonoscopy while four
lesions (three on targeted biopsy) were detected by SD
colonoscopy. SD colonoscopy has been felt to be poor at
picking up right-sided lesions at colonoscopy, which are
more likely to be flat or sessile.19 The clinical implications
of picking up a larger number of right-sided lesions needs
to be evaluated in the long term, especially in relation to
reduction in cancer-related mortality. Several studies have
now confirmed that local endoscopic resection with close
follow-up is a safe alternative to colectomy in selected
patients with dysplastic lesions detected on surveillance
colonoscopy.20,21 The increased rates of targeted yield of
dysplastic lesions by HD colonoscopy suggest that it could
help detect dysplastic lesions for endoscopic resection and
reduce the need for colectomy.
CE and electronic CE (NBI, Fujinon intelligent
chromo endoscopy and Pentax i-scan) have all been advo-
cated to improve endoscopic detection of dysplastic lesions.
All reported studies on CE in patients with IBD have used
SD endoscopes.13 Recently, HD chromocolonoscopy has
been shown to only marginally increase the yield of overall
adenoma detection compared to HD white light colono-
scopy in patients undergoing surveillance colonoscopy for
colorectal polyps.15 No differences in dysplasia detection
in patients with colitis was noted between NBI and HD
colonoscopy in a study from the Netherlands.22 Studies
TABLE 1. Patient and Colonoscopy Information for SDand HD Colonoscopy
Variable
SDColonoscopy(n ¼ 160)
HDColonoscopy(n ¼ 209)
Age (mean 6 SD) years 49.7 (14.2) 50.8 (14.9)
Gender:
Male (%) 85 (53%) 121 (58%)
Female (%) 75 (47%) 88 (42%)
Diagnosis:
Crohn’s disease (%) 59 (37%) 63 (30%)
Ulcerative colitis (%) 101 (63%) 146 (70%)
Disease extent:)
Pancolitis (%) 50 (31%) 85 (41%)
Left sided UC (%) 50 (31%) 61 (29%)
Ileocolonic CD (%) 30 (19%) 37 (18%)
Colonic CD (% 30 (19%) 26 (12%)
5-Aminosalicylate use (%)* 130 (81%) 187 (89%)
Immunomodulator orbiological drug exposure (%)
79 (49%) 86 (41%)
Disease duration (mean 6 SD) 16.3 (8.4) 17.8 (8.6)
Adequacy of bowel preparation
Poor (%) 13 (8%) 15 (7%)
Fair (%) 42 (26%) 70 (34%)
Good (%) 77 (48%) 96 (46%)
Excellent (%) 28 (18%) 28 (13%)
Scoped beyond knownextent of disease
145 (91.2%) 196 (93.8%)
Grade of colonoscopist**
Consultant (%) 97 (61%) 86 (41%)
Nurse endoscopist (%) 47 (29%) 45 (22%)
Trainee (%) 16 (10%) 78 (37%)
Mean number of biopsies (SD) 13 (6) 14 (7)
*P ¼ 0.03; **P ¼ 0.01.
Inflamm Bowel Dis Detection of Dysplastic Lesions
3
comparing HD colonoscopy with HD chromocolonoscopy
in detecting dysplasia in long-standing colitis are therefore
needed to confirm if the added time taken and experience
needed for CE has any advantages over HD colonoscopy.
The advantage of HD colonoscopy over other techniques
like CE and NBI is that it should not increase the time
taken for the procedure and is likely to have better accep-
tance among endoscopists, as it involves familiar white
light color imaging and does not involve learning a new
skill.
Similar to most studies looking at detection of dys-
plastic lesions in patients with UC, the major proportion of
lesions detected in this study were low-grade dysplasia
(LGD).13 However, both techniques picked up a similar
proportion of lesions with LGD (25/32 [78%] by HD and
8/11 [73%] by SD colonoscopy), suggesting that the
increased yield with HD colonoscopy is probably a true
reflection of the increased yield with this technique.
Of the 11 lesions detected using SD colonoscopy,
five were on random biopsies and of the 32 dysplastic
lesions detected using HD colonoscopy five were on ran-
dom biopsies, suggesting that random biopsies will still
need to be taken with either modality. These results are
similar to the yield of random biopsies in a large retrospec-
tive study from the Netherlands where 24/125 (19%) of
dysplastic lesions (all low grade) were detected on random
biopsy.23 In the Dutch study, only one of four patients in
whom dysplasia was noted on random biopsies had any
long-term clinical consequence. In our study of the two
patients with LGD detected on random biopsies alone in
the HD group, no changes in clinical outcomes were noted.
Both these patients were placed on colonoscopic surveil-
lance and of these one had LGD reconfirmed on repeat sur-
veillance colonoscopy and the other is awaiting follow-up
colonoscopy. Further studies are needed to confirm if aban-
doning random biopsies would have long-term clinical
TABLE 2. Number of Dysplastic Lesions, Number of High Grade Dysplasia’s or Cancer’s, Number of Patients with Dys-plastic Lesions Detected on Targeted Biopsies, and Number of Patients with Flat Dysplastic Lesions Detected on Tar-geted Biopsies
Standard Definition High Definition
Total number of dysplastic lesions 11 (6 on targeted biopsies) 32 (27 on targeted biopsies)
Number of patients with dysplasia 8 24
Anatomic location of lesions:
Right Colon 04 (03 on targeted biopsies) 18 (16 on targeted biopsies)
03 (03 on targeted biopsies) 10 (07 on targeted biopsies)
Sigmoid/descending/transverse rectum 04 (00 on targeted biopsies) 04 (04 on targeted biopsies)
Number of patients with high grade dysplasia /cancer High grade dysplasia: 1 High grade dysplasia: 2
Cancer: 2 Cancer: 5
Number of patients with any dysplastic lesions detectedby targeted biopsies
5 22
Number of patients with flat dysplastic lesions detectedby targeted biopsy
2 11
TABLE 3. Multiple Regression Analysis of the Association Between Type of Colonoscopy (SD or HD) and Detection ofAny Dysplastic Lesions, High Grade Dysplasia or Cancer, Endoscopically Visible Dysplastic Lesions, and EndoscopicallyVisible Flat Dysplastic Lesions on a Per Patient Basis
PrevalenceRatio (95% CI) P-value
Adjusted PrevalenceRatio* (95% CI) P-value
Any dysplasia 2.30 (1.03-5.11) 0.04 2.21 (1.09-4.45) 0.03
High grade dysplasia or cancer 1.79 (0.48-6.91) 0.4 1.56 (0.42-5.79) 0.5
Any dysplastic lesions detected by targeted biopsies 3.37 (1.28-8.89) 0.01 2.99 (1.16-7.77) 0.02
Flat dysplastic lesions detected by targeted biopsies 4.21 (0.93-18.99) 0.06 2.64 (0.66-10.51) 0.16
*Adjusted for age, gender, diagnosis, duration of disease, extent of disease, 5-aminosalicylate use, immunomodulator or biologic drug exposure, numberof biopsies taken experience of colonoscopist, and bowel preparation noted.
Inflamm Bowel DisSubramanian et al
4
consequences, but improved targeted detection with HD
colonoscopy shows promise in reducing the need for ran-
dom biopsies in this patient population.
The retrospective nature of this study is another
potential limitation. We were not able to standardize the
collection of all data, such as withdrawal time, timing of
the procedure, and several endoscopists performed the pro-
cedure. However, the retrospective design could also be
considered a strength. The study provides information
regarding the yield of HD colonoscopy in actual clinical
practice. It has been argued that randomized controlled
studies in endoscopy are naturally biased, as endoscopists
perform the procedure more diligently than in routine clini-
cal practice as they feel they are being scrutinized. This
had led to the Institute of Medicine recommendations and
priorities on ‘‘comparative effectiveness research’’ includ-
ing colorectal cancer screening strategies to be done in rou-
tine clinical practice.24 As with all nonrandomized trials, it
is not possible to entirely exclude confounding variables;
however, we used appropriate statistical methods to
describe and account for differences in the populations
such as 5-aminosalicylate use and the proportion of trainee
colonoscopists performing the procedure. After accounting
for these differences, the dysplasia detection rate remained
different for HD versus SD colonoscopy. Not all the endo-
scopists used both modalities routinely. However, when the
analysis was limited to endoscopists who used both modal-
ities, 20 lesions (in 14 patients) were detected in 118 colo-
noscopies with HD colonoscopy, while no lesion was
picked up in 46 colonoscopies with SD colonoscopy, sug-
gesting that endoscopist skill alone could not explain the
differences noted in the study.
As this was a retrospective study, endoscopists were
not blinded to the clinical data and pathologists were not
blinded to the clinical and endoscopy data. Finally, we do
not have data on the false-positive rates in both SD and
HD colonoscopy, as lesions deemed nonsignificant were
less likely to have been recorded in the endoscopy report,
potentially causing underreporting of false-positive lesions
for both modalities. The median number of biopsies taken
in this study was 13 in the SD group and 14 in the HD
group. This is less than the recommended 33 biopsies that
are needed to achieve 90% confidence to detect dysplasia.4
As noted before, during the time period of this study the
British Society of Gastroenterology recommended taking
four quadrant biopsies every 10 cm2 and this was the pro-
tocol adopted by the endoscopy unit at Nottingham. How-
ever, this was not enforced and individual clinicians were
allowed to practice according to their clinical judgment.
The self-reported number of biopsies in a survey of UK
gastroenterologists revealed that 57% take fewer than 10
biopsies per patient.25 The St Mark’s group also reported a
median number of eight biopsies in their surveillance
program.26 Our figures are therefore representative of UK
clinical practice and can be generalizable to most clinical
practices. This could have led to lower identification of
dysplasia but, conversely, many studies have shown an
extremely low yield of dysplasia from nontargeted biopsies
compared to targeted biopsies, suggesting that the incre-
mental benefit from multiple nontargeted biopsies is
small.27,28
In conclusion, dysplasia yield was significantly higher
for HD colonoscopy compared to SD colonoscopy in
patients with long-standing colonic IBD. HD colonoscopy
was three times more likely to pick up a dysplastic lesion
on targeted biopsy than SD colonoscopy. No differences
were noted for the pick-up rate of lesions harboring HGD
or early cancer or visible flat lesions. Large multicenter
randomized controlled studies are needed to confirm the
findings of this study.
ACKNOWLEDGMENTSAuthor contributions: V.S., study design, data collec-
tion and analysis, wrote the article; V.R. and E.T., data
collection and analysis, critical review of the article; J.M.,
A.J., C.J.H., and K.R., study design, critical review of the
article.
REFERENCES1. Eaden JA, Abrams KR, Mayberry JF. The risk of colorectal cancer in
ulcerative colitis: a meta-analysis. Gut. 2001;48:526–535.2. Carter MJ, Lobo AJ, Travis SP. Guidelines for the management of
inflammatory bowel disease in adults. Gut. 2004;53(Suppl 5):V1–16.3. Kornbluth A, Sachar DB. Ulcerative colitis practice guidelines in
adults (update): American College of Gastroenterology, PracticeParameters Committee. Am J Gastroenterol. 2004;99:1371–1385.
4. Rubin CE, Haggitt RC, Burmer GC, et al. DNA aneuploidy in colonicbiopsies predicts future development of dysplasia in ulcerative colitis.Gastroenterology. 1992;103:1611–1620.
5. Eaden JA, Ward BA, Mayberry JF. How gastroenterologists screen forcolonic cancer in ulcerative colitis: an analysis of performance.Gastrointest Endosc. 2000;51:123–128.
6. Itzkowitz SH, Harpaz N. Diagnosis and management of dysplasia inpatients with inflammatory bowel diseases. Gastroenterology. 2004;126:1634–1648.
7. van den Broek FJ, Fockens P, van Eeden S, et al. Endoscopic tri-modal imaging for surveillance in ulcerative colitis: randomised com-parison of high-resolution endoscopy and autofluorescence imagingfor neoplasia detection; and evaluation of narrow-band imaging forclassification of lesions. Gut. 2008;57:1083–1089.
8. Huneburg R, Lammert F, Rabe C, et al. Chromocolonoscopy detectsmore adenomas than white light colonoscopy or narrow band imagingcolonoscopy in hereditary nonpolyposis colorectal cancer screening.Endoscopy. 2009;41:316–322.
9. Le Rhun M, Coron E, Parlier D, et al. High resolution colonoscopywith chromoscopy versus standard colonoscopy for the detection ofcolonic neoplasia: a randomized study. Clin Gastroenterol Hepatol.2006;4:349–354.
10. Rubin DT, Rothe JA, Hetzel JT, et al. Are dysplasia and colorectalcancer endoscopically visible in patients with ulcerative colitis? Gas-trointest Endosc. 2007;65:998–1004.
11. Rutter MD, Saunders BP, Wilkinson KH, et al. Most dysplasia in ul-cerative colitis is visible at colonoscopy. Gastrointest Endosc. 2004;60:334–339.
Inflamm Bowel Dis Detection of Dysplastic Lesions
5
12. Hurlstone DP, Sanders DS, Atkinson R, et al. Endoscopic mucosalresection for flat neoplasia in chronic ulcerative colitis: can we changethe endoscopic management paradigm? Gut. 2007;56:838–846.
13. Subramanian V, Mannath J, Ragunath K, et al. Meta-analysis: thediagnostic yield of chromoendoscopy for detecting dysplasia inpatients with colonic inflammatory bowel disease. Aliment PharmacolTher. 2011;33:304–312.
14. Subramanian V, Mannath J, Hawkey CJ, et al. High definition colono-scopy vs. standard video endoscopy for the detection of colonicpolyps: a meta-analysis. Endoscopy. 2011;43:499–505.
15. Kahi CJ, Anderson JC, Waxman I, et al. High-definition chromocolo-noscopy vs. high-definition white light colonoscopy for average-riskcolorectal cancer screening. Am J Gastroenterol. 2010;105:1301–1307.
16. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification ofgastrointestinal epithelial neoplasia. Gut. 2000;47:251–255.
17. Barros AJ, Hirakata VN. Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly esti-mate the prevalence ratio. BMC Med Res Methodol. 2003;3:21.
18. Zou G. A modified poisson regression approach to prospective studieswith binary data. Am J Epidemiol. 2004;159:702–706.
19. Bressler B, Paszat LF, Vinden C, et al. Colonoscopic miss rates forright-sided colon cancer: a population-based analysis. Gastroenterology.2004;127:452–456.
20. Rubin PH, Friedman S, Harpaz N, et al. Colonoscopic polypectomy inchronic colitis: conservative management after endoscopic resection ofdysplastic polyps. Gastroenterology. 1999;117:1295–1300.
21. Vieth M, Behrens H, Stolte M. Sporadic adenoma in ulcerative colitis:endoscopic resection is an adequate treatment. Gut. 2006;55:1151–1155.
22. van den Broek FJ, Fockens P, van ES, et al. Narrow-band imagingversus high-definition endoscopy for the diagnosis of neoplasia inulcerative colitis. Endoscopy. 2011;43:108–115.
23. van den Broek FJ, Stokkers PC, Reitsma JB, et al. Random biopsiestaken during colonoscopic surveillance of patients with longstandingulcerative colitis: low yield and absence of clinical consequences. AmJ Gastroenterol. 2011 [Epub ahead of print].
24. Sox HC, Greenfield S. Comparative effectiveness research: a reportfrom the Institute of Medicine. Ann Intern Med. 2009;151:203–205.
25. Lynch DA, Lobo AJ, Sobala GM, et al. Failure of colonoscopic sur-veillance in ulcerative colitis. Gut. 1993;34:1075–1080.
26. Rutter MD, Saunders BP, Wilkinson KH, et al. Thirty-year analysis ofa colonoscopic surveillance program for neoplasia in ulcerative colitis.Gastroenterology. 2006;130:1030–1038.
27. Rutter MD, Saunders BP, Wilkinson KH, et al. Most dysplasia in ulcerativecolitis is visible at colonoscopy. Gastrointest Endosc. 2004;60:334–339.
28. Rutter MD, Saunders BP, Schofield G, et al. Pancolonic indigo car-mine dye spraying for the detection of dysplasia in ulcerative colitis.Gut. 2004;53:256–260.
Inflamm Bowel DisSubramanian et al
6