Factors affecting adenoma detection rate in a national flexible sigmoidoscopy screening programme: a retrospective analysis

Factors affecting adenoma detection rate in a national flexible sigmoidoscopy screening programme: a retrospective analysis

Articles Factors affecting adenoma detection rate in a national flexible sigmoidoscopy screening programme: a retrospective analysis Roisin Bevan, Ro...

110KB Sizes 0 Downloads 52 Views

Articles

Factors affecting adenoma detection rate in a national flexible sigmoidoscopy screening programme: a retrospective analysis Roisin Bevan, Roger G Blanks, Claire Nickerson, Brian P Saunders, John Stebbing, Richard Tighe, Andrew M Veitch, William Garrett, Colin J Rees

Summary

Background A national colorectal cancer screening programme started in England in 2013, offering one-off flexible sigmoidoscopy to all men and women aged 55 years in addition to the biennial faecal occult blood testing programme offered to all individuals aged 60–74 years. We analysed data from six pilot flexible sigmoidoscopy screening centres to examine factors affecting the adenoma detection rate (ADR). Methods We did a retrospective analysis of flexible sigmoidoscopy screening procedures performed in individuals aged 55 years at six pilot sites in England as part of the National Health Service Bowel Scope Screening programme. ADR (number of procedures in which at least one adenoma was removed or biopsied, divided by total number of procedures) was calculated for each site and each endoscopist. Multiple regression models were used to examine the variation in ADR with withdrawal time and extent of examination, and the effect of other factors including comfort and bowel preparation on extent of examination. Findings The analysis included 8256 procedures done between May 7, 2013, and May 6, 2014. The overall ADR was 9·1% (95% CI 8·5–9·8; 755 of 8256 procedures), varying from 7·4% (6·2–8·9) to 11·0% (9·1–13·4) by screening centre. The ADR was 11·5% (95% CI 10·6–12·5; 493 of 4299 procedures) in men and 6·6% (5·9–7·4; 262 of 3957 procedures) in women (p<0·0001). On multivariate analysis, factors associated with adenoma detection were male sex (relative risk 1·69, 95% CI 1·46–1·95; p<0·0001) and a withdrawal time from the splenic flexure of at least 3·25 min in negative procedures (1·22, 1·00–1·48; p=0·045). However, increasing the withdrawal time to 4·0 min or more did not increase the likelihood of adenoma detection (1·22, 0·99–1·51; p=0·057). Procedures not reaching the splenic flexure were associated with lower chance of adenoma detection (eg, 0·77, 0·66–0·91; p=0·0015 for procedures reaching the descending colon), but there was no additional benefit associated with reaching the transverse colon (0·83, 0·67–1·02; p=0·069). Women (0·83, 0·80–0·87; p<0·0001), individuals with adequate (0·79, 0·76–0·83; p<0·0001) or poor (0·58, 0·51–0·67; p<0·0001) bowel preparation (compared with good bowel preparation), and those with mild (0·82, 0·76–0·88; p<0·0001) or moderate or severe (0·58, 0·51–0·66; p<0·0001) discomfort (compared with no discomfort) were less likely to have a procedure reaching the splenic flexure. Interpretation Key performance indicators for flexible sigmoidoscopy screening should be defined, including standards for insertion and withdrawal times, optimal depth, and bowel preparation. ADR could be improved by recommending a withdrawal time from the splenic flexure of at least 3·25 min (ideally 3·5–4·0 min). Funding None. Copyright © 2019 Elsevier Ltd. All rights reserved.

Introduction Population screening for colorectal cancer began in England in 2006 in the form of faecal occult blood testing followed by colonic investigations for individuals with positive tests.1 The programme aims to detect colorectal cancer at an early stage to improve cure rates and length of survival. A secondary aim is to detect and remove adenomas.1 Standards are high, with adenoma detection rates (ADRs) after positive faecal occult blood testing of 29% in women and 43% in men.2 An additional programme within the English Na­tional Health Service (NHS) Bowel Cancer Screening Pro­ gramme (BCSP) started in 2013, offering one-off flex­ible sigmoid­oscopy to all individuals aged 55 years,3 with the aim of detecting and removing adenomas from the left side of the colon, reducing development of colorectal

cancer via the adenoma–carcinoma sequence.4 A large study showed a reduction in incidence of colorectal cancer of 23% and a 31% reduction in mortality in intention-totreat analyses,5 forming the rationale for the Bowel Scope Screening (BoSS) programme. Six pilot sites started this programme in 2013, with a staged roll out throughout the rest of England planned. An initial benchmark ADR for endoscopists was set at 6·8%.6 The previous trial of flexible sigmoidoscopy screening reported an ADR of 12·1%,7 although the cohort age group differed from that of BoSS (55–64 years in the trial compared with 55 years in BoSS). An ADR of 9·6% was recorded in an ex­ploratory study to test the feasibility of BoSS delivery.8 Other international flexible sigmoidoscopy screening programmes have re­ ported ADRs ranging from 15%9 to 18%,10 although again the age group of screened individuals varied. Colonoscopy

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

Lancet Gastroenterol Hepatol 2019 Published Online January 14, 2019 http://dx.doi.org/10.1016/ S2468-1253(18)30387-X See Online/Comment http://dx.doi.org/10.1016/ S2468-1253(19)30002-0 North Tees and Hartlepool NHS Foundation Trust, Stocktonon-Tees, UK (R Bevan MClinRes); Cancer Epidemiology Unit, University of Oxford, Oxford, UK (R G Blanks PhD); NHS Cancer Screening Programmes, Sheffield, UK (C Nickerson MSc); St Mark’s Hospital, London, UK (Prof B P Saunders MD); Royal Surrey County Hospital NHS Foundation Trust, Guildford, UK (J Stebbing MBBS); Norfolk and Norwich University Hospitals NHS Foundation Trust, Norwich, UK (R Tighe MD); The Royal Wolverhampton NHS Trust, Wolverhampton, UK (A M Veitch MD); Medway NHS Foundation Trust, Gillingham, UK (W Garrett MBBS); South Tyneside NHS Foundation Trust, South Shields, UK (Prof C J Rees MD); and Newcastle University, Newcastle-upon-Tyne, UK (Prof C J Rees) Correspondence to: Dr Roisin Bevan, North Tees and Hartlepool NHS Foundation Trust, Stockton-on-Tees TS19 8PE, UK [email protected]

1

Articles

Research in context Evidence before this study We searched MEDLINE and Google Scholar with the terms “flexible sigmoidoscopy” AND (“adenoma detection rate” OR “key performance indicators”) for papers published between Jan 1, 1980, and Dec 31, 2017. No restrictions on study type or language were implemented. Key performance indicators for colonoscopy within the English National Health Service Bowel Cancer Screening Programme are well established and high quality has been demonstrated. Flexible sigmoidoscopy screening reduces incidence of colorectal cancer and mortality from the disease. The practice of flexible sigmoidoscopy screening varies between different endoscopists and key performance indicators are poorly defined. We aimed to identify factors associated with adenoma detection rate (ADR) and extent of examination in flexible sigmoidoscopy. To our knowledge, our analysis is the first study to explore variations in ADR in a sigmoidoscopy screening programme. Added value of this study We did a retrospective analysis of flexible sigmoidoscopy screening procedures performed in individuals aged 55 years at six pilot sites in England as part of the NHS Bowel Scope Screening programme. The overall ADR was 9·1% (95% CI 8·5–9·8), varying from 7·4% (6·2–8·9) to

after faecal occult blood testing in the BCSP has sev­eral quality indicators against which performance is moni­ tored,11 ensuring that the programme delivers the best possible service and sup­ porting improvement of nonscreening endoscopy ser­vices, as demonstrated by overall improvements in UK endoscopic performance attrib­uted in part to the stringent standards attached to screening colonoscopy.12 Similar quality standards are required for the BoSS programme, but have not yet been defined. This study aimed to assess the findings of the first 12 months of BoSS at the six pilot sites, with regards to ADR and the procedural factors that affect this measure, and to consider how these results can be used to form quality standards.

Methods

Study design and participants In this retrospective analysis, we interrogated the online database of the NHS Bowel Cancer Screening System (BCSS), which records procedural data during screen­ ing episodes, for all flexible sigmoidoscopy procedures performed at six pilot sites in England. In BoSS, data are recorded live during the procedure and histological results are added later. The pilot sites were six BCSP screening centres that had the endoscopy unit capacity to commence the BoSS programme and represent a range of socioeconomic backgrounds and geographical areas. The sites were Norwich, South of Tyne and Wear, St Mark’s Hospital (London), Surrey, West Kent and Medway, and Wolverhampton. After receiving an invitation in the post, 2

11·0% (9·1–13·4) by screening centre. Differences in ADR exist between men and women (11·5% in men vs 6·6% in women; p<0·0001), but case mix adjustments do not explain the variation in ADR between endoscopists. On multivariate analysis, men, individuals with a withdrawal time of at least 3·25 min, and those with procedures that reached the splenic flexure were associated with increased likelihood of adenoma detection. ADR increases with increasing withdrawal time from the splenic flexure (up to a plateau at 4 min). Mean time taken to reach maximal insertion was 4 min for all segments reached, but reaching further was associated with less discomfort. Women, individuals without good bowel preparation, and those experiencing mild, moderate, or severe discomfort were less likely to have a procedure reaching the splenic flexure. Implications of all the available evidence Key performance indicators for flexible sigmoidoscopy screening should be defined, including standards for insertion and withdrawal times, optimal depth, and bowel preparation. ADR could be improved by recommending a withdrawal time from the splenic flexure of at least 3·25 min (ideally 3·5–4·0 min). The programme should maximise ADR while minimising discomfort, particularly in women, who have lower ADRs and higher levels of discomfort.

participants self-administered an enema at home before attending the department for the flexible sigmoidoscopy procedure. If the patient was unable or unwilling to administer the enema at home, it could be given in the department on arrival. The bowel preparation method was the same at all sites. Procedures were performed without sedation, using a standard colonoscope, paedi­atric colon­ oscope, or flexible sigmoidoscope. Endo­ scope imagers could be used if available and compatible with the scope in use, but their use was not mandated and not recorded in the BCSS. The option to opt in to the BoSS is made available to individuals older than 55 years but younger than 60 years and thus not yet invited to take part in the guaiac faecal occult blood testing screening programme; however, we excluded these individuals from the analyses because of small numbers and to avoid potential bias because their method of invitation was different.

Data extraction Only procedures in which an adenoma was removed or biopsied were counted as having an adenoma detected. In procedures in which one or more polyp was detected, but not removed and not biopsied (eg, because of size, total number, or patient factors such as anticoagulation), no adenoma was recorded. Cases with more than three adenomas (either as judged clinically by the endoscopist or proven on histology), a polyp with villous histology or high-grade dysplasia, a polyp more than or equal to 10 mm in size, more than 20 hyperplastic polyps above the

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

Articles

Endoscopists with Procedures ≥30 procedures Total

Men

Polyps or cancer detected (%)

Adenoma detected (%)

Adenoma detected (%)

Procedures (% of total)

Adenoma detected (%)

Centre A

3

824

188 (22·8%)

448 (54·4%)

61 (13·6%)

376 (45·6%)

Centre B

9

1398

309 (22·1%)

104 (7·4%)

717 (51·3%)

71 (9·9%)

681 (48·7%)

33 (4·9%)

Centre C

8

2098

420 (20·0%)

202 (9·6%)

1070 (51·0%)

132 (12·3%)

1028 (49·0%)

70 (6·8%)

Centre D

11

1944

431 (22·2%)

196 (10·1%)

990 (50·9%)

125 (12·6%)

954 (49·1%)

71 (7·4%)

Centre E

9

1315

230 (17·5%)

99 (7·5%)

707 (53·8%)

66 (9·3%)

608 (46·2%)

33 (5·4%)

Centre F Total

91 (11·0%)

Procedures (% of total)

Women

30 (8·0%)

4

677

161 (23·8%)

63 (9·3%)

367 (54·2%)

38 (10·4%)

310 (45·8%)

25 (8·1%)

44

8256

1739 (21·1%)

755 (9·1%)

4299 (52·1%)

493 (11·5%)

3957 (47·9%)

262 (6·6%)

Table 1: Endoscopists per screening centre and procedural results

rectum, or cases in which a polyp could not be removed because of patient factors were referred for colonoscopy,13 performed on a BCSP colonoscopy list by a BCSPaccredited colon­oscopist. Any adenomas detected during this procedure did not affect the ADR of the original procedure. The rate of colonoscopy conversion for each site was calculated. Details of BoSS endoscopists were collected from information held by the BCSP or directly from the sites, and anonymised before analysis; centres were also an­ onymised. Data analyses were restricted to endos­copists who had performed at least 30 procedures, because fewer than 30 procedures (approximately three lists) during a 12-month period was felt to reflect an inadequate number of procedures on which to perform analysis. Below this level, statistical instability and potential bias could arise within the data. Fewer than 30 procedures might indi­ cate endoscopists who do not regularly perform BoSS pro­ cedures or BoSS endoscopists at the start of their screening career, when a learning curve might affect the results. Data were extracted from BCSS at procedural level, recording age, sex, screening site, endoscopist, date of procedure, extent of examination (as reported by the endoscopist during the procedure based on their opinion of the extent of examination reached, in conjunction with external imager guidance if available), total procedure time, withdrawal time, comfort assessment, bowel prep­ aration quality, presence of polyps or adenomas, presence of cancer, and details of complications. The quality of bowel preparation was recorded during the procedure as good, adequate, or poor (subjectively reported by the endoscopist). Validated scoring systems of bowel preparation quality for flexible sigmoidoscopy are not available. To provide a numerical measure of bowel preparation, a bowel preparation score for each endoscopist was produced by weighting procedures with good pre­ paration by 2, adequate by 1, and poor by 0 (giving a final value between 0 and 2). Comfort during the procedure was assessed after the procedure by the nursing staff and recorded by the specialist screening practitioner, using the scale of no, minimal, mild, moderate, or severe discomfort.

Centre A

Centre B

Centre C

Centre D

Centre E

5

4

4

26

Centre F

Area of bowel Anus, rectum, or rectosigmoid Sigmoid colon Descending colon

2

4

47

96

47

136

179

19

188

507

447

634

326

113

Splenic flexure

193

347

954

658

494

230

Transverse colon

206

134

225

81

60

150

Total

636

1089

1677

1513

1085

516

Mean extent score* Proportion of procedures reaching splenic flexure†

4·87 63%

4·47 44%

4·80 70%

4·44 49%

4·35 51%

4·97 74%

Data are number of procedures, unless otherwise stated. *A mean negative extent score was calculated by assigning the extent of examination reached a number as follows: anus (0), rectum (1), rectosigmoid (2), sigmoid (3), descending colon (4), splenic flexure (5), and transverse colon (6); expressed as a number less than or equal to 6. †The proportion of procedures reaching the splenic flexure is the number of procedures reaching the splenic flexure plus the number of procedures reaching the transverse colon divided by the total number of procedures.

Table 2: Extent of examination in negative procedures by centre and summary measures

Statistical analysis ADR (defined as the number of procedures in which at least one adenoma was detected, divided by total number of procedures) was calculated for each site and each endoscopist. We also calculated the sex-adjusted ADR for each endoscopist (calculated as population risk for men × Nm + population risk for women × NF [Nm=number of procedures in men, NF=number in women]). Inser­ tion time was calculated by subtracting the recorded withdrawal time from the recorded total procedure time. We calculated an expected-to-observed ADR adjusted for sex. We used the population risk of finding an adenoma in women and men (taken from the overall ADR in the 12 months of the programme assessed) to predict how many adenomas each endoscopist should find on the basis of their case mix and then calculated the ratio of predicted-to-observed adenomas. The number of procedures in which no polyp, adenoma, or cancer was detected was recorded for each site. Within the BCSP, such procedures are termed negative procedures. We assessed the reported extent of examination for negative procedures (ie, the furthest

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

3

Articles

Mean estimated withdrawal time (SD), min

Number of endoscopists

Number of procedures (with adenomas)

Adenoma detection rate (95% CI)*

<2·5 min

2·12 (3·7)

9

1877 (154)

8·2% (7·0–9·5)

2·5–3·24 min

2·81 (2·8)

15

2249 (191)

8·5% (7·4–9·7)

3·25–3·99 min

3·53 (2·7)

11

2425 (240)

9·9% (8·7–11·2)

≥4 min

4·40 (3·2)

9

1705 (170)

10·0% (8·6–11·5)

*ptrend=0·021.

Table 3: Adenoma detection rate by estimated splenic flexure withdrawal time

Univariate RR (95% CI)

Role of the funding source

Multivariate p value

RR (95% CI)

p value

Sex Women

1 (ref)

Men

1·73 (1·50–2·00)

·· <0·0001

1 (ref) 1·69 (1·46–1·95)

·· <0·0001

Estimated splenic flexure withdrawal time (min) ··

1 (ref)

<2·5

1 (ref)

2·5–3·24

1·04 (0·84–1·27)

0·739

1·05 (0·86–1·29)

0·635

3·25–3·99

1·21 (0·99–1·46)

0·057

1·22 (1·00–1·48)

0·045

≥4·0

1·22 (0·99–1·50)

0·066

1·22 (0·99–1·51)

··

0·057

Transverse colon

0·89 (0·73–1·10)

0·274

0·83 (0·67–1·02)

0·069

Splenic flexure

1 (ref)

Descending colon

0·75 (0·64–0·88)

0·0004

0·77 (0·66–0·91)

0·0015

Sigmoid colon

0·54 (0·39–0·74)

0·0002

0·59 (0·43–0·81)

0·0013

Anus, rectum, or rectosigmoid

0·46 (0·15–1·40)

0·173

0·48 (0·16–1·46)

0·198

Extent of examination ··

1 (ref)

··

RR=relative risk.

Table 4: Binomial regression of risk of adenoma detection by sex, estimated splenic flexure withdrawal time, and extent of examination See Online for appendix

4

procedures. Withdrawal times were then grouped into broadly numerically equal-sized groups to allow for even comparison. Univariate and multivariate analyses of factors affect­ ing ADR were done using binomial regression to obtain the risk ratio and 95% CIs, and significance as measured by the p value. We deemed p values of less than 0·001 as significant. Relative risks were calculated as the ratio of two rates. Analyses were done with Stata version 14.

point reached during a procedure in which no polyps or cancer were detected, because detection of a malignancy or several polyps that would trigger a colonoscopy referral would likely halt the procedure at an earlier point than might have been reached if no pathology was identified). A mean negative extent score was calculated by assigning the extent of examination reached a number as follows: anus (0), rectum (1), rectosigmoid (2), sigmoid (3), de­ scending colon (4), splenic flexure (5), and transverse colon (6); expressed as a number less than or equal to 6. We calculated mean negative withdrawal time (ie, the time taken to withdraw the scope when no polyps or cancer were detected, indicating the time taken to examine the mucosa rather than the time taken to undertake polypectomy). Because withdrawal time varies depending on depth of insertion, the withdrawal time from the splenic flexure was used. Because numbers of procedures reaching exactly the splenic flexure were often too small (ie, at individual endoscopist level where few procedures were performed), the splenic flexure with­ drawal time could not be accurately measured. An estimated splenic flexure withdrawal time was calculated using a correct­ ion factor for the bowel segment reached in negative

There was no funding source for this study. The raw data were accessed by RB, CN, RGB, and CJR. The corre­ sponding author had full access to all the data in the study and had the final responsibility for the decision to submit for publication.

Results Between May 7, 2013, and May 6, 2014, 8582 participants (4471 male, 4111 female) underwent flexible sigmoidoscopy, of whom 8494 (99%) were aged 55 years. 88 individuals older than 55 years had opted in to BoSS screening and were excluded from the analyses. 238 (3%) of 8494 participants were excluded because their procedure was done by an endoscopist who had performed fewer than 30 procedures in total, leaving 8256 (4299 male, 3957 female) procedures in the final analyses. These procedures were performed by 44 endoscopists (mean procedure count 188, range 32–476). The overall ADR was 9·1% (95% CI 8·5–9·8), varying from 7·4% (6·2–8·9) to 11·0% (9·1–13·4) by screening centre (table 1). 390 cases led to a referral for colonoscopy (appendix p 1). Detailed indications for, and out­ comes from, these colonoscopies were not known. The ADR was 6·6% (95% CI 5·9–7·4) in women and 11·5% (10·6–12·5) in men (p<0·0001; table 1); therefore, the population risk of adenomas was 0·115 (11·5%) for men and 0·067 (6·7%) for women. Most endoscopists performed within the expected ranges (ie, falling within the 99·8% funnel lines; appendix p 2); one endosco­ pist had a notably low ADR despite a high volume of procedures. The percentage of negative procedures in which the splenic flexure was reached varied from 74% in centre F to 44% in centre B (p<0·0001; table 2). The mean extent score varied from 4·97 in centre F to 4·35 in centre E. The lower extent score in centre E resulted from a large number of negative procedures reaching the sigmoid colon. The mean negative withdrawal time from the splenic flexure (which can be used as a measure of inspection time) varied from 2·34 min to 4·76 min between centres (p<0·0001; appendix p 1). The mean estimated splenic flexure withdrawal time varied between centres from 2·30 min to 4·13 min (p<0·0001). The estimated splenic flexure withdrawal time for the 44 endoscopists was 1·85–5·04 min (mean 3·18 min [SD 3·2]; appendix p 2).

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

Articles

Women Procedures

Anus, rectum, or rectosigmoid Sigmoid colon

24

Men Moderate or severe Mean discomfort (%) insertion time, min (SD) 7 (29%)

2·8 (1·5)

Procedures

All Moderate or severe Mean insertion discomfort (%) time, min (SD)

21

3 (14%)

1·8 (1·0)

Procedures

45

Moderate or severe Mean insertion discomfort (%) time, min (SD) 10 (22%)

2·4 (1·4)

348

83 (24%)

4·9 (3·9)

176

22 (13%)

3·9 (3·0)

524

105 (20%)

4·6 (3·7)

Descending colon

1265

145 (11%)

4·7 (3·4)

950

58 (6%)

4·0 (2·2)

2215

203 (9%)

4·4 (3·0)

Splenic flexure

1413

109 (8%)

4·7 (3·0)

1463

47 (3%)

4·0 (3·1)

2876

156 (5%)

4·3 (3·1)

300

11 (4%)

5·2 (4·3)

556

5 (1%)

4·2 (3·5)

856

16 (2%)

4·5 (3·8)

3350

355 (11%)

4·8 (3·4)

3166

135 (4%)

4·0 (3·0)

6516

490 (8%)

4·4 (3·2)

Transverse colon All

Insertion time was calculated by subtracting the withdrawal time from the total procedure time. It is not known if pathology detected was biopsied or removed during insertion or withdrawal.

Table 5: Moderate or severe discomfort in negative procedures by extent of examination

ADR increased from 8·2% (95% CI 7·0–9·5) for endoscopists with the shortest estimated splenic flexure withdrawal time (<2·5 min) to 9·9% (3·25–3·99 min) and 10·0% (≥4 min) in the groups with the longest estimated time (ptrend=0·021; table 3). There was little dif­ference between the two groups with the longest withdrawal times (table 3). In univariate analysis, male sex had a relative risk (RR) of adenoma detection of 1·73 (95% CI 1·50–2·00; p<0·0001; table 4). The risk of adenoma detection decreased with lesser extent of examination (relative to the splenic flexure as the baseline) as would be expected, being particularly low (0·54, 95% CI 0·39–0·74) if only the sigmoid colon was reached. There is no evidence that reaching the transverse colon increases the detection rate above that of the splenic flexure (RR 0·89, 95% CI 0·73–1·10). Similar results were seen in multivariate models. Reaching the splenic flexure depended on bowel prep­ aration and sex (appendix p 1). For men with good bowel preparation, 1335 (70%) of 1904 negative procedures reached the splenic flexure, falling to 68 (43%) of 158 in men with poor bowel preparation (p<0·0001). In women these proportions were lower: the splenic flexure was reached in 1218 (56%) of 2162 negative procedures in women with good bowel preparation and in 65 (31%) of 207 in women with poor bowel preparation (p<0·0001). The extent of examination was influenced by procedural discomfort, and procedures in which a deeper extent of examination (ie, to the splenic flexure or transverse colon) was reached were rarely those for which the patient reported discomfort (table 5). Of 856 proced­ures in which the transverse colon was reached, only 16 (2%) patients had moderate or severe discomfort. By contrast, in 524 procedures in which only the sigmoid colon was reached, 105 (20%) patients had moder­ ate or severe discomfort (p<0·0001). For all procedures irrespective of depth of insertion, moder­ate or se­vere discomfort was reported by women in 355 (11%) of 3350 procedures and by men in 135 (4%) of 3166 procedures. The probability of deeper intubation to the splen­ic flex­ ure declined with poor bowel preparation and in­creasing

gesting that bowel procedural discomfort (table 6), sug­ preparation was an important factor in achieving a higher ADR. The percentage of patients with good bowel preparation varied between endoscopists from 40% to 80%, and between centres from 55% to 70% (appendix p 1). Between centres, the mean bowel prepara­ tion score varied from 1·49 to 1·69 (p<0·0001). When considering bowel preparation score among endo­scopists rather than sites, the mean score was 1·57 (SD 0·12, range 1·26–1·79).

Discussion In our retrospective analysis of data from the BoSS flexible sigmoidoscopy screening programme for people aged 55 years in England, the overall ADR was 9·1%— lower than that reported in the previous trial of flexible sigmoidoscopy screening7 and similar to that of the exploratory project,8 but substantially higher than the benchmark of 6·8%. Our findings show a signifi­ cant difference between adenoma detection in men and women. ADR varied substantially between endoscopists, some of whom had very low detection rates. We have demonstrated the relation between depth of insertion and ADR, and between withdrawal time and ADR. Quality of bowel preparation was also key to ensuring that ADR was optimised. The maximum ADR was achieved in men in whom the extent of the examination reached the splenic flexure, and withdrawal time from negative procedures was 3·5–4·0 min. The difference in ADR between men and women is expected, and has been shown in other flexible sig­ moidoscopy screening studies (16·1% in men vs 8·1% in women in the UK Flexible Sigmoidoscopy Screening Trial;7 22% vs 14% [all neoplasia] in the NORCCAP study from Norway,9 and 16·9% vs 12·6% in the SCORE trial from Italy10). This difference between men and women has also been shown in the BCSP faecal occult blood test screening programme (43% vs 29%).2 Because ADR is one of the major benchmarks for performance in screening programmes, this difference between men and women must be taken into account when considering

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

5

Articles

Univariate RR (95% CI)

Multivariate p value

RR (95% CI)

p value

Sex Men

1 (ref)

Women

0·80 (0·76–0·84)

·· <0·0001

1 (ref) 0·83 (0·80–0·87)

·· <0·0001

Bowel preparation Good

1 (ref)

Adequate

0·80 (0·76–0·84)

<0·0001

··

0·79 (0·76–0·83)

1 (ref)

<0·0001

··

Poor

0·58 (0·51–0·67)

<0·0001

0·58 (0·51–0·67)

<0·0001

Discomfort None

1 (ref)

Minimal

0·92 (0·88–0·96)

<0·0001

··

0·96 (0·92–1·00)

1 (ref)

0·055

··

Mild

0·77 (0·72–0·83)

<0·0001

0·82 (0·76–0·88)

<0·0001

Moderate or severe

0·55 (0·49–0·63)

<0·0001

0·58 (0·51–0·66)

<0·0001

RR=relative risk.

Table 6: Binomial regression of risk of reaching splenic flexure in negative procedures by sex, bowel preparation, and comfort

quality measures. A sex-adjusted ratio could be used when analysing the performance of individual endo­ scopists to ensure that an accurate representation of their screening practice is assessed. Analysis of our dataset also suggests a difference in the experience of the screen­ ing procedure, with higher levels of discomfort in women than in men and, as such, a balance between comfort and pathology detection needs to be considered. Our analysis suggests that extent of examination and withdrawal time are associated with ADR. Unlike col­ onoscopy screening, in which there is only one point of complete examination (caecal intubation), there is no defined endpoint in flexible sigmoidoscopy screening. In BoSS, the recommendation is to examine the rectum and lower sigmoid colon and not to go past the splenic flexure, with the comfort of participants during the procedure being paramount; however, intubation to the transverse colon is also acceptable.13 Early data on flexible sigmoid­ oscopy screening showed reductions in the develop­ ment of colorectal cancer after proctosig­moid­oscopy with clearance of adenomas, which suggests that reaching beyond the sigmoid was not necessary.14 In the UK Flexible Sigmoidoscopy Screening Trial, a 60 cm sigmoidoscope was inserted “as far as could be achieved without causing undue pain or distress (normally to the junction of the sigmoid and descending colon)”.7 The current screening programme largely uses colonoscopes inserted to the acceptable depth. Our data show no difference in insertion time between procedures reaching the sigmoid only and those reaching the splenic flexure or transverse colon, suggesting that, in practice, insertion time is constant irrespective of extent of examination. It should be noted that the accuracy of depth of insertion is not reliable; studies have shown correct identification of extent of examination in 50–75% of cases.15,16 If we assume that the reported extent of examination is accurate, our data are 6

consistent with the concept that an acceptable depth is the extent of examination that is reached within a reasonable time without causing excessive discomfort. In general, easier procedures continue to the splenic flexure or transverse colon and more difficult procedures to the sigmoid colon only. Our findings show that women report a much higher frequency of difficult procedures, evidenced by longer insertion times and greater discomfort scores. ADR was higher when extent of examination was greater, with optimal depth insertion to the splenic flexure. Lower ADR was seen with extents of examination distal to the splenic flexure, but there was no additional gain if the extent of examination reached the transverse colon. Interpretation of these results is complex because only a small subset of patients had an investigation that reached the transverse colon, and the reporting of extent of examination might lack accuracy. Recording the use of a scope imager was not routine at the time of data collection, and more detailed analyses comparing extents of exam­ ination in procedures with and without a scope imager have not been done. Extent of examination needs to be coupled with adequate time spent inspecting the mucosa on withdrawal, and is only appropriate if comfort levels are maintained. We suggest an aim of reaching the splenic flexure to maximise ADR, a marginally greater extent of examination than suggested by other studies.9,14,17 This extent of examination should only be reached when comfort is maintained. The current data suggest that, if scopes can be comfortably inserted to the splenic flexure, this target should be achievable. Persistent (probably unsuccessful) attempts at deep insertion are likely to cause discomfort, and extent of examination should be minimised in these procedures. Our findings show that the extent of examination in negative procedures correlates to reporting of moderate or severe discomfort; when such discomfort is reported the procedure is stopped at a lesser extent of examina­tion, presumably because of the discomfort. This course of action is reflected in BoSS guidance for endoscopists.13 In this study, the transverse colon was reached in 856 (10%) procedures, with most participants not reporting moderate or severe discomfort, suggesting that these procedures might have reached the transverse colon precisely because no discomfort was reported. How to minimise discomfort in all procedures needs to be con­ sidered, particularly in women, in whom fewer adenomas are detected and in whom the procedure is less well tolerated. One Austrian study of colonoscopic colorectal cancer screening suggested that a similar number needed to screen to detect advanced adenomas was reached in men aged 50–54 years and in women a decade older at 60–64 years.18 Screening women at a slightly older age (eg, 57 years or 58 years) might be appropriate to optimise the use of BoSS in both men and women. In colonoscopy practice, a withdrawal time of at least 6 min has been associated with improved ADR,19 and

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

Articles

extending withdrawal time to 10 min in patients with positive faecal occult blood test screening further improves ADR.20 No recommendations for withdrawal time exist in flexible sigmoidoscopy, in part because of the vary­ ing extent of examination. We have created a measure of withdrawal time, corrected for depth of insertion (the estimated splenic flexure withdrawal time), to assess its effect on ADR. These data suggest that a splenic flexure withdrawal time of 3·25–4·00 min is necessary, although more work is needed to establish whether certain seg­ ments of the bowel require longer inspection times than others, rather than an overall withdrawal time. There is an assumption with these data that withdrawal was at a steady speed, and that all segments of the bowel re­ quired the same pace of withdrawal to perform an ac­ curate assessment of the mucosa. Therefore, the time of 3·25 min to withdraw should be considered the minimum standard, with 3·5–4·0 min being the aspirational standard. The estimated splenic flexure withdrawal time is a tool that could be used to review individual endo­ scopist practice, even at fairly low numbers. In this study, 24 endoscopists had withdrawal times of less than 3·25 min—the level below which ADR dropped significantly—suggesting room to improve ADR by slower examination of the distal colon. By increasing withdrawal time to 4 min, there could be a 20% increase in ADR at endoscopist level, with an overall ADR increase of 10%; in this study, overall ADR would increase from 9·1% to 10·0%. The estimated splenic flexure withdrawal time has limitations; it is a novel concept, not yet tested for validity, relies on the assumption that withdrawal is done in a smooth manner at a constant rate, and does not account for periods of time spent inspecting but not moving within the bowel. However, this measurement tool provides a starting point for assessment of with­ drawal times in screening sig­moidoscopy, and warrants further work. Quality of bowel preparation is important because it affects extent of examination reached,21 and is likely to be one of the most important factors in detecting adenomas (because poor preparation will mean reduced ability to accurately inspect the colonic mucosa). The effect on ADR in colonoscopy is unclear, with no difference in ADR between good and fair preparation in a large US review,22 and a possible reduction in proximal ADR in cases with poor preparation.22–24 UK data on bowel preparation in flexible sigmoidoscopy screening showed improved ADR with better quality bowel preparation.25 That study also suggested that endoscopists with higher ADRs are more likely to be critical of bowel preparation quality.25 We suggest that a bowel preparation score should be calculated for all endoscopists and centres, and a minimum standard score of 1·4 and a recommended score of at least 1·5 should be introduced. Bowel preparation scores varied between centres, and might reflect list timings and preparation instructions (the preparation was a phosphate enema at all sites, but the

Panel: Suggested recommendations for the Bowel Scope Screening (BoSS) programme • The adenoma detection rate (ADR) benchmark of 6·8% should be reviewed. An aspirational ADR should be set at 9%. Funnel plots will demonstrate endoscopists performing outside the expected levels. • A sex-adjusted ADR can be used to ensure case mix is taken into account when reviewing ADR. • Endoscopists may aim to reach the splenic flexure during screening sigmoidoscopy, but within the constraints of patient comfort. Greater depth of insertion should not be at the expense of a positive patient experience. • Withdrawal time from the splenic flexure should not be less than the minimum of 3·25 min, and ideally 3·5–4·0 min in negative examinations. Where necessary, estimated splenic flexure withdrawal time can be used to ensure appropriate time is being spent on mucosal inspection. • Insertion time should be limited to 4·0 min. • Good quality bowel preparation should be achieved. Centres should aim for 70% of procedures to have good bowel preparation with a minimum of 60%. • The patient experience of undergoing a BoSS procedure might affect future acceptance of other screening modalities—ie, currently faecal occult blood testing, which is soon to change to faecal immunochemical test. The uptake of screening at the next screening round should be compared between individuals who have and have not undergone BoSS to assess the effect on overall screening. • Further studies are required on how to minimise discomfort in women, the type and progression potential of adenomas detected in women, and whether the age at invitation for women should be raised to increase adenoma detection yield and improve the benefit-to-harm ratio.

instructions sent out were developed by each individual site and so could vary from site to site), and should be considered for review in the future. We also acknowledge that reporting of bowel preparation is subjective, and such assessments are made by the endoscopist rather than the nurse or the specialist screening practitioner, and in the future, more objective measures might be more appropriate. We used funnel plots to demonstrate individual endo­ scopist performance. Visual aids allowing comparison of performance are increasingly used because they allow for varying numbers of procedures, are simple to understand, and prevent the use of league tables to assess per­ formance.26–28 The plots of these data show that most endo­ scopists performed within an acceptable ADR range, and highlight the few that might need further investigation to improve performance levels. We also used an expected-toobserved ratio that allows endoscopists to evaluate their performance against predicted outcomes. This method of assessment has been increasingly used to assess

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

7

Articles

per­ formance in various aspects of health-care delivery, often in terms of hospital mortality.28–30 Our findings highlight a range of factors that affect ADR. Other sigmoidoscopy screening programmes and studies have shown similar variations.31–33 The UK Flexible Sigmoidoscopy Screening Trial33 reported increased ADR in men compared with women, and reported other endoscopist-related factors affecting ADR, but did not examine procedure-related factors such as withdrawal time or extent of examination. We acknowledge that our study included a limited number of screening centres, and these centres were selected on a willingness to participate, and a performance history and capacity to deliver the pilot. Therefore, these sites might not be completely representative of all screening centres, and performance of flexible sigmoidoscopy might be height­ ened because of an observer effect during this pilot phase. However, the six sites reflect the range of socio­economic groups in the UK and, as such, these findings should be generalisable at the patient level. ADR continues to be the most widely recognised key performance indicator for colorectal cancer screening and, as such, should remain the main method to assess quality. However, it is important to look beyond headline ADR to understand what other factors might be affecting quality. Understanding these factors allows for inter­ ventions to improve the quality of flexible sig­moidoscopy and thus improve the quality and yield of a screening programme (panel). Contributors CJR and RB developed the concept for the project. RB coordinated the data collection, liaison with units, and writing of the report. CN analysed the BCSS database, compiled the data, and contributed to analysis and writing. RGB developed the statistical analysis and performed the analyses. BPS, JS, RT, AMV, and WG contributed additional data from units, and contributed to development and review of the report. Declaration of interests We declare no competing interests. Acknowledgments No funding was received for this work. We thank the staff at the six pilot screening centres of St Mark’s Hospital (London), South of Tyne and Wear, West Kent and Medway, Wolverhampton, Norwich, and Surrey. Special acknowledgment goes to Richard Loke (West Kent and Medway Screening Centre), who passed away during the writing of this report. References 1 Rees CJ, Bevan R. The National Health Service Bowel Cancer Screening Program: the early years. Expert Rev Gastroenterol Hepatol 2013; 7: 421–37. 2 Logan RF, Patnick J, Nickerson C, Coleman L, Rutter MD, von Wagner C. Outcomes of the Bowel Cancer Screening Programme (BCSP) in England after the first 1 million tests. Gut 2012; 61: 1439–46. 3 Mayor S. UK committee recommends flexible sigmoidoscopy to screen for bowel cancer. BMJ 2011; 342: d2325. 4 Leslie A, Carey FA, Pratt NR, Steele RJ. The colorectal adenoma-carcinoma sequence. Br J Surg 2002; 89: 845–60. 5 Atkin WS, Edwards R, Kralj-Hans I, et al. Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010; 375: 1624–33. 6 NHS Cancer Screening Programmes. Letter to all Screening Centre Directors. 2015.

8

7

8

9

10 11 12

13

14 15 16 17

18 19 20

21

22 23

24

25

26

Atkin WS, Cook CF, Cuzick J, Edwards R, Northover JM, Wardle J. Single flexible sigmoidoscopy screening to prevent colorectal cancer: baseline findings of a UK multicentre randomised trial. Lancet 2002; 359: 1291–300. Bevan R, Rubin G, Sofianopoulou E, Patnick J, Rees CJ. Implementing a national flexible sigmoidoscopy screening program: results of the English early pilot. Endoscopy 2015; 47: 225–31. Gondal G, Grotmol T, Hofstad B, Bretthauer M, Eide TJ, Hoff G. The Norwegian Colorectal Cancer Prevention (NORCCAP) Screening Study: baseline findings and implementations for clinical work-up in age groups 50-64 years. Scand J Gastroenterol 2003; 38: 635–42. Segnan N, Senore C, Andreoni B, et al. Baseline findings of the Italian multicenter randomized controlled trial of “once-only sigmoidoscopy” - SCORE. J Natl Cancer Inst 2002; 94: 1763–72. Lee TW, Rutter MD, Blanks RG, et al. Colonoscopy quality measures: experience from the NHS Bowel Cancer Screening Programme. Gut 2012; 61: 1050–57. Gavin DR, Valori RM, Anderson JT, Donnelly MT, Williams JG, Swarbrick ET. The national colonoscopy audit: a nationwide assessment of the quality and safety of colonoscopy in the UK. Gut 2013; 62: 242–49. Public Health England. Bowel scope screening: standard operating procedures. https://www.gov.uk/government/publications/bowelcancer-screening-bowel-scope-standard-operating-procedures/ bowel-scope-screening-standard-operating-procedures (accessed Nov 26, 2018). Atkin WS, Cuzick J, Northover JM, Whynes DK. Prevention of colorectal cancer by once-only sigmoidoscopy. Lancet 1993; 341: 736–40. Maslekar S, Waudby P, Avery G, Monson JR, Duthie GS. Quality assurance in flexible sigmoidoscopy: medical and nonmedical endoscopists. Surg Endosc 2010; 24: 89–93. Adam IJ, Ali Z, Shorthouse AJ. How accurate is the endoscopist’s assessment of visualization of the left colon seen at flexible sigmoidoscopy? Colorectal Dis 2000; 2: 41–44. Segnan N, Patnick J, von Karsa L. European guidelines for quality assurance in colorectal cancer screening and diagnosis. 2010. http://www.kolorektum.cz/res/file/guidelines/CRC-screeningguidelines-EC-2011-02-03.pdf (accessed Nov 26, 2018). Ferlitsch M, Reinhart K, Pramhas S, et al. Sex-specific prevalence of adenomas, advanced adenomas, and colorectal cancer in individuals undergoing screening colonoscopy. JAMA 2011; 306: 1352–58. Barclay RL, Vicari JJ, Doughty AS, F JJ, Greenlaw RL. Colonoscopic withdrawal times and adenoma detection during screening colonoscopy. N Engl J Med 2006; 355: 2533–41. Lee TW, Blanks RG, Rees CJ, et al. Longer mean colonoscopy withdrawal time is associated with increased adenoma detection: evidence from the Bowel Cancer Screening Programme in England. Endoscopy 2013; 45: 20–26. Froehlich F, Weitlisbach V, Gonvers JJ, Burnand B, Vader JP. Impact of colonic cleansing on quality and diagnostic yield of colonoscopy: the European Panel of Appropriateness of Gastrointestinal Endoscopy European multicenter study. Gastrointest Endosc 2005; 61: 378–84. Harewood GC, Sharma VK, de Garmo P. Impact of colonoscopy preparation quality on detection of suspected colonic neoplasia. Gastrointest Endosc 2003; 58: 76–79. Anderson JC, Butterly LF, Robinson CM, Goodrich M, Weiss JE. Impact of fair bowel preparation quality on adenoma and serrated polyp detection: data from the New Hampshire colonoscopy registry by using a standardized preparation-quality rating. Gastrointest Endosc 2014; 80: 463–70. Clark BT, Rustagi T, Laine L. What level of bowel prep quality requires early repeat colonoscopy: systematic review and meta-analysis of the impact of preparation quality on adenoma detection rate. Am J Gastroenterol 2014; 109: 1714–23. Thomas-Gibson S, Rogers P, Cooper S, et al. Judgement of the quality of bowel preparation at screening flexible sigmoidoscopy is associated with variability in adenoma detection rates. Endoscopy 2006; 38: 456–60. Spiegelhalter DJ. Funnel plots for comparing institutional performance. Statist Med 2005; 24: 1185–202.

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

Articles

27 Neuburger J, Cromwell DA, Hutchings A, Black N, van der Meulen JH. Funnel plots for comparing provider performance based on patient-reported outcome measures. BMJ Qual Saf 2011; 20: 1020–26. 28 Noyez L. Control charts, Cusum techniques and funnel plots. A review of methods for monitoring performance in healthcare. Interact Cardiovasc Thorac Surg 2009; 9: 494–99. 29 Metnitz PG, Lang T, Vesely H, Valentin A, Le Gall JR. Ratios of observed to expected mortality are affected by differences in case mix and quality of care. Intensive Care Med 2000; 26: 1466–72. 30 Cevasco M, Ashley SW. Quality measurement and improvement in general surgery. Perm J 2011; 15: 48–53.

31 Pinsky PF, Schoen R, Weissfeld JL, Kramer B, Hayes RB, Yokochi L. Variability in flexible sigmoidoscopy performance among examiners in a screening trial. Clin Gastroenterol Hepatol 2005; 3: 792–97. 32 Fracchia M, Senore C, Armaroli P, et al. Assessment of the multiple components of the variablity in the adenoma detection rate in sigmoidoscopy screening, and lessons for training. Endoscopy 2010; 42: 448–55. 33 Atkin W, Rogers P, Cardwell C, et al. Wide variation in adenoma detection rates at screening flexible sigmoidoscopy. Gastroenterol 2004; 126: 1247–56.

www.thelancet.com/gastrohep Published online January 14, 2019 http://dx.doi.org/10.1016/S2468-1253(18)30387-X

9