Nearly 150,000 cases of colorectal cancer are diagnosed in

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5: Outcome of Follow-up Colon Examination Among a Population-Based Cohort of Colorectal Cancer Patients STEPHEN J. RULYAK,* DAVID A. LIEBERMAN, EDWARD H. WAGNER, and MARGARET T. MANDELSON *University of Washington, Seattle, Washington; Oregon Health and Science University, Portland, Oregon; and Group Health Cooperative, Seattle, Washington See CME exam on page 407. Background & Aims: The benefit of colonoscopy in the follow-up of colorectal cancer survivors is uncertain, and findings of surveillance colonoscopy are not well-characterized. We sought to estimate survival among colorectal cancer patients according to receipt of a follow-up colon examination and to describe the findings of such exams. Methods: We studied health maintenance organization enrollees with colorectal cancer who underwent surgical resection. Mortality was estimated by using survival analysis, and findings of colon examinations were determined by review of pathology reports. Results: One thousand two patients were eligible for study; 5-year survival was higher (76.8%) for patients who had at least one follow-up exam than for patients who did not undergo follow-up (52.2%, P <.0001). In multivariate analysis, colon examination remained independently associated with improved survival (hazard ratio, 0.58; 95% confidence interval, ). Twenty patients (3.1%) were diagnosed with a second colorectal cancer, including 9 cancers detected within 18 months of initial cancer diagnosis. Advanced neoplasia was more common (15.5%) among patients followed up between months after diagnosis compared with patients followed up within 18 months (6.9%, P.02). History of adenomas was associated with advanced neoplasia on follow-up (P.002). Patients with advanced neoplasia on initial follow-up were at high risk for advanced neoplasia on subsequent examinations (13/16, 81%). Conclusions: After colorectal cancer resection, patients have a high risk of interval cancers, some of which represent missed lesions at initial diagnosis. Therefore, surveillance colonoscopy within 1 year of initial diagnosis is warranted. After adjusting for key variables, endoscopic surveillance is associated with improved survival. Nearly 150,000 cases of colorectal cancer are diagnosed in the United States each year, 1 and three fourths of such patients will undergo a surgical resection with curative intent. 2 Nevertheless, approximately one third of resected patients will develop a recurrence of colorectal cancer, and most of these patients will ultimately die of this disease. 3 5 Colorectal cancer patients are also at increased risk to develop metachronous colorectal cancers and adenomas. 6 9 For these reasons, published guidelines recommend surveillance colonoscopy as part of colorectal cancer follow-up care, although guidelines differ with respect to recommended timing of such exams The efficacy of intensive surveillance of patients with colorectal cancer with physical examination, laboratory tests, radiographic studies, and endoscopy is unknown. Randomized trials and meta-analyses 24,25 of these trials have reported inconsistent results, with some reporting benefit and others observing no effect Nonetheless, colonoscopy is regarded as a key component in the surveillance of colon cancer survivors, although neither the precise survival benefit of endoscopic surveillance nor the optimal timing of such exams has been defined. 26,27 Moreover, the yield and outcomes of surveillance colonoscopy have not been characterized in a population-based setting. The purpose of this study was to evaluate the relation between surveillance examination and survival in a well-defined population-based cohort of colorectal cancer patients and to determine the rate and time interval for newly discovered interval cancers and adenomas in patients who have had colon cancer resection. Methods Setting and Study Design This study was conducted at Group Health Cooperative (GHC), a large health maintenance organization (HMO) serving approximately 500,000 individuals in Washington State. The study cohort included all patients diagnosed with a first adenocarcinoma or adenocarcinoma-variant tumor of the colon or rectum between January 1, 1993 December 31, 1999 (n 1292). Colorectal cancer cases were identified by linkage between the GHC enrollment database and the Seattle-Puget Sound Surveillance, Epidemiology, and End Results (SEER) cancer registry. 28 The study included cases of colorectal cancer with American Joint Commission on Cancer stages 0, I, II, or III. 29 Cases with metastatic disease at presentation (stage IV cancers) (n 190) were excluded because such patients are generally not candidates for colon surveillance. Patients with stage 0 disease (also known as carcinoma in situ) were included because colon surveillance after initial resection has been recommended with similar frequency in these patients as for patients with more advanced tumors. 30,31 Subjects were ineligible for study if their stage of disease was missing or unknown (n 21), or if they were not treated with surgical or endoscopic resection (n 23). We also excluded patients who underwent total proctocolec- Abbreviations used in this paper: CI, confidence interval; GHC, Group Health Cooperative; HMO, health maintenance organization; HR, hazard ratio; SEER, Surveillance Epidemiology and End Results by the AGA Institute /07/$32.00 doi: /j.cgh

2 April 2007 OUTCOMES OF COLON SURVEILLANCE 471 tomy for ulcerative colitis (n 8), Crohn s disease (n 8), or other indications (n 11). Finally, any remaining patients who survived less than 6 months after diagnosis were excluded from the cohort (n 29). No member of the cohort had a diagnosis of a familial colon cancer syndrome in the HMO automated database or their medical record. Study patients were followed from the date of diagnosis until one of the following events: death, loss to follow-up, or the end of the study period (December 31, 2001). For subjects who were lost to follow-up, the end of the follow-up interval was defined as the date of the last clinical encounter recorded in the GHC databases. Receipt of Colon Examinations Dates and types of colon examinations after diagnosis were ascertained from electronic medical records of ambulatory and inpatient encounters at GHC. Colon examinations were identified by using Current Procedural Terminology 4 procedure codes 32 and included either colonoscopy or sigmoidoscopy and barium enema. Flexible sigmoidoscopy without a barium enema was not considered to constitute an adequate colon surveillance examination. Colon examinations performed within 2 months after diagnosis were excluded from our surveillance analysis, because these exams are often conducted to evaluate postoperative complications or to exclude synchronous lesions. GHC does not have a formal guideline for postdiagnostic colon surveillance and does not restrict access to colon examination procedures. Findings of Colon Examinations We ascertained pathologic outcomes of colon examinations by linking procedures among study participants to electronic pathology reports. Reports were reviewed by a single gastroenterologist (S.J.R.) for the following histologic findings: carcinoma, adenoma (tubular, villous, unspecified), and hyperplasia. Lesion size was coded as the actual or aggregate size of the lesion given on the pathology report. Pathologic size estimation was chosen because electronic clinical data systems of endoscopic reports were not available for the entire study period. In addition, previous studies have reported that endoscopic size estimates might be less reliable than pathologic estimates. 33 Neoplasia was classified according to the most advanced histologic finding noted in the pathology report. Advanced neoplasia was defined by 1 or more of the following: adenoma 10 mm in size, adenoma with villous architecture, adenoma with high-grade dysplasia, or carcinoma. The location of the lesion within the colon was also determined from the pathology report where possible. When no pathology data corresponding to an endoscopic procedure were available, the exam was considered to be negative for dysplasia because a previous pilot study found complete ascertainment for the pathology database when compared with findings described in the endoscopy report (data not shown). Covariates Covariates of interest were selected a priori and were drawn from the SEER registry and GHC databases. These included patient age at diagnosis, sex, race, tumor stage, tumor location, receipt of adjuvant chemotherapy or radiotherapy, marital status, socioeconomic status, and medical comorbidity. History of adenomatous polyps was also ascertained by review of the pathology database and by abstraction of medical records from the date of HMO enrollment until the date of diagnosis. A family history of colorectal cancer was also determined by abstraction of medical records from date of HMO enrollment until the date of diagnosis. Family history was considered missing if there was no mention of either a positive or negative family history of colon cancer in the medical record. Medical comorbidity was estimated by using the Chronic Disease Score, which uses automated pharmacy data to estimate health status. 34 The Chronic Disease Score is a validated measure of comorbidity that has been shown to have similar predictive validity to other comorbidity scores including the Charlson Index We calculated the Chronic Disease Score by using automated pharmacy data from third through ninth month after the date of diagnosis to limit the influence of early postoperative morbidity on the comorbidity estimate. Data Analysis Outcomes of the cohort were estimated by using survival analysis, with between-group comparisons based on the log-rank test. 39 Because the receipt of colon examination is dependent on patient survival, the prognostic importance of colon examination was modeled as a time-dependent covariate. 40 The independent effect of covariates on mortality, measured by the associated hazard ratio (HR), was estimated by using Cox proportional hazards models. 41,42 The final proportional hazards model excluded covariates if they were neither independent predictors nor influenced the risk estimates for other predictors. We tabulated findings of colon examinations of the cohort according to the length of time between diagnosis and the examination, personal history of adenomas, or a family history of colorectal cancer. The findings of flexible sigmoidoscopies conducted during follow-up of the subset of patients with rectal cancer were also examined. Statistical comparisons were performed by using the 2 test, Fisher exact test, or log-rank test, as appropriate. All data analyses were performed with Stata 8.0 (Stata Corp, College Station, TX), and the type I error level ( ) was.05. Results Patient Characteristics A total of 1002 patients met inclusion criteria for the study, and characteristics of these patients are given in Table 1. The study sample consisted of equal proportions of men and women, and most patients (78%) were 60 years of age or older at diagnosis. Most patients were white (93%), reflecting the racial composition of the Puget Sound geographic area. Two thirds of patients (66%) had local disease at diagnosis, and 25% had nodal disease at diagnosis. A minority of patients received adjuvant chemotherapy (35%) or radiation (13%). Follow-up after diagnosis ranged from 6 months 8.7 years (median, 3.6 years). Reasons for censoring included death (n 217; 22%), loss to follow-up (n 183; 18%), or the end of the study period (n 602; 60%). Sixty-five percent of the study cohort (n 652) underwent 1 or more colon examinations during the study period. A total of 936 exams were conducted on members of the cohort, with an average of 1.4 (standard deviation, 0.6) exams per patient during follow-up. This included 730 exams among 505 patients

3 472 RULYAK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 with colon cancer and 206 exams among 146 patients with rectal cancer. The majority of examinations performed were colonoscopies (97%), and relatively few patients underwent examination with flexible sigmoidoscopy combined with barium enema (3%). Using survival analysis to account for variability in follow-up time, 61% (95% confidence interval [CI], 58% 64%) of the cohort underwent a colon examination within 18 months of diagnosis, and 80% (95% CI, 77% 84%) underwent a colon examination within 5 years of diagnosis. Patient Survival Seven hundred four patients (70%) were alive at the end of the study period, and the cumulative survival of the study cohort was 96% (95% CI, 94% 97%) at 1 year and 68% (95% CI, 64% 71%) at 5 years. As expected, stage at diagnosis was strongly associated with survival (Table 2). In addition, colorectal cancer patients who underwent 1 or more colon examinations during follow-up had improved survival compared with patients who did not undergo colon examination (P.0001; Figure 1). Estimated 5-year survival was similar regardless of whether the initial colon examination was performed within 18 months of diagnosis, between months of diagnosis, or between months of diagnosis (78.0% [95% CI, 72.4% 82.7%]; 75.7% [95% CI, 63.9% 84.2%]; 77.3% [95% CI, 51.9% 90.4%], respectively). In comparison, 5-year survival was at least 45% lower (52.2%; 95% CI, 45.4% 58.6%) among members of the cohort who did not undergo a colon examination during follow-up. Table 1. Characteristics of 1002 Subjects Diagnosed With Colorectal Cancer, a Characteristic Number % Gender (n 1002) Male Female Age at diagnosis (y) (n 1002) Race (n 995) White Black Other b AJCC tumor stage (n 1002) I II III Tumor location (n 1002) Colon Rectum Received adjuvant therapy Radiotherapy (n 998) Chemotherapy (n 991) AJCC, American Joint Commission on Cancer. a Cases with missing data excluded from column totals. b Other races include Asian or Pacific Islander (n 29), Latino (n 5), Native American (n 3), other (n 1). Table 2. Factors Associated With Mortality Among Colorectal Cancer Patients With Local or Regional Disease at Diagnosis (n 1002) Factor HR 95% CI Receipt of surveillance colonoscopy 0.58 a Age b 1.03 a Female gender 0.68 a Race White 1.00 Black 1.92 a Other Chronic disease score c 1.11 a AJCC stage I II 1.94 a III 3.81 a Tumor site Colon 1.00 Rectum Receipt of adjuvant chemotherapy 0.50 a Receipt of adjuvant radiotherapy 1.78 a NOTE. Cox proportional hazards model , 12 df, P a P.05. b Per 1-year increase in age. c Per 10% change in chronic disease score. Receipt of a colon examination was a significant independent predictor of decreased overall mortality (HR, 0.58; 95% CI, ) after adjustment for other prognostic variables (Table 2). In addition, female gender (HR, 0.68; 95% CI, ) and the receipt of adjuvant chemotherapy (HR, 0.50; 95% CI, ) were associated with decreased mortality. Increased mortality was seen among black patients (HR, 1.92; 95% CI, ) as well as among older patients, patients with greater medical comorbidity, patients with more advanced tumor stage, and among patients treated with radiotherapy. Figure 1. Overall survival of colorectal cancer patients undergoing 1 or more colon exam (solid line) compared with patients not undergoing colon exam (dotted line). Patients who underwent colon surveillance had significantly improved overall survival (P.0001). Number of patients: colon surveillance (n 652); no colon surveillance (n 350).

4 April 2007 OUTCOMES OF COLON SURVEILLANCE 473 Table 3. Most Advanced Finding on Initial and Subsequent Follow-up Colonoscopy All colonoscopies (n 936) Initial exam (n 652) Subsequent exams a (n 284) Finding No. % of exams No. % of exams No. % of exams Carcinoma Adenoma Tubular Villous Unspecified Advanced neoplasia Hyperplastic polyp a Distribution of subsequent follow-up exams: 2 exams (n 242), 3 exams (n 33), 4 exams (n 7), 5 exams (n 2). Findings of Follow-up Colonoscopy The most significant findings on follow-up colonoscopies included carcinoma (n 10, 1.1%), adenoma (n 206, 22.0%), or hyperplastic polyp (n 159, 17.0%) (Table 3). Advanced neoplasia was identified in 7.3% (n 68) of follow-up colonoscopies, including 23 exams detecting an adenoma 10 mm and 35 exams detecting adenomas with villous architecture. Most patients underwent their initial follow-up colonoscopy within 18 months of initial diagnosis (n 484), with fewer patients undergoing initial follow-up colonoscopy between 18 and 36 months since initial diagnosis (n 110) or between 36 and 60 months since initial diagnosis (n 58). Advanced neoplasia was more commonly identified when the initial postdiagnostic surveillance examination was deferred until months after diagnosis (n 9, 15.5% of exams) compared with initial exams performed within 18 months of diagnosis (n 33, 6.9% of exams) (P.02). Diagnosis of Interval Cancers Overall, 20 interval cancers were found during followup, including 9 cancers diagnosed within 18 months of the initial cancer diagnosis (Table 4). A total of 10 cancers were diagnosed on follow-up colonoscopy, and the majority of these were found on the initial follow-up exam. Most cancers diagnosed on follow-up colonoscopy were found in different segments of the colon than the original cancer (n 6), although 2 cancers were found at the anastomotic site, and 2 were found in the same segment of the colon as the anastomosis. In addition, 44% (n 98) of rectal cancer patients in the cohort underwent a total of 209 flexible sigmoidoscopies during follow-up, and these examinations diagnosed 10 (4.8%) additional carcinomas arising in the rectum and 11 (5.3%) other advanced neoplastic lesions. The overall incidence of cancer detected by follow-up colonoscopy or sigmoidoscopy was 5.0 per 1000 patient-years of follow-up. Survival of patients diagnosed with carcinoma by either surveillance colonoscopy or flexible sigmoidoscopy was 64%, 53%, and 40% at 1, 3, and 5 years after diagnosis of the second cancer, respectively. There were no differences in survival among patients diagnosed with carcinoma by sigmoidoscopy compared with colonoscopy (P.95). Factors Associated With Neoplasia on Follow-up Colonoscopy Almost one half (48%) of all patients who underwent surveillance colonoscopy had a history of adenomas. These patients (n 315) were more likely to have adenomas (27.0% vs 18.1%, P.006) and advanced neoplasia (12.1% vs 5.3%, P.002) on follow-up colonoscopy compared with patients without a history of adenomas (n 337). In contrast, we observed no difference in the adenoma prevalence at surveillance colonoscopy among patients with and without a family history of colon cancer (n 116 and n 279, respectively; adenoma prevalence, 22.4% and 23.7%; P.79). There was an association between family history of colon cancer and the prevalence of advanced neoplasia at surveillance colonoscopy, but this finding did not achieve statistical significance (n 6, 5.2% vs n 30, 10.8%; P.08). Approximately one third of the cohort (n 235) underwent more than 1 colonoscopy during follow-up. Patients with no Table 4. Interval Cancers Detected on Initial Surveillance Exam With Colonoscopy or Sigmoidoscopy Site of original cancer No. of patients No. with follow-up colonoscopy or sigmoidoscopy Follow-up interval Initial follow-up exam No. of interval cancers Subsequent follow-up exams Colon cancer mo (n 373) mo (n 86) mo (n 46) 1 1 Rectal cancer mo (n 105) mo (n 48) mo (n 33) 1 1 Total interval cancers (n 20) 13 7

5 474 RULYAK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 adenomas found on the initial colonoscopy (n 181) rarely had adenomas (n 11, 6.1%) or advanced neoplasia (n 1, 0.6%) on subsequent exams, despite a median of 35.8 months (95% CI, months) between examinations. Patients with non-advanced adenomas on the initial follow-up examination (n 54) were frequently found to have adenomas on the second surveillance examination (n 36, 67%) but rarely had advanced neoplasia on repeat surveillance (n 3, 1.9%). However, patients with advanced neoplasia on the initial exam (n 16) were frequently found to have advanced neoplasia on subsequent exams (n 13, 81.2%), with a median of 24.0 months (95% CI, months) between examinations in this group. Discussion The present study found that colorectal cancer survivors who undergo follow-up colon examination have improved survival compared with patients who do not, even after adjustment for other important variables including tumor stage and medical comorbidity. To further evaluate the contribution of surveillance colonoscopy to this survival benefit, we examined the pathologic findings of follow-up colonoscopies performed in our cohort. Our results suggest that this survival benefit associated with colonoscopy cannot be explained solely by the finding of curable neoplasia. Relatively few patients were diagnosed with curable colorectal cancer on colonoscopy, and much of the survival benefit appears to precede the interval when even advanced neoplastic lesions would be expected to evolve into a lethal cancer. 43,44 Although we did not find additional survival benefit associated with colonoscopy performed earlier than 3 years after resection, the prevalence of adenomas and advanced neoplasia does increase as the interval between resection and surveillance lengthens. Moreover, 9 of 20 (45%) interval cancers in our cohort were diagnosed within 18 months of initial cancer diagnosis, and we speculate that many of the cancers and advanced neoplastic lesions diagnosed on follow-up might have represented missed synchronous lesions that were present at the time of initial cancer diagnosis. Our results do suggest that certain patients such as those with a personal history of other adenomas and those with advanced neoplasia at the initial surveillance examination are at further increased risk of neoplasia and might warrant more intensive surveillance compared with patients without these risk factors. Interestingly, a family history of colon cancer was not significantly associated with additional risk of adenomas or advanced neoplasia in our cohort, although a trend toward an association was noted. A retrospective study of 3546 U.S. veterans also found improved survival among patients undergoing surveillance colonoscopy, although data on tumor stage were not available in that study. 45 Our findings in a non-veteran population are similar and also show that the survival benefit appears to be independent of tumor stage. Nevertheless, given that relatively few cancers and advanced neoplasms were found during follow-up in this study, reasons for improved survival are unclear. It is possible that this benefit might be attributable to the use of other radiographic or laboratory surveillance tests performed concurrently with surveillance endoscopy. Alternatively, this result might reflect confounding by unmeasured comorbid illness that could have resulted in both decreased use of follow-up colonoscopy and poorer survival. Although we used a validated measure to control for comorbidity, previous studies have shown that all comorbidity measures determined from administrative data are imperfect in their ability to control for confounding by concurrent illness. 35,37 However, we do not believe that our findings can be attributed to lack of follow-up among patients who did not have a colonoscopy because these patients had a mean of 50.4 outpatient visits during follow-up. Several other factors, in addition to colon surveillance, were associated with survival after colorectal cancer diagnosis. We found that older patients, those with more advanced tumor stage at diagnosis, and those with greater medical comorbidty had poorer survival. We also found poorer survival among black patients, in keeping with the findings of prior studies Although the reasons for poorer survival in blacks have been the subject of debate, our results suggest that this finding is independent of the receipt of follow-up colonoscopy, tumor stage at diagnosis, and comorbidity. Female sex was also an independent predictor of survival, although there were no significant differences in the prevalence of recurrent cancers or advanced neoplasia among the women in our study cohort (data not shown). Last, we found that the receipt of adjuvant chemotherapy was associated with improved survival, whereas the need for radiotherapy was associated with poorer survival, perhaps because the latter therapy was offered to patients with suboptimal surgical outcomes or more advanced pathology. Flexible sigmoidoscopy has been recommended for the early postoperative surveillance of rectal cancer patients because of the greater propensity of this population to develop local recurrence. 15,16 In our study, the yield of flexible sigmoidoscopy in the evaluation of rectal cancer patients was substantial and included 10 rectal carcinomas that likely represented local recurrences. However, it is important to emphasize that fewer than half of the rectal cancer patients in our population underwent sigmoidoscopy during follow-up, and we cannot be certain whether the indication for these exams was surveillance or symptom-directed, so it is difficult to conclude definitively that the uniform use of surveillance sigmoidoscopy is warranted on the basis of the present study. Some additional limitations to the present study deserve mention. First, although colonoscopies performed in this study are referred to as surveillance, the retrospective nature of our study did not permit us to determine whether the exams were performed for surveillance of asymptomatic patients or to evaluate symptoms. However, if symptoms of recurrent or metachronous cancer did lead to the performance of colonoscopy, we would expect to observe poorer survival in the surveillance group. Nonetheless, misclassifying symptomatic examinations as surveillance exams could result in an overestimate of the prevalence of adenomas and carcinomas. Second, although it is possible that some early interval cancers in this study were present at diagnosis but undetected because of incomplete colonoscopy as a result of tumor obstruction or technical difficulty, none of the patients with interval cancer in this study were diagnosed within 6 months after their initial colorectal cancer, suggesting that this potential limitation did not have a large impact on study findings. In addition, the prevalence of advanced neoplasia in this cohort might have been underestimated, because aggregate polyp size was not available in 30% of pathology reports reviewed.

6 April 2007 OUTCOMES OF COLON SURVEILLANCE 475 We considered family history data missing for a large proportion of study subjects because analysis of family history was restricted to patients with a mention of family history, positive or negative, in their medical record. Although the associated reduction in sample size limited our ability to detect a statistically significant association between family history and adenoma or advanced neoplasia as postdiagnostic surveillance, this method minimized the high potential for misclassification of family history on the basis of medical record review. However, we found no evidence of increased risk in our analysis on the basis of the limited sample for whom family history could be accurately ascertained. Finally, it is possible that variations in surgical practices could also be associated with patient survival, although it is not clear that such variations would influence the use or timing of surveillance colonoscopy. In conclusion, the results of the present study support growing evidence that patients with colorectal cancer who undergo colonoscopic surveillance during follow-up appear to have improved survival. After resection of colorectal cancer, patients are at high risk of interval cancers and advanced adenomas, some of which are likely to represent lesions missed at the time of initial cancer diagnosis. Although many current published guidelines recommend initial surveillance colonoscopy between 3 5 years from the time of resection, our results support the routine use of colonoscopic surveillance at 1 year after resection as suggested by recent guidelines published by the American Cancer Society and the U.S. Multi-Society Task Force on Colorectal Cancer. 17 Thereafter, more intensive surveillance might be warranted in patients with advanced neoplasia on initial surveillance examination as well as among patients with a history of adenomas. Last, it appears that flexible sigmoidoscopy might have additional utility for detection of local recurrence after resection for rectal cancer, although this area warrants further investigation. References 1. Jemal A, Murray T, Ward E, et al. Cancer statistics, CA Cancer J Clin 2005;55: Easson AM, Cotterchio M, Crosby JA, et al. A population-based study of the extent of surgical resection of potentially curable colon cancer. Ann Surg Oncol 2002;9: Galandiuk S, Wieand HS, Moertel CG, et al. Patterns of recurrence after curative resection of carcinoma of the colon and rectum. Surg Gynecol Obstet 1992;174: Safi F, Link KH, Beger HG. Is follow-up of colorectal cancer patients worthwhile? Dis Colon Rectum 1993;36: Read TE, Mutch MG, Chang BW, et al. Locoregional recurrence and survival after curative resection of adenocarcinoma of the colon. J Am Coll Surg 2002;195: Cali RL, Pitsch RM, Thorson AG, et al. Cumulative incidence of metachronous colorectal cancer. Dis Colon Rectum 1993;36: Togashi K, Konishi F, Ozawa A, et al. Predictive factors for detecting colorectal carcinomas in surveillance colonoscopy after colorectal cancer surgery. Dis Colon Rectum 2000;43(Suppl): S47 S Chen F, Stuart M. Colonoscopic follow-up of colorectal carcinoma. Dis Colon Rectum 1994;37: 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: Fleischer DE, Goldberg SB, Browning TH, et al. Detection and surveillance of colorectal cancer. JAMA 1989;261: Engstrom PF, Benson AB 3rd, Cohen A, et al. NCCN colorectal cancer practice guidelines: the National Comprehensive Cancer Network. Oncology 1996;10(Suppl): Richard CS, McLeod RS. Follow-up of patients after resection for colorectal cancer: a position paper of the Canadian Society of Surgical Oncology and the Canadian Society of Colon and Rectal Surgeons. Can J Surg 1997;40: Winawer SJ, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997; 112: Winawer S, Fletcher R, Rex D, et al. Colorectal cancer screening and surveillance: clinical guidelines and rationale-update based on new evidence. Gastroenterology 2003;124: Desch CE, Benson AB 3rd, Smith TJ, et al. Recommended colorectal cancer surveillance guidelines by the American Society of Clinical Oncology. J Clin Oncol 1999;17: Benson AB 3rd, Desch CE, Desch CE, et al update of American Society of Clinical Oncology colorectal cancer surveillance guidelines. J Clin Oncol 2000;18: Rex DK, Kahi CJ, Levin B, et al. Guidelines for colonoscopy surveillance after cancer resection: a consensus update by the American Cancer Society and the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology 2006;130: Makela JT, Laitinen SO, Kairaluoma MI. Five-year follow-up after radical surgery for colorectal cancer: results of a prospective randomized trial. Arch Surg 1995;130: Ohlsson B, Breland U, Ekberg H, et al. Follow-up after curative surgery for colorectal carcinoma: randomized comparison with no follow-up. Dis Colon Rectum 1995;38: Kjeldsen BJ, Kronborg O, Fenger C, et al. A prospective randomized study of follow-up after radical surgery for colorectal cancer. Br J Surg 1997;84: Schoemaker D, Black R, Giles L, et al. Yearly colonoscopy, liver CT, and chest radiography do not influence 5-year survival of colorectal cancer patients. Gastroenterology 1998;114: Pietra N, Sarli L, Costi R, et al. Role of follow-up in management of local recurrences of colorectal cancer. Dis Colon Rectum 1998;41: Secco GB, Fardelli R, Gianquinto D, et al. Efficacy and cost of risk-adapted follow-up in patients after colorectal cancer surgery: a prospective, randomized and controlled trial. Eur J Surg Oncol 2002;28: Renehan AG, Egger M, Saunders MP, et al. Impact on survival of intense follow-up after curative resection of colorectal cancer: systematic review and meta-analysis of randomized trials. BMJ 2002;324: Jeffery GM, Hickey BE, Hider P. Follow-up strategies for patients treated for non-metastatic colorectal cancer. Cochrane Database Syst Rev 2002;1:CD Figueredo A, Rumble RB, Maroun J, et al. Follow-up of patients with curatively resected colorectal cancer: a practice guideline. BMC Cancer 2003;3: Audisio RA, Setti-Carraro P, Segala M, et al. Follow-up in colorectal cancer patients: a cost-benefit analysis. Ann Surg Oncol 1996;3: Ries LAG, Eisner MP, Kosary CL, et al (eds). SEER Cancer Statistics Review, , National Cancer Institute. Bethesda, MD. Available at: Accessed September Fleming ID, Cooper JS, Henson DE, et al. American Joint Committee on Cancer cancer staging manual. 5th ed. Philadelphia: Lippincott-Raven, Ehrinpreis MN, Kinzie JL, Jaszewski R, et al. Management of the malignant polyp. Gastroenterol Clin North Am 1988;17: Masaki T, Mori T, Matsuoka H, et al. Colonoscopic treatment of colon cancers. Surg Oncol Clin N Am 2001;10:

7 476 RULYAK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No American Medical Association. Current procedural terminology. Chicago: American Medical Association, Schoen RE, Gerber LD, Margulies C. The pathologic measurement of polyp size is preferable to the endoscopic estimate. Gastrointest Endosc 1997;46: Clark DO, Von Korff M, Saunders K, et al. A chronic disease score with empirically derived weights. Med Care 1995;33: Schneeweiss S, Maclure M. Use of comorbidity scores for control of confounding in studies using administrative databases. Int J Epidemiol. 2000;29: Schneeweiss S, Seeger JD, Maclure M, et al. Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data. Am J Epidemiol 2001;154: Perkins AJ, Kroenke K, Unutzer J, et al. Common comorbidity scales were similar in their ability to predict health care costs and mortality. J Clin Epidemiol 2004;57: Boulos DL, Groome PA, Brundage MD, et al. Predictive validity of five comorbidity indices in prostate carcinoma patients treated with curative intent. Cancer 2006;106: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: Hosmer DW, Lemeshow S. Applied survival analysis: regression modeling of time to event data. New York: John Wiley & Sons, Inc, Cox DR. Regression models and life tables. J Royal Stat Soc 1972;B34: Prentice RL, Williams BJ, Petersen AV. On the regression of multivariate failure time data. Biometrika 1981;68: Winawer SJ, Zauber AG, O Brien MJ, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps: the National Polyp Study Work Group. N Engl J Med 1993; 28: Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy: the National Polyp Study Work Group. N Engl J Med 1993;329: Fisher DA, Jeffreys A, Grambow SC, et al. Mortality and follow-up colonoscopy after colorectal cancer. Am J Gastroenterol 2003; 98: Shavers VL, Brown ML. Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst 2002;94: Govindarajan R, Shah RV, Erkman LG, et al. Racial differences in the outcome of patients with colorectal carcinoma. Cancer 2003; 97: Rabeneck L, Souchek J, El-Serag HB. Survival of colorectal cancer patients hospitalized in the Veterans Affairs Health Care System. Am J Gastroenterol 2003;98: Address requests for reprints to: Stephen J. Rulyak, MD, MPH, University of Washington, Division of Gastroenterology, Harborview Medical Center, 325 Ninth Ave, HH-642, Box , Seattle, Washington 98104; Fax: (206) Supported by NIH GI training grant DK07742 and the Applied Research Program, NCI Division of Cancer Control and Population Sciences. Presented at Digestive Disease Week, May 15-20, 2004, New Orleans, LA.