Nomograms to predict risk of in-hospital and post-discharge venous thromboembolism after abdominal and thoracic surgery: an American College of Surgeons National Surgical Quality Improvement Program analysis

Abstract Background Postoperative venous thromboembolism (VTE) is increasingly viewed as a quality of care metric, although risk-adjusted incident rates of postoperative VTE and VTE after hospital discharge (VTEDC) are not available. We sought to characterize the predictors of VTE and VTEDC to develop nomograms to estimate individual risk of VTE and VTEDC. Methods Using the American College of Surgeons National Surgical Quality Improvement Program database, we identified 471,867 patients who underwent inpatient abdominal or thoracic operations between 2005 and 2010. We excluded primary vascular and spine operations. We built logistic regression models using stepwise model selection and constructed nomograms for VTE and VTEDC with statistically significant covariates. Results The overall, unadjusted, 30-d incidence of VTE and VTEDC was 1.5% and 0.5%, respectively. Annual incidence rates remained unchanged over the study period. On multivariate analysis, age, body mass index, presence of preoperative infection, operation for cancer, procedure type (spleen highest), multivisceral resection, and non-bariatric laparoscopic surgery were significant predictors for VTE and VTEDC. Other significant predictors for VTE, but not VTEDC, included a history of chronic obstructive pulmonary disease, disseminated cancer, and emergent operation. We constructed and validated nomograms by bootstrapping. The concordance indices for VTE and VTEDC were 0.77 and 0.67, respectively. Conclusions Substantial variation exists in the incidence of VTE and VTEDC, depending on patient and procedural factors. We constructed nomograms to predict individual risk of 30-d VTE and VTEDC. These may allow more targeted quality improvement interventions to reduce VTE and VTEDC in high-risk general and thoracic surgery patients.