[Validation of a deep vein thrombosis prediction rule in primary care]

Abstract

Objective: Suspected deep vein thrombosis (DVT) of lower limbs (LL) may require different tools to rule out or confirm the diagnosis. Clinical probability provides help to select useful tests, interpret their results, and decide to treat the patient meanwhile. Clinical prediction rules that risk stratify patients with suspected DVT can be established from inpatients, but no prediction rule not requiring laboratory tests has been established from primary care patients. We previously derived and internally validated such a prediction rule. The aim of this study is to externally validate this score.

Patients and methods: The score was applied to Optimev outpatients with suspected LL-DVT, and without suspected pulmonary embolism. Sensitivity and specificity were calculated for proximal and distal DVT, according to each score. The area under the ROC curve was calculated for each kind of DVT, in order to assess the validity of the score on predicting the presence or absence of DVT.

Results: Among 3523 outpatients prospectively included in the Optimev study for suspected LL DVT, overall prevalence of DVT was 29.7% (n=1046), ranging from 21.7% in the non-high score probability, to 61.4% in the high score probability. The area under the ROC curve was 0.79 [CI 95%, 0.77-0.80]. With subgroup analysis, the area under curve was 0.83 [CI 95%, 0.82-0.85] for proximal DVT, and 0.75 [CI 95%, 0.73-0.77] for distal DVT.

Conclusion: This score reliably identifies primary care patients with LL DVT, whether proximal or distal.