Accuracy of Polyp Detection by Gastroenterologists and Nurse Endoscopists During Flexible Sigmoidoscopy: A Randomized Trial

Transcription

1 GASTROENTEROLOGY 1999;117: Accuracy of Polyp Detection by and Nurse Endoscopists During Flexible Sigmoidoscopy: A Randomized Trial PHILIP SCHOENFELD,*, STEVEN LIPSCOMB,*, JENNIFER CROOK,* JONATHAN DOMINGUEZ,*, JAMES BUTLER,*, LINDA HOLMES,* DAVID CRUESS, and DOUGLAS REX *Division of Gastroenterology, National Naval Medical Center, Bethesda, Maryland; Departments of Internal Medicine and Epidemiology and Biostatistics, Uniformed Services University of Health Sciences, Bethesda, Maryland; and Division of Gastroenterology, Indiana University Medical Center, Indianapolis, Indiana See editorial on page 486. Background & Aims: The use of nurse endoscopists to perform flexible sigmoidoscopy is expanding, increasing the availability of colorectal cancer screening. However, the effectiveness of this practice has not been studied in randomized trials. The aim of this trial was to examine the miss rate of polyps, the depth of sigmoidoscope insertion, and the incidence of complications during flexible sigmoidoscopy performed by nurse endoscopists and by gastroenterologists. Methods: Three hundred twenty-eight patients were randomized to undergo screening flexible sigmoidoscopy performed by a nurse endoscopist or by a gastroenterologist. Frequency of missed polyps was determined by repeat sigmoidoscopy, performed by a gastroenterologist blinded to the identity of the first endoscopist. Multiple logistic regression analysis identified characteristics associated with missed polyps. Results: and nurse endoscopists had equivalent miss rates for adenomatous polyps (20% vs. 21%, respectively; P 0.91). No complications occurred in any patient. inserted the sigmoidoscope further than nurse endoscopists (61 vs. 55 cm, respectively; P F ). Polyp location in the descending colon (odds ratio, 4.1; 95% confidence interval, ) was highly associated with missed polyps. Conclusions: No differences in detection of adenomatous polyps or frequency of complications were found. These data suggest that experienced nurse endoscopists may perform screening flexible sigmoidoscopy as safely and as effectively as gastroenterologists. Colorectal cancer is the second most common cause of cancer death in the United States, accounting for more than 130,000 new cases per year and more than 55,000 deaths per year. 1 Several case-control studies have found that screening flexible sigmoidoscopy (FS) is associated with a 55% 70% reduction in colorectal cancer mortality. 2 4 By the year 2000, more than 50 million Americans will be eligible for screening FS, which could lead to the performance of 10 million FS procedures per year. 5 However, only 30% of the eligible U.S. population have been screened by FS, 6 partly because of a lack of trained endoscopists. To increase the availability of FS, many medical centers have trained nurses to perform screening FS. Nurses have performed sigmoidoscopy since Observational studies and case series 7 12 indicate that nurses do not differ from physicians in depth of insertion of the sigmoidoscope, 10,11 patients perception of discomfort during FS, 9 patient satisfaction, 10 or percentage of patients with adenomatous polyps detected during FS. 8,10 12 Observational studies and case series 7 12 also indicate that nurses have performed more than 10,000 FS procedures without colon perforation. However, the inferences that can be drawn from these trials are limited by the inherent methodological weaknesses of observational studies and case series. A single randomized controlled trial 13 found that untrained nurses do not differ from internal medicine residents and family practice residents in the number of FS procedures required to attain certification. However, no randomized controlled trial has compared trained nurse endoscopists with trained physician endoscopists. The Society of Gastroenterology Nurses and Associates, 14 the American Society for Gastrointestinal Endoscopy, 15 and the British Society of Gastroenterology 16 support the performance of FS by nurses. However, 30% of U.S. state Boards of Nursing expressly prohibit registered nurses (RNs) from performing screening FS, partly because of limited data about the effectiveness of Abbreviations used in this paper: FS, flexible sigmoidoscopy; RN, registered nurse by the American Gastroenterological Association /99/$10.00

2 August 1999 ACCURACY OF POLYP DETECTION BY RNs AND MDs 313 this practice. 17 Although the U.S. Congress passed a Medicare bill in 1997 providing reimbursement for screening FS, 18 the Health Care Financing Administration has ruled that only physicians will be reimbursed for performing screening FS to ensure that [the procedures]... are performed as safely and accurately as possible. 19 We performed a randomized controlled trial to compare the safety and accuracy of screening FS performed by gastroenterologists and by nurse endoscopists. Frequency of complications, depth of sigmoidoscope insertion, patient satisfaction, and patient discomfort during FS are important markers of procedural effectiveness, but accurate identification of polyps is the most important marker of procedural effectiveness. Previous studies 20,21 show that back-to-back colonoscopies can estimate the frequency of missed polyps during the first colonoscopic procedure. Therefore, we conducted a randomized controlled trial to determine the frequency of missed polyps during screening FS by trained RN endoscopists and by gastroenterologists, using back-to-back FS to quantify the frequency of missed polyps. Mean depth of sigmoidoscope insertion and frequency of complications were also recorded for both groups of endoscopists. Materials and Methods Study Design All patients referred for screening FS at the Division of Gastroenterology, National Naval Medical Center, Bethesda, Maryland, during a 5-month period were offered enrollment in this study. The study was approved by the Institutional Review Board at our institution, and all participating patients provided informed consent. During informed consent, patients were instructed that they could withdraw from the study after the first FS if the first FS was too uncomfortable. Patients were randomly assigned to have their first FS performed by a gastroenterologist or by a nurse endoscopist based on a computer-generated 1:1 randomization table in blocks of 16. Using the randomization table, a gastroenterology technician placed index cards stating assignment of patients to nurse endoscopists or gastroenterologists into sequentially numbered opaque envelopes. These envelopes were opened only after patients gave informed consent. Exclusion Criteria Patients were excluded from the study if (1) in the opinion of the endoscopist, too much stool was retained in the colon to complete FS; (2) in the opinion of the endoscopist, the patient s general medical condition made the performance of back-to-back FS potentially harmful; or (3) the patient was inappropriately referred for screening FS (e.g., history of adenomatous polyps). Training of Nurse Endoscopists and All nurse endoscopists (n 3) at the National Naval Medical Center participated in this study. Each nurse endoscopist is an RN with more than 2 years of experience as a gastroenterology nurse. All nurse endoscopists have passed the certifying examination for gastroenterology nurses offered by the Certification Board for Gastroenterology Nurses and Associates. All nurse endoscopists completed 100 supervised FS procedures in a standard training program, 10 demonstrating the ability routinely to insert the sigmoidoscope at least 50 cm and to complete the procedure within 15 minutes. All nurse endoscopists had independently performed at least 50 FS procedures (range, 50 to more than 300) before the start of the study. All gastroenterologists (n 4) at the National Naval Medical Center participated in this study. were board-eligible or board-certified with 1 4 years of experience after a gastroenterology fellowship and had performed more than 1000 endoscopic procedures (range, 1000 to more than 3000). A senior gastroenterology fellow (2 years of endoscopic experience and completion of more than 1000 endoscopic procedures) and a colorectal surgeon (more than 4 years of experience after a colorectal surgery fellowship and completion of more than 2000 endoscopic procedures) also participated in this study. Study Protocol for Performance of FS Bowel preparation for the procedure consisted of a clear liquid diet for 16 hours before the procedure, consumption of 300 ml of magnesium citrate on the night before the procedure, and self-administration of Fleet enemas (C. B. Fleet, Lynchburg, VA) 4 hours and 1 hour before the procedure. All examinations were performed with a 70-cm Olympus CF 100S flexible sigmoidoscope using an Olympus CDV-U20 video display system. The first endoscopist was instructed to pass the sigmoidoscope to 70 cm. Depth of insertion of the FS was limited at the discretion of the endoscopist if the patient reported discomfort or if retained stool prevented further insertion of the sigmoidoscope. The first endoscopist was instructed to perform biopsies of all polyps during withdrawal of the sigmoidoscope. The only exception was multiple ( 5) tiny ( 3 mm) polyps in the rectum. In this case, the first endoscopist was instructed to perform biopsies of several of these lesions and document the presence of multiple polyps in the rectum. Polyp size was estimated by comparison with an open biopsy forceps (8 mm in diameter), and polyp size was classified as 1 5 mm, 6 9 mm, or 10 mm. Polyp shape (sessile or pedunculated), polyp location in the rectum/sigmoid colon/descending colon (estimated based on anatomic landmarks and insertion depth), presence of diverticulosis, depth of sigmoidoscope insertion, patient s age and sex, and adequacy of colon preparation were recorded. Adequacy of colon preparation was defined as (1) excellent, no retained stool in colon; (2) adequate, small amounts of stool found intermittently in colon;

3 314 SCHOENFELD ET AL. GASTROENTEROLOGY Vol. 117, No. 2 or (3) poor, large or small amounts of stool found diffusely in colon. Complications were defined as perforation of the colon, postprocedure infection, or any symptom (e.g., abdominal pain) requiring evaluation within 48 hours of completion of FS. After completion of the first FS, verbal consent to proceed with the second FS was obtained, and the second FS was started within 5 minutes of completion of the first FS. The gastroenterologist performing the second FS was blinded to the identity of the endoscopist who had performed the first FS. All patients were instructed to refrain from commenting on the name, age, sex, or profession of the first endoscopist during the second FS. The gastroenterologist performing the second FS followed the same procedure as the first endoscopist. Any nonbleeding polyp identified during the second FS was scored as a missed polyp, and a biopsy was performed. During post hoc analysis, procedure logs were reviewed to estimate duration of FS and mean procedure time for each group of endoscopists. Also during post hoc analysis, the size, shape, and location of polyps identified during second FS were matched against the size, shape, and location of polyps identified during first FS to ensure that no polyp was misclassified as a missed polyp. Statistical Analysis Based on previous studies, individual gastroenterologists miss 12% 48% of polyps during colonoscopy. Arbitrarily, we decided that it would be clinically significant if nurse endoscopists missed twice as many polyps as gastroenterologists. We estimated that gastroenterologists would miss 15% of all polyps and nurse endoscopists would miss 30% of all polyps on FS. Based on these estimates, 122 polyps had to be found in patients who underwent FS performed by nurse endoscopists and 122 polyps had to be found in patients who underwent FS performed by gastroenterologists to give the study a power of 0.80 at a 5% level of significance ( 0.05) with a two-tailed test. cannot consistently differentiate hyperplastic polyps from adenomatous polyps based on their endoscopic appearance. 24 Therefore, endoscopists were instructed to perform biopsies on all polyps, which is the standard of practice at our institution. Before the miss rates of adenomatous polyps and hyperplastic polyps were combined into a summary measure of total polyps missed and found, the Breslow Day test of heterogeneity was performed to ensure that statistically significant differences in miss rates for adenomatous and hyperplastic polyps were not present. Statistical analysis was performed with SAS version 12.0 for Windows (SAS Institute Inc., Cary, NC). The Student t test was used to compare interval and ratio data, and the 2 test for proportions was used for nominal and ordinal data. Analysis of variance and 2 analysis were both performed to determine if any individual gastroenterologist missed significantly more polyps than the other gastroenterologists, and similar analyses were performed for nurse endoscopists. Polyp shape, polyp histology, polyp size, polyp location, presence of diverticulosis, and adequacy of colon preparation were entered into multiple logistic regression analysis to identify unique characteristics associated with missed polyps. Results Patient Characteristics at Baseline During the study period, 350 patients were referred for screening FS (Table 1). Three hundred twenty-eight patients (94%) gave informed consent to participate in the trial. Fifteen patients were eliminated after randomization because they met exclusion criteria: too ill for back-to-back FS (n 2), too much retained stool in colon (n 10), and inappropriate referral for screening FS (n 3). Study groups were similar with respect to age, sex, percentage of patients with hyperplastic polyps, percentage of patients with adenomas, percentage of patients with any polyps, and percentage of patients with multiple polyps (Table 2). Twenty percent (64/313) of study patients refused the second FS and withdrew from the trial. Equivalent percentages of patients refused the second FS whether their first FS was performed by a gastroenterologist or by an RN (36/162, or 22%, and 28/151, or 19%, respectively; P 0.42). Frequency of Missed Polyps by and Nurse Endoscopists and RNs missed adenomas with similar frequency (20% vs. 21%, respectively; P 0.91). The characteristics of found polyps and missed polyps by study group are described in Tables 3 and 4. No colorectal cancers were diagnosed during the study. missed 2 large (1 cm in diameter) Table 1. Patient Flow-Through Trial Patients referred for screening FS (n 350) Patients who gave informed consent and were randomized (n 328) 12 MD group patients (n 166) excluded because of medical condition (n 1) excluded because of poor preparation (n 3) MD group patients in study (n 162) MD group patients who withdrew after first FS (n 36) MD group patients who completed trial (n 126) RN group patients (n 162) excluded because of medical condition (n 1) excluded because of poor preparation (n 7) excluded because of inappropriate referral (n 3) RN group patients in study (n 151) RN group patients who withdrew after first FS (n 28) RN group patients who completed trial (n 123)

4 August 1999 ACCURACY OF POLYP DETECTION BY RNs AND MDs 315 Table 2. Baseline Characteristics of Patients Enrolled in the Trial (n 162 Nurse endoscopists (n 151 Age yr yr 0.23 Sex (M:F ratio) 67:33 66: Patients with hyperplastic polyps 31% 37% 0.20 Patients with adenomas 16% 11% 0.17 Patients with anatomic polyps (i.e., hyperplastic or adenomatous polyps) 43% 45% 0.83 Patients with multiple anatomic polyps 17% 22% 0.25 adenomas, and the RNs did not miss any large adenomas (Table 4). missed 1 villous adenoma, and nurse endoscopists did not miss any villous adenomas. and RNs performed biopsies of normal mucosa, mistaken for diminutive polyps, with similar frequencies (7% vs. 11%, respectively; P 0.27). When total polyps missed and found are examined, gastroenterologists missed 29% (41/139) of all polyps, and RNs missed 17% (22/128) of all polyps (P 0.02). This difference resulted from the significant difference in the miss rate of hyperplastic polyps by gastroenterologists and RNs (32% vs. 17%, respectively; P 0.01). The Breslow Day test of heterogeneity did not identify statistically significant heterogeneity in the miss rates for adenomatous and hyperplastic polyps (P 0.26), indicating that combining miss rates of hyperplastic and adenomatous polyps into a summary outcome measure (i.e., total polyps missed and found) was statistically acceptable. Table 3. Missed Polyps in Patients Who Completed Both FS Procedures Based on Histology of Polyp (n 126 (missed/total polyps) Nurse endoscopists (n 123 (missed/total polyps) All polyps 29% (41/139) 17% (22/128) 0.02 All hyperplastic polyps 32% (35/109) 17% (19/114) 0.01 All adenomatous polyps 20% (6/30) 21% (3/14) 0.91 NOTE. Range of FS procedures performed by individual gastroenterologists, Range of FS procedures performed by individual nurse endoscopists, The second or gold standard FS was performed by a single gastroenterologist in 78% (194/249) of patients. Another gastroenterologist performed the second or gold standard FS in 21% (52/249) of patients. P P Table 4. Missed Adenomatous Polyps in Patients Who Completed Both FS Procedures Based on Size of Polyps Adenomatous polyp size (n 126 (missed/total polyps) Nurse endoscopists (n 123 (missed/total polyps) 1 5 mm 2/15 3/8 6 9 mm 2/5 0/2 10 mm 2/10 0/4 The miss rate of polyps by individual gastroenterologists ranged from 24% to 41%. Neither 2 analysis (P 0.60) nor analysis of variance (P 0.33) identified significant differences in the miss rates of polyps by any individual gastroenterologist compared with other gastroenterologists. The miss rates for polyps by RNs ranged from 11% to 23%. Neither 2 analysis (P 0.17) nor analysis of variance (P 0.12) identified significant differences in the miss rates for polyps by any individual RNs compared with other RNs. Per-patient analysis of miss rates for polyps and adenomas during first FS is described in Table 5. More patients whose first FS procedures were performed by gastroenterologists had polyps found during second FS (i.e., missed polyps) than did patients whose first FS procedures were performed by nurse endoscopists (23% vs. 14%, respectively; P 0.06). However, if a patient had an adenoma identified during the first FS, missing another polyp might be insignificant because the patient would be referred for a colonoscopy because of the finding of the initial adenoma. The percentage of patients with no adenomas found on the first FS and adenoma found on the Table 5. Percent of Patients With Polyps Identified on First and Second FS (n 126 Nurse endoscopists (n 123 Patients with any anatomic polyps 51% (64/126) 47% (58/123) 0.57 Patients with polyp found on 1st FS 40% (50/126) 41% (51/123) 0.78 Patients with polyp found on 2nd FS 23% (29/126) 14% (17/123) 0.06 Patients with no polyp found on 1st FS and polyp found on 2nd FS 12% (14/126) 6% (7/123) 0.12 Patients with no adenoma found on 1st FS and adenoma found on 2nd FS 3% (4/126) 2% (2/123) 0.43 NOTE. Data are only from patients who underwent both FS procedures. P

5 316 SCHOENFELD ET AL. GASTROENTEROLOGY Vol. 117, No. 2 second FS was very small for patients whose first FS procedures were performed by gastroenterologists and by nurse endoscopists (3% vs. 2%, respectively; P 0.43). Other Markers of Procedural Effectiveness inserted the sigmoidoscope further than nurse endoscopists (61 10 cm vs cm, respectively; P ). According to study criteria, no complications occurred. However, 1 study patient who underwent FS performed by a nurse presented 4 days after the procedure with intermittent abdominal discomfort. Abdominal x-ray results were normal, and the patient had total resolution of discomfort 48 hours after evaluation. Review of procedure logs showed that mean procedure time for both nurse endoscopists and gastroenterologists was less than 15 minutes, although the exact duration of each procedure was not measured. Polyp Characteristics Associated With Missed Polyps Table 6 shows the effect of variables on frequency of missed polyps. Absence of diverticulosis (odds ratio, 2.5; 95% confidence interval [CI], ; P 0.01) and polyp location in the descending colon (odds ratio, 4.1; 95% CI, ; P 0.004) were significantly associated with missed polyps. If the sigmoidoscope was only inserted to 35 cm during the first FS, then descending colon polyps were frequently missed. The first endoscopist inserted the scope at least 50 cm in 90% (223/249) of patients. In these patients, only 3 additional polyps were found when the sigmoidoscope was inserted further during the second FS. Depth of sigmoidoscope insertion was limited to 35 cm in 4.4% (11/249) of patients. Of these patients, 27% (3/11) had additional polyps identified when the scope was inserted further during the second FS. Ten polyps were present in these 11 patients, and 30% (3/10) of these polyps were missed because they were distal to 35 cm from the anus. Table 6. Multiple Logistic Regression Analysis of Effects of Polyp/Colon Characteristics on the Likelihood of a Missed Polyp Odds ratio 95% CI Polyp shape Polyp diameter (mm) Polyp histology Descending colon polyp Adequacy of colon preparation Absence of diverticulosis Association Between Size and Histology of Polyps In some institutions, biopsies are not performed during FS, and the mere identification of a polyp during FS triggers a colonoscopy. Data on the association between size and histology of polyps may clarify whether this policy is appropriate. Among patients who completed both FS procedures (n 249), 267 polyps were found. All polyps 10 mm or greater in diameter were adenomas (14/14). Fifty-nine percent (10/17) of polyps 6 9 mm in diameter were hyperplastic. Ninety percent (213/236) of polyps 1 5 mm in diameter were hyperplastic. Among patients who only had 1 9-mm polyps (n 108), 81% (88/108) did not have adenomas and only had hyperplastic polyps. Discussion As the U.S. population ages, the demand for screening FS may exceed available resources, 5 and many medical centers have trained nurses to perform screening FS to meet this perceived increase in demand. 17 However, partly due to the relative paucity of data on the accuracy and safety of this practice, some state Boards of Nursing prohibit RNs from performing FS. 17 The Health Care Financing Administration currently prohibits reimbursement of screening FS performed by nonphysicians. 19 This is the first randomized controlled trial to compare the accuracy and safety of screening FS performed by trained nurse endoscopists and gastroenterologists. We found that gastroenterologists and nurse endoscopists missed adenomas at similar rates (20% vs. 21%, respectively; P 0.91), and we did not find any differences in procedure-related complications. The miss rate of adenomas essentially replicates the miss rate of adenomas by gastroenterologists described in previous studies. 20 These data suggest that experienced, well-trained nurses can perform FS as safely and accurately as gastroenterologists. Because previous trials 24 have indicated that endoscopists cannot reliably differentiate hyperplastic polyps from adenomas based on endoscopic appearance, total polyps missed and total found was the primary end point of this trial. An ideal multicenter study would assess the miss rate of adenomas as the primary study end point, but the sample size for this ideal trial would exceed 1500 patients. To determine if combining miss rates of hyperplastic polyps and adenomas was statistically acceptable, the Breslow Day test of heterogeneity was performed (P 0.26), and this test ruled out significant heterogeneity in the miss rates of hyperplastic polyps and adenomas. Nevertheless, the difference in miss rates of total polyps by gastroenterologists and by RN endoscopists (29% vs.

6 August 1999 ACCURACY OF POLYP DETECTION BY RNs AND MDs %, respectively; P 0.02) occurred because gastroenterologists missed significantly more hyperplastic polyps than RN endoscopists (32% vs. 17%, respectively; P 0.01), limiting the clinical significance of miss rates of total polyps. The unexpected difference in miss rates of hyperplastic polyps may be attributable to two factors. Despite instructions to perform biopsies on all polyps, gastroenterologists may have accurately differentiated some hyperplastic polyps from adenomas, based on their endoscopic appearance, and then ignored these hyperplastic polyps. Also, review of procedure logs indicated that the mean procedure time was less than 15 minutes for both groups of endoscopists, although exact duration of FS was not measured. If mean procedure time for RN endoscopists were 2 3 minutes longer than for gastroenterologists, this difference could partly account for the difference in polyp miss rates. Regardless of the reason for the difference in polyp miss rates, our data clearly indicate that RN endoscopists identify polyps at least as accurately as gastroenterologists at our institution. Several additional explanations for the difference in polyp miss rates are also possible. First, the senior gastroenterology fellow and the colorectal surgeon in this trial could have increased the miss rate for the group of gastroenterologists. However, the senior gastroenterology fellow and the colorectal surgeon were experienced endoscopists (i.e., each had performed more than 1000 endoscopic procedures), and both had lower miss rates (32% vs. 25%, respectively) than board-certified gastroenterologists who participated in the study. Second, it could be argued that the gastroenterologists who participated in this study performed poorly, but the range of miss rates for gastroenterologists in this study (24% 41%) is similar to miss rates for polyps among individual gastroenterologists (12% 48%) reported in previous studies Finally, it could be argued that the miss rate of polyps by RN endoscopists in this trial was minimized by selecting only the most skilled RN endoscopists to participate or by having the RN endoscopists perform biopsies on every mucosal irregularity. However, all RN endoscopists at our institution participated in this trial, and all certified gastroenterology nurses at our institution have been trained as RN endoscopists. Also, RN endoscopists and gastroenterologists had similar frequencies of normal mucosa found on pathological examination of their biopsies (11% vs. 7%, respectively; P 0.27). In assessment of the importance of missed polyps, per-patient analysis may be more significant clinically. Certainly, if a patient had an adenoma identified during FS, missing another polyp would not be as clinically significant because the patient would still be referred for colonoscopy. It is reassuring that only 2.5% (6/249) of patients had no adenoma identified on first FS and then had an adenoma identified during the second FS. Both groups of endoscopists inserted the sigmoidoscope to a mean of 55 cm or greater, but gastroenterologists inserted the sigmoidoscope further than RN endoscopists (61 cm vs. 55 cm, respectively; P ). However, when the first endoscopist inserted the sigmoidoscope at least 50 cm, only 3 additional polyps were found when the sigmoidoscope was inserted further during the second FS. Thus, the 6-cm difference in mean insertion of sigmoidoscope between RN endoscopists and gastroenterologists does not appear to be clinically significant. Insertion of the scope only 35 cm appears to increase polyp miss rates. The sigmoidoscope was only inserted to 35 cm during the first FS in 4.4% (11/249) of patients. Among these 11 patients, 30% (3/10) of polyps were missed because the polyp was distal to 35 cm from the anus. This finding supports previous evidence 25 that 20% of polyps in the left side of the colon are identified when the sigmoidoscope is inserted past 35 cm to an insertion depth of cm. Our multiple logistic regression analysis, which indicated that polyp location in the descending colon was significantly associated with missed polyps (odds ratio, 4.1; 95% CI, ), also supports this finding. To our knowledge, no guideline states that depth of insertion of the sigmoidoscope to less than 50 cm is inadequate. These data support the development of a new guideline for adequate depth of insertion of the sigmoidoscope, although patient safety or patient discomfort may be compromised if less experienced primary care providers feel compelled to insert the sigmoidoscope to 50 cm. In some institutions, biopsies are not performed during FS, and the mere observation of a polyp during FS triggers a colonoscopy. This practice may be appropriate if the polyp is 10 mm in diameter but may be inappropriate for 1 9-mm polyps. Among patients who underwent both FS procedures, all polyps 10 mm were adenomas (14/14). However, 59% (10/17) of 6 9-mm polyps were hyperplastic, and 90% (213/236) of 1 5-mm polyps were hyperplastic. These data were also analyzed on a per-patient basis: among patients who only had 1 9-mm polyps (n 108), 81% (88/108) did not have adenomas and only had hyperplastic polyps. Clearly, most polyps found on FS are hyperplastic, and mere observation of 1 9-mm polyps during FS should not be cause for colonoscopy. Several methodological limitations of our study will need to be resolved in the future. Because this study was conducted at a single center, further studies are needed before these results can be generalized to all RN endosco-

7 318 SCHOENFELD ET AL. GASTROENTEROLOGY Vol. 117, No. 2 pists. Complications from FS are rare ( 1 in 10,000), so this study did not have an adequate sample size to identify a significant difference in complication rates between RN endoscopists and gastroenterologists. Nevertheless, the data were collected for use in a future meta-analysis. Patient satisfaction and pain associated with FS are important outcomes that were not measured in this study. Because back-to-back FS procedures were performed in this study, we believed that the second FS would affect patient satisfaction and perception of pain, biasing measurement of these outcomes within this trial. We trained RNs to perform FS at our center because we wanted to maximize the availability of colorectal cancer screening services. These RNs were experienced gastroenterology nurses who had completed an extensive FS training program and become accomplished endoscopists before the onset of this trial. Screening FS may be uniquely suited for performance by nurses. Complications from FS are rare, and nurses only need to differentiate normal anatomy from abnormal anatomy (e.g., polyps), which is a cornerstone of nursing practice. When our nurses identify other abnormalities (e.g., colitis), the patient is immediately referred to a gastroenterologist. We believe this team approach maximizes medical and nursing contributions to patient care while expanding our colon cancer screening services. Our results support the use of nurse endoscopists to perform screening FS. References 1. American Cancer Society. Cancer facts and figures, Atlanta, GA: American Cancer Society, 1996; publication Selby JV, Friedman GD, Quesenberry CP Jr, Weiss NS. A case control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992;326: Newcomb PA, Norfleet RG, Storer BE, Surawicz T, Marcus PM. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst 1992;84: Muller AD, Sonnenberg A. Protection by endoscopy against death from colorectal cancer. Arch Intern Med 1995;155: Ransohoff D, Lang C. Sigmoidoscopic screening in the 1990 s. JAMA 1993;269: Vernon SW. Participation in colorectal cancer screening: a review. J Natl Cancer Inst 1997;89: Spencer RJ, Ready RL. Utilization of nurse endoscopists for sigmoidoscopic examinations. Dis Colon Rectum 1977;20: Maule WF. Screening for colorectal cancer by nurse endoscopists. N Engl J Med 1994;330: Schroy PC, Wiggins T, Winawer SJ, Diaz B, Lightdale CJ. Video endoscopy by nurse practitioners: a model for colorectal cancer screening. Gastrointest Endosc 1988;34: Schoenfeld P, Cash B, Piorkowski M, Kita J, Ransohoff D. Effectiveness and patient satisfaction with nurse-performed sigmoidoscopy. Gastrointest Endosc 1999;49: Khandelar S, DiSario J, Teitze C, et al. Similar results at screening sigmoidoscopy by nurses and gastroenterologists (abstr). Am J Gastroenterol 1997;92:A Wallace MB, Ho KY, Trnka Y, Henderson C, Kemp JA, Farraye F. Comparison of nurse practitioners/physician assistants to gastroenterologists in performing flexible sigmoidoscopy (abstr). Gastroenterology 1998;114:A DiSario JA, Sanowski RA. Sigmoidoscopy training for nurses and resident physicians. Gastrointest Endosc 1993;39: Society of Gastroenterology Nurses and Associates Practice Committee. Performance of flexible sigmoidoscopy by registered nurses for the purpose of colorectal cancer screening. Gastroenterol Nurs 1997;20:S1 S Practice Parameter Committee, American Society for Gastrointestinal Endoscopy. Endoscopy by non-physicians. Manchester, MA: ASGE publication 1035, British Society of Gastroenterology Endoscopy Section Working Party. The nurse endoscopist. Gut 1995;36: Cash B, Schoenfeld P, Ransohoff D. Licensure, training, and utilization of paramedical personnel to perform screening flexible sigmoidoscopy. Gastrointest Endosc 1999;49: Martin S. Medicare expanding to pay for prevention. Am Med News 1997;40: Federal Register Rules and Regulations. Oct 31, 1997;62: Rex D, Cutler C, Lemmel G, et al. Colonoscopic miss rates of adenomas determined by back-to-back colonoscopies. Gastroenterology 1997;112: Hixson LJ, Fennerty MB, Sampliner RE, Mcgee D, Garewal HS. Prospective study of the frequency and size distribution of polyps missed by colonoscopy. J Natl Cancer Inst 1990;82: Waye JD, Braunfield S. Surveillance intervals after colonoscopic polypectomy. Endoscopy 1982;14: Hoff G, Vatn M. Epidemiology of polyps in the rectum and sigmoid colon: endoscopic evaluation of size and location of polyps. Scand J Gastroenterol 1985;20: Norfleet RG, Ryan ME, Wyman JB. Adenomatous and hyperplastic polyps cannot be reliably differentiated by their appearance through the fiberoptic sigmoidoscope. Dig Dis Sci 1988;33: Dubow RA, Katon RM, Benner KG, et al. Short (35-cm) versus long (60-cm) flexible sigmoidoscopy: a comparison of findings and tolerance in asymptomatic patients screened for colorectal neoplasia. Gastrointest Endosc 1985;31: Received December 15, Accepted April 19, Address requests for reprints to: Philip Schoenfeld, M.D., M.S.Ed., M.Sc.(Epi), Division of Gastroenterology, National Naval Medical Center, 8901 Wisconsin Avenue, Bethesda, Maryland Fax: (301) The opinions and assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the policies of the Department of the Navy or the Department of Defense. The Chief, Navy Bureau of Medicine and Surgery, Washington, D.C., Clinical Investigation Program sponsored this study B The authors thank Mark Johnston, M.D., Marjorie Piorkowski, C.G.R.N., and Jane Allaire, C.G.R.N., for participating in this study. They also thank the patients who participated in this study and endured back-to-back FS.