Development and validation of a nomogram for identifying venous thromboembolism following oral and maxillofacial oncological surgery with simultaneous reconstruction

Abstract

ObjectivesThis study aimed to establish and validate a novel predictive model for venous thromboembolism (VTE) in patients undergoing oral and maxillofacial oncological surgery with simultaneous reconstruction. Material and methodsA total of 372 patients were selected, and their demographic data, comorbidities, medical history, laboratory variables, postoperative Caprini scores, perioperative indicators, and procedures were recorded and analyzed to build the model. The predictive model is displayed as a nomogram. ResultsThe incidence of VTE was 20.7% (77/372). Several factors were found to be significantly associated with VTE, including age (67 vs. 56 years, P < 0.001), preoperative level of D-dimer (0.56 vs. 0.36 mg/L, P < 0.001), proportion of female patients (46.8% vs. 33.6%, P = 0.032), hypertension (33.8% vs. 21%, P = 0.019). The predictive model was composed of age, gender, and preoperative D-dimer level, with good discriminative ability, as reflected by an area under the curve (AUC) of 0.756, the 95% confidence interval (CI) was 0.696–0.816. Moreover, it showed favorable diagnostic performance compared with both the 2005 (AUC 0.646, 95% CI = 0.578–0.714) and 2010 (AUC 0.627, 95% CI = 0.559–0.694) versions of the Caprini risk assessment model. For patients with malignant tumor, neoadjuvant chemotherapy was also an independent risk factor. ConclusionsThis novel predictive model consists of three readily available clinical variables that show good diagnostic performance in predicting postoperative VTE.