CAN WE Predict Risk Factors of Venous Thromboembolism Among Cancer Inpatients

Abstract Introduction Increased risk of venous thromboembolism (VTE) has been noted among cancer patients as compared to non-cancer. VTE identified as leading cause of death among those with cancer. Cancer associated thrombosis caused increased hospitalizations, increased inpatient/outpatient medical and prescription claims, and increased total health care costs per patient. Our objective was to study demographic, clinical and laboratory risk factors for venous thromboembolism (VTE) among hospitalized cancer patients and built a predictive model for VTE risk. Methods Ours was a retrospective cohort study focused on patients with VTE and cancer from January 2013 - September 2015. Univariate and multivariate logistic regression analysis using stepwise approach was performed. A final predictive model was derived using receiver-operating characteristics (ROC) curves and concordance indices (c-statistics). Results N=3948 cancer inpatients were identified which was split into a derivation cohort and a validation cohort, each with 1957. Mean age 65.9±13.8 years; 52.6% were male; 85.6% Caucasian, 7% African Americans; 15.5% were obese; common comorbidities were hypertension (46%), pulmonary disease (34.5%), diabetes (22.9%), renal disease (20.9%) and congestive heart failure (10.4%). Overall, there was 152 (3.9%) events of VTE with 77 (3.9%) in derivation and 75 (3.8%) in validation cohort. On univariate analysis, comorbidities such as infection and renal diseases, laboratory findings such as low hemoglobin and low albumin was associated with high VTE risk. The derivation set had a c-statistic or AUC of 0.668 while the validation set had an AUC of 0.65. Conclusions Infection, renal disease (comorbidity) and low albumin levels were associated with a higher risk of VTE. Digestive and respiratory cancers were associated with higher VTE risk. We identified three clinical and laboratory variable that was associated with increased risk of VTE in addition to the cancer group. Future research could use this analysis as a basis for forming a risk score that could be used by clinicians to identify those cancer patients at risk for VTE. Disclosures No relevant conflicts of interest to declare.