Coding accuracy of abdominal aortic aneurysm repair procedures in administrative databases - a note of caution

Background Administrative databases have been used to compare methods used for abdominal aortic aneurysm (AAA) repair. This requires the use of procedural codes whose accuracy has not been established. In this study we measured the accuracy of procedural codes for open AAA repair and endovascular aneurysm repair (EVAR) in administrative databases. Methods Between April 2000 and July 2005, we identified all surgeries of non-ruptured AAA using open or EVAR technique at a tertiary-care teaching hospital. During the same time period, we identified all patients who were coded with either an open AAA repair or EVAR. Results During the study period, 514 people had an elective AAA repair or were coded with one. Coding quality of open AAA repair was poor (sensitivity 48.1%; specificity 77.4%; accuracy 52.9%) while that for EVAR was slightly better (sensitivity 58.2%; specificity 100%; accuracy 93.6%). We developed an algorithm that included similar procedures and considered anaesthetic type to improve the identification of both open repair (sensitivity 97.7%; specificity 86.9%; accuracy 95.9%) and EVAR (sensitivity 84.8%; specificity 99.5%; accuracy 97.3%). Conclusion Administrative database codes that are routinely used to identify open AAA repairs or EVARs are inaccurate. However, slight modifications to the coding algorithms permit the use of administrative databases to study AAA repair.