Colorectal cancer (CRC) is a major medical and public. Effect of Screening Colonoscopy on Colorectal Cancer Incidence and Mortality

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2009;7: Effect of Screening Colonoscopy on Colorectal Cancer Incidence and Mortality CHARLES J. KAHI,*, THOMAS F. IMPERIALE,*, BETH E. JULIAR, and DOUGLAS K. REX* *Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine; The Richard L. Roudebush VA Medical Center; Regenstrief Institute, Inc; and Department of Biostatistics, Indiana University, Indianapolis, Indiana This article has an accompanying continuing medical education activity on page 711. Learning Objectives After completing this CME activity, the learner should be able to understand the public impact of colon cancer in the United States and to study the impact of colonoscopy and polypectomy on colon cancer incidence and mortality. Podcast interview: see CME exam on page 711; see Editorial on page 714. BACKGROUND & AIMS: Colonoscopy is used widely for colorectal cancer (CRC) screening; however, its long-term impact on the incidence and mortality of CRC is not known. METHODS: We assessed CRC incidence and mortality in a group of asymptomatic average-risk patients who underwent screening colonoscopy between 1989 and 1993 at a university hospital. By using standardized incidence ratios and standardized mortality ratios, we compared our observed CRC rates with expected rates from the Surveillance, Epidemiology, and End Results (SEER) data. RESULTS: The cohort comprised 715 patients (mean age, y; 59% male; 95% Caucasian) with 10,492 patient-years of follow-up. There were 12 cases of CRC: 5 found at baseline and 7 found after a median follow-up period of 8 years (range, 3 16 y). When the first 2 years of follow-up were excluded, there were 7 incident cases of CRC (95% confidence interval [CI], 2 13) over 9075 person-years of follow-up. The expected number based on SEER data was 21. The incidence rate was 0.77 cases per 1000 person-years, and the standardized incidence ratio was 0.33 (95% CI, ), consistent with a relative risk reduction in CRC incidence of 67%. Three patients died from CRC (95% CI, 0 9). The expected number of deaths based on SEER data was 9. The mortality rate was 0.29 per 1000 person-years, and the standardized mortality ratio was 0.35 (95% CI, ), consistent with a relative reduction in CRC death of 65%. CONCLUSIONS: In this average-risk cohort, CRC incidence and mortality were reduced after screening colonoscopy. These results provide additional evidence for the effectiveness of colonoscopy as a primary CRC screening modality. To view this article s video abstract, go to the AGA s YouTube Channel. Colorectal cancer (CRC) is a major medical and public health challenge: with nearly 150,000 new cases and 50,000 deaths expected in 2008, it is the third most common and second leading cause of cancer deaths in the United States. 1 A significant proportion of CRC cases and deaths can be prevented through screening. Since 2001, colonoscopy has been an accepted modality for screening in average-risk individuals. 2,3 Several factors have contributed to this trend, foremost of which is that colonoscopy has high sensitivity for detecting CRC and precancerous adenomatous polyps, and allows complete examination of the colon and rectum and removal of polyps at the time of detection. 4 7 More recently, reimbursement by third-party payers including Medicare, 8 high-profile endorsement by celebrities, 9 and heightened media attention have increased public acceptance and demand for screening colonoscopy. 10 Recent data have shown a decline in CRC incidence in the United States, 11 and the emergence of colonoscopy as a primary screening modality is associated temporally with detection of earlier-stage disease. 12 Despite these observations and the increased use of colonoscopy for screening, there are surprisingly few data that quantify the impact of colonoscopy on CRC incidence and mortality in average-risk populations. The National Polyp Study reported a 76% to 90% reduction in CRC incidence in patients with adenomas who underwent clearing colonoscopy, 13 and a substantial reduction in mortality after long-term follow-up. 14 However, other adenoma cohort studies, including dietary prevention and chemoprevention trials, have shown lower levels of protection after colonoscopy and polypectomy The applicability of these mixed findings to the general population is problematic because the individuals enrolled in these studies had adenomas and underwent polypectomy, and may constitute a subgroup at higher risk for developing CRC than patients without colon neoplasms. Conversely, a recent Canadian claims-based study found that the risk of developing CRC was 60% to 70% of the risk in the general population after a negative colonoscopy (defined as colonoscopy without biopsy or polypectomy), and remained decreased for more than 10 years after the procedure. 18 However, these findings may not apply to patients who have adenomatous polyps and require polypectomy and subsequent surveillance. In daily clinical practice, average-risk patients who undergo screening colonoscopy are a heterogeneous group: most have no colon neoplasms, a smaller proportion has adenomas and requires subsequent surveillance after polypectomy, and a minority has Abbreviations used in this paper: CI, confidence interval; CRC, colorectal cancer; SEER, Surveillance, Epidemiology, and End Results; SIR, standardized incidence ratio by the AGA Institute /09/$36.00 doi: /j.cgh

2 July 2009 SCREENING COLONOSCOPY, CRC INCIDENCE AND MORTALITY 771 CRC. Screening colonoscopy conceptually benefits all of these subgroups: patients with no neoplasms are identified and followed up at relatively long intervals, patients with adenomas undergo polypectomy, which confers protection against CRC, and patients with CRC are detected at an early, treatable stage. We report the long-term follow-up of a cohort of such patients after screening colonoscopy. Nearly 2 decades ago, one of the authors (D.K.R.) performed one of the earliest studies of screening colonoscopy in selected average-risk persons 19 ; this report describes the long-term impact of such screening on the incidence and mortality of CRC for up to 18 years after the initial procedure. Methods Patients This was a follow-up study of subjects enrolled in a previous screening colonoscopy study. 19 It was approved by the Institutional Review Board at Indiana University Purdue University at Indianapolis as an amendment to the original protocol. The methods and findings of the original screening study have been reported previously. 19,20 In summary, between 1989 and 1993 a written invitation was sent to 5000 physicians and dentists, 12,000 nurses, and all of their spouses, offering a screening colonoscopy. Exclusion criteria included age younger than 50 years, previous colon cancer or adenoma, barium enema or colonoscopy within 3 years for any indication, inflammatory bowel disease, Peutz Jeghers syndrome, previous breast or uterine cancer, abdominal radiation, coagulopathy or prosthetic heart valve, family history of colonic neoplasia in 2 or more first-degree relatives, and family history of colon cancer in a first-degree relative before the age of 40. All potential subjects completed a detailed questionnaire to confirm their asymptomatic status. All subjects underwent fecal occult blood testing using Hemoccult II (Beckman Coulter, Fullerton, CA) during the week before colonoscopy. In the original study, subjects with positive test results underwent colonoscopy but were not included in the analysis. 19 In the current study, subjects were included irrespective of their baseline fecal occult blood testing status because other subsequent studies of screening colonoscopy have not prescreened patients with fecal occult blood testing. 4,6,7,21 Subjects underwent complete colonoscopy to the cecum, and all visualized polyps were resected completely. The histology of all lesions was reviewed by a single pathologist with a special interest in gastrointestinal pathology. Data Collection Follow-up information was obtained by several methods. All subjects were contacted by telephone, or interviewed personally at the time of clinic or colonoscopy follow-up evaluation at Indiana University. Many subjects who had polyps either at the original examination or at a follow-up examination still were being followed up at Indiana University Hospital; their information was retrieved by chart review. Other subjects underwent follow-up evaluation at other institutions; they also were contacted by telephone and permission was obtained to contact their medical providers and to receive copies of follow-up procedures, including pathology results. In instances in which a subject had died, permission was obtained from the next of kin, and the cause of death was determined by contacting their primary health care providers and reviewing death certificates when possible. Reference Groups We compared the observed incidence and mortality rates of CRC in our study group with the rates expected based on the Surveillance, Epidemiology, and End Results (SEER) program. SEER monitors the incidence of CRC and the mortality rate from this malignancy in 10 registries in the United States. 22 For this study, we used data from the 9 SEER registries of Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco Oakland, Seattle Puget Sound, and Utah. We used age- and sex-specific rates for the calendar years 1989 through 2007 because this period overlapped with the accrual and follow-up phases of the study. For the calendar years 2005, 2006, and 2007, we used the rates for the calendar year 2004 because these were the most recent available data. Statistical Analysis The index colonoscopy was considered the study entry date; all subjects were followed up until death or study censure date of September 15, The number of person-years at risk was calculated for each patient as the interval from the index colonoscopy to diagnosis of CRC, death, or study censure date; subjects with incomplete follow-up were censored at the time of their last documented follow-up evaluation. The median time to CRC diagnosis and cumulative incidence curves were calculated. Incidence rates were calculated as the number of CRCs divided by person-years of follow-up. The 95% confidence intervals (CIs) for the number of observed cancers were constructed using a Poisson distribution. We compared the overall CRC incidence rate in our study group with SEER rates as follows. We stratified person-years at risk according to age, sex, and calendar year; the age strata were divided into quintiles starting at age 50. Incidence rates of CRC in SEER corresponding to each age stratum, sex category, and calendar year were multiplied by the number of person-years at risk in that stratum, yielding an expected number of cases. We then summed the expected number of cases per stratum to yield a total number of expected cases. The standardized incidence ratio (SIR) of the number of observed CRC cases compared with the number of expected cases based on SEER then was calculated. The reduction in the incidence of colorectal cancer was calculated as 100 (1 - SIR). To calculate the SIR using only incident cases that occurred after 2 years of follow-up, we repeated the calculations after excluding patient-years at risk during the first 2 years after the index colonoscopy. A standardized mortality ratio of the number of observed CRC-specific deaths compared with the number expected based on SEER data was calculated in a manner similar to the SIR. For all observed cases and deaths, 95% CIs were calculated using a Poisson distribution, according to the method of Breslow and Day. 23 The CIs for the SIR and standardized mortality rate were obtained by dividing the upper and lower 95% confidence limits of the observed number of cases by the expected number of cases. All statistical analyses were performed using SAS version 9.0 (SAS Institute Inc, Cary, NC). Results Of the original 733 patients, follow-up information could not be obtained in 16 patients; 2 patients declined to participate and were excluded from further analysis. The re-

3 772 KAHI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 7 Table 1. Characteristics of Patients Diagnosed With CRC Patient Age, y Sex Time to diagnosis, mo a Presentation Cancer location Dukes stage Treatment Outcome 1 64 M 0 Asymptomatic Ascending A Right hemicolectomy b Alive 2 74 M 0 Asymptomatic, Sigmoid B Low anterior resection Died c FOBT 3 62 M 0 Asymptomatic Sigmoid A Endoscopic resection Alive 4 50 F 0 Asymptomatic, Transverse B Right hemicolectomy Alive FOBT 5 50 M 0 Asymptomatic Sigmoid A Right hemicolectomy, Alive sigmoidectomy d 6 56 M 95 Asymptomatic Proximal A Right hemicolectomy Alive ascending/ic valve 7 63 M 53 Asymptomatic Cecum A Right hemicolectomy Died 8 51 M 36 Abdominal pain Sigmoid B Sigmoidectomy Alive 9 56 M 193 Hematochezia Cecum A Right hemicolectomy Alive M 99 Anemia, FOBT Hepatic flexure D Right hemicolectomy, Alive excision of liver metastases, gastrojejunostomy M 164 RLQ pain, melena, Ascending colon/ D Right hemicolectomy, Died c anemia hepatic flexure partial omentectomy M 138 Anemia, FOBT Cecum C Right hemicolectomy Died c FOBT, fecal occult blood testing; IC, ileocecal; RLQ, right lower quadrant. a Patients 1 5 were diagnosed at the index colonoscopy (prevalent CRCs). b Malignant polyp was removed endoscopically; subsequent right hemicolectomy showed no evidence of residual cancer. c Cause of death was CRC. d Right hemicolectomy was performed for multiple right-sided adenomatous polyps. maining 715 patients comprise the study population. The mean age at study entry was years (range, y), 59% were male, and 95% were Caucasian. A total of 190 patients (26.6%) had at least one adenoma at baseline (excluding patients with CRC at the index colonoscopy). Of these 190 patients, 38 (20%) had 3 or more adenomas, 22 (12%) had an adenoma 1 cm or larger, 25 (13%) had an adenoma with villous histology, and 2 (1%) had an adenoma with high-grade dysplasia (some patients had adenomas with more than 1 feature). Five CRCs were diagnosed at the index colonoscopy; 7 incident CRCs were discovered after a median of 8 years (range, 3 16 y) and 10,492 person-years of follow-up. The clinical features of these 12 patients with CRC are detailed in Table 1. The colonoscopic surveillance history of the 7 patients who developed CRC after the index colonoscopy is shown in Figure 1. Patients with adenomas at baseline underwent 714 surveillance colonoscopies over 2685 person-years of follow-up, equivalent to 1 procedure every 3.8 years. Patients without adenomas at baseline underwent 1401 colonoscopies over 7745 personyears of follow-up, equivalent to 1 procedure every 5.5 years (P.0001; nonparametric Mann Whitney U test). Standardized Incidence and Mortality Ratios We observed 12 total cases of CRC (95% CI, 5 19) over 10,492 person-years of follow-up. The expected number of CRC cases based on SEER data was 23. The incidence rate was 1.14 cases per 1000 person-years, and the corresponding SIR was 0.52 (95% CI, ), consistent with a relative risk reduction in colorectal cancer incidence of 48%. When the first 2 years of follow-up were excluded, there were 7 incident cases of CRC (95% CI, 2 13) over 9075 person-years of follow-up. The expected number of cases based on SEER data was 21. The incidence rate was 0.77 cases per 1000 person-years, and the corresponding SIR was 0.33 (95% CI, ), consistent with a relative risk reduction in CRC incidence of 67%. During the study follow-up period, 3 patients died from CRC (95% CI, 0 9). The expected number of CRC deaths based on SEER data was 9, yielding a mortality rate of 0.29 per 1000 person-years. The standardized mortality ratio was 0.35 (95% CI, ), consistent with a relative reduction in CRC death of 65%. Figure 1. Colonoscopic screening and surveillance history of patients (n 7) with incident CRC.

4 July 2009 SCREENING COLONOSCOPY, CRC INCIDENCE AND MORTALITY 773 Discussion This long-term follow-up study shows that in an average-risk general population undergoing screening colonoscopy, the risk of developing CRC is reduced by 48% to 67%, and the risk of death from CRC is reduced by 65% compared with a reference population. These findings add to the growing body of literature supporting the effectiveness of colonoscopy as a primary tool for CRC screening, and show that the benefit extends for up to 18 years after screening has occurred. To our knowledge, there are no studies of long-term follow-up after screening colonoscopy to which our results can be compared directly. However, there are several studies that have assessed the impact of colonoscopy and polypectomy on CRC incidence and mortality. The National Polyp Study reported that patients with adenomas who underwent colonoscopic polypectomy experienced a 76% reduction in CRC incidence compared with a SEER reference population, 13 and the incidence rate was 0.6 per 1000 person-years. European studies have reported comparable reductions in CRC incidence after colonoscopy with polypectomy However, the Polyp Prevention Trial and Wheat-Bran Fiber Trial have reported a higher risk of incident CRC after colonoscopy with polypectomy, ranging from 2.2 to 2.4 per 1000 person-years. 15,27 In the recently published pooled analysis of 3 chemoprevention trials, the incidence of CRC was 1.7 per 1000 person-years, and was not significantly different than the rate expected based on SEER data. 16 The reasons for these conflicting results are not clear, but likely reflect differences in study methodology. Patients enrolled in the National Polyp Study underwent thorough colonoscopic clearing before randomization (13% received at least 2 examinations to clear the colon), and the procedures were performed by experts. Furthermore, patients with adenomas 3 cm or greater were excluded from the National Polyp Study, which was not the case in some of the other trials. 2 In contrast with these postpolypectomy studies, Singh et al 18 reported that the CRC incidence in a cohort with negative baseline colonoscopy was 1.1 per 1000 person-years. Our incidence rate was 1.14 per 1000 person-years overall, and 0.77 per 1000 person-years when the first 2 years of follow-up were excluded. Although these rates appear closer to those reported in the National Polyp study and in the study by Singh et al, 18 several factors make a direct comparison difficult. An obvious reason is different study methodology and enrollment criteria because we did not restrict enrollment to patients with baseline adenomas or patients with negative colonoscopies. Another factor that may account for differences among the studies is the unclear distinction between prevalent and incident cancers in the comparison group. Ad hoc registries, such as SEER, collect data on new cases of cancer reported in their corresponding catchment regions. This depends on the identification of clinically overt cancer cases, but may not accurately account for cases of slowgrowing cancers that have not yet become clinically evident, and for which the catchment population has not undergone thorough and uniform screening. 28 This is usually not problematic for diseases with a rapid course and high fatality, but in the case of a cancer with relatively long latency such as colorectal cancer, the statistical and epidemiologic consequences of this misclassification may be substantial. Because adequate screening in the population has not occurred, clinically silent cases of CRC will not be accounted for, leading to an underestimate of true prevalence. Conversely, when these cases eventually become overt, they are counted as new, leading to an overestimate of true incidence. 28 This issue has affected many studies reporting cancer statistics, and their calculated rates, which usually are assembled from reports of overtly detected cases of cancer, represent neither incidence nor prevalence. 28 Studies reporting SIRs of observed incidence compared with SEER should ideally make adjustments for prevalent cases, especially with diseases with long latency. In our study, we addressed this problem by using 2 approaches: the first was to report all cases of CRC (prevalent cases at baseline colonoscopy, and incident cases that occurred during follow-up evaluation), and compared this incidence rate with SEER data. The advantage of this approach is that all cancers are accounted for, and highlights one of the benefits of screening colonoscopy: all 5 prevalent CRCs were treatable clinically silent early stage cancers, 2 of which were resectable by endoscopic means. The main limitation is that the reported incidence is actually a prevalence-incidence aggregate. A more appropriate descriptor would have been CRC occurrence; however, we wanted to keep our nomenclature consistent with the published literature. In the second approach, we reported incident cases that occurred after 2 years of follow-up, and repeated the calculations after excluding patient-years at risk during the first 2 years after the index colonoscopy. This assumes that the prevalent cancers would have declared themselves clinically after 2 years, had uniform screening not occurred for all patients. The advantages of this approach is that it is intuitively a better estimate of observed incidence (because prevalent cancers are accounted for and removed from the equation), and it has been used in previous studies. 13,29 Both approaches convey important information, and underscore some of the difficulties in determining a true incidence reduction, given the limitations of the available data sources. The bottom line is that screening colonoscopy effectively reduces the incidence of CRC, but a precise determination of the magnitude of the reduction depends on a clear distinction between cancer prevalence and incidence in the comparison population. We found that CRC-specific mortality was reduced by 65% after screening colonoscopy. The result was not statistically significant because of the small number of deaths (3 over 10,492 person-years of follow-up evaluation), resulting in wide CIs. Our findings complement the long-term follow-up results of the National Polyp Study 14 : after a mean follow-up period of 14 years and 48,473 person-years at risk, CRC mortality was reduced for all patients with adenomas by 69% (95% CI, 46% 84%). The reduction for short-term follow-up (the first 10 years after initial colonoscopy) was 81% (95% CI, 52% 95%) and 54% (95% CI, 9% 80%) for long-term ( 10 y) follow-up. 14 Our study had several limitations. First, we focused on the impact of the initial colonoscopy on CRC incidence and mortality. Most of the patients underwent colonoscopic surveillance at varying intervals during the study period, and some had adenomas that were removed during follow-up examinations. Although it is possible that these patients received additional protection against CRC by having metachronous adenomas detected and removed, current literature suggests that most of the benefit is actually derived from the initial colonoscopy. A modeling analysis based on the National Polyp Study showed that 90% of the impact of colonoscopy was owing to the initial polypectomy, rather than to surveillance colonoscopy, for the 10 years after polypectomy. 30 This likely also applies to

5 774 KAHI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 7 patients who have no adenomas at baseline because these individuals remain at low risk of developing CRC several years after their negative initial colonoscopy. 18 Second, our study cohort consisted mostly of Caucasian volunteers who agreed to undergo screening colonoscopy, and this may be a group with healthier lifestyles, more health-seeking behaviors, or better access to other health care resources. However, this selection bias should not affect polyp prevalence rates and CRC incidence rates; these largely depend on the presence of certain symptoms, 31 and patients with symptoms were carefully excluded from our study. Furthermore, from the standpoint of baseline colorectal neoplasia prevalence, our cohort was similar to populations described in recent large studies of screening colonoscopy: in these reports, the prevalence of adenomas has ranged from 14% to 36.5%, and the prevalence of CRC has ranged from 0.1% to 1.3% 4,6,7,21,32 34 ; the corresponding rates in our study were 26.6% and 0.7%, respectively. Third, we excluded patients with significant family history and those with inflammatory bowel disease; however, such patients are included in SEER, and this may have led to a small overestimate of the benefit of colonoscopy. Fourth, we were unable to measure the impact of other risk and protective factors for colorectal neoplasia, such as smoking status or nonsteroidal anti-inflammatory drug use during the study period. Routine use of nonsteroidal antiinflammatory drugs, as well as other health-promoting cointerventions, likely biased our results in favor of colonoscopy. Fifth, we had initially intended to report cancer SIRs for the cohort stratified by whether neoplasia was present or absent at baseline. This would have allowed more direct comparison of subgroup SIRs with studies that enrolled subjects with adenomas (such as the National Polyp Study), or studies that enrolled subjects without colon neoplasms. However, the relatively small sample size and number of incident cancers precluded this analysis. Finally, many surveillance procedures were performed by physicians not affiliated with Indiana University, and the impact of differences in colonoscopy technique and performance quality cannot be measured. We conclude that in this average-risk cohort, CRC incidence and mortality were reduced after screening colonoscopy. These results provide additional evidence for the effectiveness of colonoscopy as a primary CRC screening modality. Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi: /j.cgh References 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, CA Cancer J Clin 2008;58: Rex DK, Eid E. Considerations regarding the present and future roles of colonoscopy in colorectal cancer prevention. Clin Gastroenterol Hepatol 2008;6: Seeff LC, Richards TB, Shapiro JA, et al. How many endoscopies are performed for colorectal cancer screening? Results from CDC s survey of endoscopic capacity. Gastroenterology 2004; 127: Lin OS, Kozarek RA, Schembre DB, et al. Screening colonoscopy in very elderly patients: prevalence of neoplasia and estimated impact on life expectancy. JAMA 2006;295: Regula J, Rupinski M, Kraszewska E, et al. Colonoscopy in colorectal-cancer screening for detection of advanced neoplasia. N Engl J Med 2006;355: Schoenfeld P, Cash B, Flood A, et al. Colonoscopic screening of average-risk women for colorectal neoplasia. N Engl J Med 2005; 352: Strul H, Kariv R, Leshno M, et al. The prevalence rate and anatomic location of colorectal adenoma and cancer detected by colonoscopy in average-risk individuals aged years. Am J Gastroenterol 2006;201: Harewood GC, Wiersema MJ, Melton LJ 3rd. A prospective, controlled assessment of factors influencing acceptance of screening colonoscopy. Am J Gastroenterol 2002;97: Cram P, Fendrick AM, Inadomi J, et al. The impact of a celebrity promotional campaign on the use of colon cancer screening: the Katie Couric effect. Arch Intern Med 2003;163: Prajapati DN, Saeian K, Binion DG, et al. Volume and yield of screening colonoscopy at a tertiary medical center after change in Medicare reimbursement. Am J Gastroenterol 2003; 98: Sedjo RL, Byers T, Barrera E Jr, et al. A midpoint assessment of the American Cancer Society challenge goal to decrease cancer incidence by 25% between 1992 and CA Cancer J Clin 2007;57: Gross CP, Andersen MS, Krumholz HM, et al. Relation between Medicare screening reimbursement and stage at diagnosis for older patients with colon cancer. JAMA 2006;296: Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup. N Engl J Med 1993;329: Zauber AG, Winawer SJ, O Brien MJ, et al. Significant long term reduction in colorectal cancer mortality with colonoscopic polypectomy: findings of the National Polyp Study. Gastroenterology 2007;132:A Alberts DS, Martinez ME, Roe DJ, et al. Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas. Phoenix Colon Cancer Prevention Physicians Network. N Engl J Med 2000;342: Robertson DJ, Greenberg ER, Beach M, et al. Colorectal cancer in patients under close colonoscopic surveillance. Gastroenterology 2005;129: Schatzkin A, Lanza E, Corle D, et al. Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas. Polyp Prevention Trial Study Group. N Engl J Med 2000;342: Singh H, Turner D, Xue L, et al. Risk of developing colorectal cancer following a negative colonoscopy examination: evidence for a 10-year interval between colonoscopies. JAMA 2006;295: Rex DK, Lehman GA, Ulbright TM, et al. Colonic neoplasia in asymptomatic persons with negative fecal occult blood tests: influence of age, gender, and family history. Am J Gastroenterol 1993;88: Rex DK, Cummings OW, Helper DJ, et al. 5-year incidence of adenomas after negative colonoscopy in asymptomatic averagerisk persons. Gastroenterology 1996;111: Regula J, Rupinski M, Kraszewska E, et al. Colonoscopy in colorectal-cancer screening for detection of advanced neoplasia. N Engl J Med 2006;355: Ries LKC, Hankey B, Harras A, et al. SEER cancer statistics review, : tables and graphs. Bethesda, MD: U.S. Department of Health and Human Services, National Cancer Institute, Breslow NE, Day N. Statistical methods in cancer research. Volume II. The design and analysis of cohort studies. Lyon: International Agency for Research on Cancer, Citarda F, Tomaselli G, Capocaccia R, et al. Efficacy in standard

6 July 2009 SCREENING COLONOSCOPY, CRC INCIDENCE AND MORTALITY 775 clinical practice of colonoscopic polypectomy in reducing colorectal cancer incidence. Gut 2001;48: Dove-Edwin I, Sasieni P, Adams J, et al. Prevention of colorectal cancer by colonoscopic surveillance in individuals with a family history of colorectal cancer: 16 year, prospective, follow-up study. BMJ 2005;331: Thiis-Evensen E, Hoff GS, Sauar J, et al. Population-based surveillance by colonoscopy: effect on the incidence of colorectal cancer. Telemark Polyp Study I. Scand J Gastroenterol 1999; 34: Pabby A, Schoen RE, Weissfeld JL, et al. Analysis of colorectal cancer occurrence during surveillance colonoscopy in the dietary Polyp Prevention Trial. Gastrointest Endosc 2005;61: Feinstein AR, Esdaile JM. Incidence, prevalence, and evidence. Scientific problems in epidemiologic statistics for the occurrence of cancer. Am J Med 1987;82: Green RJ, Metlay JP, Propert K, et al. Surveillance for second primary colorectal cancer after adjuvant chemotherapy: an analysis of Intergroup Ann Intern Med 2002;136: Zauber AG VI, van Ballegooijen M, Boer R, et al. Impact of initial polypectomy versus follow-up surveillance colonoscopy on colorectal cancer incidence in post-polypectomy patients. Gastroenterology 2006;130:A Rex DK. Colonoscopy: a review of its yield for cancers and adenomas by indication. Am J Gastroenterol 1995;90: Morikawa T, Kato J, Yamaji Y, et al. A comparison of the immunochemical fecal occult blood test and total colonoscopy in the asymptomatic population. Gastroenterology 2005;129: Imperiale TF, Wagner DR, Lin CY, et al. Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N Engl J Med 2000;343: Lieberman DA, Weiss DG, Bond JH, et al. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. Veterans Affairs Cooperative Study Group 380. N Engl J Med 2000;343: Reprint requests Address requests for reprints to: Charles J. Kahi, MD, MSc, Assistant Professor of Clinical Medicine, Department of Medicine, Division of Gastroenterology and Hepatology, Indiana University School of Medicine, Roudebush VAMC C-7055, 1481 W 10th Street, Indianapolis, Indiana ckahi2@iupui.edu; fax: (317) Acknowledgments The authors would like to acknowledge Joanne Chou for her expert work with Surveillance, Epidemiology, and End Results STAT to derive expected incidence and mortality rates based on Surveillance, Epidemiology, and End Results data. Conflicts of interest The authors disclose no conflicts. Funding Supported in part by National Institutes of Health grant K24 DK02756 (to T.F.I.).