Development and validation of a risk assessment model for hospital-acquired venous thrombosis in medical in-patients with cancer.

BACKGROUND

Existing risk assessment models (RAM) for hospital-acquired venous thromboembolism (VTE) have not been widely assessed in people with cancer or have underperformed.

OBJECTIVES

We aimed to develop and validate a RAM for hospital-acquired VTE for patients with active cancer admitted to medical services.

METHODS

We included admissions to medical services of adults ≥ 18 years with active cancer at 6 academic health systems from January 1, 2016, to December 31, 2022. We excluded patients with VTE at admission. Four systems formed the development cohort and 2 the validation cohort. First, we refit a prior RAM for medical patients by adding cancer type. Then, we then developed a de-novo RAM using Bayesian least absolute shrinkage and selection operator logistic regression.

RESULTS

Among 65 626 development and 58 124 validation admissions, hospital-acquired VTE occurred in 411 (0.6%) and 298 (0.5%), respectively. After refitting the existing RAM, the area under the receiver operating curve and calibration slope were 0.60 and 0.60, respectively. The de-novo model identified predictors (coefficient): prior VTE (1.08), malnutrition (0.42), hematologic malignancy excluding aggressive lymphomas (0.35), lung cancer (0.33), inflammatory comorbidity (0.41), age [≥75, -0.55; 60-75 years, -0.07]), anticoagulation level at admission (full, 0.45; prophylactic, -0.06), and chronic obstructive pulmonary disease (-0.04). The area under the receiver operating curve and calibration slope were 0.68 and 1.07 in the development cohort and 0.64 and 0.84 in the validation cohort, respectively.

CONCLUSION

The hospital-acquired RAM developed and validated in this study for patients with cancer performed better than a noncancer-specific model. This tool will support individualized risk assessment in medical in-patients with cancer.