Ascertainment of colonoscopy indication using administrative data.

BACKGROUND: Administrative procedure code data can estimate colonoscopy utilization; however, determining colonoscopy indication is more difficult as procedure codes do not inherently reflect the purpose (screening, surveillance, diagnosis) of the colonoscopy. AIM: To improve the reported sensitivity (70%) and specificity (72%) of a published algorithm for identifying screening colonoscopies using Veterans Health Administration (VHA) administrative data. METHODS: We validated three algorithms for determining colonoscopy indication using medical records as the gold standard in a national sample of 650 patients. Algorithms used International Classification of Diseases, 9th Revision (ICD-9) and Current Procedural Terminology (CPT) codes. Medical records were manually abstracted using standardized protocols. RESULTS: The best algorithm had 83% sensitivity and 76% specificity for screening indication. Over 99% of colonoscopy CPT codes corresponded to a colonoscopy in the medical record. CONCLUSIONS: VHA procedure codes are very accurate for colonoscopy utilization; however, algorithms to ascertain indication have only moderate accuracy.

Duke Scholars

Author Cynthia Jan Coffman Biostatistics & Bioinformatics, Division of Biostatistics