Colorectal cancer is the third most commonly diagnosed cancer in the United States.1 Despite several advances in colorectal cancer screening, including fecal immunochemical testing and fecal DNA assays, colonoscopy remains the most frequently implemented method of colorectal cancer screening. Colonoscopy uniquely holds the potential to prevent colorectal cancer when adequate detection and removal of precancerous polyps is achieved. Access to colonoscopy for routine colorectal cancer screening widened in 2001 when Medicare expanded coverage to include colonoscopies.
Despite the overall successes of the colorectal cancer screening initiatives in the United States, much room remains for improvement. In 2015, only 62% of the eligible US population received either a fecal test or colonoscopy for screening.2 Limited access to colonoscopy can present a major barrier, and although some models demonstrate sufficient available resources for colorectal cancer screening, this does not take into account geographic distribution of colonoscopy capacity.3 The use of fecal testing for colorectal cancer screening can improve the ability of primary care providers to offer screening, but patients with positive fecal testing who ultimately require a colonoscopy may still face disparity in access. This has been demonstrated in rural settings, where the geographic density of gastroenterologists is far lower than in urban areas.4 In order to address this gap in access to endoscopic procedures, nonphysician endoscopists were trained as early as the 1970s to perform sigmoidoscopy.5-8 These initial studies suggested that following appropriate training, the skills of these nonphysician endoscopists were comparable with that of physician counterparts.9-13 Yet, guidelines published by the American Society for Gastrointestinal Endoscopy (ASGE) support the performance of flexible sigmoidoscopy by clinicians other than physicians, citing a paucity of evidence to support colonoscopy by these clinicians.14,15
At the St. Louis (Mo.) VA Medical Center (VASTLHCS), physician assistants (PAs) have been performing diagnostic and colorectal cancer screening colonoscopies for almost 2 decades. This study was conducted to examine colonoscopy metrics and quality indicators among supervised PAs compared with attending gastroenterologists and gastroenterology (GI) fellows under supervision. Based on our clinical experiences, we hypothesized that this study would demonstrate PAs to be competent providers of screening colonoscopy using established quality metrics. In turn, this study held the potential to highlight the PA endoscopist model as a potential means of closing colonoscopy provider gaps, particularly if adapted in underserved areas with recognized shortages of trained gastroenterologists.
A total of 743 consecutive patients undergoing routine, average-risk screening colonoscopy at VASTLHCS in a 12-month period (July 2015 to June 2016) were identified. Exclusion criteria included the performance of colonoscopy for reasons other than first time average-risk screening (for example, family history or symptoms of colorectal cancer), and inadequate bowel prep, which was described qualitatively as fair or poor, or a Boston Bowel Prep Score (BBPS) of 1 or less in any colon segment and/or a total BBPS of less than 6. Colonoscopy procedures were performed by seven board certified or eligible gastroenterologists, five PAs, and 32 GI fellows from two academic affiliates (St. Louis University and Washington University in St. Louis). The PAs at our center undergo similar training to the GI fellows from our academic affiliates, initially performing procedures with one-on-one attending supervision. After completing the minimum number of colonoscopy procedures for competency as designated by the ASGE, the PAs at our facility are then permitted to perform colonoscopy under minimal supervision, with an attending gastroenterologist available in the endoscopy suite to provide input and assist during the procedure as needed.15 The PAs at VASTLHCS generally only request assistance when performing large polypectomies. This practice is in accordance with Veterans Health Administration (VHA) directives.16
All endoscopy procedures were classified into one of three groups by primary endoscopist type: attending gastroenterologists, supervised PAs, or supervised GI fellows. In the PA endoscopist group, the procedure was performed in its entirety by the PA; cases where attending gastroenterologist assistance during the procedure was documented were excluded from further analysis in order to achieve a more objective comparison of skill levels between the groups. Attending gastroenterologist input into the interpretation of endoscopic findings was permitted, however.
The GI fellow cases were performed with a greater degree of physician supervision, and endoscopic assistance was permitted in this group. Colonoscopy procedure and pathology reports (in cases where polypectomies were performed) were reviewed by a single investigator (CS) who was not involved in the performance of any of the procedures. Procedural data collected included intubation time, withdrawal time, bowel prep quality, colon segment reached, and total numbers of polyps removed. Polyps were further subclassified based on pathology as adenomas, advanced adenomas (that is, greater than 1 cm in size, high-grade dysplasia, or villous histology), or cancer based on the interpretation by a GI pathologist. Withdrawal time was only collected if the procedures were without any biopsy or polypectomy. Any documentation of procedure-related adverse events was noted, though this was a secondary endpoint, and the study was not powered to assess for statistically significant differences in these outcomes. This project was undertaken as a VASTLHCS GI section quality initiative project.
Grouped values were reported as mean and standard deviation. Across group differences in continuous variables among the endoscopy providers was assessed using analysis of variance (ANOVA). In cases where differences were identified on ANOVA, between-group comparisons were performed using student t-tests for continuous variables. Chi-square or Fisher exact analyses were carried out on binomial data as indicated. In each case, P < .05 was required for statistical significance. All statistical analyses were conducted using SPSS version 22. Based on our center's historical adenoma detection rates (the percentage of completed colonoscopies in which one or more adenomatous polyps were removed), we assumed an overall rate of 40%, and assuming 0.1 proportional difference in the PA and physician adenoma detection rate (0.35 versus 0.45, -0.10 defining our noninferiority margin), and a 2:1 PA to physician case ratio, we calculated a needed sample size of 336 total colonoscopies performed by PAs and physicians to have power of 80% to yield a statistically significant result when the criterion for significance was set at alpha of 0.05 and a two-tailed testing was applied.
A total of 597 of the 743 consecutive patients undergoing routine screening colonoscopy at VASTLHCS in a 12-month period (July 2015 to June 2016) were scrutinized after excluding patients who were not average risk, or those with inadequate or incompletely documented bowel preparation. No significant difference was found in average patient age or race based on primary endoscopist provider type. The average age of patients was 59 years in the group with attending gastroenterologists as the primary endoscopist, 59.4 years with PAs as the primary endoscopist, and 59.7 years with GI fellows as the primary endoscopist (P = .59 across groups). The percentage of black patients was 52% in the group with attending gastroenterologists as the primary endoscopist, 48% with PAs as the primary endoscopist, and 45% with GI fellows as the primary endoscopist (P = .73). A slightly increased proportion of men was noted in the group with GI fellows as the primary endoscopist (92.2%) compared with PAs (85.2%) and attending gastroenterologists (86.5%) (P = .04).
Colonoscopy summary data
The 597 screening procedures meeting inclusion criteria (attending gastroenterologist, n = 119; PA, n = 169; fellow, n = 309) were further scrutinized. The physicians in this study had a median of 15.5 years of endoscopy experience (range, 4 to 31 years) and the PAs had a median of 10.8 (range, 1 to 24) years of experience. Overall, cecal intubation was successful in 588 (98.5%) cases, with a mean intubation time of 9.9 ± 7.9 minutes. Of the 597 screening procedures in the study, 200 (33.5%) were cases without biopsy or polypectomy and were used to calculate withdrawal time. The mean colonoscope withdrawal time was 12.3 ± 5.4 minutes. Collectively, adenomas were detected in 44.6% of patients, with 13.4% patients having one or more adenomas larger than 1 cm. Only one average-risk screening colonoscopy revealed colorectal cancer. For the colonoscopies included in the analysis, no adverse reactions were reported related to anesthesia or the endoscopic procedure up to 30 days postprocedure.
Colonoscopy data by endoscopist experience level
All included endoscopists were grouped by their number of years of experience in order to more accurately compare attending gastroenterologist and PA performance (fellows were excluded from this portion of the analysis). Three attending gastroenterologists with 33 colonoscopies and one PA with 46 colonoscopies had 5 or fewer years of experience. Two attending gastroenterologists with 66 colonoscopies and two PAs with 80 colonoscopies had 6 to 15 years of experience. Two attending gastroenterologists with 20 colonoscopies and two PAs with 43 colonoscopies had more than 15 years of experience.
Figure 1 includes colonoscope intubation time organized by endoscopist type and years of experience. In the groups with 5 or fewer years of endoscopy experience, no statistically significant difference was found in mean intubation time between attending gastroenterologists and PAs (8.8 min versus 7.6 min; P = .46). Among those with 6 to 15 years of experience, no significant difference was found in intubation time (7.2 min versus 8.1 min; P = .38). However, the group of PAs with more than 15 years of experience had shorter mean intubation times than the two attending gastroenterologists with similar experience (15.6 min versus 7.5 min; P = .002).
Colonoscope withdrawal time was greater than 6 minutes for attending gastroenterologists and for PAs. The withdrawal time for attending gastroenterologists with 3 to 5 years of experience was 12.7 ± 2.8 minutes; for physicians with more than 5 years of experience, the time was 11.3 ± 3.1 minutes. The withdrawal time for attending gastroenterologists with 1 to 2 years of experience could not be calculated because they had no procedures without biopsy or polypectomy.
The colonoscope withdrawal time for PAs with 1 to 2 years of experience was 10.8 ± 4.2 minutes; for PAs with more than 5 years of experience, the time was 9.1 ± 3.1 minutes. Withdrawal time for PAs with 3 to 5 years of experience could not be calculated because they had no procedures without biopsy or polypectomy.
No statistically significant differences in adenoma detection rates in any groups were identified when compared based on years of experience (Figure 2). In the groups with 5 or fewer years of endoscopy experience, the adenoma detection rate was identical for attending gastroenterologists and PAs at 46%. The adenoma detection rates for attending gastroenterologists and PAs with 6 to 15 years of experience were 54% and 53%, respectively (P = .87). For clinicians with more than 15 years of experience, the adenoma detection rates were 24% for attending gastroenterologists and 35% for PAs (P = .4).
Fellow performance was also examined based on years of total endoscopy training. The postgraduate year (PGY)-4 fellows (first year of fellowship training) had a mean intubation time of 17.2 minutes, a mean withdrawal time of 17.4 minutes, an adenoma detection rate of 50%, and reached the cecum 100% of the time with attending gastroenterologist assistance permitted. PGY-5 fellows had a mean intubation time of 14.3 minutes, a mean withdrawal time of 12.6 minutes, an adenoma detection rate of 41%, and reached the cecum 100% of the time with attending gastroenterologist assistance permitted. PGY-6 fellows had a mean intubation time of 11.4 minutes, a mean withdrawal time of 10.6 minutes, an adenoma detection rate of 45%, and reached the cecum 99% of the time with attending gastroenterologist assistance permitted. The differences in intubation time across the three GI fellow groups reached statistical significance (P < .001).
Colonoscopy data by endoscopist type
Colonoscopy performance also was examined with endoscopists grouped by provider type regardless of years of experience. These data revealed that PAs performed superior to GI fellows with regard to mean intubation time (7.8 min versus 13.2 min, P < .001) and were found to have a shorter mean withdrawal time (9.6 min versus 11.5 min). No significant difference was found between the intubation time of PAs and attending gastroenterologists (7.8 min versus 8.8 min, respectively, P = .25).
PAs had higher cecal intubation rates than attending gastroenterologists (98.8% versus 94.8%, respectively, P = .039), but no difference was found when PAs were compared with GI fellows (P = .25), although attending gastroenterologist assistance was permitted if a fellow experienced difficulty with cecal intubation. PAs achieved an average adenoma detection rate of 46.7%, which was comparable with both attending gastroenterologists (adenoma detection rate of 43.5%, P = .59) and GI fellows (44.2%, P = .89).
In this study of an academic supervised endoscopy practice, PAs performed better than supervised GI fellows and comparably to attending gastroenterologists on established colonoscopy metrics and quality indicators. When accounting for number of years of experience, our comparison of attending gastroenterologists and trained PAs did not demonstrate any significant differences in these same metrics. Notably, PAs achieved a higher cecal intubation rate than their physician counterparts. Further, we observed that PAs, as a group, were superior to supervised GI fellows with regard to cecal intubation time and achieved comparable rates of adenoma detection with both GI fellows and attending gastroenterologists in this average risk patient population. All groups had withdrawal times longer than the recommended 6-minute minimum withdrawal time.
Though the study was primarily intended to assess differences in adenoma detection rates between the different endoscopy provider types, some additional interesting differences in cecal intubation times and rates were observed between PAs and physicians. We speculate that the PAs in this study, most of whom are exclusively clinical in focus, may have been more likely than the experienced attending gastroenterologists to implement some newer endoscopic techniques such as water immersion and cap endoscopy, which have been shown to improve colonoscopy efficiency and effectiveness.17
Training nonphysician endoscopists to perform colorectal cancer screening was first described in the early 1970s, when using these clinicians to perform flexible sigmoidoscopy was proposed as a way to make colorectal cancer screening available to a greater number of patients than would be possible with physician endoscopists alone.5 The subsequent studies on this topic focused exclusively on flexible sigmoidoscopy, in line with the colorectal cancer screening guidelines at the time, and uniformly found that nonphysician endoscopists were able to perform sigmoidoscopy adequately with sufficient training.7-13,18-22 To date, the literature on clinicians other than physicians performing colonoscopy largely has focused on nurse-endoscopists, and most of these studies were positive. In a prospective feasibility study by Koornstra and colleagues in 2009, two nurses and one first-year GI fellow were concurrently trained in colonoscopy with subsequent examination of their first 150 consecutive colonoscopies after completion of training.23 Consistent improvement in cecal intubation rates and times was demonstrated by nurses and the first-year fellows alike. Subsequent publications demonstrated adequate performance and patient satisfaction after the implementation of nurse-endoscopists.24,25 Most recently, a prospective randomized study published in 2015 by Hui and colleagues examined the performance of a group of nurse-endoscopists following training in colonoscopy and compared it with a group of more experienced physician endoscopists.26 They found the nurse-endoscopists required a longer withdrawal time, but were ultimately able to achieve adenoma detection rates comparable with their physician counterparts. Of note, only one of the previous studies comparing nonphysician endoscopists to physicians was conducted in the United States.27
The present study is novel in two ways: it is the first to examine the performance of PAs, with their own distinct training, as endoscopists; and second, to our knowledge, it is the only study to document the colonoscopy skill level attainable by a group of experienced PA or nurse endoscopists in comparison to physicians. VASTLHCS implemented PAs in the endoscopy program more than 2 decades ago, and the PAs in this study had 54 years of collective endoscopic experience (ranging from 1 to 24 years), all far exceeding the minimum number of required colonoscopies to be deemed competent.15 A previous study compared a single experienced nurse-endoscopist with a group of physician endoscopists and showed similar ability, but this comparison does not provide insight into the endoscopy experiences with nonphysician endoscopists of different training backgrounds.27 Other studies on this topic have compared nurse-endoscopists with physicians as both groups go through initial training in colonoscopy, demonstrating that nurse aptitude during and immediately after training meets or exceeds that of physician trainees.23,28 Again, this approach does not shed light on the potential long-term performance of experienced PA or nurse-endoscopists. The fact that our study group of PAs was able to exceed the quality metrics of supervised GI fellows and perform comparably to experience-matched physicians illustrates the potential for continued improvement in PA endoscopy skills over the course of their careers, matching the skill level of seasoned physician endoscopists in performing average-risk screening colonoscopy.
In 2014, a national campaign was launched in the United States with the goal of increasing colorectal cancer screening rates to 80% or more of adults aged 50 to 75 years by 2018.2 A group of CDC researchers published a study in 2016 using mathematical modeling to estimate the number of colonoscopies or fecal immunochemical tests required to reach this goal. They found that 5.1 to 13 million colonoscopies would need to be performed each year, depending on the uptake of fecal immunochemical testing as a screening alternative. Indeed, in 2012 about 15 million colonoscopies were performed in the United States, and the study suggested that US providers annually could perform an additional 10.5 million colonoscopies, if needed.3 Although this study suggests that our present healthcare system has the capacity to meet future colonoscopy demands, most respondents (68%) were from practices located in urban settings, suggesting these results likely are not generalizable to the entire country, where many rural regions deal with shortages of available endoscopists.4
PAs and NPs might be particularly well positioned to perform as endoscopists in order to extend our ability to perform screening colonoscopy in resource-poor settings. With foresight, implementation of PA endoscopic training programs similar to that at VASTLHCS could serve to not only close gaps in colorectal cancer screening goals via the provision of screening colonoscopy but, as demonstrated in other countries, may offer colonoscopy at a substantial cost savings if Medicare and insurers begin reimbursement for this model.28
This study has some noteworthy limitations. As a retrospective study without a formal randomization process, this investigation may be subject to the potential biases inherent in any study of this design, including selection bias. Although VASTLHCS has no formal mechanism to funnel more “straightforward” colonoscopy cases to PAs or GI fellows, case assignment ultimately occurs at the discretion of the attending gastroenterologist, and an element of such triage conceivably could exist within these data. Nevertheless, we believe the potential for this phenomenon was minimized by the fact that the patients included in this study were all average risk and many were undergoing their first colonoscopy, making it less likely that more challenging cases could be anticipated before the procedure.
A comparison of cecal intubation rates between the three groups was complicated by the fact that attending gastroenterologists were available to help GI fellows when necessary, but these cases were excluded from analysis in the PA group. In general, the PAs at VASTLHCS rarely request assistance with cecal intubation, and will typically only seek assistance in performing large polypectomies. GI fellows required assistance with cecal intubation far more often, especially in their first year of fellowship. We did not have data available to compare PA endoscopists in training with GI fellows in training, but this would be an important comparison for consideration. Given that this study was conducted at a large, academically affiliated VA medical center with a male-predominant patient population, the generalizability of these findings to community settings is somewhat limited. Despite this, we feel that this study of a large number of average-risk screening colonoscopies performed by a large group of PAs and physicians provides insight into what a real-world application of this endoscopist model would resemble. As only VA medical records were reviewed, any patient who may have presented to a non-VA hospital with a delayed postprocedural adverse reaction would not have been identified. Finally, patient body mass index and American Society of Anesthesiologists physical status, two factors known to affect procedure complexity, were not examined.
We found that experienced PA endoscopists performed better than gastroenterology trainees and comparably to their GI physician counterparts at screening colonoscopy. Multicenter, longitudinal studies would provide greater insight into the effect a PA or NP endoscopist program might have on our medical system's ability to achieve colorectal cancer screening goals. In the interim, we suggest that these data support the implementation of supervised PA endoscopists in the provision of average-risk screening colonoscopy following a proper training program. This approach may be particularly relevant to underserved populations and resource-poor areas where access to and cost of colonoscopy limits the optimization of colorectal cancer screening strategies.
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin
3. Joseph DA, Meester RG, Zauber AG, et al Colorectal cancer screening
: estimated future colonoscopy need and current volume and capacity. Cancer
4. Aboagye JK, Kaiser HE, Hayanga AJ. Rural-urban differences in access to specialist providers of colorectal cancer
care in the United States: a physician workforce issue. JAMA Surg
5. Strode JE. Is the concept of proctosigmoidoscopic examinations in asymptomatic individuals by a trained technician justifiable. Ann Surg
6. Spencer RJ, Ready RL. Utilization of nurse endoscopists for sigmoidoscopic examinations. Dis Colon Rectum
7. Rosevelt J, Frankl H. Colorectal cancer screening
by nurse practitioner using 60-cm flexible fiberoptic sigmoidoscope. Dig Dis Sci
8. Weissman GS, Winawer SJ, Baldwin MP, et al Multicenter evaluation of training of non-endoscopists in 30-cm flexible sigmoidoscopy. CA Cancer J Clin
9. DiSario JA, Sanowski RA. Sigmoidoscopy training for nurses and resident physicians. Gastrointest Endosc
10. Maule WF. Screening for colorectal cancer
by nurse endoscopists. N Engl J Med
11. Duthie GS, Drew PJ, Hughes MA, et al A UK training programme for nurse practitioner flexible sigmoidoscopy and a prospective evaluation of the practice of the first UK trained nurse flexible sigmoidoscopist. Gut
12. Schoenfeld P, Lipscomb S, Crook J, et al Accuracy of polyp detection by gastroenterologists and nurse endoscopists during flexible sigmoidoscopy: a randomized trial. Gastroenterology
13. Wallace MB, Kemp JA, Meyer F, et al Screening for colorectal cancer
with flexible sigmoidoscopy by nonphysician endoscopists
. Am J Med
14. American Society for Gastrointestinal Endoscopy. Endoscopy by non-physicians: guidelines for clinical application. From the ASGE. Gastrointest Endosc
15. ASGE Standards of Practice Committee. Faulx AL, Lightdale JR, Acosta RD, et al. Guidelines for privileging, credentialing, and proctoring to perform GI endoscopy. Gastrointest Endosc
16. VHA directive 1063: utilization of physician assistants
(PA). 2013. http://www.va.gov/vhapublications/publications.cfm?pub=1
. Accessed January 22, 2020.
17. Vemulapalli KC, Rex DK. Evolving techniques in colonoscopy. Curr Opin Gastroenterol
18. Norfleet RG, Johnson SE, Mulholland DD, et al Proctosigmoidoscopic examinations by gastroenterology assistants. J Clin Gastroenterol
19. Schroy PC, Wiggins T, Winawer SJ, et al Video endoscopy by nurse practitioners: a model for colorectal cancer screening
. Gastrointest Endosc
20. Goodfellow PB, Fretwell IA, Simms JM. Nurse endoscopy in a district general hospital. Ann R Coll Surg Engl
21. Maruthachalam K, Stoker E, Nicholson G, Horgan AF. Nurse led flexible sigmoidoscopy in primary care--the first thousand patients. Colorectal Dis
22. Shapero TF, Hoover J, Paszat LF, et al Colorectal cancer screening
with nurse-performed flexible sigmoidoscopy: results from a Canadian community-based program. Gastrointest Endosc
23. Koornstra JJ, Corporaal S, Giezen-Beintema WM, et al Colonoscopy training for nurse endoscopists: a feasibility study. Gastrointest Endosc
. 2009;69(3 Pt 2):688–695.
24. Maslekar S, Hughes M, Gardiner A, et al Patient satisfaction with lower gastrointestinal endoscopy: doctors, nurse and nonmedical endoscopists. Colorectal Dis
25. van Putten PG, Ter Borg F, Adang RP, et al Nurse endoscopists perform colonoscopies according to the international standard and with high patient satisfaction. Endoscopy
26. Hui AJ, Lau JY, Lam PP, et al Comparison of colonoscopic performance between medical and nurse endoscopists: a non-inferiority randomised controlled study in Asia. Gut
27. Limoges-Gonzalez M, Mann NS, Al-Juburi A, et al Comparisons of screening colonoscopy performed by a nurse practitioner and gastroenterologists: a single-center randomized controlled trial. Gastroenterol Nurs
28. Massl R, van Putten PG, Steyerberg EW, et al Comparing quality, safety, and costs of colonoscopies performed by nurse vs physician trainees. Clin Gastroenterol Hepatol