S1011 The Burden of Digestive Diseases in Korea

Introduction: We previously utilized an administrative dataset, created for the purpose of recording diagnoses and visit/test charges at patient encounters, to identify that just 22% of over 20,000 screen eligible patients in the safety-net health system serving Fort Worth, Texas were up to date with colorectal cancer(CRC) screening(Gupta ‘09). While this study suggested that administrative data may be used to efficiently assess screening, the validity this approach has not been extensively studied. Further, test indication(screening vs. nonscreening), and the quality of test performance are not discernible from administrative data, which do not serve as a complete medical record. Understanding the validity of test use measurements, and quality of test use are critical for optimizing screening. We hypothesized that while administrative coding would be accurate for test use capture, many tests would be for non-screening indications, and further, that direct chart review would provide insight into test performance quality. Primary aims: 1) Determine validity of using administrative data coding for CRC test completion to measure screening test use, and 2) Evaluate test performance quality. Methods: Test use was determined by administrative coding for: 1) Fecal occult blood test(FOBT), 2) Barium enema, 3) Sigmoidoscopy, 3) Colonoscopy(COLO), or 5) No screening over a 5-year period(Gupta CEBP, DDW ‘09). 100 subjects from each category were randomly selected for medical chart abstraction, which served as the validation gold standard. A tool validated for determining accuracy of Medicare screening test claims data(Schenck ‘07) was adapted to abstract test indication(screening vs. sign/symptom evaluation), presence of test completion, and quality parameters. Agreement was estimated by the Kappa(κ) statistic. Descriptive statistics were used to characterize test indication and quality. Results: Agreement between the administrative database and chart abstraction was almost perfect for FOBT(κ=0.83, 95%CI:0.75-0.90) and COLO(κ=0.91, 95%CI:0.85-0.96), and fair for sigmoidoscopy(κ=0.39, 95%CI:0.28-0.49) and barium enema(κ=0.21, 95%CI:0.12-0.29). 40% of tests were for the indication of screening. Test quality, and record of quality parameters were variable. For example, among subjects with positive FOBT, follow-up COLO was noted as recommended for 46%, and bowel preparation quality and cecal intubation were undocumented for 42% and 17% of COLO, respectively. Conclusions: Administrative data have validity for CRC test completion measurement. For assessment of test indication and quality, medical chart review, or implementation of alternate data systems are required.