Evaluating longitudinal data accuracy in postpartum venous thromboembolism risk assessments: a retrospective observational study

Introduction Venous thromboembolism (VTE) risk assessment (VTERA) tools can be used to guide clinical decisions on postpartum thromboprophylaxis. However, the appropriateness of these decisions depends on the accuracy of the data input into VTERA tools. Objectives The primary objective was to assess whether the accuracy of postpartum VTE risk assessments differed between the study years (2019–2025), across nine predefined risk factors. The secondary objectives included evaluating the accuracy of all risk factors, and thromboprophylaxis recommendations, in a 2025 sample. Methods In this retrospective observational study, the accuracy of data entered into a postpartum VTERA tool was assessed, through comparison to electronic health record (EHR) data. Differences in accuracy across study years (2019–2025), for nine structured risk factors, were evaluated using Poisson regression. For the primary analysis, all deliveries and bookings, with a matched VTE risk assessment between January 2019 and August 2025 were included. For the secondary analysis, 283 patients from 2025 were manually reviewed, to assess all 23 risk factors and thromboprophylaxis recommendations. Results Primary analysis included 47 489 deliveries and bookings, with a matched risk assessment. BMI was the risk factor with the lowest accuracy overall (94.3%), and remained the lowest across each year. Compared with 2019, postpartum VTE risk factor recording was significantly more accurate in 2020, 2021, 2022, 2023, 2024 and 2025. Secondary analysis showed that 6.0% of risk assessments produced inaccurate thromboprophylaxis recommendations, regarding indication and duration, and 4.9% yielded inaccurate dose recommendations. Conclusions Future research must focus on enhancing the accuracy of VTE risk assessments. This may involve auto-population of VTERA tools using EHR data.