Article Text


Original article
Comparison of colonoscopic performance between medical and nurse endoscopists: a non-inferiority randomised controlled study in Asia
  1. Aric J Hui1,
  2. James Y Lau2,
  3. Phyllis P Y Lam3,
  4. Alman O M Chui3,
  5. Alice S H Fan3,
  6. Thomas Y T Lam2,
  7. Yee-kit Tse2,
  8. Raymond S Y Tang2,
  9. Siew C Ng2,
  10. Justin C Y Wu2,
  11. Jessica Y L Ching2,
  12. Martin C S Wong4,
  13. Francis K L Chan2,
  14. Joseph Sung2
  1. 1Department of Medicine, Alice Ho Miu Ling Nethersole Hospital, Hong Kong, Hong Kong
  2. 2Institute of Digestive Disease, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  3. 3Combined Endoscopy Unit, Alice Ho Miu Ling Nethersole Hospital, Hong Kong, Hong Kong
  4. 4School of Public Health and Primary Care, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  1. Correspondence to Dr Aric Josun Hui, Department of Medicine, Block J, 6/F, Alice Ho Miu Ling Nethersole Hospital, 11 Chuen On Road, Tai Po, Hong Kong, Hong Kong; arichui{at}


Objective To test the hypothesis that trained nurse endoscopists are not inferior to medical endoscopists in finding adenomas during colonoscopy.

Design This is a prospective, randomised, single-blind, non-inferiority study comparing nurses with medical endoscopists in performing screening colonoscopy. The nurse endoscopists had been trained according to the British Joint Advisory Group on GI Endoscopy curriculum and had completed at least 140 colonoscopic procedures prior to the study. The primary endpoint was the adenoma detection rate. Secondary endpoints included the caecal intubation rate, intubation time, complication rate, patient pain and satisfaction scores.

Results We enrolled and analysed a total of 731 patients over a 15-month period. At least one adenoma was found in 159 (43.8%) of 363 patients by nurse endoscopists and 120 (32.7%) of 367 patients by medical endoscopists and a proportion difference of +11.1% compared with the medical endoscopists (95% CI 4.1% to 18.1%). The withdrawal time was, however, significantly longer among nurses (998 vs 575 s, p<0.001). After adjusting for differences in a regression analysis, colonoscopy by nurses was associated with a lower adenoma detection rate (OR 0.475: 95% CI 0.311 to 0.725). Nurse endoscopists had a lower caecal intubation rate (97.3% vs 100%), received better pain and satisfaction scores and had a high rate of patient acceptance.

Conclusions In this pragmatic trial, nurses can perform screening colonoscopy but require a longer procedural time to achieve a comparable adenoma detection rate as medical endoscopists.

Trial registration number NCT01923155.

  • colonoscopy
  • nurse endoscopist
  • colorectal cancer screening

Statistics from

Significance of this study

What is already known about this subject?

  • Nurse colonoscopists are well established in the UK and many European countries but are not recognised in America and Asia.

  • There are preliminary data showing that nurses can acquire colonoscopic skills and perform colonoscopy as safely and accurately as medical endoscopists.

  • Prospective surveys showed that nurse endoscopists were capable of performing colonoscopy up to international standards.

What are the new findings?

  • Endoscopy nurses who have received a structured training programme and performed at least 140 colonoscopies under supervision can perform colonoscopy up to international standards.

  • Nurses require a longer procedural time to achieve a comparable adenoma detection rate as medical endoscopists.

  • Nurses had a high acceptance rate by patients.

  • Medical endoscopists can provide adequate supervision to nurse endoscopists in a parallel room design.

How might it impact on clinical practice in the foreseeable future?

  • Nurses can be trained to safely perform screening colonoscopy, but they may take longer to achieve a comparable adenoma detection rate and hence a longer withdrawal time. This has implication on resource allocation.


Nurse endoscopists were introduced to the USA in 1977 to meet the increasing demand for flexible sigmoidoscopy.1 In a randomised study, Maule et al2 showed that nurses can perform flexible sigmoidoscopy as safely as performed by an experienced gastroenterologist. In 1994, the British Society of Gastroenterology (BSG) endorsed the development of nurse endoscopists in their Working Party Report but initially restricted them to gastroscopies and flexible sigmoidoscopies.3

Interestingly, the subsequent development of nurse endoscopists diverged significantly between the USA and the UK; an update by the BSG Working Party in 2005 stated that nurse endoscopists were an integral part of national GI service with over 200 nurses providing endoscopic service including colonoscopy in the UK.4 In contrast, the American Society of Gastrointestinal Endoscopy (ASGE) guidelines stated that unlike flexible sigmoidoscopy, which required less training to achieve competency, required no sedation and was inherently safer, there were insufficient data to support non-physicians performing colonoscopy.5

A series of papers studying the clinical effectiveness of nurse colonoscopists have become available. In one study, nurse trainees were found to be able to acquire colonoscopy skills at least as quickly as a physician trainee and achieved a caecal intubation rate of over 90% during the first 150 procedures.6 Limoges-Gonzalez et al7 performed a prospective, randomised pilot study which showed that a nurse practitioner (who had both a higher adenoma detection rate and patient satisfaction score) can perform colonoscopy as safely, accurately and satisfactorily as a gastroenterologist. Van Putten et al8 prospectively collected the colonoscopic performance data of 10 nurse endoscopists performing 100 colonoscopies each and concluded that nurse endoscopists can perform colonoscopies to internationally recognised standards.

In this study, we hypothesised that trained nurse endoscopists are not inferior to medical endoscopist in performing colonoscopy as measured by rate of adenoma detection.


Qualification of nurse endoscopists

The three nurse endoscopists involved in this study each had over 10 years’ experience as endoscopy assistants and were taking care of patients during the peri-procedural period. Furthermore, they have completed a structured training programme over a 1-year period involving procedural observation, hands-on experience and a documented assessment of proficiency based on the Joint Advisory Group (JAG) on Gastrointestinal Endoscopy requirements.9 The three nurse endoscopists had respectively performed 160, 200 and 230 colonoscopies under supervision prior to the current study. The ASGE training guidelines recommend that approximately 140 procedures are required for an endoscopist to gain an acceptable level of proficiency.10 ,11 Prior to this trial, nurse endoscopists were allowed to perform selected polypectomies under direct supervision by experienced endoscopists.

Qualifications of medical endoscopists

All five endoscopists involved in this study had received their respective fellowships (from USA, UK or Hong Kong) and had been performing at least 300 colonoscopy procedures annually for at least 2 years.

Study population

Asymptomatic subjects between the ages of 50 and 70 years were recruited with full informed consents to trial participation. These subjects volunteered to participate in a local colorectal cancer (CRC) screening programme and were scheduled to receive ambulatory colonoscopy at the Combined Endoscopy Unit of the Alice Ho Miu Ling Nethersole Hospital. Exclusion criteria included subjects who had undergone colonoscopy in the past 5 years, subjects with prior colorectal surgery, subjects with a personal history of inflammatory bowel disease, colonic adenoma or CRC, subjects with a family history of familial adenomatous polyposis syndrome or familial non-polyposis syndromes, pregnant or lactating women and subjects unable to provide consent.

Study design

This is a prospective, randomised, single-blind, one-sided, non-inferiority study. This study was approved by the Joint Chinese University of Hong Kong and New Territories East Cluster Clinical Research Ethics Committee.

Subjects were randomised on a 1:1 basis to receive colonoscopy either by nurse or by medical endoscopists. A randomisation sequence was generated using a computer programme in a concealed allocation fashion in block sizes of 4 and stored in sequentially numbered, sealed, opaque envelopes. Subjects were randomly allocated by opening the next numbered sealed envelope. All subjects were blinded to the identity of the endoscopists.

All subjects received polyethylene glycol (Klean-Prep, Helsinn Birex Pharmaceuticals, Ireland) as bowel preparation. Subjects received a standard sedation regime consisting of midazolam 2.5 mg (Groupe Panpharma, France) and pethidine 25 mg (Martindale Pharmaceuticals, UK) administered intravenously. Further doses of midazolam and pethidine were given according to the subject's level of discomfort. Air insufflation was used for all the colonoscopies in this study.

Nurse endoscopists would perform the procedure independently without direct observation. A medical endoscopist would usually be performing colonoscopy in a parallel room design. The nurse endoscopists were required to call for assistance for any therapeutic manoeuvres such as polypectomy or if they encountered any difficulties. All polypectomies would be performed under the direct observation of medical endoscopists who could take over as required. We stipulated a minimum withdrawal time of 6 min.12 The withdrawal time was measured using a stop watch from when the endoscopist started to withdraw the endoscope (after reaching either the terminal ileum or caecum) to when the anal verge was reached. We collected data on the quality of bowel preparation (satisfactory bowel preparation is defined as either good or excellent), colonoscopy findings, caecal intubation time, terminal ileum intubation time, withdrawal times and any immediate complications.

After the subjects had fully recovered from sedation and were able to perform the 100–7 subtraction test correctly for three times, they were asked to rate their satisfaction in terms of bedside manner, technique, level of explanation and overall experience on a scale of 1 (poor) to 5 (excellent). This is based on a validated patient satisfaction questionnaire for GI endoscopy (modified GHAA-9).13 Moreover, subjects were asked to rate both the intraprocedural and postprocedural pain on a scale of 0 (no pain) to 10 (severe pain).

All study subjects were provided with a telephone hotline after the colonoscopy procedure. Subjects were educated on the symptoms of potential complications of colonoscopy including abdominal pain and per rectal bleeding and instructed to contact Unit staff should they develop such symptoms. All the polyp histology reports were reviewed by co-investigators to calculate the adenoma detection rate.

The primary endpoint for this study is the colon adenoma detection rate, which was defined as the number of patients found to have at least one adenoma out of the total number of complete colonoscopies performed. Secondary endpoints include the caecal intubation rate (defined as reaching the caecum without any assistance), intubation time, total procedural withdrawal time, complication rate, subject's pain score and satisfaction score.

Sample size calculation and statistical analysis

Our group published a paper on colorectal screening in 2003 in which the adenoma detection rate was 29%.14 If nurse endoscopists could achieve an adenoma detection rate of at least 20%, they would still meet international standards. Therefore, we based our calculations on an adenoma detection rate of 29% with a one-sided non-inferiority margin of 10%. A sample size of 646 evaluable subjects randomised in a 1:1 ratio (323 subjects in each arm) would provide 80% power to demonstrate non-inferiority of colonoscopic performance of nurse endoscopists compared with that of medical endoscopists, using a two-sided 95% CI. The CI was calculated according to the Wald method without continuity correction. Furthermore, category variables were compared using χ2 or Fisher's exact tests, and continuous variables were compared using Student t test or Mann–Whitney U test as appropriate. Poisson log-linear regression model was used for the analysis of colonic polyp findings. The analyses were performed using SPSS V.20 for Windows (SPSS, Chicago, Illinois, USA). Two-tailed p value of <0.05 was considered significant. Results were expressed as mean and SD or median and IQR, if data were not normally distributed for continuous data, while categorical data were presented as frequency and percentage.

We performed further statistical analysis to examine the effect of the withdrawal time on the adenoma detection rate. Specifically, a multiple logistic regression was used to estimate adjusted OR and its 95% CI with withdrawal time, age, sex, endoscopists group, level of bowel preparation, diabetes, family history of CRC and body mass index entered as covariates, the presence or absence of one or more adenomas was considered response variable. The optimal model was selected based on Akaike Information Criterion (AIC) values (ie, the optimal model had the smallest AIC value). The SAS V.9.3 version (SAS Institute, Cary, North Carolina, USA) with macro %ic_logistic was used to compute the AIC.15


We randomised a total of 787 subjects over a 15-month period (March 2012–June 2013), but only 364 subjects in the nurse group and 367 subjects in the medical group were used for analysis due to protocol deviation. There were no significant differences in their basic demographics (table 1).

Table 1

Demographic data

The colonoscopy findings are tabulated in table 2. Significantly, more colonic polyps were identified in the nurse group compared with the medical endoscopist group (58.2% vs 43.3%, p<0.001). A detailed description of the colonic polyps is shown in online supplementary table S1. The higher polyp detection rate by the nurse group was mainly for tubular adenomas or serrated adenomas that were <5 mm and did not follow any clear pattern in terms of polyp location or morphology. There was no difference in the detection of advanced colonic neoplasia between the study arms. Both arms had a similar proportion of subjects with a good or excellent bowel preparation; however, significantly more subjects were found to have poor bowel preparation in the medical endoscopist group. There was only one incomplete colonoscopy encountered during the clinical trial; colonoscopy was attempted initially by a nurse and then a medical endoscopist.

Table 2

Colonoscopy findings (per patient analysis)

The study outcomes are tabulated in table 3. Looking at the primary endpoint, the nurse endoscopist group had a significantly higher adenoma detection rate per procedure (43.8% vs 32.7%) and was associated with a proportion difference of +11.1% (95% CI 4.1% to 18.1%). The lower limit of this CI was therefore higher than the pre-specified non-inferiority margin (ie, −10%). Subgroup analysis of patients with average risk for CRC16 ,17 (ie, excluded subjects with family history of CRC), the adenoma detection rate was 131/309 (42.4%) for the nurse group and 98/300 (32.7%) for the medical endoscopist group, respectively.

Table 3

Study results

The nurse group had a significantly longer caecal intubation time (380 vs 314 s, p<0.001) and withdrawal time (998 vs 575 s, p<0.001). Subgroup analysis of negative colonoscopies (ie, no polyps found) showed that the nurse group still had a significantly longer withdrawal time (818 vs 453 s, p<0.001). The nurse group had a slight but significantly lower caecal intubation rate (97.3% vs 100%, p=0.001) but a significantly higher terminal ileum intubation rate (92.3% vs 80.1%, p<0.001).

Results of multivariable regression analysis are shown in table 4. Factors affecting adenoma detection rate include age, sex, withdrawal time and quality of bowel preparation. After adjusting for these factors, the nurse group had a significantly lower adenoma detection rate with an adjusted OR of 0.47 (95% CI 0.31 to 0.71). Furthermore, we charted the adenoma detection rate against the withdrawal time for both study arms as shown in figure 1. There is a clear pattern that the nurse group required a significantly longer withdrawal time to achieve a comparable adenoma detection rate as medical endoscopists.

Table 4

Multivariable regression analysis on factors affecting adenoma detection rate (the best-fitted model based on the smallest AIC value)

Figure 1

Adenoma detection rate against the withdrawal time.

Both arms had a similar complication rate (9.3% vs 9.0%, p=0.899). The majority of complications were hypotension or hypoxia due to midazolam or pethidine which required interventions such as oxygen supplementation, intravenous fluid challenge or injection of flumazenil. There were no cases of perforation, cardiac or respiratory failure and there was no meaningful difference in post-polypectomy bleeding (0.5% vs 0.3%). One subject in the medical endoscopist group developed chest pain during the procedure, which completely resolved while in recovery.

The nurse group had significantly fewer subjects requiring extra sedatives/painkillers (11.5% vs 21.3%, p<0.001) and a lower post-procedural pain score (0.6 vs 0.8). Finally, the nurse group had better subject satisfaction scores with 98.9% of subjects willing to be re-scoped by the same endoscopist.

Nurse endoscopists did not require any assistance in 91.2% of procedures and performed over 90% of the polypectomies. The type of assistance required is shown in online supplementary table S2.


The main finding of this study is that nurse endoscopists had a higher adenoma detection rate per procedure but the rate becomes significantly lower than that of medical endoscopists after adjusting for the longer withdrawal time. Moreover, when plotting the adenoma detection rate against the withdrawal time, nurse endoscopists were shown to consistently require a longer withdrawal time to achieve a comparable adenoma detection rate as their medical counterparts. We believe this may be accounted for by the disparity in experience between the nurse and medical endoscopists. Our study suggests that the inferior adenoma detection rate of the nurse endoscopists in our study may be corrected by recommending a longer minimum withdrawal time. More importantly, nurses in our study had well exceeded international standards with a caecal intubation rate exceeding 95%, an adenoma detection rate exceeding 35% and a mean withdrawal time exceeding 6 min (see online supplementary table S3 summarises the international standards on colonoscopy performance).

The complication rates of the nurse endoscopists were comparable with those of the medical endoscopists (9.3% vs 9.0%); the relatively high complication rate in this study is due to the stringent reporting of even minor adverse events such as transient desaturation or hypotension after conscious sedation. Our nurse endoscopists also performed well in terms of patient's pain and satisfaction scores. As over 98% of patients were willing to have a repeat colonoscopy by the same endoscopist, our nurse endoscopists had a high level of patient acceptance.

The main limitations of this study stem from the highly operator-dependent nature of colonoscopy for which reliability studies to measure inter/intra-observer agreement are often not feasible. The small number of endoscopists limits the generalisability of the study findings. Furthermore, as this is a pragmatic trial designed to reflect real-life practice, endoscopists were given discretion on how much effort was required to clean the colonic mucosa and were allowed to finish the procedure when they were satisfied they had performed a thorough examination of the colon without any time constraints. This resulted in a difference in withdrawal times, which was an important confounding factor. Other limitations include information bias as the endoscopists could not be blinded to the study intervention, which may potentially influence their performance; this includes difficult-to-measure parameters such as the enthusiasm or beliefs of the endoscopists towards the study. As our study population is derived from a community-based CRC screening programme, our findings in regard to nurse endoscopist performance may not be generalised to other patient populations, specifically in-patients with severe co-morbidities or those requiring endoscopic therapy such as haemostasis. As such, we do not recommend nurse endoscopists to perform colonoscopy in high-risk patient populations.

There are two main implications from our study. First, nurses with proper training and supervision can competently perform screening colonoscopy. Proper training in this case means completion of well-established training programmes such as JAG programme. Adequate supervision, we believe, can be provided by medical endoscopists performing colonoscopy in a parallel room design as over 90% of colonoscopies performed by nurses in our study did not require any direct assistance. Second, particularly for relatively inexperienced nurse endoscopists, a longer withdrawal time should be allowed to ensure a comparable adenoma detection rate.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors AJH wrote the study protocol, implemented the study, analysed and interpreted the data, and wrote the manuscript draft. He is a guarantor. JYL provided training to the endoscopy nurses, implemented the study, interpreted the data and revised the manuscript draft. He is a guarantor. PPL, AOC, ASF and RST were responsible for implementation of the study. SC-N assisted in training the endoscopy nurses and implemented the study. TYL, YKT and JYC were responsible for acquisition of data, statistical analysis and data interpretation. JCW, MCW, FKC and JJS were responsible for critical revision of the manuscript and site supervision.

  • Competing interests None.

  • Ethics approval Joint Chinese University of Hong Kong and New Territories East Cluster Clinical Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional unpublished data has been shown to anyone.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.