PURPOSE: Postoperative deep venous thrombosis (DVT) and pulmonary embolism (PE) are associated with high morbidity and mortality. Previous risk scores/calculators are limited by the number of variables and patients studied. Our objective was to study the current national burden of postoperative DVT and PE, identify factors associated with their development within 30 days of surgery, and develop a predictive risk calculator.
METHODS: Patients undergoing surgery were identified from the ACS’ 2007 NSQIP - a multicenter, prospective database. Multivariate logistic regression analyses were performed. A risk calculator was created and subsequently validated using the 2008 NSQIP dataset.
RESULTS: Of 211,399 patients, 1525(0.72%) developed DVT, while 654(0.31%) developed PE. Thirty-day mortality was higher in patients developing DVT than those not (7.74% vs. 1.70%, p<0.0001) and in those developing PE than those not (8.56% vs. 1.72%, p<0.0001). Preoperative risk factors significantly associated on multivariate analysis with DVT and PE include increasing age, higher ASA class, emergency case, impaired sensorium, prior operation within 30 days, and increasing weight. Corticosteroid use, dependent functional status, general anesthesia, recent myocardial infarction, preoperative sepsis, stroke, preoperative transfusion, >10% weight loss, African American/Caucasian vs. Hispanic race, and abnormal albumin and hematocrit were risk factors for DVT, but not PE. Disseminated cancer, dyspnea, neoadjuvant radiation, and African American vs. Asian/Hispanic/Caucasian race were risk factors for PE, but not DVT. Elevated PTT was protective for both. Dialysis dependence was protective for DVT, while diabetes and previous cardiac surgery were protective for PE (p<0.05 for all the above risk factors; c-statistic: DVT-0.811; PE-0.765). These factors were used to develop the risk calculator. Validation using the 2008 dataset showed similar c-statistics (DVT-0.821; PE-0.775) indicating similar model performance between the training and validation sets.
CONCLUSION: Postoperative DVT and PE, while uncommon, are associated with increased 30-day mortality. NSQIP data can be used to develop a risk calculator that adequately predicts postoperative DVT and PE.
CLINICAL IMPLICATIONS: The risk calculator is anticipated to aid in tailoring DVT prophylaxis, risk reduction, and informed patient consent.
DISCLOSURE: Prateek Gupta, No Financial Disclosure Information; No Product/Research Disclosure Information