Development of a prediction model for deep vein thrombosis in a retrospective cohort of patients with suspected deep vein thrombosis in primary care.

Objective: Early and accurate prediction and diagnosis of deep vein thrombosis (DVT) is essential to allow for immediate treatment and reduce potential complications. However, all potentially strong risk factors have not been included in pretest probability assessments such as the Wells score. In addition, the Wells score might not be suitable for use in primary care because it was developed for secondary care. We hypothesized that the addition of more risk factors for DVT to existing diagnostic approaches could improve the prediction of DVT.
Methods: All consecutive patients suspected of having DVT from 2004 to 2016 in a primary care setting were included in our retrospective study. All the patients had undergone Wells score, D-dimer, and duplex ultrasound assessments. The available recorded data of the patients were used to develop a model to predict DVT.
Results: Of 3381 eligible patients, 489 (14.5%) had confirmed DVT. The developed model, which included the D-dimer level, Wells score, gender, anticoagulation use, age, and family history of venous thrombosis, was able to distinguish patients with DVT among those with suspected DVT with a sensitivity of 82% (95% confidence interval, 78%-86%) and specificity of 82% (95% confidence interval, 80%-83%).
Conclusions: The proposed model was able to predict for the presence of DVT among all patients with suspected DVT in a primary care setting with reasonable accuracy. Further validation in prospective studies is required.
(Copyright © 2022 The Authors. Published by Elsevier Inc. All rights reserved.)