Concordance among venous thromboembolism risk assessment models

There are a lot of venous thromboembolism risk assessment models with the aim of predicting the individual risk of venous thromboembolism (VTE), although most have not been externally validated. The objective of our study was to analyse the concordance among the Padua, PRETEMED, IMPROVE and MEDENOX scores.

Observational cohort study with 602 patients admitted to Medical Services of the Hospital Universitario Santa Lucía. Concordance was calculated using the kappa index (KI) between the four risk prediction scales and the VTE occurring within 90 days of hospitalisation.

Patients considered were those with a high risk of VTE, 78%, 56%, 69% and 59% according to Padua, PRETEMED, IMPROVE and MEDENOX scores respectively. A KI=0.60 was observed among Padua-IMPROVE scores; KI=0.44 among IMPROVE-PRETEMED, KI=0.43 among PRETEMED-MEDENOX, KI=0.33 among Padua-PRETEMED, KI=0.27 among IMPROVE-MEDENOX and a KI=0.24 among Padua-MEDENOX. During follow-up, nine cases of VTE (1.5%) were observed.

The highest degree of agreement was observed among the Padua and IMPROVE scores. There is no good agreement between any of the predictive risk scales for VTE in medical patients. The Padua score was the best predictor of VTE episodes at 90 days among high risk patients.