Colorectal cancer (CRC) is a common malignancy and. Complications of Screening Flexible Sigmoidoscopy

Transcription

1 GASTROENTEROLOGY 2002;123: Complications of Screening Flexible Sigmoidoscopy THEODORE R. LEVIN,*, CAROL CONELL,* JEAN A. SHAPIRO, SHELLA G. CHAZAN,* MARION R. NADEL, and JOE V. SELBY* *Kaiser Permanente Division of Research, Oakland, California; Gastroenterology Department, Kaiser Permanente Medical Center, Walnut Creek, California; and Centers for Disease Control and Prevention, Atlanta, Georgia Background & Aims: Flexible sigmoidoscopy (FS) is recommended for mass screening for colorectal cancer (CRC), yet little is known about the risk of adverse events when FS is used in general clinical practice.we aimed to determine the incidence of gastrointestinal complications and acute myocardial infarction (MI) after screening FS. Methods: Northern California Kaiser Permanente Medical Care Program members of average risk for CRC (n 107,704) who underwent screening FS during 1994 to 1996 (109,534 FS), as part of the Colorectal Cancer Prevention (CoCaP) program.the main outcome measure was hospitalization for gastrointestinal complications or acute MI within 4 weeks of FS. Results: The mean age of subjects was 61 years, and 48.8% were female. Nongastroenterologist physicians, nurses, or physician assistants performed 72% of FS. Overall, 24 persons were hospitalized for a gastrointestinal complication.of these, 7 were serious (2 perforations, 2 episodes of diverticulitis requiring surgery, 2 cases of bleeding requiring transfusion, and 1 episode of unexplained colitis).in multivariate models, complications were significantly more common in men than in women (odds ratio, 3.34; 95% confidence interval [CI], ). MI occurred in 33 persons within 4 weeks of FS, but the incidence for this period was similar to that in the subsequent 48 weeks (rate ratio, 0.8; 95% CI, ). Conclusions: The risk of serious complications after screening FS in this setting appears to be modest. Although MI occurs after FS, the risk during the 4 weeks after the procedure appears to be similar to expectations for persons of screening age. Colorectal cancer (CRC) is a common malignancy and usually develops from adenomatous polyps, which are easily removed benign lesions with a long premalignant phase. Accordingly, CRC is an ideal target for early detection and prevention through screening. Flexible sigmoidoscopy (FS) screening is associated with at least a 30% reduction in colorectal cancer related mortality in case-control studies. 1 3 Moreover, FSscreening appears to be as cost-effective as other well-accepted cancer screening strategies such as mammography 4,5 and may be cost-saving under certain conditions. 6 Still, the benefits of screening accrue only to the fraction of individuals who have colorectal cancer or an adenomatous polyp, whereas all people who are screened are at risk for harms from the procedure. Thus, the magnitude of the risk and the severity of the harms are important for informing the decision to screen for this disease. 7 Potential gastrointestinal complications of FSinclude colitis from chemicals used for endoscope sterilization, bowel perforation, bleeding, and infection. Acute diverticulitis may also be related to FSbecause it is a microscopic perforation of the colon, and perforation may be induced by mechanical or pneumatic trauma during FS. To date, however, there has been little description of the rate of these complications when FSis used for large-scale screening with a diverse group of examiners such as nurses, physician assistants, primary care physicians, and gastroenterologists. A recent expert review identified a serious complication rate of 1 in 10,000 but was based on just a few studies and did not distinguish screening from diagnostic FS. 8 Other recently reported rates range from 0 to 4 per 1000, 9 11 but these rates are derived from small series and are imprecise. A recent larger study of perforations from FSreported a rate of 1/25,000, 12 but this study also did not distinguish screening from diagnostic examinations. Symptomatic patients undergoing diagnostic examinations may be expected to have a greater prevalence of lesions requiring removal and decreased colonic wall integrity leading to a higher perforation rate than asymptomatic screenees. Myocardial infarction (MI) has been reported among persons undergoing FS, 9 and in 1 recent randomized controlled trial excess cardiovascular deaths were largely Abbreviations used in this paper: CI, confidence interval; CRC, colorectal cancer, FS, flexible sigmoidoscopy; ICD-9, International Classification of Diseases,Ninth Revision; KFHP, Kaiser Foundation Health Plan; KFHP/H, Kaiser Foundation Health Plan and Hospitals; KP, Kaiser Permanente; MI, myocardial infarction by the American Gastroenterological Association /02/$35.00 doi: /gast

2 December 2002 SCREENING SIGMOIDOSCOPY COMPLICATIONS 1787 responsible for an elevated overall mortality rate in the screening FSgroup. 13 This trial was small, and risk factor information is not available for the control group; thus, further observational studies are needed. The present study takes advantage of the electronic databases maintained by Kaiser Permanente to report the incidence of hospitalization for gastrointestinal complications and acute MI in a large population screened with FSfor CRC by a diverse set of examiners, including gastroenterologists, nongastroenterologist physicians, and nurses. Materials and Methods The institutional review boards of the Kaiser Foundation Research Institute and the Centers for Disease Control and Prevention approved this retrospective cohort study. Study Population and Setting The details of the Kaiser Permanente Colorectal Cancer Prevention program have been published elsewhere. 14,15 In brief, screening FSis available to all Kaiser Foundation Health Plan (KFHP) members aged 50 and older in northern California as part of routine medical care; gastroenterologists supervise the FSfacilities, but nongastroenterologist physicians or nurses perform most of the examinations. The study population for the present study consisted of KFHP members aged years who were of average risk for CRC and received a screening FSat a Kaiser Permanente (KP) facility in Northern California between January 1, 1994, and December 31, Examinations for screening were identified by the endoscopist s assessment of the reason for the FS, which was recorded either as screening or consult/diagnostic. If this assessment was missing (3.1% of all eligible subjects), self-reported absence of symptoms was considered to indicate a screening examination. To limit this study to a low-risk, screening population, we excluded the few people with a serious family history of colorectal cancer (defined as 2 or more first degree relatives or at least one first degree relative diagnosed before age 55) or a history of colorectal polyps or cancer. Persons who had a colonoscopy on the same day as FS(n 71) were excluded because subsequent complications would more likely be attributable to the colonoscopy. After these exclusions, 107,704 persons remained with 109,534 eligible screening FSexaminations. Repeat examinations were most often performed because of incomplete bowel preparation on the initial screening. Data Sources During the study period, standardized sigmoidoscopy and colonoscopy reports from each Northern California KP facility were routinely entered into a computer database. Demographic information was obtained from health plan databases. Information on hospitalization was extracted from Kaiser Foundation Health Plan and Hospitals (KFHP/H) databases of discharge diagnoses from KFHP/H-owned hospitals and from files of claims from non-kfhp/h hospitals for both referred and emergency hospitalizations. Information on mortality was retrieved from the California State registry. Observation Periods We searched for gastrointestinal complications for up to 28 days (4 weeks) after FS, beginning with the day of the procedure. Hospitalizations and deaths from MIs were sought for 364 days (52 weeks), including the day of the FS. The incidence of MI during the first 4 weeks was compared with the rate during the remainder of the year (48 weeks). The observation period was censored if participants had a colonoscopy subsequent to the FS, if another FS was performed, or if they ceased to be a member of KFHP. Complete follow-up for the full 4 weeks was possible after 97.8% of examinations. Censoring of follow-up most often occurred because a colonoscopy or another FSoccurred. Complete, yearlong follow-up for MI using electronic databases was possible in 88.6% of examinations. Identification of Eligible Events Hospitalizations within 4 weeks of FSat both KFHP/H and non-kphp/h hospitals were identified using electronic databases. We reviewed charts for all cases in which the subject was hospitalized urgently or emergently for at least 24 hours and at least 1 discharge diagnosis or procedure indicated a gastrointestinal complication. We also reviewed all available charts when the principal discharge diagnosis was MI. Deaths within 28 days of FSwere identified through linkage with the California State Registry, with manual review of available records to confirm identities. Gastrointestinal Complications Three diagnoses formed an a priori list of gastrointestinal complications for this analysis: gastrointestinal bleeding, colonic perforation, and diverticulitis. For the present study, however, we initially cast a broad net by including all hospitalizations within 28 days of the FS, regardless of diagnosis. For the 240 persons hospitalized within 4 weeks chart review was completed on 167 (70%) (Figure 1). Full charts were not reviewed for the other 73 because neither primary nor secondary discharge diagnoses could plausibly be related to the FS. Chart reviews included all cases in which the discharge diagnoses included 1 of the following: bowel perforation, complication of procedure, abdominal pain, lower gastrointestinal bleeding, fever, colitis of unclear etiology, diverticulitis, or anemia (International Classification of Diseases, Ninth Revision [ICD-9] codes 998.2, , 588.9, 578.1, , 780.6, , , 280, ). Review also included all cases in which the electronic database indicated 1 or more of the following procedures: suture of laceration of large intestines, partial excision of large intestines, total intra-abdominal colectomy, laparotomy, colostomy, ileostomy, and abdominal resection of rectum (ICD-9 codes 4675, , 458, , , and 48.5). Chart review indicated that in

3 1788 LEVIN ET AL. GASTROENTEROLOGY Vol.123, No.6 Figure 1. Flow chart for chart review of gastrointestinal complications (n 109,534 screening FS examinations). 81 cases the patient was hospitalized for nongastrointestinal causes; in 65 of these cases, anemia was the only possibly related discharge diagnosis, but chart review indicated no overt evidence of gastrointestinal bleeding. Among the other 86 cases with chart review, in 12 a colonoscopy had intervened between the FSand the hospitalization and in 50 cases hospitalization was solely for treatment of colorectal cancer. The remaining 24 cases were identified as probable FScomplications. Of these 24 probable complications, 7 were serious (colonic perforation, lower gastrointestinal bleeding leading to blood transfusion, colitis that was not present on the index FS, and diverticulitis requiring surgery). The 17 nonserious complications included fever of uncertain etiology, unexplained abdominal pain, and lower gastrointestinal bleeding without transfusion (n 9). Acute Myocardial Infarction To evaluate the risk for MI after FS, we searched electronic databases of discharge diagnoses and the state death registry to identify principal discharge diagnoses of MI and deaths because of MI during the 4 weeks after FS, beginning with the day of the procedure. Chart review was conducted for all MIs in the first 4 weeks to confirm the electronic diagnosis of MI and to determine whether hospitalization was for an elective cardiac procedure (e.g., coronary revascularization or angiography). An acute MI was defined at chart review as the presence of characteristic electrocardiographic changes or documented elevation of the creatine kinase myocardial band fraction. Chart review confirmed 28 of 33 diagnoses of MI coded in the electronic data. The unconfirmed MI cases were either elective hospitalizations for cardiac procedures where MI was listed as the principal diagnosis or a delayed hospitalization for an MI in which the timing of the actual MI could not be identified. Based on the success of our validation study of electronic MI diagnosis and prior research at KP, 16 we relied on electronic data to calculate rates of acute MI in the study population over 52 weeks (including the first 4 weeks), beginning with the day of a screening FS. An acute MI was defined as a nonroutine hospitalization with a principal discharge diagnosis of myocardial infarction (ICD-9 code 410.x) that resulted in either death or a hospital stay of 24 hours or more. We also included confirmed deaths outside the hospital, if the principal cause was identified as MI in the state death registry. All MIs during the 52-week study period recorded in the electronic databases

4 December 2002 SCREENING SIGMOIDOSCOPY COMPLICATIONS 1789 Table 1. Characteristics of the Examination, Patient Demographics, and Endoscopist Type Cases in study population Cases with complications Examination All (N 109,534) Male (N 56,320) Female (N 53,214) Any (N 24) Serious (N 7) Polyps present, N (%) 21,280 (19.4) 13,148 (23.4) 8132 (15.3) 9 (37.5) 1 (14.3) Biopsy or polypectomy, N (%) 16,991 (15.5) 10,547 (18.6) 6545 (12.3) 9 (37.5) 1 (14.3) Limited by pain/angulation, N (%) a 26,473 (24.2) 8930 (15.8) 17,543 (33.0) 7 (29.2) 3 (42.9) Mean depth of insertion, cm ( SE) a 53.6 ( 10.4) 55.6 ( 18.9) 51.6 ( 11.5) 51.9 ( 14.7) 44.3 ( 18.2) Patient demographics Mean age, yr ( SE) 61.0 ( 17.6) 60.9 ( 17.7) 61.2 ( 17.7) 61.5 ( 9.3) 65.4 ( 8.8) Age50 59, N (%) 51,529 (47.0) 27,042 (48.0) 24,487 (46.0) 12 (50.0) 2 (29.0) Age60 69, N (%) 39,868 (36.4) 20,301 (36.1) 19,567 (36.8) 6 (25.0) 3 (43.0) Age70 79, N (%) 18,137 (16.6) 8977 (15.9) 9160 (17.2) 6 (25.0) 2 (29.0) Female, N (%) 53,214 (48.6) 5 (20.8) 3 (42.9) Endoscopist type Gastroenterologist, N (%) 30,374 (27.9) 15,645 (28.0) 14,729 (27.9) 5 (20.8) 2 (28.6) Other physician, N (%) 47,226 (43.4) 25,271 (45.2) 21,951 (41.6) 13 (54.2) 2 (28.6) Nonphysician, N (%) 31,136 (28.6) 15,004 (26.8) 16,132 (30.6) 6 (25.0) 3 (42.9) a As defined by the endoscopist on standardized procedure reports. and the death registry were included; 6.6% were second MIs within the 52-week study period. Statistical Analysis The gastrointestinal complication rates per 100,000 persons were calculated for the 28 days (4 weeks) beginning with the day of the FSprocedure, along with a 95% confidence interval (CI) based on a binomial distribution. We used logistic regression to determine whether the complication rate varied by the age and gender of the subject, the type of provider, and whether tissue was removed at FS(by biopsy or polypectomy). Because complications are rare events, the odds ratios provide a reasonable estimate for the relative risk. We used Poisson regression to determine whether the rate of MIs in the period shortly after FSwas elevated relative to the rate during the remainder of the 52-week period. Specifically, we checked for elevations in the number of MIs per 1000 person years during 3 successively longer periods beginning with the day of the FS(day 0), the first 3 days (days 0 2), the first week (days 0 6), and the first 4 weeks (days 0 27). We then compared the rates during these periods with the rate observed during the remaining 48 weeks (days ) after FS. Comparing rates in the same population at different time periods avoids problems caused by the voluntary nature of screening. Individuals with better health habits may be more likely to undergo screening FSthan those with preexisting health problems. To avoid possible problems caused by attrition or repeat events, we also compared the rate of MI in the first 3 days and the first week with the rate during the remainder of the first 4 weeks, both by using events confirmed by chart review and by using unconfirmed electronically identified events. Results A total of 24 persons had gastrointestinal complications of FSwithin 28 days, and 7 of these were serious. Overall, the complication rate was 21.9/100,000 procedures. Serious complications included 2 perforations requiring surgery (1 was a perforated colorectal cancer presenting 11 days after the index FSthat was not seen on FSdue to a poor preparation limiting sigmoidoscope insertion), 2 episodes of lower gastrointestinal bleeding requiring transfusion, 2 episodes of diverticulitis requiring surgery, and 1 episode of unexplained colitis reported by the treating physician to be ischemic. Episodes of nonserious complications included 4 of fever, 4 of unexplained abdominal pain, and 9 of self-limited lower gastrointestinal bleeding that did not require a transfusion. The mean interval between FSand hospitalization for bleeding was 7 days (range, 2 13 days). Bleeding after hot biopsy or snare polypectomy averaged 9 days after the date of FS. Just under half (48.6%) of the study subjects were women. The mean age for both men and women was 61 years (Table 1). Nongastroenterologist physicians or nurse endoscopists performed 72% of the eligible screening FS. On average, the gastroenterologists performed 344 FSs per year, the nonphysicians performed 482 FS per year, and the nongastroenterologist physicians performed 155 FSs per year. Men were significantly more likely than women to have polyps identified and more likely to undergo biopsy or polypectomy (P ). The mean depth of insertion was similar for persons with

5 1790 LEVIN ET AL. GASTROENTEROLOGY Vol.123, No.6 Table 2. Rate of Complications of Flexible Screening Sigmoidoscopy by Age, Gender, Presence of Biopsy or Polypectomy, and Endoscopist Type Complications (N) Complication rate (per 100,000) (95% CI) Odds ratios (95% CI) Age(yr) ( ) 1.0 (Referent) ( ) 0.7 ( ) ( ) 1.4 ( ) Gender Female ( ) 1.0 (Referent) Male ( ) 3.6 ( ) Biopsy or polypectomy No ( ) 1.0 (Referent) Yes ( ) 2.5 ( ) Endoscopist type Gastroenterologist ( ) 1.0 (Referent) Other physician ( ) 1.7 ( ) Nonphysician ( ) 1.2 ( ) Total ( ) and without complications. Cases with serious complications were more likely to be limited by pain or angulation of the colon, as reported by the endoscopist. Male gender and biopsy/polypectomy were significantly associated with increased risk of complications (P 0.05) (Table 2). The risk of a complication was 3.6 times as high for men as for women and 2.5 times as high when biopsy or polypectomy occurred. Neither the professional background of the examiner nor age affected risks significantly (Table 2). Including all of the 4 factors in Table 2 in a multivariate logistic regression model (results not shown) did not change any of the relative risk estimates substantially. In particular, the greater frequency of biopsy or polypectomy in male subjects did not account for the higher risks they experienced. Biopsy was associated with a complication rate of 19 per 100,000 (95% CI, per 100,000), polypectomy was associated with a complication rate of 68 per 100,000 (95% CI, ). There were a total of 9 events associated with biopsy or polypectomy, 1 with polypectomy and 8 with biopsy. Hospitalizations for MI were identified in 33 subjects in the 4 weeks after FS, including 7 in the first week. The rate of MI during the first 2 days after FSwas 4.5 per 1000 person-years of observation (95% CI, ); in days 0 6 (first week) after FS, the rate was 3.3 per 1000 person-years (95% CI, per 1000) (Table 3). Rates during the first week and first month after FSwere similar to the rates for the remainder of the year (rate ratios, 0.7 [95% CI, ]) and 0.8 (95% CI, ), respectively. The MI rate was not increased during days 0 6 for those who had tissue removed. There was also no significant elevation of the MI rate in the first 3 days or in the first week in a comparison with the remainder of the first 4 weeks, using either events confirmed by chart review or unconfirmed electronically identified events. Ten subjects died in the 4 weeks after FS, including 5 from cardiovascular causes. Deaths from other causes appeared unrelated to the FS. Discussion The principal finding of this study is that the rate of complications after FSis modest. Approximately 1 in 5000 screening subjects was hospitalized for a gastrointestinal complication, and 1 in 16,000 was hospitalized for a serious complication. Colonic perforations, serious bleeding, and diverticulitis leading to surgery each occurred in this population less often than 1 in 50,000 examinations. Of particular note is the finding that nonphysicians perform FSas safely as gastroenterologists and nongastroenterology physicians. Our observed FSperforation rate (1 in 50,000) was similar in magnitude to that of the 10-year experience of the Mayo Clinic-Scottsdale in Arizona (1 in 25,000). 12 Table 3. Incidence Rates and Rate Ratios for Acute Myocardial Infarction Days after FS MIs Rateper 1000 person-years (95% CI) Rateratios (95% CI) relative to days ( ) 0.9 ( ) ( ) 0.7 ( ) ( ) 0.8 ( ) ( )

6 December 2002 SCREENING SIGMOIDOSCOPY COMPLICATIONS 1791 Our rate is substantially lower, however, than the 1 in 500 to 1 in 1300 rate of perforations seen in 2 recent colonoscopy series, including both screening and nonscreening examinations. 12,17 The VA Cooperative screening colonoscopy study reported no perforations in approximately 3000 colonoscopy examinations, but it reported a hospitalization rate for serious complications of 1 in A common site of colonic perforation in both colonoscopy and FSseries is the sigmoid colon, which is traversed by the endoscope in both examinations. One possible reason for the relatively greater safety of FScompared with colonoscopy is that the unsedated FSpatient is able to provide sensory feedback to the endoscopist, limiting endoscope insertion in the event of barotrauma or endoscope abrasion that disrupts the integrity of the colonic wall. The sedation of patients for colonoscopy may blunt this feedback. 12 The more frequent use of polypectomy in colonoscopy may also contribute to the apparent higher complication rates. Men were more likely to experience a complication than women, and our data lack detail to shed light on this phenomenon. One possible explanation for this is that the women, on average, had a shorter depth of insertion than did the men, possibly because of anatomic differences between the sexes. This greater depth of insertion may have been associated with higher levels of barotrauma and mechanical trauma in men than in women. Although acute MIs occurred in the first days and weeks after FS, the rate of MIs was not significantly greater than one might expect in screening-age subjects. Because the background rate for MI in this age group is relatively high, an appropriate comparison group is needed to interpret the event rate in the early postprocedure period. Because screening is voluntary, screenees may differ in important ways from nonparticipating but otherwise eligible KFHP members or from the general population. In this analysis the rate of MI in the screened population in the first 4 weeks after the FSwas compared with the rate of MI in the same group in the remainder of the year after the FSin a cross-over type design that should overcome problems with differences between patients who volunteer for screening and those who are not screened. 19 We relied on MI as the marker of cardiovascular events because it is easily verified using objective laboratory or electrocardiographic criteria and because electronic discharge diagnoses are generally accurate. 16 Other events, such as unstable angina, cannot be identified reliably without chart review, making it impractical to observe for an extended period. Moreover, the diagnosis of unstable angina is often subjective, depending on the judgment of the treating physician, making retrospective evaluation of risk more problematic. A potential limitation of this study is our reliance on administrative data to detect complication-related events. However, these databases capture 100% of admissions to KFHP hospitals. There is slightly less certainty regarding stays in non KFHP-owned hospitals. A small proportion of members have dual insurance, which would pay for care in outside hospitals. However, it seems highly likely that hospitals would bill KP for services provided to KFHP members, particularly if they were admitted for a complication of a procedure performed at KP. Another concern is that our study did not assess minor side effects that do not lead to hospitalization but may carry direct and indirect costs for screening participants, such as abdominal discomfort or inability to work. On the other hand, we attributed all possible events to an FS-related complication, even though events such as fever, abdominal pain, or diverticulitis could occur spontaneously in the absence of an FS. Given the pathophysiology of diverticulitis, for example, any procedure that increased trauma to the colonic wall could be considered as a plausible potential cause of this condition. We also excluded hospitalizations that occurred after a colonoscopy because the colonoscopy was a likely cause for those complications. If those colonoscopy-related admissions were all complications that were caused by FSour estimate would be increased to approximately 1 in 3000 (from 1 in 5000). The primary strengths of this study are the size of the study population, the diversity of the examiners performing FS, and the variety of settings in which the examinations were performed. In addition, by combining our FSdatabase with an electronic record of hospitalizations and claims, we achieved over 97% 4-week follow-up. The 4 weeks of follow-up also allow the detection of hospitalizations for delayed biopsy and polypectomy bleeding. In summary, the complication rate after FSappears to be quite low and substantially below that reported elsewhere for colonoscopy. From the standpoint of risk relative to value, FSis a worthwhile alternative to colonoscopy for colorectal cancer screening in average risk individuals because it detects from two thirds to four fifths of the advanced neoplasia found by colonoscopy 18,20 while carrying only a small fraction of its risk. Because patient and provider preferences vary considerably, information on risk of complications from screening tests

7 1792 LEVIN ET AL. GASTROENTEROLOGY Vol.123, No.6 should be incorporated into any attempts at informed shared decision-making regarding screening for colorectal cancer. References 1. Selby JV, Friedman GD, Quesenberry CP, Weiss NS. A casecontrol study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992;326: Newcomb PA, Norfleet RG, Storer BE, Surawicz TS, 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. A case-control study among veterans. Arch Intern Med 1995;155: Frazier AL, Colditz GA, Fuchs CS, Kuntz KM. Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 2000;284: Salzmann P, Kerlikowske K, Phillips K. Cost-effectiveness of extending screening mammography guidelines to include women 40 to 49 years of age. Ann Intern Med 1997;127: Loeve F, Brown ML, Boer R, van Ballegooijen M, van Oortmarssen GJ, Habbema JD. Endoscopic colorectal cancer screening: a costsaving analysis. J Natl Cancer Inst 2000;92: Woolf SH. The best screening test for colorectal cancer a personal choice. N Engl J Med 2000;343: WayeJD, Kahn O, Auerbach ME. Complications of colonoscopy and flexible sigmoidoscopy. Gastrointest Endosc Clin N Am 1996; 6: Atkin WS, Hart A, Edwards R, McIntyreP, Aubrey R, WardleJ, Sutton S, Cuzick J, Northover JM. Uptake, yield of neoplasia, and adverse effects of flexible sigmoidoscopy screening. Gut 1998; 42: Jentschura D, Raute M, Winter J, Henkel T, Kraus M, Manegold BC. Complications in endoscopy of the lower gastrointestinal tract. Therapy and prognosis. Surg Endosc 1994;8: Verne JE, Aubrey R, Love SB, Talbot IC, Northover JM. Population based randomised study of uptake and yield of screening by flexible sigmoidoscopy compared with screening by faecal occult blood testing. BMJ 1998;317: Anderson ML, Pasha TM, Leighton JA. Endoscopic perforation of the colon: lessons from a 10-year study. Am J Gastroenterol 2000;95: Thiis-Evensen E, Hoff GS, Sauar J, Langmark F, Majak BM, Vatn MH. Population-based surveillance by colonoscopy: effect on the incidence of colorectal cancer. Telemark Polyp Study I. Scand J Gastroenterol 1999;34: Palitz AM, Selby JV, Grossman S, Finkler LJ, Bevc M, Kehr C, Conell CA. The Colon Cancer Prevention Program (CoCaP): rationale, implementation, and preliminary results. HMO Practice 1997;11: Levin TR, Palitz A, Grossman S, Conell C, Finkler L, Ackerson L, Rumore G, Selby JV. Predicting advanced proximal colonic neoplasia with screening sigmoidoscopy. JAMA 1999;281: Selby JV, Fireman BH, Lundstrom RJ, Swain BE, Truman AF, Wong CC, Froelicher ES, Barron HV, Hlatky MA. Variation among hospitals in coronary-angiography practices and outcomes after myocardial infarction in a large health maintenance organization. N Engl J Med 1996;335: Farley DR, Bannon MP, Zietlow SP, Pemberton JH, Ilstrup DM, Larson DR. Management of colonoscopic perforations. Mayo Clin Proc 1997;72: Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec G. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. Veterans Affairs Cooperative Study Group 380. N Engl J Med 2000;343: Louis TA, Lavori PW, Bailar JC III, Polansky M. Crossover and self-controlled designs in clinical research. N Engl J Med 1984; 310: Imperiale TF, Wagner DR, Lin CY, Larkin GN, Rogge JD, Ransohoff DF. Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N Engl J Med 2000;343: Leard LE, Savides TJ, Ganiats TG. Patient preferences for colorectal cancer screening. J Fam Pract 1997;45: Dominitz JA, Provenzale D. Patient preferences and quality of life associated with colorectal cancer screening. Am J Gastroenterol 1997;92: Pignone M, Bucholtz D, Harris R. Patient preferences for colon cancer screening. J Gen Intern Med 1999;14: Lewis JD, Asch DA, Ginsberg GG, Hoops TC, Kochman ML, Bilker WB, Strom BL. Primary care physicians decisions to perform flexible sigmoidoscopy. J Gen Intern Med 1999;14: Received March 20, 2002.Accepted September 5, Address requests for reprints to: Theodore R. Levin, M.D., Kaiser Permanente Division of Research, 2000 Broadway, 2nd Floor, Oakland, California Theodore.Levin@kp.org; fax: (510) Supported by The Centers for Disease Control and Prevention, through a Task Order to the Alliance of Community Health Plans.