Predicting deep venous thrombosis and pulmonary embolism in community patients with superficial venous thrombosis: a model development study.

BACKGROUND

In general practice, superficial venous thrombosis (SVT) is considered a benign condition that does not require anticoagulant treatment. However, a small but not insignificant subgroup of patients will develop serious thrombotic complications, such as deep venous thrombosis (DVT) and/or pulmonary embolism (PE). Tailored anticoagulant treatment in high-risk patients might prevent these complications, yet tools for clinical decision-making are currently lacking.

OBJECTIVES

To develop a multivariable prediction model to identify specific SVT subgroups at higher or lower risk of developing DVT/PE to aid clinical decision-making.

METHODS

Data from general practices in the longitudinal routine care UK Clinical Practice Research Datalink Aurum database were used in this study, covering the period from 2000 to 2021. A multivariable prediction model was developed for adult SVT patients to predict the outcome of DVT and/or PE within 90 days after the first consultation for SVT. Prespecified predictors were biological sex, age, (history of) cancer, history of DVT/PE, and absence of varicose veins. The model's performance was evaluated using bootstrapping and internal-external cross-validation.

RESULTS

Of all identified patients with SVT (N = 293 166; median age, 61 years [IQR, 46-73]; 33% male), 10 301 (3.5%) developed DVT or PE within 90 days. The risk was highest in the first days after the SVT consultation: 51% of DVTs/PEs were diagnosed within 1 week, and 77% within 30 days. The C-statistic of the model was 0.66 (95% CI, 0.66-0.67), and the model showed good calibration (intercept = 0.04, slope = 1.01). High-risk patients with a predicted risk of ≥5% (9.7% of the total population) had a history of DVT/PE or were male with (a history of) cancer.

CONCLUSION

In community SVT patients, the average risk of DVT/PE development within 90 days of follow-up is 3.5%. This simple prediction model showed good performance in predicting the short-term risk of DVT/PE and could be used to rule out thrombotic complications in this setting, but also to identify a subgroup of patients at higher risk in whom further diagnostic testing for concomitant DVT/PE and/or short-term preventive anticoagulant treatment should be considered to prevent clot progression.