As shown in Table 1, 82 patients (14 men and 68 women, aged 40.1±12.8 years) who diagnosed with CD were enrolled in our study. All patients received ETS and 74 patients (90.24%) relieved after surgery, of whom 71 patients had immediate remission and 3 patients showed delayed remission; whereas the remaining 8 patients didn’t achieve remission. Nineteen patients (19/82, 23.17%) developed VTE, among which 5 cases were before surgery, 10 within 1-week after ETS, and the remaining 4 were within 40 days after ETS. After excluded those who developed VTE before surgery, the left 77 patients were then divided into two groups: postoperative VTE group (pVTE, n=14) and non-VTE group (non-pVTE, n=63). In the pVTE group, 2 patients had PE, while the remaining 12 were calf DVT events. Importantly, only 2 patients with PE showed symptoms of shortness of breath and decreased oxygen saturation post-surgery (3 and 14 days, respectively) and then diagnosed of coexistent DVT and PE, whereas the remaining 12 patients were all asymptomatic. Ten patients (71.4%) developed VTE within 1-week after surgery, and the remaining 4 were within 40 days postoperatively (14, 33, 35, and 40 days, respectively) (Table 1). Moreover, the postoperative mortality rate in our study was 1.22% (1/82), whose direct cause of death was brain hemorrhage and perioperative VTE events (diagnosed of DVT before surgery and DVT plus PE after surgery).
Baseline and clinical data for patients with postoperative VTE events
Patients’ clinical, biochemical features, and comorbidities are summarized in Table 2-1. The mean age of patients in the pVTE group was 50 (range 37-58) versus 34 (range 29-43) in the non-pVTE group (p<0.001). Current infections and diabetes were more common among pVTE patients (p<0.05). Hematocrit, 2h insulin, and C-peptide in OGTT test, and alanine aminotransferase (ALT) levels were lower in the pVTE group (p<0.05); whereas HbA1c was higher (p<0.05).
Table 2-2 summarizes surgery-related data, hormonal, and coagulation data. The remission rate was 78.57% (11/14) in pVTE group. Though without statistical significance, a higher remission rate was observed in the non-pVTE group (59/63, 93.65%, p>0.05). It is shown that bed time, usually represented patients’ statis status after surgery, was significantly longer in pVTE group than non-pVTE one (p<0.01). However, we didn’t observe statistical difference between groups when it comes to cerebrospinal fluid rhinorrhea rates, nadir cortisol levels after surgery within one week, as well as the post-surgery cortisol reduction value. In the meantime, no difference was observed regarding hormonal and coagulation parameters.
Correlations of postoperative VTE incident with clinical parameters in CD patients
Furthermore, we performed the correlation analysis to obtain the potent predictive parameters for VTE events in CD patients. As shown in Table 3, a strong positive correlation was found between VTE and age, course of disease, current infection, diabetes, blood urea nitrogen (BUN), HbA1c, postoperative bed time, and reduced mobility after surgery (defined as bed time>3days). Meanwhile, VTE was negatively correlated with ALT, 2h insulin, and C-peptide in OGTT.
Accuracy of CS-VTE score, Pauda score, and Caprini score in predicting VTE in our cohort
When we verified the CS-VTE score raised by Zilio M and his colleagues, the accuracy was about 76.83%, with a sensitivity of 10.53% and a specificity of 96.83%; which indicated that this scoring system might not enough to make an accurate prediction of VTE in Chinese CD patients. We also evaluated our entire cohort with the Padua prediction score and Caprini risk assessment model, which are wildly used in non-surgical and surgical inpatients to evaluate the risk of VTE, respectively17. In the low-risk and high-risk group of Pauda score, 20.83% (15/72) and 40% (4/10) patients had VTE, with a sensitivity of 21.1% and a specificity of 90.5%. In the low-risk, medium risk, high risk and extremely high group of Caprini score, 14% (7/50), 26.32% (5/19), 50% (6/12) and 100% (n=1) patients had VTE, thus the sensitivity was 36.8% and the specificity was 91.9%. In conclusion, these risk models can partially predict VTE in CD patients, but far from satisfying.
Establishment of a venous thromboembolism risk assessment model
By using stepwise regression analysis, we obtained 4 independent risk factors for VTE: age, 2-h insulin in OGTT, current infection, and postoperative bedtime. The nomogram method of the postoperative risk assessment model was then shown in Figure 1A. When evaluating patients with this model, we determined the points of each risk factor and then did the vertical lines to get score for each item. After summing up the total points, we can get the risk of VTE. The AUC of this model was 0.899 (95% CI, 0.787-0.999) (Figure 1B).