A Validation Study of a Retrospective Venous Thromboembolism Risk Scoring Method

TLDR

Assessing patients for venous thromboembolism risk is essential to initiate appropriate prophylaxis and reduce mortality and morbidity from deep vein thrombosis and pulmonary embolism. The study validates a retrospective VTE risk scoring method derived from the Caprini model and evaluates how VTE prophylaxis confounds its performance. Logistic regression and a bivariate probit model were used to estimate odds ratios for 30‑day VTE by risk factor and risk level, controlling for prophylaxis adherence and other covariates, with risk levels distributed as 52.1 % highest, 36.5 % high, 10.4 % moderate, and 0.9 % low. The retrospective scoring method proved valid, with a 1.4 % overall 30‑day VTE incidence that rose to 1.94 % in the highest risk group, and risk level, pregnancy, sepsis, malignancy, prior VTE, and central venous access were significantly associated with VTE, while non‑adherence to prophylaxis guidelines correlated with higher VTE probability (ρ = 0.299).

Abstract

Validate a retrospective venous thromboembolism (VTE) risk scoring method, which was developed at the University of Michigan Health System and based on the Caprini risk assessment model, and assess the confounding effects of VTE prophylaxis.Assessing patients for risk of VTE is essential to initiating appropriate prophylaxis and reducing the mortality and morbidity associated with deep vein thrombosis and pulmonary embolism.VTE risk factors were identified for 8216 inpatients from the National Surgical Quality Improvement Program using the retrospective scoring method. Logistic regression was used to calculate odds ratios (OR) for VTE within 30 days after surgery for risk factors and risk level. A bivariate probit model estimated the effects of risk while controlling for adherence to prophylaxis guidelines.Distribution of the study population by risk level was highest, 52.1%; high, 36.5%; moderate, 10.4%; and low, 0.9%. Incidence of VTE within 30 days was overall 1.4%; by risk level: highest, 1.94%; high, 0.97%; moderate, 0.70%; low, 0%. Controlling for length of hospitalization (>2 d) and fiscal year, pregnancy or postpartum (OR = 8.3; 1.0-68, P < 0.05), recent sepsis (4.0; 1.4-10.9, P < 0.01), malignancy (2.3; 1.5-3.3, P < 0.01), history of VTE (2.1; 1.1-4.1, P < 0.05), and central venous access (1.8; 1.1-3.0, P < 0.05) were significantly associated with VTE. Risk level was significantly associated with VTE (1.9; 1.3-2.6, P < 0.01). The bivariate probit demonstrated significant correlation between the probability of VTE and lack of adherence to prophylaxis guidelines (rho = 0.299, P = 0.013).The retrospective risk scoring method is valid and supports use of individual patient assessment of risk for VTE within 30 days after surgery.