Venous thrombosis risk assessment in medical inpatients: the medical inpatients and thrombosis (MITH) study

Background We sought to define the risk factors present at admission for venous thromboembolism (VTE) in medical inpatients and develop a risk model for clinical use. Methods Between January 2002 and June 2009, 299 cases of hospital-acquired VTE were frequency matched to 601 controls. Records were abstracted using a standard form for characteristics of the thrombosis, medical conditions and other risk factors. Weighted logistic regression and survey methods were used to develop a risk model for hospital-acquired VTE that was validated by bootstrapping. Results VTE complicated 4.6 per 1000 admissions. Two risk assessment models were developed, one using laboratory data available at admission (Model 1) and the other excluding laboratory data (Model 2). Model 1 consisted of the following risk factors (points): history of congestive heart failure (5), history of inflammatory disease (4), fracture in the past 3months (3), history of VTE (2), history of cancer in the past 12months (1), tachycardia (2), respiratory dysfunction (1), white cell count 11x109/L (1), and platelet count 350x109/L (1). Model 2 was similar, except respiratory dysfunction had 2 points and white cell and platelet counts were removed. The c-statistic for Model 1 was 0.73 (95% CI 0.70, 0.77) and for Model 2 0.71 (95% CI 0.68, 0.75). Conclusions We present a VTE risk assessment model for use in medical inpatients. The score is simple and relies on information known at the time of admission and typically collected in all medical inpatients. External validation is needed.