Male Sex and Smoking Have a Larger Impact on the Prevalence of Colorectal Neoplasia Than Family History of Colorectal Cancer

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8: Male Sex and Smoking Have a Larger Impact on the Prevalence of Colorectal Neoplasia Than Family History of Colorectal Cancer MICHAEL HOFFMEISTER,* STEPHANIE SCHMITZ,* ELISABETH KARMRODT,* CHRISTA STEGMAIER, ULRIKE HAUG,* VOLKER ARNDT,* and HERMANN BRENNER* *Division of Clinical Epidemiology and Aging Research, German Cancer Research Center, Heidelberg; and Saarland Cancer Registry, Saarbrücken, Germany See related article, Campbell PT et al, on page 1138 in Gastroenterology. BACKGROUND & AIMS: Screening recommendations for colorectal cancer (CRC) commonly take family history but no other risk factors into account. We compared and assessed risk factors of colorectal polyps in a large population undergoing screening colonoscopy. METHODS: We conducted a population-based cross-sectional study that included 3349 subjects, 55 years or older (mean ages of men and women, 63.6 and 63.4 years, respectively), who underwent colonoscopy for the first time within the nationwide colonoscopy screening program in Germany. We calculated prevalences of colorectal polyps and estimated multivariate prevalence ratios (PRs) and population attributable fractions (PAFs). RESULTS: Overall, 654 subjects had hyperplastic polyps (20%), 675 had non-advanced adenomas (20%), 343 had advanced adenomas (10%), and 40 had CRC (1%). Risk factor prevalences and adjusted PRs were higher for male gender and smoking than for family history of CRC. PAFs for prevalence of non-advanced and advanced neoplasia were highest for male gender (23% and 23%, respectively), followed by smoking (7% and 9%, respectively), and family history of CRC (2% and 4%, respectively). CONCLUSIONS: Male gender and smoking have a larger impact on the prevalence of colorectal neoplasia than family history, suggesting an extensive evaluation of additional risk stratification in population-based screening, particularly by sex. Keywords: Gender; Attributable Risk; Relative Risk; Bootstrap. Colonoscopy is recommended by various expert committees as a primary screening tool for the prevention and early detection of colorectal cancer (CRC) in the general population, mostly from age 50 or 55 years on. 1,2 Current guidelines for CRC screening already include specific recommendations for high-risk individuals with family history of CRC, with inflammatory bowel disease, or with rare genetic syndromes. 1,3,4 However, higher rates of colorectal neoplasia have also been observed in other risk groups that might benefit from more targeted screening offers. 5 In observational studies, smoking was almost consistently associated with about 2-fold increased risk for hyperplastic and adenomatous polyps of the colon and rectum Studies on the association with CRC risk were less consistent. 12 A recent meta-analysis, however, showed a moderate increase in CRC incidence and mortality, 13 and it was suggested that screening recommendations should be customized for smokers. 6,13 In another recently published meta-analysis, male gender was associated with almost 2-fold higher risk of advanced colorectal neoplasia, and it was concluded that men should undergo screening examinations earlier than women. 14 From a public health point of view and for the design of risk-adapted screening offers or other targeted prevention measures, it is not sufficient to compare individual disease risks associated with the exposure factor in terms of relative risk. Also, the prevalence of the exposure needs to be considered. In this regard, results from large population-based studies are more directly relevant to clinical practice than studies conducted at individual and specialized medical centers. We conducted a population-based study with 3349 first-time participants of colonoscopy within a real screening setting in Germany and aimed to investigate prevalences, prevalence ratios (PRs), and population attributable fractions (PAFs) of colorectal polyps associated with smoking, gender, and family history of CRC. Methods Study Design and Study Population This cross-sectional analysis is based on data of averagerisk subjects undergoing screening colonoscopy in 33 gastroenterology practices between May 2005 and December 2007 in Saarland, a small state ( 1,000,000 inhabitants) located in the southwest of Germany (KolosSal study). Almost all practices conducting screening colonoscopies in Saarland agreed to recruit patients for this study. To be eligible for participation, patients undergoing screening colonoscopy had to be 55 years or older and residents of Saarland. Patients with history of CRC or with previous colorectal surgery were not eligible. The data ascertained are baseline data of a statewide cohort study initiated in 2005 with the aim of monitoring long-term reduction in CRC incidence and mortality among screening colonoscopy participants. More information about the study design and about quality assurance within the German screening colonoscopy program were described elsewhere. 15 The KolosSal study was approved by ethics committees of the University of Heidelberg and of the Medical Association of Saarland. Written informed consent was obtained from each participant. In this study, patients undergoing colonoscopy for the first time were included only, because patients with a previous Abbreviations used in this paper: CRC, colorectal cancer; PAF, population attributable fraction; PR, prevalence ratio by the AGA Institute /$36.00 doi: /j.cgh

2 October 2010 PREVALENCE OF COLORECTAL POLYPS 871 colonoscopy might have negative or normal findings as a result of an earlier polypectomy. Participants were also excluded if the cecum was not reached during colonoscopy (incomplete colonoscopy) or if the colorectum could not be fully overlooked as a result of poor bowel preparation (ie, some parts could not be inspected during endoscopy because stool was not rinsed out properly). Data Collection Patients were asked to fill out a standardized questionnaire on potential risk factors of CRC, including sociodemographic and lifestyle factors, own medical history and family history of CRC, and to return the completed questionnaire to the study center by the appointed date of colonoscopy. However, some participants returned their questionnaire after colonoscopy by mail, and another minority of participants could not be recruited before colonoscopy because of practices work overload and were invited to participate by mail shortly after colonoscopy. Screening colonoscopy reports and histology reports were requested and transferred into a standardized form by double entry of 2 independent trained investigators who were blinded with respect to questionnaire data. The items recorded included range of colonoscopy, number and size of polyps, and their histologic classification. Discrepancies in the record entries were discussed and resolved in consensus. Exposure Assessment Self-reported current and former smoking of at least 1 cigarette per day for at least 1 year was assessed including total duration of active smoking, average number of cigarettes smoked during the years of smoking, current number of cigarettes, age at beginning, and among former smokers, age at smoking cessation. Former smoking was defined as smoking cessation at least 2 years ago. Information on family history of CRC in a first-degree relative was ascertained for parents and siblings separately. Statistical Analysis First, we described characteristics and findings of participants undergoing colonoscopy for the first time. Patients were then classified according to the prevalence of any of the following findings: no polyps, unspecified polyps, hyperplastic polyps, non-advanced adenoma (equivalent to non-advanced neoplasia), advanced adenoma (defined as presence of an adenoma with at least 1 of the following features: 1 cm in size, tubulovillous or villous histology, severe dysplasia), or CRC. Log-binomial regression was used to estimate PRs of the association of risk factors and hyperplastic polyps, non-advanced neoplasia, and advanced neoplasia (ie, any advanced colorectal adenoma or CRC), respectively. 16 Apart from age at colonoscopy, gender, cigarette smoking status (never, former, current), and family history of CRC (any parent or sibling and any age of diagnosis), the following covariates were included in the model because they were differentially distributed between polyp-positive and polyp-free patients (P.20): school education ( 9, 10 11, years), body mass index, physical activity (quartiles of metabolic equivalents), alcohol consumption (no alcohol and quartiles of amount of ethanol in grams), daily consumption of red meat (yes/no), ever use of hormone replacement therapy (yes/no). Other potential confounding factors such as previous diagnosis of diabetes or inflammatory bowel disease or ever use of NSAIDs did not meet this definition and were not included in the model. The adjusted PAF describes the proportion of disease cases in the population under study that is attributable to the exposure (or the expected proportional reduction in disease risk under the more favorable exposure level). Estimation of adjusted PAFs was based on the method by Bruzzi et al 17 by using the formula of Miettinen, 18 PAF P dis (E) (PR-1)/(PR), where PAF represents the adjusted PAF, P dis (E) is the proportion of the exposed among the diseased (ie, among those with hyperplastic polyps, non-advanced or advanced neoplasia, respectively), and PR is the adjusted PR estimated from log-binomial regression. Although formulated for case-control designs, Benichou 19 suggested that the method by Bruzzi et al might as well be used in cross-sectional studies. Bootstrapping was used to estimate 95% confidence intervals. In detail, the PAF was computed in each of the bootstrap samples (n 1000), and the percentile method was used to determine the lower and upper confidence limits (2.5th and 97.5th percentiles of computed PAFs). 20 All analyses were carried out with the statistical software package SAS 9.2 (SAS Institute Inc, Cary, NC). Results Overall, 5181 participants of screening colonoscopy were recruited. After exclusion of subjects with previous endoscopy of the large bowel (n 1507), with missing information about previous endoscopies (n 12), with imperfect cleansing of the colorectum (n 257), and with incomplete colonoscopy (cecum not reached, n 56), a total of 3349 participants undergoing colonoscopy for the first time were included in this study. Men and women were represented equally in the study sample, and mean age was 63.5 years overall (Table 1). Only onethird of participants had more than 9 years of school education, and more than two-thirds were overweight or obese. CRC in a first-degree relative was reported by 12% of men and 11% of women. About 11% of men and 9% of women reported current smoking. In addition, 49% of men and 23% of women were former smokers. A smoking history of 20 pack-years and more was observed more often in men than in women among both current and former smokers. Overall, 47% of men and 63% of women had a negative colonoscopy (no polyps found). Hyperplastic polyps, non-advanced adenomas, advanced adenomas, and cancer were more common in men (23%, 25%, 13%, 1%, respectively) than in women (16%, 16%, 8%, 1%, respectively). PRs of all types of polyps were consistently higher in men compared with women, in individuals with family history of CRC compared with those with negative family history, and in smokers compared with non-smokers (Table 2). Also, PRs of current smokers were consistently higher than those of former smokers. Compared with women who never smoked, former and current smoking in women was associated with nonsignificantly higher PRs of neoplasia. However, prevalence of neoplasia in non-smoking men was even higher than in women with a smoking history, and prevalence was highest in currently smoking men (more than twice that of non-smoking women). In women, family history of CRC was associated with significantly higher prevalence of non-advanced neoplasia only (PR,

3 872 HOFFMEISTER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Table 1. Characteristics of 3349 Participants at First-Time Colonoscopy Women (N 1674) Men (N 1675) Total (N 3349) N (%) N (%) N (%) Age (y) (35) 552 (33) 1135 (34) (23) 374 (22) 760 (23) (24) 454 (27) 855 (25) (11) 197 (12) 387 (12) (7) 98 (6) 212 (6) Mean ( SD) 63.4 ( 6.6) 63.6 ( 6.4) 63.5 ( 6.5) Education (y) (72) 1076 (65) 2266 (68) (18) 219 (13) 521 (16) (10) 353 (21) 519 (15) Missing 16 (1) 27 (2) 43 (1) Body mass index (kg/m 2 ) (37) 404 (24) 1024 (31) (20) 398 (24) 727 (22) (22) 506 (30) 866 (26) (20) 343 (20) 681 (20) Missing 27 (2) 24 (1) 51 (2) Family history of CRC No/unknown 1469 (88) 1460 (87) 2929 (87) Yes 189 (11) 202 (12) 391 (12) Missing 16 (1) 13 (1) 29 (1) Cigarette smoking Never 1098 (66) 594 (35) 1692 (51) Former 377 (23) 822 (49) 1199 (36) 20 pack-years 290 (17) 415 (25) 705 (21) 20 pack-years 79 (5) 380 (23) 459 (14) Missing 8 ( 1) 27 (2) 35 (1) Current 156 (9) 181 (11) 337 (10) 20 pack-years 53 (3) 28 (2) 81 (2) 20 pack-years 75 (4) 122 (7) 197 (6) Missing 28 (2) 31 (2) 59 (2) Missing 8 ( 1) 16 (1) 24 (1) All findings prevalent at colonoscopy Negative 1056 (63) 790 (47) 1846 (55) Hyperplastic polyp 271 (16) 383 (23) 654 (20) Non-advanced adenoma 261 (16) 414 (25) 675 (20) Advanced adenoma 127 (8) 216 (13) 343 (10) Cancer 17 (1) 23 (1) 40 (1) Polyp not specified 56 (3) 99 (6) 155 (5) Most advanced finding at colonoscopy Negative 1056 (63) 790 (47) 1846 (55) Hyperplastic polyp 194 (12) 211 (13) 405 (12) Non-advanced adenoma 254 (15) 400 (24) 654 (20) Advanced adenoma 127 (8) 216 (13) 343 (10) Cancer 17 (1) 23 (1) 40 (1) Polyp not specified 26 (2) 35 (2) 61 (2) SD, standard deviation. 1.42; 95% confidence interval, ) (Table 2). Polyp prevalences of men with no family history of CRC were generally higher than those of women with positive family history (eg, advanced neoplasia: PR 1.46, ). Compared with women with negative family history, the prevalence of advanced neoplasia was even higher for the combination of male gender and positive family history (PR 2.22, ). In the studied population, PAFs of non-advanced and advanced neoplasia were highest for male gender (23% and 23%, respectively), followed by history of smoking (7% and 9%, respectively) (Table 3). PAFs of non-advanced and advanced neoplasia associated with family history of CRC were low (2% and 4%, respectively). For hyperplastic polyps, PAFs were highest for smoking (13%), followed by male gender (10%), and negligible for family history of CRC. Discussion In this population-based study with first-time participants of colonoscopy, prevalences of colorectal neoplasia were

4 October 2010 PREVALENCE OF COLORECTAL POLYPS 873 Table 2. Association of Gender, Smoking Status, and First-Degree Family History of CRC With Prevalence of Findings at First- Time Colonoscopy Hyperplastic polyps Non-advanced neoplasia Advanced neoplasia N (prevalence) PR (95% CI) a N (prevalence) PR (95% CI) a N (prevalence) PR (95% CI) a Gender Women 271 (16%) (16%) (9%) 1.00 Men 383 (23%) 1.21 ( ) 414 (25%) 1.60 ( ) 239 (14%) 1.57 ( ) Smoking Never smokers 273 (16%) (17%) (10%) 1.00 Ever smokers 378 (23%) 1.28 ( ) 379 (23%) 1.15 ( ) 211 (13%) 1.19 ( ) Former smokers 260 (22%) 1.18 ( ) 268 (22%) 1.10 ( ) 142 (12%) 1.05 ( ) Current smokers 93 (28%) 1.63 ( ) 86 (26%) 1.27 ( ) 51 (15%) 1.56 ( ) Family history of CRC No 567 (19%) (20%) (11%) 1.00 Yes 80 (21%) 1.03 ( ) 94 (24%) 1.17 ( ) 63 (16%) 1.33 ( ) Compared with women who never smoked Women, never smokers 157 (14%) 1.00 Reference 154 (14%) 1.00 Reference 91 (8%) 1.00 Reference Women, former smokers 66 (18%) 1.16 ( ) 70 (19%) 1.26 ( ) 32 (8%) 1.13 ( ) Women, current smokers 37 (24%) 1.73 ( ) 33 (21%) 1.28 ( ) 15 (10%) 1.35 ( ) Men, never smokers 116 (20%) 1.22 ( ) 139 (23%) 1.72 ( ) 75 (13%) 1.57 ( ) Men, former smokers 194 (24%) 1.44 ( ) 198 (24%) 1.79 ( ) 110 (13%) 1.62 ( ) Men, current smokers 56 (31%) 1.92 ( ) 53 (29%) 2.15 ( ) 36 (20%) 2.59 ( ) Compared with women with no family history of CRC Women, no family history of 235 (16%) 1.00 Reference 218 (15%) 1.00 Reference 122 (8%) 1.00 Reference CRC Women, family history of CRC 33 (17%) 1.07 ( ) 37 (20%) 1.42 ( ) 21 (11%) 0.96 ( ) Men, no family history of CRC 332 (23%) 1.22 ( ) 355 (24%) 1.68 ( ) 197 (14%) 1.46 ( ) Men, family history of CRC 47 (23%) 1.23 ( ) 57 (28%) 1.76 ( ) 42 (21%) 2.22 ( ) CI, confidence interval. a PRs adjusted for age at colonoscopy, gender, school education, first-degree family history of CRC, ever use of hormone replacement therapy, body mass index, physical activity, alcohol consumption, and red meat consumption. higher in men than in women, irrespective of smoking status and family history of CRC. Smoking and, in particular, current smoking were associated with higher prevalence of all polyp types. A history of CRC in a parent or a sibling was associated with higher prevalence of non-advanced neoplasia in women and of advanced neoplasia in men. The population fraction of non-advanced or advanced neoplasia attributable to a single risk factor was highest for male gender. Only a small proportion of neoplastic polyps was attributable to history of CRC in a first-degree relative (3%), and no association was found with prevalence of hyperplastic polyps. Even by accounting for higher risks that were reported for CRC Table 3. PAFs of Prevalent Findings at First-Time Colonoscopy Associated With Male Gender, History of Smoking, and History of CRC in First-Degree Relative Proportion of cases with risk factor PAF a (95% CI) Cases attributable to risk factor b Male gender Hyperplastic polyps 59% 10% (0% 20%) 65 of 654 cases Non-advanced neoplasia 61% 23% (14% 32%) 155 of 675 cases Advanced neoplasia 62% 23% (9% 36%) 79 of 343 cases Smoking Hyperplastic polyps 58% 13% (5% 20%) 85 of 654 cases Non-advanced neoplasia 56% 7% ( 1% to 15%) 47 of 675 cases Advanced neoplasia 56% 9% ( 2% to 20%) 31 of 343 cases Family history of CRC Hyperplastic polyps 12% 0% ( 3% to 3%) 1 of 654 cases Non-advanced neoplasia 14% 2% ( 1% to 5%) 14 of 675 cases Advanced neoplasia 16% 4% ( 1% to 8%) 14 of 343 cases CI, confidence interval. a PAFs adjusted for age at colonoscopy, gender, school education, first-degree family history of CRC, ever use of hormone replacement therapy, body mass index, physical activity, alcohol consumption, and red meat consumption. b [Total number of cases PAF] out of total number of cases.

5 874 HOFFMEISTER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 (not including advanced adenomas) in a meta-analysis (relative risk 2.25, ), 21 the fraction with advanced neoplasia attributable to a first-degree family history of CRC (9%) would not exceed the fraction attributable to smoking or male gender, which is due to the much higher prevalences of smoking (former and current) and male gender in the population. However, because individuals with positive family history are at increased risk of developing CRC earlier than the general population, 4 special screening recommendations were established for this risk group. 1,3 Like in previous studies, smoking and, in particular, continued smoking were associated with higher prevalences of both hyperplastic and neoplastic polyps. 6,9 Because of the higher risk of neoplasia among smokers, some authors have suggested that smokers might undergo CRC screening earlier than non-smokers. 6,11,22 We found that PAFs for neoplasia associated with male gender (23%) were even higher than PAFs for smoking (8%), so that gender-specific optimization of screening might be given a higher priority. In a comparative analysis of age-specific and sex-specific CRC incidence and mortality data from multiple countries in different parts of the world, it was demonstrated that women consistently reach equivalent levels of 10-year cumulative CRC incidence and mortality 4 8 years later than men at age 50, 55, or In another study, age-specific and sex-specific transition rates from advanced adenomas to CRC were estimated; transition rates of the subsequent 10 years were increasing with age but were similar for both sexes. 24 Taken together, these results, along with results from other studies, 14 provide evidence of an earlier risk of CRC in men than in women. In the joint guidelines of American expert associations, further customization of CRC screening was postponed. Arguments against further differentiated screening recommendations included the lack of data indicating a benefit on incidence and mortality of CRC and the increasing complexity of screening offers that might result in confusion and declining participation rates. 1 Still, the characterization of groups at higher risk and those at lower risk will be a future challenge, and the incorporation of findings of different risks in defined groups into screening recommendations can help avoid unnecessary procedures or fears in people at lower risk and might lead to more efficient use of limited resources. Lieberman 5 pointed out that screening guidelines are already complex, including consideration of age, family history of CRC, inflammatory bowel disease, and findings at previous endoscopic examinations, and that computer-based algorithms might help facilitate identification of high-risk patients in primary care. Furthermore, because other cancer screening measures are gender-specific anyway (prostate, breast, cervical), differentiation by gender might not be too complex for patients. 23 For an extensive evaluation of gender-specific screening, however, many more criteria need to be considered, including the higher prevalence of advanced neoplasia in the proximal colon in women than in men, 25 recent findings of differential effectiveness of colonoscopy in the proximal and distal parts of the colorectum, gender-specific participation rates, 30 different life expectancies of women and men, and cost-effectiveness. 31 To compare the impact a risk factor has on the prevalence of colorectal polyps, we estimated PAFs. PAFs usually describe the hypothetical proportion of disease that could be avoided if the exposure was completely eliminated. Elimination of exposure is, of course, not possible with regard to male gender and familial predisposition to CRC and not realistic in the case of smoking. 32 However, by considering both disease risk and prevalence of exposure, PAFs can be used to quantify the impact of a risk factor on the population level and to identify the most promising strategies for prevention. 33 The concept of partial attributable fractions (or risks) might provide even better estimates when comparing multiple exposures, but no convenient algorithms or software have been developed for use in cross-sectional studies yet. 34 In this article, 3 different outcomes were considered, ie, hyperplastic polyps, non-advanced neoplasia, and advanced neoplasia. With respect to CRC risk and potential implications for screening, advanced neoplasia is the most relevant outcome. However, the consistency of results for non-advanced and advanced colorectal neoplasia is reassuring, because both outcomes are considered to be part of the multi-stage model of colorectal carcinogenesis. 35 Hyperplastic polyps might also have malignant potential through the serrated pathway, but so far no special screening or surveillance options have been suggested. 1,36 However, an association was found with smoking like previously reported. 9 There are several strengths and limitations to consider in the interpretation of this study. A major strength is the populationbased design in a real screening setting. The individuals of this study undergoing colonoscopy for the first time likely represent an average-risk population for the most part. Also, it was possible to differentiate between individuals with and without a previous colonoscopy. Although based on self-reports, information about previous large bowel endoscopies was found to be highly valid. 37 Finally, the size of the study allowed for calculation of prevalences in subgroups and for the estimation of PRs associated with risk factor combinations. Because of its observational design this study is prone to confounding. Information on smoking, family history of CRC, and body mass index was based on self-reports and might include some misclassification. In particular, underreporting of family history might have resulted in underestimation of the PR. 38,39 People undergoing colonoscopy might be more health conscious than people not undergoing colonoscopy, potentially leading to selection bias. On the other hand, a colonoscopy screening program might particularly attract individuals with specific or unspecific symptoms. Thus, prevalences might not fully represent prevalences that would be observed in an asymptomatic population. However, a study from the United Kingdom conducted in a population with positive fecal occult blood test results suggests that gastrointestinal symptoms are not predictive of colonoscopic findings. 40 Use of NSAIDs, particularly in higher doses and after use for 5 or 10 years, was associated with lower risk of colorectal neoplasia in previous studies. 41,42 In this study, no differential distribution of NSAID use among participants with and without polyps or neoplasia was observed. This could be due to the fact that NSAID use was regular use of low-dose aspirin for the most part that was not restricted to long-term users only. Still, for the estimation of adjusted PRs and PAFs, we were able to include major confounding factors that were associated with the risk of colorectal polyps in previous studies. However, the possibility of confounding by unmeasured factors cannot be excluded. In particular, the higher PR and PAF observed for male gender likely

6 October 2010 PREVALENCE OF COLORECTAL POLYPS 875 include effects of confounding as a result of sex-specific differences in lifestyle (eg, general health behavior, nutrition, amount of alcohol and cigarettes, body fatness) and might not be explained by biological differences between women and men only. Some PAFs in Table 3 have negative lower limits that would correspond to an increase of cases after elimination of the risk factor. However, negative lower limits in our study rather reflect statistically borderline or nonsignificant PRs and can best be interpreted as PAFs close to zero. 43 In conclusion, prevalences of colorectal polyps, in particular neoplastic polyps, were higher in men than in women, irrespective of smoking status and family history of CRC. Male gender and smoking have a larger impact on the prevalence of colorectal neoplasia than family history, suggesting an extensive evaluation of additional risk stratification in population-based screening, particularly by gender. In future studies (including genetic studies), PAFs should be reported more often for better communication of the importance of risk factors on the population level. References 1. Levin B, Lieberman DA, McFarland B, et al. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps, 2008: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. CA Cancer J Clin 2008; 58: Schmiegel W, Reinacher-Schick A, Arnold D, et al. [Update S3- guideline colorectal cancer 2008]. Z Gastroenterol 2008;46: Schmiegel W, Pox C, Adler G, et al. [S3-Guidelines Conference Colorectal Carcinoma 2004]. Z Gastroenterol 2004;42: Brenner H, Hoffmeister M, Haug U. Family history and age at initiation of colorectal cancer screening. Am J Gastroenterol 2008;103: Lieberman D. Race, gender, and colorectal cancer screening. Am J Gastroenterol 2005;100: Botteri E, Iodice S, Raimondi S, et al. Cigarette smoking and adenomatous polyps: a meta-analysis. Gastroenterology 2008; 134: Shrubsole MJ, Wu H, Ness RM, et al. Alcohol drinking, cigarette smoking, and risk of colorectal adenomatous and hyperplastic polyps. Am J Epidemiol 2008;167: Ji BT, Weissfeld JL, Chow WH, et al. Tobacco smoking and colorectal hyperplastic and adenomatous polyps. Cancer Epidemiol Biomarkers Prev 2006;15: Lieberman DA, Prindiville S, Weiss DG, et al. Risk factors for advanced colonic neoplasia and hyperplastic polyps in asymptomatic individuals. JAMA 2003;290: Giovannucci E, Colditz GA, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women. J Natl Cancer Inst 1994;86: Giovannucci E, Rimm EB, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. men. J Natl Cancer Inst 1994;86: Verla-Tebit E, Lilla C, Hoffmeister M, et al. Cigarette smoking and colorectal cancer risk in Germany: a population-based case-control study. Int J Cancer 2006;119: Botteri E, Iodice S, Bagnardi V, et al. Smoking and colorectal cancer: a meta-analysis. JAMA 2008;300: Nguyen SP, Bent S, Chen YH, et al. Gender as a risk factor for advanced neoplasia and colorectal cancer: a systematic review and meta-analysis. Clin Gastroenterol Hepatol 2009;7: e Brenner H, Haug U, Arndt V, et al. Low risk of colorectal cancer and advanced adenomas more than 10 years after negative colonoscopy. Gastroenterology 2010;138: Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol 2005;162: Bruzzi P, Green SB, Byar DP, et al. Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol 1985;122: Miettinen OS. Proportion of disease caused or prevented by a given exposure, trait or intervention. Am J Epidemiol 1974;99: Benichou J. A review of adjusted estimators of attributable risk. Stat Methods Med Res 2001;10: Miller DP. Bootstrap 101: obtain robust confidence intervals for any statistic. Proceedings of the 29th annual SAS users group international conference. Vol 29. Montréal, Canada: SAS Institute Inc, Johns LE, Houlston RS. A systematic review and meta-analysis of familial colorectal cancer risk. Am J Gastroenterol 2001;96: Anderson JC, Attam R, Alpern Z, et al. Prevalence of colorectal neoplasia in smokers. Am J Gastroenterol 2003;98: Brenner H, Hoffmeister M, Arndt V, et al. Gender differences in colorectal cancer: implications for age at initiation of screening. Br J Cancer 2007;96: Brenner H, Hoffmeister M, Stegmaier C, et al. Risk of progression of advanced adenomas to colorectal cancer by age and sex: estimates based on 840,149 screening colonoscopies. Gut 2007;56: Schoenfeld P, Cash B, Flood A, et al. Colonoscopic screening of average-risk women for colorectal neoplasia. N Engl J Med 2005; 352: Regula J, Rupinski M, Kraszewska E, et al. Colonoscopy in colorectal-cancer screening for detection of advanced neoplasia. N Engl J Med 2006;355: Brenner H, Chang-Claude J, Seiler CM, et al. Potential for colorectal cancer prevention of sigmoidoscopy versus colonoscopy: population-based case control study. Cancer Epidemiol Biomarkers Prev 2007;16: Baxter NN, Goldwasser MA, Paszat LF, et al. Association of colonoscopy and death from colorectal cancer. Ann Intern Med 2009;150: Brenner H, Hoffmeister M, Arndt V, et al. Protection from rightand left-sided colorectal neoplasms after colonoscopy: population-based study. J Natl Cancer Inst 2010;102: Brenner H, Hoffmeister M, Brenner G, et al. Expected reduction of colorectal cancer incidence within 8 years after introduction of the German screening colonoscopy programme: estimates based on 1,875,708 screening colonoscopies. Eur J Cancer 2009;45: Lansdorp-Vogelaar I, van Ballegooijen M, Zauber AG, et al. Individualizing colonoscopy screening by sex and race. Gastrointest Endosc 2009;70:96 108, 108 e Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable fractions. Am J Public Health 1998;88: Gefeller O. The concept of attributable risk in epidemiology: yesterday, today and tomorrow. Stat Methods Med Res 2001; 10: Ramsch C, Pfahlberg AB, Gefeller O. Point and interval estimation of partial attributable risks from case-control data using the R-package parccs. Comput Methods Programs Biomed 2009; 94: Winawer SJ. Natural history of colorectal cancer. Am J Med 1999;106:3S 6S; discussion 50S 51S.

7 876 HOFFMEISTER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No Bond JH. Polyp guideline: diagnosis, treatment, and surveillance for patients with colorectal polyps Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol 2000;95: Hoffmeister M, Chang-Claude J, Brenner H. Validity of self-reported endoscopies of the large bowel and implications for estimates of colorectal cancer risk. Am J Epidemiol 2007;166: Murff HJ, Spigel DR, Syngal S. Does this patient have a family history of cancer? An evidence-based analysis of the accuracy of family cancer history. JAMA 2004;292: Patrick DL, Cheadle A, Thompson DC, et al. The validity of selfreported smoking: a review and meta-analysis. Am J Public Health 1994;84: Ahmed S, Leslie A, Thaha MA, et al. Lower gastrointestinal symptoms are not predictive of colorectal neoplasia in a faecal occult blood screen-positive population. Br J Surg 2005;92: Dube C, Rostom A, Lewin G, et al. The use of aspirin for primary prevention of colorectal cancer: a systematic review prepared for the U.S. Preventive Services Task Force. Ann Intern Med 2007; 146: Rostom A, Dube C, Lewin G, et al. Nonsteroidal anti-inflammatory drugs and cyclooxygenase-2 inhibitors for primary prevention of colorectal cancer: a systematic review prepared for the U.S. Preventive Services Task Force. Ann Intern Med 2007;146: Daly LE. Confidence limits made easy: interval estimation using a substitution method. Am J Epidemiol 1998;147: Reprint requests Address requests for reprints to: Michael Hoffmeister, PhD, MSc, Division of Clinical Epidemiology and Aging Research, German Cancer Research Center, Bergheimer Strasse 20, D Heidelberg, Germany. m.hoffmeister@dkfz.de; fax: (49) Acknowledgments The authors gratefully acknowledge the excellent cooperation of the physicians conducting screening colonoscopies in patient recruitment and the excellent contributions of Isabel Lerch, Silvia März, and Natalia Zumkeller in data collection, monitoring, and documentation. Conflicts of interest The authors disclose no conflicts. Funding This study was supported in part by grants from the Central Research Institute of Ambulatory Health Care in Germany, Berlin, Germany, and from the German Cancer Aid (no ).