What Predicts Venous Thromboembolism Among Cancer Inpatients: A Proposed Predictive Model

Abstract Introduction Increased risk of venous thromboembolism (VTE) has been noted among cancer patients. Cancer associated thrombosis caused three folds increased hospitalizations, increased inpatient/outpatient medical and prescription claims, and increased total health care costs per patient. Our objective was to identify risk factors among demographic, clinical and laboratory risk for VTE among hospitalized cancer patients. We recognized the predictors of VTE and have proposed a predictive model. Methods Ours is a retrospective cohort study in patients with VTE and cancer from January 2013 to September 2015. Univariate (odds ratio with 95% confidence interval) and multivariate logistic regression analysis using stepwise approach was performed. An accurate ROC analysis with cross-validated probabilities of VTE was utilized to form a predictive model. By random partition, we formed a validation and derivative cohort to evaluate our model. Results A total of 3918 cancer inpatients were identified. Mean age was 66±14 years; 53% were males; 85% Caucasians, 7% African Americans; and 7% were obese. Common cancer types were 30% prostate, 16% lung, 11% cervical, and 11% breast. Common comorbidities were 46% hypertension, 35% pulmonary diseases, 21% renal diseases, 23% diabetes, 11% congestive heart failure, 9% anemia etc. Total 141 (3.87%) cancer patients had VTE. On univariate analysis, cancer patients with renal disease (95% CI: 1.42-2.91; p 0.0001), anemia (95% CI: 1.142-2.875; p 0.0116), and infection (95% CI: 0.96-2.04; p 0.078) were associated with increased VTE risk. On multivariate analysis, cancer group (Wald Chi-square: 9.86, p 0.0794), anemia (Wald Chi-square: 9.86, p 0.0794), infection (Wald Chi-square: 2.77, p 0.096) and renal disease (Wald Chi-square: 15.74; p Conclusion VTE was a common occurrence among male cancer inpatients. Anemia, renal diseases and infection increased VTE risk among cancer inpatients. Breast cancer, genitourinary cancer, anemia, infection, and renal disease formed the final predictive model. Download : Download high-res image (242KB) Download : Download full-size image Disclosures Dhakal: Graduation Medical Education, Inc., Michigan State University: Research Funding. Rayamajhi: Michigan State University, College of Education: Research Funding; Sparrow/MSU Center for Innovation and Research: Research Funding.