Skip to main content
BMC is moving to Springer Nature Link. Visit this journal in its new home.
Download PDF

External validation of four venous thromboembolism risk assessment models after colorectal cancer surgery: a retrospective study

Thrombosis Journal volume 23, Article number: 92 (2025) Cite this article

Abstract

Background

See More

Caprini score, the most commonly used assessment tool for predicting postoperative venous thromboembolism (VTE) risk has shown poor predictive value in colorectal cancer surgery. Recent risk assessment models (RAMs) Sir-Run-Run-Shaw VTE RAM, Risk of Venous Thromboembolism Algorithm (RVTA) score, and Colorectal Cancer - Venous Thromboembolism (CRC-VTE) score, which were specific for colorectal cancer, were developed and had good VTE predictive performance. We sought to externally validate for their generalizability and accuracy in Chinese patients undergoing colorectal cancer surgery.

Materials and methods

A retrospective analysis was conducted to predict the 6-month postoperative VTE risk in patients undergoing colorectal cancer surgery from January 2020 to December 2023. Demographic characteristics, clinical data, and 6-month postoperative VTE status of the patients were collected based on Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score, and Caprini score. We estimated the four VTE RAMs’ discrimination of 6-month postoperative VTE risk by using the area under the receiver operating characteristic curve (AUROC). Calibration plots, Hosmer-Lemeshow test, and decision curve analysis were also explored to assess the predictive performance of the four VTE RAMs.

Results

A total of 323 patients were included. The median age of our cohort was 66 years (range, 58–73 years), and 182 (56.3%) patients were male. VTE occurred in 68 (21.1%) cases within 6 months after operation, with 5 cases of pulmonary embolism and 63 cases of deep vein thrombosis, of which 45 (66.2%) cases experienced VTE within 4 weeks after operation. Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score and Caprini score demonstrated possibly helpful discrimination, with AUCs of 0.691 (95%CI: 0.624–0.758), 0.638 (95%CI: 0.564–0.713), 0.728 (95%CI: 0.663–0.793), and 0.661 (95%CI: 0.596–0.725), respectively. The Hosmer-Lemeshow test indicated a lack of fit for Sir-Run-Run-Shaw VTE RAM, RVTA score, and CRC-VTE score (P < 0.05). Furthermore, decision curve analysis revealed that CRC-VTE score provided greater net benefits than the other VTE RAMs.

Conclusion

External validation of the four VTE RAMs for predicting postoperative VTE in a real-world cohort of colorectal cancer patients showed that CRC-VTE score outperformed the other VTE RAMs. It can help clinicians identify patients with high risk of VTE, thereby facilitating timely prophylactic interventions and close monitoring.

Highlights

• 68 (21.1%) patients experienced VTE events within 6 months after colorectal cancer surgery, including 5 cases of PE and 63 cases of DVT.

• CRC-VTE score had an AUC of 0.728 (95%CI: 0.663–0.793), which was higher than the AUCs of the other VTE RAMs.

• The results of the comprehensive model discrimination, calibration, and clinical utility showed that CRC-VTE score outperformed the other VTE RAMs. It can identify patients with high risk of VTE and guide appropriate thromboprophylaxis.

Introduction

Venous thromboembolism (VTE), including deep vein thrombosis (DVT) and pulmonary embolism (PE), is the second leading cause of death in cancer patients [1]. The risk of VTE is 22-fold higher in surgical patients than that in non-surgical patients [2], and 9-fold higher in cancer patients than that in the general population [3]. Therefore, the risk of postoperative VTE in colorectal cancer patients is notably elevated, with incidence rates varying from 3.1–24.3% [4,5,6,7]. Currently, the National Comprehensive Cancer Network (NCCN), European Society for Medical Oncology (ESMO), and American Society of Clinical Oncology (ASCO) have proposed routine use of evidence-based VTE risk assessment models (RAMs), and recommended mechanical or pharmacologic thromboprophylaxis based on individual patient, disease, or treatment-related risk factors [8,9,10]. Clinicians can use validated VTE risk assessment tools to evaluate the risk of VTE and choose appropriate prophylactic anticoagulation strategies based on risk stratification. This approach aims to provide thromboprophylaxis for patients with high risk of VTE, and meanwhile it could avoid excessive prevention in patients with low risk of VTE, which may lead to an increased risk of bleeding.

Caprini score is the most commonly used VTE RAM for general surgery, and its external validation in patients undergoing colorectal cancer surgery has shown an area under the receiver operating characteristic curve (AUROC) ranging from 0.59 to 0.70, indicating low to moderate discrimination [6, 7, 11]. Because of its high sensitivity, most patients are classified as high risk of VTE, which may result in excessive preventive anticoagulation and bleeding events [12, 13]. Consequently, it may have limitations in predicting postoperative VTE risk in colorectal cancer patients, particularly among Asian populations, which also reported by Lee et al. [14]. These limitations are primarily reflected in the BMI threshold and the inclusion of certain risk factors that are less prevalent in Asian patients, such as Factor V Leiden and positive prothrombin20210A.

Several different VTE RAMs have been developed specifically for Chinese patients with colorectal cancer, such as CRC-VTE score [6], Sir-Run-Run-Shaw VTE RAM [11], and RVTA score [15]. CRC-VTE was developed through a multicenter prospective study, while the other two scores were derived from single-center retrospective cohorts. RVTA score was specifically designed for outpatients with colorectal cancer, whereas the other two scores were applied to postoperative colorectal cancer patients. They incorporated four to seven predictors associated with comorbidities, cancer stage, blood transfusion, D-dimer levels, etc., demonstrating strong predictive capabilities in assessing the risk of VTE in Chinese patients with colorectal cancer. Compared to Caprini score, the three VTE RAMs were all considered simpler and more efficient for predicting VTE risk. However, they have not been externally validated widely, and whether they are capable of adequately assessing the risk of postoperative VTE in colorectal cancer patients remains uncertain.

The performance of VTE RAMs can be evaluated from three aspects: discrimination, calibration, and clinical utility. Discrimination assesses the model’s capacity to accurately categorize patients into high-risk and low-risk groups for VTE, but it does not provide an accurate probability of VTE occurrence. Calibration assesses the consistency between actual probabilities and predicted probabilities of VTE. A model with high discrimination may not necessarily calibrate well. Therefore, if one model has better discrimination and the other better calibration, decision curve analysis (DCA) is more helpful for clinical decision making [16]. The purpose of this study was to externally validate the performance of Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score, and compare to Caprini score through a retrospective analysis of Chinese patients undergoing colorectal cancer surgery.

Materials and methods

Data source and study population.

Data from patients treated with colorectal cancer surgery from January 2020 to December 2023 were collected. The inclusion criteria were as follows: (1) patients aged 18 years or older; (2) patients treated with radical surgery for colorectal cancer; and (3) patients with complete clinical data. The exclusion criteria included patients with: (1) recurrent colorectal cancer; (2) diagnosis of VTE within 6 months prior to surgery; (3) long-term use of anticoagulants or antiplatelet agents prior to surgery; (4) hospitalization due to stroke, acute coronary syndrome, congestive heart failure, or acute respiratory failure; (5) without postoperative lower extremity venous ultrasound; and (6) treated with emergency surgery. We assessed the patient’s VTE risk using the Caprini score and evaluated bleeding risk to guide the selection of mechanical (e.g., graduated compression stockings) and/or pharmacological (e.g., low molecular weight heparin) prophylaxis for postoperative thrombosis. The decision to administer chemotherapy should be based on pathological stage and individual patient factors.

Ethical approval for this retrospective study was obtained from Beijing Friendship Hospital, Capital Medical University (number: 2023-P2-177-01). Informed consent was waived because of the retrospective design of the study. The work has been reported in line with the STROCSS criteria [17].

Study outcomes

Patients received at least one lower extremity ultrasound examination prior to discharge, with some patients undergoing additional unscheduled ultrasound examinations during the 6-month postoperative follow-up period. The primary outcome was image-confirmed asymptomatic or symptomatic VTE, (including proximal DVT, distal DVT, and acute symptomatic PE), which occurred within six months after operation. The diagnosis of DVT was confirmed by color ultrasound of lower extremity veins, with endovascular thrombosis. The diagnosis of PE was confirmed incidentally by computed tomography pulmonary angiography (CTPA) on follow-up, with filling defect of pulmonary artery and its branches.

VTE RAMs.

The four VTE RAMs externally validated in our study were as follows:

  1. (1)

    Sir-Run-Run-Shaw VTE RAM [11]: Logit(P) = −3.161 + (0.978*Age ≥ 69 years) + (0.859*Preoperative plasma D-dimer ≥ 0.49 mg/L) + (0.956*Transfusion) + (0.929*Cancer stage IV).

  2. (2)

    RVTA score [15]: Logit(P) = − 4.047 + (2.421*D-dimer ≥ 1.7 µg/mL) + (0.938*Multiple chemotherapy regimens) + (1.298*Platelet count ≥ 350 × 109/L) + (1.835*Metastasis) + (1.197*Blood transfusion history).

  3. (3)

    CRC-VTE score [6]: Logit(P) = −1.981 + (0.622*Female) + (1.000*Age ≥ 70 years) + (1.001*Varicose veins of lower extremity) + (0.949*Cardiac insufficiency) + (0.707*Preoperative bowel obstruction) + (0.495*Preoperative bloody stool/tarry stool) + (0.622*Anesthesia time ≥ 180 min).

  4. (4)

    Caprini score was routinely used to assess the risk of VTE in colorectal cancer patients upon admission, 24 h post-operation, and at the time of discharge. And the patient’s risk score for VTE, as documented in their electronic medical records, was calculated based on type of operation, age, and the presence of additional risk factors in this model [18].

Statistical analysis

Statistical analyses were performed using SPSS software, version 27.0 (IBM Corp, Armonk, NY, USA) and the R software, version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were presented as means ± standard deviation (SD) or medians with interquartile range (IQR), and categorical variables were reported as numbers and percentages. The associations among risk factors and VTE events were examined by using the χ² test or Fisher’s exact test. To evaluate the predictive performance, we assessed the discrimination, calibration and clinical utility of the four VTE RAMs. AUC and the corresponding 95% confidence interval (CI) were calculated to assess the discrimination ability of the model, while the Hosmer-Lemeshow test and brier score were used to assess model calibration. Additionally, DCA was performed to assess the clinical utility of the model. AUC of 0.5 to 0.6, 0.6 to 0.75 and more than 0.75 were considered as poor, possibly helpful and clearly useful discrimination [19]. Hosmer-Lemeshow test was used to assess the goodness of fit between observations and predictions, with P < 0.05 suggesting that the model was not well calibrated [20]. Brier score evaluated model calibration ranging from 0 to 1, with lower scores representing higher predictive accuracy [20]. The comparisons between VTE RAMs were performed by comparing ROC curves using the DeLong’s test. And we also calculated the sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) based on the optimal cut-off points. Our study was registered with www.chictr.org.cn (number ChiCTR2400092069) on 8 November 2024.

Results

Characteristics of the study cohort

A total of 323 patients who underwent colorectal cancer surgery were included in our study. The characteristics of these patients were described in Table 1. The median age was 66 years (range, 58–73 years), 182 (56.3%) patients were male and the mean (SD) body mass index (BMI) was 24.0 (3.6) kg/m2. In terms of disease history, the most common comorbidities were hypertension (48.3%), diabetes (28.8%) and cardiac insufficiency (28.5%). Preoperative bloody stool/tarry stool and preoperative bowel obstruction were present in 134 (41.5%), and 34 (10.5%) patients, respectively. The median intraoperative blood loss was 50 ml (range, 50–100 ml), and only 11 patients received intraoperative blood transfusion. Chemotherapy was administered in 48 (14.9%) cases. The majority of tumors were located in the sigmoid colon/rectum (81.7%), with tumor staging predominantly in stage II-III (67.8%).

Table 1 Characteristics of the study cohort
Full size table

Table 2 provided an overview of the predictors in the external validation cohort and derivation cohorts of CRC-VTE score, Sir-Run-Run-Shaw VTE RAM and RVTA score. Age and gender distribution of the patients were similar across the cohorts. However, our study included a significantly higher proportion of patients with cardiac insufficiency and preoperative bowel obstruction compared to the CRC-VTE score derivation cohort. In comparison to the Sir-Run-Run-Shaw VTE RAM derivation cohort, the patients in our study had lower preoperative plasma D-dimer levels and a lower percentage of cancer stage IV (4.3% vs. 9.1%). Lastly, significant differences were observed between our study and the RVTA score derivation cohort in terms of metastasis, plasma D-dimer levels, platelet count, multiple chemotherapy regimens, and blood transfusion history.

Table 2 Demographic and clinical characteristics in the validation cohort and derivation cohorts of patients
Full size table

VTE outcomes

68 (21.1%) patients experienced VTE events within 6 months after colorectal cancer surgery, including 5 cases of PE and 63 cases of DVT (3 had proximal DVT and 60 had distal DVT). The incidence of VTE was predominantly observed in the early postoperative period, with 29 (9.0%) and 45 (13.9%) patients developing VTE events within the first postoperative week and first month, respectively. Among the predictors included in the three VTE RAMs, only the following 5 risk factors were significantly associated with VTE outcomes (Table 3): age ≥ 69 years (P < 0.001), preoperative plasma D-dimer ≥ 0.49 mg/L (P = 0.001), D-dimer ≥ 1.7 mg/L (P < 0.001), age ≥ 70 years (P < 0.001), preoperative bloody stool/tarry stool (P < 0.001). The remaining risk factors showed no significant correlation with the occurrence of VTE, including cancer stage IV (P = 0.298), transfusion (P = 0.890), multiple chemotherapy regimens (P = 0.233), platelet count ≥ 350 × 109/L (P = 1.000), metastasis (P = 0.136), blood transfusion history (P = 0.510), female (P = 0.693), varicose veins of lower extremity (P = 0.110), cardiac insufficiency (P = 0.622), preoperative bowel obstruction (P = 0.413), anesthesia time ≥ 180 min (P = 0.807).

Table 3 Incidence of VTE events under different risk factor subgroups in the study cohort
Full size table

Model discrimination, calibration, and clinical utility

As shown by the ROC curves in Fig. 1, all four VTE RAMs demonstrated possibly helpful discrimination in the external validation cohort. CRC-VTE score had an AUC of 0.728 (95%CI: 0.663–0.793), which was higher than the AUCs of Sir-Run-Run-Shaw VTE RAM (0.691, 95%CI: 0.624–0.758, P = 0.216), RVTA score (0.638, 95%CI: 0.564–0.713, P = 0.065), and Caprini score (0.661, 95%CI: 0.596–0.725, P = 0.067). However, the differences were not statistically significant. The specificity, sensitivity, NPV, and PPV of the four VTE RAMs were presented in Table 4. The sensitivity of CRC-VTE score was slightly higher than that of the other three VTE RAMs, while their specificity was comparable.

Fig. 1
figure 1

Receiver operating characteristic (ROC) curves of Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score and Caprini score in predicting the risk of VTE

Full size image
Table 4 ROC curve comparison of Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score and Caprini score
Full size table

In terms of model calibration, the Hosmer-Lemeshow test revealed that Sir-Run-Run-Shaw VTE RAM (P = 7.631e-10), RVTA score (P < 2.2e-16), and CRC-VTE score (P < 2.2e-16) were not well calibrated. The brier scores of Sir-Run-Run-Shaw VTE RAM, RVTA score, and CRC-VTE score were 0.165, 0.176 and 0.206, respectively, suggesting that Sir-Run-Run-Shaw VTE RAM demonstrated relatively higher accuracy. Additionally, the calibration curves exhibited poor consistency between the predicted probability and observed probability of VTE for the three VTE RAMs (Fig. 2).

Fig. 2
figure 2

Calibration curves of Sir-Run-Run-Shaw VTE RAM (A), RVTA score (B) and CRC-VTE score (C)

Full size image

Besides discrimination and calibration, DCA seems to provide more insight for clinical decision making. As shown in Fig. 3, “None” refers to no intervention for anyone, while “All” refers to intervention for everyone. A model is considered clinically useful at a certain threshold probability if it has a higher net benefit (NB) than “None” and “All“ [16]. Compared to the other three VTE RAMs, the use of CRC-VTE score for predicting the risk of VTE yielded the highest NB (Fig. 3).

Fig. 3
figure 3

Decision curve analysis (DCA) of Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score and Caprini score for predicting the risk of postoperative VTE in colorectal cancer patients

Full size image

Discussion

Our study was the first to conduct external validation of four VTE RAMs in Chinese patients who have undergone colorectal cancer surgery. Four VTE RAMs were Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score and Caprini score. The incidence of VTE within 6 months after colorectal cancer surgery was 21.1%, with DVT events accounting for 92.6%. These models had possibly helpful discrimination and poor calibration. Among them, CRC-VTE score showed relatively high discrimination capacity and Sir-Run-Run-Shaw VTE RAM could be calibrated relatively well. However, based on DCA, CRC-VTE score had the highest NB, which could lead to the best clinical decisions. The four VTE RAMs could be further validated in the future through large-scale, multicenter, prospective cohorts.

One study showed that for patients in the perioperative period of colorectal surgery, the most common VTE risk assessment tool used was Caprini score (78.89%), followed by the hospital-owned score (29.26%), Rogers score (17.04%), Padua score (10.37%), and Khorana score (8.15%) [21]. Caprini score for surgical inpatients and Khorana score for outpatients with cancer were both widely recommended by clinical guidelines to assess the risk of VTE in cancer patients [1, 10, 22]. Caprini score was obtained by summarizing the VTE risk factors based on clinical experience and published research results. It contained 39 scoring components such as patients’ demographic characteristics, surgical status and hereditary factors [18]. For Capini score, patients at high risk of VTE with a score of ≥ 5 require perioperative anticoagulants, and all patients in our study had a 24-hour postoperative Capini score of ≥ 5. But whether all patients are needed to receive anticoagulation to prevent VTE is controversial, because excessive prophylaxis may lead to bleeding events. Previously, a retrospective study found that up-regulation of the cutoff of Capirni score and introduction of D-dimer significantly increased the AUC (0.750 vs. 0.845), which could more accurately guide the use of anticoagulant therapy in postoperative colorectal cancer patients [23]. Khorana score, which was another guideline-often-recommended predictive score for VTE in cancer patients, has been shown to be ineffective in predicting postoperative VTE risk in colorectal cancer patients [24]. A meta-analysis found that only 23.4% of patients who experienced VTE events were categorized as high risk according to Khorana score [25]. Therefore, Caprini score and Khorana score were considered not applicable for VTE risk assessment in postoperative colorectal cancer patients.

All three VTE RAMs externally validated in our study were compared to either Caprini score or Khorana score in the original studies, yielding superior AUC values (Sir-Run-Run-Shaw VTE RAM vs. Caprini score: 0.769 vs. 0.656; CRC-VTE score vs. Caprini score: 0.72 vs. 0.59; RVTA score vs. Khorana score: 0.825 vs. 0.709) [6, 11, 15]. Our study also showed the consistent results that Sir-Run-Run-Shaw VTE RAM (AUC 0.691 vs. 0.661) and CRC-VTE score (AUC 0.728 vs. 0.661) were superior to Caprini score in predicting the risk of VTE after colorectal cancer surgery. However, RVTA score had inferior discrimination capacity to Caprini score (AUC 0.638 vs. 0.661), which may be related to the fact that RVTA score was primarily developed based on outpatients with colorectal cancer, encompassing individuals undergoing surgical interventions as well as those receiving systemic anticancer therapies. Consequently, RVTA score was not appropriate for use in postoperative colorectal cancer patients. Furthermore, our study found that 21.1% of patients developed VTE, while the original studies of Sir-Run-Run-Shaw VTE RAM and CRC-VTE score found that the incidence of postoperative VTE in colorectal cancer patients was between 12.0% and 11.2%. The discrepancy in VTE incidence between our study and the other two original studies can be attributed to the differences in follow-up durations; our study employed a follow-up period of six months after operation, whereas the other studies assessed patients within one week or one month postoperatively. In our study, the incidence rates of postoperative VTE within the first week and month following colorectal cancer surgery were 9.0% and 13.9%, respectively. Consequently, the extended follow-up period in our study likely captured a greater number of VTE cases.

Sir-Run-Run-Shaw VTE RAM, RVTA score, and CRC-VTE score differed in terms of discrimination, calibration, and clinical utility. The higher AUC for CRC-VTE score may be explained by the fact that the original study was prospective and multicenter, making this model more generalizable. The risk factors of CRC-VTE score mainly included patients’ demographic characteristics, comorbidities, and surgical status, which were easily accessible clinically. Therefore, it was convenient for the prediction of VTE risk after colorectal cancer surgery and was also the most clinically valuable VTE RAM. This model took account of preoperative bowel obstruction and preoperative bloody stool/tarry stool, which were clinical symptoms that may occur in patients with colorectal cancer. Risk factors of Sir-Run-Run-Shaw VTE RAM and RVTA score were similar, both considering the predictive value of D-dimer levels, stage of cancer, and transfusion for VTE events, which was similar to the results of several studies [26,27,28]. D-dimer, as a fibrin degradation product, is one of the most common biomarkers studied for the occurrence of VTE in cancer patients. Elevated levels of D-dimer may respond to hypercoagulability in vivo, but the influence of inflammatory settings cannot be excluded. Besides, D-dimer levels increase with age, making it difficult to develop standard reference levels [29]. All of the above may hinder the use of D-dimer in the actual prediction of VTE risk. Finally, unlike Sir-Run-Run-Shaw VTE RAM and CRC-VTE score, study population of RVTA score was outpatients, whose severity of disease and living status differed from that of surgical inpatients. Therefore, the results of using this model to predict VTE risk after colorectal cancer surgery were not satisfactory.

Our study identified advanced age, plasma D-dimer and preoperative bloody stool/tarry stool as independent variables predicting the risk of VTE. However, the remaining risk factors such as metastasis, intraoperative blood transfusion, multiple chemotherapy regimens, platelet count, blood transfusion history, sex, varicose veins of lower extremity, cardiac insufficiency, preoperative bowel obstruction and anesthesia time did not show a correlation with the occurrence of VTE. Possible reasons for this were as follows. Firstly, validation cohort had significantly fewer patients with advanced tumor (stage Ⅳ: 4.3% vs. 9.1%; stage Ⅲ-Ⅳ: 35.0% vs. 66.5%), intraoperative blood transfusion (3.4% vs. 12.2%), platelet count ≥ 350 × 109/L (2.5% vs. 11.0%), blood transfusion history (5.3% vs. 10.2%) than the original derivation cohort, which may have result in difficulties in obtaining statistically significant differences between the VTE and non-VTE groups in terms of these predictors. Secondly, the sample sizes of validation cohort and CRC-VTE score derivation cohort were vastly different, which may lead to inconsistent results. Thirdly, the study population of RVTA score was outpatients with antitumor regimens that differed from preoperative neoadjuvant chemotherapy regimens. The risk of VTE varies among different cytotoxic drugs. A meta-analysis showed that compared to bevacizumab plus irinotecan group, bevacizumab plus oxaliplatin group had a lower risk of VTE (RR = 0.60, 95%CI: 0.46 ~ 0.79, P = 0.0002) [30]. The common chemotherapy drug combinations of RVTA score derivation cohort were XELOX (capecitabine; oxaliplatin), XELIRI (capecitabine; irinotecan), FOLFOX (oxaliplatin; calcium folinate; fluorouracil), and FOLFIRI (irinotecan; calcium folinate; fluorouracil). The chemotherapy regimens of validation cohort were predominantly capecitabine or capecitabine combined with oxaliplatin, contributing to fewer VTE events.

Our study has some limitations. Firstly, patients with partially missing clinical data were not included in the study, making the sample size relatively small. It may affect the performance of the VTE RAMs. Secondly, external validation of the four VTE RAMs was conducted in a single center, and whether they could be widely used in different populations requires further research. Thirdly, not all patients received postoperative thromboprophylaxis. 48.3% received no prophylaxis, 15.2% received mechanical prophylaxis alone, while the remaining patients received combined mechanical and pharmacological prophylaxis. Among these, 22.9% underwent prophylaxis for 1–6 days postoperatively, and 13.6% for ≥ 7 days. This variation in prophylaxis regimens may contribute to differences in postoperative VTE incidence. Finally, due to the retrospective study design, lower extremity ultrasound was not routinely performed for patients undergoing colorectal cancer surgery. The exclusion of patients without preoperative or postoperative lower extremity ultrasound examinations may have introduced selection bias, which warrants further validation through prospective multicenter studies.

Our study conducted an external validation of Sir-Run-Run-Shaw VTE RAM, RVTA score, CRC-VTE score, and Caprini score for predicting postoperative VTE in a real-world cohort of patients with colorectal cancer. CRC-VTE score outperformed other VTE RAMs, as it is more effective in identifying patients with high risk of VTE who require aggressive pharmacologic thromboprophylaxis. In contrast to Caprini score targeting all surgical patients, RVTA score focusing on outpatients with colorectal cancer, and Sir-Run-Run-Shaw VTE RAM developed in a single center, CRC-VTE score was derived from a multicenter prospective cohort specifically for patients undergoing colorectal cancer surgery, a population for which perioperative pharmacologic thromboprophylaxis is strongly recommended in guidelines. In the future, multicenter prospective validation is needed in order to identify a unified postoperative VTE RAM for Chinese colorectal cancer patients.

Data availability

No datasets were generated or analysed during the current study.

References

  1. Farge D, Frere C, Connors JM, et al. 2022 international clinical practice guidelines for the treatment and prophylaxis of venous thromboembolism in patients with cancer, including patients with COVID-19[J]. Lancet Oncol. 2022;23(7):e334–47.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Heit JA, O’Fallon WM, Petterson TM, et al. Relative impact of risk factors for deep vein thrombosis and pulmonary embolism: A Population-Based Study[J]. Arch Intern Med. 2002;162(11):1245–8.

    Article  PubMed  Google Scholar 

  3. Mulder FI, Horváth-Puhó E, van Es N, et al. Venous thromboembolism in cancer patients: a population-based cohort study. Blood. 2021;137(14):1959–69.

    Article  CAS  PubMed  Google Scholar 

  4. McKenna NP, Bews KA, Behm KT, et al. Postoperative venous thromboembolism in colon and rectal cancer: do tumor location and operation Matter?[J]. J Am Coll Surg. 2023;236(4):658–65.

    Article  PubMed  Google Scholar 

  5. Wu D, Gu H, Tang Y, et al. Predictive factors on postoperative venous thromboembolism after minimally invasive colorectal cancer surgery: a retrospective observational study. BMC Surg. 2023;23(1):85.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Wei Q, Wei ZQ, Jing CQ, et al. Incidence, prevention, risk factors, and prediction of venous thromboembolism in Chinese patients after colorectal cancer surgery: a prospective, multicenter cohort study. Int J Surg. 2023;109(10):3003–12.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Lu X, Zeng W, Zhu L, et al. Application of the Caprini risk assessment model for deep vein thrombosis among patients undergoing laparoscopic surgery for colorectal cancer. Medicine. 2021;100(4):e24479.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Streiff MB, Holmstrom B, Angelini D, et al. Cancer-Associated venous thromboembolic disease, version 2.2021, NCCN clinical practice guidelines in Oncology[J]. J Natl Compr Canc Netw. 2021;19(10):1181–201.

    PubMed  Google Scholar 

  9. Falanga A, Ay C, Di Nisio M, et al. Venous thromboembolism in cancer patients: ESMO clinical practice guideline. Ann Oncol. 2023;34(5):452–67.

    Article  CAS  PubMed  Google Scholar 

  10. Key NS, Khorana AA, Kuderer NM, et al. Venous thromboembolism prophylaxis and treatment in patients with cancer: ASCO guideline update. J Clin Oncol. 2023;41(16):3063–71.

    Article  CAS  PubMed  Google Scholar 

  11. Yao J, Lang Y, Su H, et al. Construction of risk assessment model for venous thromboembolism after colorectal cancer surgery: a Chinese single-center study. Clin Appl Thromb Hemost. 2022;28:10760296211073748.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Barber EL, Clarke-Pearson DL. The limited utility of currently available venous thromboembolism risk assessment tools in gynecological oncology patients. Am J Obstet Gynecol. 2016;215:e4451-9.

    Article  Google Scholar 

  13. McAlpine K, Breau RH, Mallick R, et al. Current guidelines do not sufficiently discriminate venous thromboembolism risk in urology[J]. Urol Oncol. 2017;35:e4571–8.

    Article  Google Scholar 

  14. Lee L, Liew NC, Teoh YJ, et al. Efficacy of Caprini risk assessment model in predicting venous thromboembolism risks among Asian surgical patients[J]. Med J Malaysia. 2024;79(3):320–5.

    CAS  PubMed  Google Scholar 

  15. Chen Y, Wang Y, Xie S, et al. A risk of venous thromboembolism algorithm as a predictor of venous thromboembolism in patients with colorectal cancer. Clin Appl Thromb Hemost. 2021;27:10760296211064900.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Van Calster B, Wynants L, Verbeek JFM, et al. Reporting and interpreting decision curve analysis: a guide for investigators. Eur Urol. 2018;74(6):796–804.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Rashid R, Sohrabi S, Kerwan A, et al. The STROCSS 2024 guideline: strengthening the reporting of cohort, cross-sectional, and case-control studies in surgery[J]. Int J Surg. 2024;110(6):3151–65.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Caprini JA. Thrombosis risk assessment as a guide to quality patient care. Dis Mon. 2005;51(2–3):70–8.

    Article  PubMed  Google Scholar 

  19. Alba AC, Agoritsas T, Walsh M, et al. Discrimination and calibration of clinical prediction models: users’ guides to the medical literature. JAMA. 2017;318(14):1377–84.

    Article  PubMed  Google Scholar 

  20. Haider H, Hoehn B, Davis S, et al. Effective ways to build and evaluate individual survival distributions[J]. JMLR. 2020;21(85):1–63.

    Google Scholar 

  21. Yao HW, Han JG, Zhang ZT. Questioniare analysis of the status of venous thromboembolism after colorectal surgery in China[J]. Chin J Practical Surg. 2020;40(5):551–6.

    Google Scholar 

  22. Liu FL, Zhang TP. Chinese guidelines for prevention and management of perioperative thrombosis in general surgery. Chin J Pract Surg. 2016;36(05):469–74.

    Google Scholar 

  23. Zhang W, Sun R, Hu X, et al. Caprini risk assessment model combined with D-dimer to predict the occurrence of deep vein thrombosis and guide intervention after laparoscopic radical resection of colorectal cancer. World J Surg Oncol. 2023;21(1):299.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Barni S, Rosati G, Lonardi S, et al. Khorana score and thromboembolic risk in stage II-III colorectal cancer patients: a post hoc analysis from the adjuvant TOSCA trial[J]. Ther Adv Med Oncol. 2020;12:1758835919899850.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Mulder FI, Candeloro M, Kamphuisen PW, et al. The khorana score for prediction of venous thromboembolism in cancer patients: a systematic review and meta-analysis. Haematologica. 2019;104(6):1277–87.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Qin L, Liang Z, Xie J, et al. Development and validation of machine learning models for postoperative venous thromboembolism prediction in colorectal cancer inpatients: a retrospective study. J Gastrointest Oncol. 2023;14(1):220–32.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Aoki J, Sakamoto K, Takahashi R, et al. Current status of venous thromboembolism development during the perioperative period for colorectal cancer, its prevention with enoxaparin, and monitoring methods[J]. Ther Clin Risk Manag. 2019;15:791–802.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Xu D, He Y, Liao C, et al. Combining KRAS gene status with preoperative D–dimer levels as a predictive marker of venous thromboembolism risk in patients with resectable colorectal cancer: a prospective cohort study. Biomed Rep. 2024;20(6):96.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Kim AS, Khorana AA, McCrae KR. Mechanisms and biomarkers of cancer-associated thrombosis. Transl Res. 2020;225:33–53.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Ren T, Wang S, Shen Z, et al. Efficacy and safety of bevacizumab plus oxaliplatin- or irinotecan-based doublet backbone chemotherapy as the first-line treatment of metastatic colorectal cancer: a systematic review and meta-analysis. Drug Saf. 2021;44(1):29–40.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This study was supported by the Beijing Research Association for Chronic Diseases Control and Health Education 2024 Pharmacy Special Research Project (MBZX0082024001) and Beijing Pharmaceutical Association Clinical Pharmacy Research Program (LCYX-2022-16).

Author information

Author notes

  1. Ying Zhang and Ying Zhao contributed equally to this work.

Authors and Affiliations

  1. Department of Pharmacy, Beijing Friendship Hospital, Capital Medical University, Beijing, China

    Ying Zhang, Ying Zhao, Xiaozhu Zhou, Yi Wu & Xiangli Cui

  2. Department of Pharmacy, Beijing Luhe Hospital, Capital Medical University, Beijing, China

    Ying Zhang & Shicai Chen

  3. Department of General Surgery, Beijing Friendship Hospital, Capital Medical University, Beijing, China

    Jun Cai & Lei Niu

Authors
  1. Ying Zhang
    View author publications

    Search author on:PubMed Google Scholar

  2. Ying Zhao
    View author publications

    Search author on:PubMed Google Scholar

  3. Jun Cai
    View author publications

    Search author on:PubMed Google Scholar

  4. Lei Niu
    View author publications

    Search author on:PubMed Google Scholar

  5. Xiaozhu Zhou
    View author publications

    Search author on:PubMed Google Scholar

  6. Yi Wu
    View author publications

    Search author on:PubMed Google Scholar

  7. Shicai Chen
    View author publications

    Search author on:PubMed Google Scholar

  8. Xiangli Cui
    View author publications

    Search author on:PubMed Google Scholar

Contributions

Y. Zhang and Y. Zhao wrote the main manuscript text; J. Cai, L. Niu, X.-Z. Zhou and Y. Wu contributed to the data acquisition and analysis; S.-C. Chen and X.-L. Cui contributed to the important guidance for this study. All authors reviewed the manuscript.

Corresponding authors

Correspondence to Shicai Chen or Xiangli Cui.

Ethics declarations

Ethics approval and consent to participate

Ethical approval for this retrospective study was obtained from Beijing Friendship Hospital, Capital Medical University.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, Y., Zhao, Y., Cai, J. et al. External validation of four venous thromboembolism risk assessment models after colorectal cancer surgery: a retrospective study. Thrombosis J 23, 92 (2025). https://doi.org/10.1186/s12959-025-00778-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12959-025-00778-w

Keywords

Thrombosis Journal

ISSN: 1477-9560

Contact us