International comparison of registry data within vascular surgery has previously been published by Vascunet. One of the limitations of such comparisons is data validity and completeness, and meaningful interpretation of differences between countries can only be made if the data are robust within each of the countries studied. The Vascunet collaboration has therefore embarked on a validation exercise of international vascular registry data.
Five out of 20 hospitals performing vascular surgery in Sweden were visited by two international validators. Independent evaluation of the procedures of carotid endarterectomy and infrarenal abdominal aortic aneurysm repair was performed, and local hospital administrative data were compared with Swedvasc registry data. External validation compared the numbers of cases in these two systems of data collection and internal validation compared data accuracy and completeness within individual patient records.
Hospital records identified 335 carotid and 393 abdominal aortic aneurysm (AAA) procedures, whereas Swedvasc identified 331 carotid and 359 AAAs. Nine carotid procedures and 64 AAA procedures were found in hospital administrative data but not in Swedvasc, and 14 carotids and 30 AAAs were found in Swedvasc but not in hospital data. External validity was 100% (95% CI 98.8-100%) for carotids and 98.8% (95% CI 96.9-99.5%) for AAAs. In internal validation, 0.8% of variables were missing in hospital data compared with Swedvasc and 4.2% were missing in Swedvasc compared with hospital data. Data contained within the data fields of Swedvasc and hospital data were the same in 97.4% (95% CI 96.3-98.3%) for carotids and 96.2% (CI 94.9-97.2%) for AAAs.
This study has provided a template for international validation of registry data and has demonstrated that Swedvasc is a highly accurate system of data collection for Swedish vascular surgery.