Internal validation of the Portuguese National Registry of Vascular Procedures – abdominal aortic aneurysm module

INTRODUCTION: Medical registries are key tools for monitoring healthcare quality. In 2019, the Portuguese Society of Angiology and Vascular Surgery (SPACV) launched the Registo Nacional de Procedimentos Vasculares (RNPV), to collect nationwide data on vascular surgical procedures within the Portuguese healthcare network. The initial chapter focused on abdominal aortic aneurysms (AAAs) and has been in existence for five years. The value of interpreting such data is directly tied to its validity and completeness. Therefore, we aim to present the findings of the first internal validation of the RNPV- AAA module. METHODS: Four of the 20 institutions included in the RNPV were visited by three independent validators. Data from procedures performed between 2021- 2022 for AAA were evaluated. Registry data from 27 variables of 15 randomly selected cases per centre were compared with data from individual patient records for internal validation. The primary endpoint was internal validity. Secondary endpoints were data discrepancy and missing data. RESULTS: Sixty AAA cases across 4 institutions, totalling 405 data fields per centre and 1620 data fields across all cases, were compared with the original registration. An internal validity of 96.0% was observed. Overall, 1.1% data fields were missing, and 3.0% of data discrepancies were observed. Baseline characteristics (4.6%) presented the highest rate of mismatch, followed by procedure- related (3.6%) and outcome data (1.4%). Mismatch was more common in continuous data, with 15.0% mismatch in pre- operative creatinine and 16.7% in the largest aortic diameter. Follow- up data (not included in this internal validation) were missing in up to 54% of patients. CONCLUSION: The first validation of the RNPV- AAA chapter revealed reassuring results with a high degree of correspondence. Although conducted across a comprehensive set of variables, the missingness in long- term data remains noteworthy. Continuous data are more likely to be related to mismatched records, though the clinical implications are uncertain. Further external validation is anticipated.