287 Effect of a Pre-Specified Minimum Colonoscopic Withdrawal Time On Adenoma Detection Rates During Screening Colonoscopy Robert L. Barclay, Joseph J. Vicari,John F. Johanson, Roger L. Greenlaw Background: Screening colonoscopy provides protection against colorectal cancer through the removal of precursor lesions, adenomatous polyps. Therefore, efforts to improve colonoscopic detection of adenomas may enhance the overall utility of colonoscopy. Recently, we demonstrated a direct, linear relationship between colonoscopic withdrawal times and adenoma detection rates among endoscopists performing screening colonoscopy. These data support the concept of a threshold time for instrument withdrawal in order to adequately assess for adenomatous polyps. Aim: The aim of this study was to assess the effect of a minimum pre- specified colonoscopic withdrawal time on the rates of adenoma detection during screening colonoscopy. We hypothesized that adherence to a minimum withdrawal time would result in improved adenoma detection rates. Methods: The study population consisted of 818 consecutive average risk subjects undergoing screening colonoscopy performed by one of 12 experienced, Board-certified community- based gastroenterologists. During colonoscopic withdrawal from the cecum, a timer with an audible alarm feature was used to signal 2-minute intervals and an 8- minute cecum-to-anus withdrawal time to the endoscopist. Endoscopists were instructed to adhere to a minimum procedural withdrawal time of 8 minutes, and to employ careful inspection techniques to try to optimize polyp detection. Withdrawal times were recorded by an endoscopy nurse. Adenoma detection rates and withdrawal times among subjects undergoing ‘‘minimum withdrawal time colonoscopy’’ were compared to those measured in an average risk control population (N Z 2559) that had previously undergone screening colonoscopy by the same group of endoscopists. Results: Mean withdrawal time among study subjects was 12.5 G 0.3 minutes, vs. 9.3 G 0.1 minutes in control subjects (p ! 0.0001). Overall adenoma detection rates among subjects undergoing minimum withdrawal time colonoscopy were significantly greater than in control subjects: 52% vs. 25% of subjects with adenoma(s); 0.73 vs. 0.58 adenomas per subject screened (p ! 0.001). For all endoscopists, instrument withdrawal times and adenoma detection rates increased significantly compared to their corresponding values in control subjects. Conclusion: Implementation of a pre- specified cecum-to-anus withdrawal time results in substantial improvements in adenoma detection rates during screening colonoscopy. 288 Impact of a Virtual Colonoscopy Screening Program On Optical Colonoscopy in Clinical Practice: One Year Data Darren C. Schwartz, Kevin J. Dasher, Adnan Said, Deepak V. Gopal, Mark Reichelderfer, David H. Kim, Perry J. Pickhardt, Andrew J. Taylor, Patrick R. Pfau Background: The potential impact of Virtual Colonoscopy (VC), also known as CT Colonography, on Optical Colonoscopy (OC) has been the topic of much speculation. Quantitative mathematical models have predicted that screening VC will lead to decreased demand for OC. Some hypothesize that VC will shift OC away from colon cancer screening (CRC) to mainly therapeutic applications (eg. polypectomy). Madison, WI is the only region in the United States where average risk CRC screening with VC is reimbursed by third-party payers. We report the first ‘‘real life’’ data of the impact of VC screening on OC in routine practice. Methods: VC and OC referral and productivity (ie. numbers performed) data were collected prospectively. We reviewed and compared data from the 3 month period before the introduction of screening VC to our institution (T1), the period 3 months before to 5 months after VC was introduced (T2), the 14 month period after introduction of VC (T3), and the period 6 to 14 months after VC screening was initiated (T4). The latter period was selected because it was felt to represent a ‘‘steady state’’ achieved following an initial incremental rise in VC productivity after its introduction. Data analyzed were monthly referrals for VC and OC, number of VC performed monthly, and number of monthly screening OC, total OC, and OC with polypectomy. Results: The mean number of overall [401.9 (T2) vs. 402 (T4), p Z 0.99] and screening [149.5 (T2) vs. 168.8 (T4), p Z 0.19] OC performed per month did not change significantly after screening VC was introduced. The mean percentage of OC with polypectomy remained constant during all study periods [42.2% (T1) vs. 41.7% (T3), p Z 0.83 and 40.5% (T2) vs. 42.9% (T4), p Z 0.21]. Of 119.8 mean screening VC performed per month during study period T4, 15.1% had polyps detected, and 7.1% (8.4 per month) were referred for OC with polypectomy. Mean monthly referrals for screening OC fell significantly from a mean of 247 (T2) to 203.6 (T4) (p Z 0.004) after institution of VC screening. In addition, 136 patients initially referred for OC subsequently opted for VC during the entire study period. Conclusions: 1) The initiation of a screening VC program did not result in a decrease in the number of OC performed in the 14 months since its introduction, but a significant reduction in referrals for OC did occur and may be an early indicator of decreased demand for OC. 2) Only a small percentage of VC were referred for OC with polypectomy, therefore a VC screening program does not appear to increase the overall number of therapeutic colonoscopies performed. 289 Risk Stratification for Advanced Colonic Neoplasia: A Screening Strategy Using Colonoscopy and Computerized Tomographic Colonography Otto Lin, Richard Kozarek, Jason Dominitz Background: The role of computerized tomographic colonography (CTC) in colon cancer screening remains undefined. One approach is to screen high-risk patients with colonoscopy while low-risk patients are screened with CTC. We developed and assessed a risk stratification index to identify such low-risk patients. Methods: 3005 asymptomatic persons aged 50C years who had undergone screening colonoscopy were retrospectively randomized to derivation and validation groups. We developed a risk index (using age, sex and family history) from the distribution of neoplastic lesions in the derivation group. The expected results of 3 screening strategies - universal colonoscopy, universal CTC, and a stratified strategy of colonoscopy for high-risk and CTC for low-risk patients - on our patients were then compared. Outcomes for the 3 strategies were projected from the known colonic findings in each patient, using sensitivity and specificity values for CTC and colonoscopy based on the medical literature. Referral for colonoscopy was assumed if a patient had a polyp O5 mm in size on CTC. Endpoints included the overall sensitivity for advanced neoplasia and the number of colonoscopies and CTCs required for each strategy. Results were validated in the validation group. Sensitivity analysis was performed by varying CTC sensitivity, specificity and threshold for colonoscopy referral. Results: In the derivation group (n Z 1512), screening colonoscopy had a sensitivity of 94% for advanced neoplasia. Universal CTC was projected to detect only 69% of advanced neoplasia and resulted in the largest total number of procedures (1723) and number of patients undergoing both procedures (211 or 14%). The stratified strategy detected 92% of advanced neoplasia, requiring colonoscopy in 1025 (68%) and CTC in 546 (36%) patients, with only 59 (4%) undergoing both procedures. In the validation group (n Z 1493), universal CTC detected 74% of advanced neoplasia, while the stratified strategy detected 90%, requiring colonoscopy in 960 (64%) and CTC in 600 (40%) patients, with 67 (4%) patients requiring both procedures. Unlike universal CTC, the stratified strategy was relatively independent of assumptions for CTC sensitivity, specificity and threshold for colonoscopy during sensitivity analysis. Conclusions: Compared with universal colonoscopy, the stratified screening strategy based on our risk index resulted in almost the same sensitivity for advanced neoplasia while reducing the number of needed colonoscopies by one-third, optimizing the return on colonoscopic resources. Universal CTC had low sensitivity for advanced neoplasia even under the most favorable assumptions in sensitivity analysis. 309 Development of Esophageal Adenocarcinoma in Patients with Barrett’s Esophagus and High Grade Dysplasia Undergoing Survilence: A Meta-Analysis and Systematic Review Srinivas R. Puli, Amit Rastogi, Sharad Mathur, Ajay Bansal, Prateek Sharma Purpose: Histologically, patients with Barretts Esophagus (BE) are stratified as those with no dysplasia, low grade dysplasia (including indeterminate for dysplasia), and high grade dysplasia (HGD). HGD patients are at the highest risk for development of esophageal cancer. The reported rate of cancer development in HGD patients undergoing surveillance is varied, making management of these patients very controversial i.e. surgery vs. endoscopic therapy vs. observation. Our aim was to determine the exact cancer incidence in patients with HGD undergoing surveillance endoscopy. Method: We searched for articles in Medline, Pubmed, Ovid journals, Cumulative index to nursing & Allied health literature, International pharmaceutical abstracts, old Medline, Medline non-indexed citations, and Cochrane control trial registry. Articles which met the following inclusion criteria were selected: patients with histologically proven BE and HGD, patients not having undergone endoscopic ablation or surgical therapy, no esophageal cancer at the time of enrollment or within 6 months, and follow up reported in person-time. All studies with a mean or median follow-up duration of less than six months were excluded. Weighted mean event rate for a study population was utilized to obtain a summary standardized rate, which was expressed as the direct standardized incidence rate (DSR). Approximate confidence intervals for DSR were calculated by both binomial model and Chiang’s normal approximation to Poisson rate sums. A test of heterogeneity was performed using the Mantel-Haenszel’s method. Results: Data from four articles which met the inclusion criteria were analyzed. This included 236 patients with HGD who were followed for 1240.5 patient years. A total of 69 esophageal cancers occurred in this group, giving a crude incidence rate of 5.57 per 100 patient-years. The direct standardized cancer incidence rate was 6.58 per 100 patient-years (95% confidence interval Z 4.97 to 8.19). The p value for the test of heterogeneity was 0.02. Conclusions: In BE patients with HGD undergoing surveillance, the standardized incidence rate of esophageal cancer development is 6.5 per 100 patient years and is higher than the crude rate. These data will help better inform physicians and patients in management decisions and also help compare the extent of any reduction in cancer incidence after intervention therapy for HGD. Abstracts www.giejournal.org Volume 63, No. 5 : 2006 GASTROINTESTINAL ENDOSCOPY AB83