ORIGINAL ARTICLE VASCULAR. Xiaohan Liu a,, Chengyuan Liu a,, Xi Chen a, Wenwen Wu b and Gendi Lu a, * Abstract INTRODUCTION

Transcription

1 Interactive CardioVascular and Thoracic Surgery 23 (2016) doi: /icvts/ivw158 Advance Access publication 13 June 2016 ORIGINAL ARTICLE VASCULAR Cite this article as: Liu X, Liu C, Chen X, Wu W, Lu G. Comparison between Caprini and Padua risk assessment models for hospitalized medical patients at risk for venous thromboembolism: a retrospective study. Interact CardioVasc Thorac Surg 2016;23: Comparison between Caprini and Padua risk assessment models for hospitalized medical patients at risk for venous thromboembolism: a retrospective study a b Xiaohan Liu a,, Chengyuan Liu a,, Xi Chen a, Wenwen Wu b and Gendi Lu a, * Department of Nursing, Changzheng Hospital, Second Military Medical University, Shanghai, China Department of Cardiothoracic Surgery, Changzheng Hospital, Second Military Medical University, Shanghai, China * Corresponding author. Department of Nursing, Changzheng Hospital, Second Military Medical University, No. 415 Fengyang Road, Shanghai , China. Tel: ; fax: ; gendilugdl@163.com (G. Lu). Received 6 December 2015; received in revised form 22 March 2016; accepted 9 April 2016 Abstract OBJECTIVES: This study aimed to evaluate the validity of the risk assessment model (RAM) of Caprini and Padua in identifying venous thromboembolism (VTE) among hospitalized medical patients. METHODS: This retrospective study reviewed a total of 320 VTE and 320 non-vte patients. Baseline demographics and clinical data of these patients were all recorded. The Caprini and Padua RAMs were implemented and the individual scores of each risk factor were summed to generate a cumulative risk score. Meanwhile, the sensitivity, specificity, and positive and negative predictive values of these two models were analysed. Receiver operating characteristic (ROC) curve was plotted to calculate the area under the curve (AUC) and the Youden index. RESULTS: Significant differences were observed in risk factors between VTE and non-vte patients. More VTE patients were classified into the high superhigh risk level by the Caprini RAM than the Padua RAM (70.9 vs 23.4%, P < 0.01). The sensitivity and positive and negative predictive values in the Caprini RAM were higher than those in the Padua RAM (P < 0.05). However, the specificity of the Caprini RAM was lower than that of the Padua RAM (P < 0.01). The AUC and the Youden index were higher in the Caprini RAM than in the Padua RAM (P <0.01), whereas the Youden index in the Padua RAM at critical point 4 was lower than that at critical point 3 (0.010 vs 0.140, P < 0.05). CONCLUSIONS: The Caprini RAM was suggested to be more effective than the Padua RAM for identification of hospitalized medical patients at risk for VTE. Keywords: Venous thromboembolism Risk assessment model Caprini Padua INTRODUCTION Venous thromboembolism (VTE), which includes deep vein thrombosis (DVT) and pulmonary embolism (PE), is a common complication among hospitalized patients associated with considerable long-term morbidity, functional disability and mortality [1]. The incidence rate of VTE ranges from 10 to 40% in medical and general surgery populations [2]. Although the risk of VTE is substantially higher among patients undergoing surgery, the number of hospitalizations for medical illness is much greater than the number of admissions for surgery [3]. Therefore, more attention should be paid to appropriate thrombosis prophylaxis in hospitalized medical patients. In the clinic, less attention on the prevention of VTE may be caused by the complexity of disease itself and various risk factors of medical inpatients [4]. Therefore, risk assessment is considered to be Xiaohan Liu and Chengyuan Liu should be regarded as co-first authors. important in improving VTE prophylaxis [5]. Nowadays, risk assessment models (RAMs) have been developed and used clinically, which could elucidate individual patient and procedure characteristics to provide a more comprehensive and precise analysis of risk by stratifying patients into different categories [6 8]. In a RAM, a list of predisposing risk factors (genetic and clinical characteristics) and exposing risk factors (presenting illness or procedure) are evaluated, and each factor exhibits an assigned relative risk score. These scores are summed and finally used to determine the onset, intensity, type and duration of prophylaxis clinically [9, 10]. To help stratify the risk of VTE in hospitalized patients, several RAMs have been suggested [1]. As reported, a four-element RAM developed by Woller et al. [11] was identified to be effective in evaluating the risk of VTE and improving thromboprophylaxis. The prevalence rate of VTE was found to be 93.2, 52.9 and 12% in the high-risk, moderate-risk and low-risk category of the Yale RAM, respectively [12]. Significantly higher scores were exhibited on DVT patients by an assessment scale raised by McCaffrey et al. [13]. The Author Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.

2 X. Liu et al. / Interactive CardioVascular and Thoracic Surgery 539 A computer-based electronic tool was successful in identification of the thrombosis risk in medical and surgical patients [14]. Recently, the Caprini and Padua RAMs have been used in evaluating the risk of VTE clinically, which are adopted in Antithrombotic therapy and prevention of thrombosis, 9th ed: American college of chest physicians evidence-based clinical practice guidelines. The Caprini RAM was first reported in 1980 and then it was regularly updated to a relatively mature model in 2010 [15]. The Caprini RAM includes a total of 39 risk factors and validly underscores the importance of assessing individual patients at risk for VTE [15]. The Padua RAM was developed by the integration of the Kucher model in 2010, which includes 11 risk factors and has been revealed to be able to improve stratification of the thromboembolic risk in hospitalized medical patients [8]. In this study, a retrospective analysis including 320 VTE and 320 non-vte patients was carried out, and the validities of the Caprini and Padua RAMs were evaluated. Our findings may identify the more effective RAM in hospitalized medical patients between them, which could not only improve the efficiency of VTE prophylaxis, but also reduce the incidence of VTE and the waste of medical resources. MATERIALS AND METHODS Patients This study reviewed medical patients who developed VTE (DVT and PE) during their hospital stay in Shanghai Changzheng Hospital from September 2011 to September Every hospitalized patient had a unique patient ID number. They were encouraged to take active and passive exercises during their hospital stay, and no thromboprophylaxis was given routinely to patients. DVT was diagnosed as intraluminal blocking or filling defects in deep veins by CT or catheter venography or evidence of thrombus in the deep vein confirmed by colour Doppler ultrasonography. PE was diagnosed as intraluminal blocking or filling defects in the pulmonary artery by pulmonary angiography; computed tomographic pulmonary arteriography or magnetic resonance imaging; or multiple segmental perfusion defects in the lung by radionuclide ventilation/perfusion lung scan. The exclusion criteria included: an age of less than 18 years, length of hospital stay <2 days and superficial vein thrombosis. Finally, a total of 320 patients with VTE were included. A total of 320 non-vte patients were randomly selected by a random digit table from all hospitalized patients admitted to the same departments homochronously as VTE patients according to their ID numbers. The inclusion criteria were as follows: a discharge diagnosis of no VTE and no evidence of DVT by colour Doppler ultrasonography and D-dimer value. Patients of age <18 years, with superficial vein thrombosis or with hospitalization time <2 days were excluded from this study. This study was approved by the Ethics Committee of Shanghai Changzheng Hospital. Clinical data collection Baseline demographics and clinical data of these patients in the VTE and the non-vte group were all recorded, including age, gender, body mass index (BMI), haemoglobin, haematocrit, leucocyte, platelet and hospitalization time. In addition, partial risk factors of VTE between VTE patients and non-vte patients were compared, including the incidences of malignancy, history of thrombosis, pneumonia, heart failure, varicose veins, central venous access, stroke, hypertension, diabetes, coronary heart disease, myocardial infarction, swollen legs (current), inflammatory bowel diseases, patient s confinement to bed (>72 h), smoking history and history of prior major surgery (<1 month). Risk assessment models Both the improved Caprini RAM and Padua RAM were used to evaluate VTE in this study. This rating scale of the Caprini RAM Table 1: from [9]) Risk factors The Caprini risk assessment model (reproduced Age years 1 Minor surgery planned 1 History of prior major surgery (<1 month) 1 Varicose veins 1 History of inflammatory bowel diseases 1 Swollen legs (current) 1 Obesity (BMI 25) 1 Acute myocardial infarction 1 Congestive heart failure (<1 month) 1 Sepsis (<1 month) Serious lung disease incl. pneumonia (<1 month) 1 Abnormal pulmonary function (COPD) 1 Medical patient currently at bed rest 1 Other risk factors 1 Age years 2 Arthroscopic surgery 2 Malignancy (present or previous) 2 Major surgery (>45 min) 2 Laparoscopic surgery (>45 min) 2 Patient confined to bed (>72 h) 2 Immobilizing plaster cast (<1 month) 2 Central venous access 2 Age over 75 years 3 History of DVT/PE 3 Family history of thrombosis a 3 Positive Factor V Leiden 3 Positive Prothrombin 20210A 3 Elevated serum homocysteine 3 Positive lupus anticoagulant 3 Elevated anticardiolipin antibodies 3 Heparin-induced thrombocytopenia (HIT) 3 Other congenital or acquired thrombophilia 3 If yes: Type Elective major lower extremity arthroplasty 5 Hip, pelvis or leg fracture (<1 month) 5 Stroke (<1 month) 5 Multiple trauma (<1 month) 5 Acute spinal cord injury (paralysis) (<1 month) 5 For women only Oral contraceptives or hormone replacement therapy 1 Pregnancy or postpartum (<1 month) 1 History of unexplained stillborn infant, recurrent 1 spontaneous abortions ( 3), premature birth with toxaemia or growth-restricted infant Total risk factor score Score DVT: deep vein thrombosis; PE: pulmonary embolism; BMI: body mass index; COPD: chronic obstructive pulmonary disease. a Most frequently missed risk factor. ORIGINAL ARTICLE

3 540 X. Liu et al. / Interactive CardioVascular and Thoracic Surgery includes a total of 39 risk factors (Table 1)[9]. The individual scores of each risk factor were summed to generate a cumulative risk score that defined the patient s VTE risk level: low risk 0 1, middle risk 2, high risk 3 4 and superhigh risk 5 [15]. In addition, another RAM raised by Padua was also performed. In the Padua RAM, a total of 11 risk factors were listed (Table 2) [8]. The risk of VTE was defined as low risk (total scores <4) and high risk (total scores 4) [8]. Accuracy analysis of risk assessment models considered to be VTE-positive, and patients with Caprini score <3 and Padua score <4 were considered to be VTE-negative. In addition, confirmed VTE patients were considered as true positive patients, whereas the non-vte patients were considered as true positive negative patients. Then receiver operating characteristic (ROC) curves were plotted referring to the sensitivity and specificity of the two models, and the areas under the curve (AUCs) and 95% confidence intervals (CIs) were calculated. In addition, the Youden index, which could also reflect the accuracy of the RAM, was calculated as sensitivity + specificity 1. To evaluate the accuracies and validities of the Caprini and Padua RAMs, the sensitivity, specificity, and positive and negative predictive values were analysed. During these assessment processes, patients with Caprini score 3 and Padua score 4 were Table 2: from [8]) Baseline features The Padua risk assessment model (reproduced Active cancer a 3 Previous VTE (with the exclusion of superficial vein thrombosis) 3 Reduced mobility b 3 Already known thrombophilic condition c 3 Recent ( 1 month) trauma and/or surgery 2 Elderly age ( 70 years) 1 Heart and/or respiratory failure 1 Acute myocardial infarction or ischaemic stroke 1 Acute infection and/or rheumatological disorder 1 Obesity (BMI 30) 1 Ongoing hormonal treatment 1 Score BMI: body mass index; VTE: venous thromboembolism. a Patients with local or distant metastases and/or in whom chemotherapy or radiotherapy had been performed in the previous 6 months. b Bed rest with bathroom privileges (either due to patient s limitations or on physician s order) for at least 3 days. c Carriage of defects of antithrombin, protein C or S, factor V Leiden, G20210A prothrombin mutation, antiphospholipid syndrome. Table 3: Statistical analysis Statistical analysis was performed by SPSS version 19.0 (SPSS Inc., Chicago, IL). Quantitative data were expressed as mean ± standard deviation and analysed by the independent t-test. Qualitative data were expressed as percentage (%) and analysed by the χ 2 test. A P-value of <0.05 was considered to be statistically significant. RESULTS Baseline demographics and clinical data of venous thromboembolism patients In this study, the VTE patients were identified originating from various hospital departments, including Respiratory Medicine (65, 20.3%), Oncology (59, 18.4%), Obstetrics and Gynaecology (41, 12.8%), Gastroenterology (26, 8.1%), Internal Neurology (23, 7.2%), Cardiology (21, 6.6%), Haematology (19, 5.9%), Nephrology (18, 5.6%), Contagion (13, 4.1%), Endocrinology (10, 3.1%), Immunology and Rheumatology (9, 2.8%), and Intensive Care Unit (16, 5.0%). Baseline demographics and clinical data of patients in the VTE and non-vte groups are summarized in Table 3. No significant differences were witnessed in gender, haemoglobin and haematocrit (all P > 0.05). However, the age, BMI, and leucocyte and platelet counts were significantly higher in VTE patients than in the non-vte patients (all P < 0.05), and VTE patients stayed longer in hospital than non-vte cases (P =0.001,Table3). Table 4 presents the partial VTE risk factors of patients in the VTE and non-vte groups. As a result, the prevalence of malignancy, history of thrombosis, pneumonia, Baseline demographics and clinical data of patients in venous thromboembolism (VTE) and non-vte groups VTE Non-VTE χ 2 /t P-value Gender (male/female), n (%) a 172/148 (53.8/46.3) 161/159 (50.3/49.7) Mean age (years) a ± ± BMI >25 kg/m 2, n (%) b 158 (54.7) 92 (32.2) Haemoglobin (g/l) c ± ± Leucocyte ( 10 9 /l) c 7.97 ± ± Haematocrit (%) c ± ± Platelet ( 10 9 /l) c ± ± Hospitalization time (days) a 9.63 ± ± BMI: body mass index. a 320 patients in the VTE group and 320 patients in the non-vte group. b 289 patients in the VTE group and 286 patients in the non-vte group. c 307 patients in the VTE group and 313 patients in the non-vte group. Measurement data were expressed as mean ± standard deviation (SD) and analysed by the independent t-test. Numeration data were shown as n (%) and analysed by the χ 2 test. P < 0.05 was considered statistically significant.

4 X. Liu et al. / Interactive CardioVascular and Thoracic Surgery 541 heart failure, varicose veins, central venous access, stroke, patient s confinement to bed (>72 h) and history of prior major surgery (<1 month) were significantly higher in VTE patients than in the non-vte patients (all P < 0.05). However, no significant differences were found in smoking history, hypertension, diabetes, coronary heart disease, myocardial infarction, swollen leg and inflammatory bowel diseases (Table 4). Comparisons between risk assessment models of Caprini and Padua The distributions of patients by risk level and cumulative risk score in the Caprini and Padua RAMs are presented in Table 5. The results Table 4: The partial VTE risk factors of patients in venous thromboembolism (VTE) and non-vte group Risk factors VTE Non-VTE χ 2 P-value History of prior major 68 (21.3) 16 (5.0) surgery (<1 month) Malignancy 78 (24.4) 37 (11.6) History of thrombosis 18 (5.6) 0 (0) Smoking history 58 (18.1) 48 (15.0) Hypertension 102 (31.9) 98 (30.6) Diabetes 46 (14.4) 37 (11.6) Coronary heart 24 (7.5) 15 (4.7) disease Pneumonia 22 (6.9) 8 (2.5) Heart failure 10 (3.1) 2 (0.6) Myocardial infarction 6 (1.9) 2 (0.6) Varicose veins 17 (5.3) 3 (0.9) Swollen legs (current) 24 (7.5) 13 (4.1) Central venous access 26 (8.1) 2 (0.6) Patient confined to 33 (10.3) 11 (3.4) bed (>72 h) Stroke 16 (5.0) 5 (1.6) Inflammatory bowel diseases 8 (2.5) 5 (1.6) Numeration data were shown as n (%) and analysed by the χ 2 test. P < 0.05 was considered statistically significant. Table 5: Number of patients by risk level and cumulative risk scores in the Caprini and Padua risk assessment models (RAMs) showed that the risk scores of VTE patients were significantly higher than those of non-vte patients based on both the Caprini and Padua RAMs (both P < 0.01). In addition, according to the Caprini RAM, the majority of the VTE patients (70.9%) were classified to the high superhigh risk level, whereas most of non-vte patients (73.4%) were classified as low moderate risk (P = 0.001). As for the Padua RAM, only 23.4% of VTE patients and 14.4% of non-vte patients were considered to be at high risk for VTE (P = 0.003). The sensitivity and positive and negative predictive values of the Caprini RAM were also higher than that of the Padua RAM (all P <0.05, Table 6). However, the specificity of the Caprini RAM was found to be lower than that of the Padua RAM (P =0.001). Receiver operating characteristic curve of Caprini and Padua risk assessment models As shown in Fig. 1, the ROC curve analysis determined that the AUC and CI (95%) of the Caprini RAM were significantly Table 6: Sensitivity and specificity of the Caprini and Padua risk assessment models (RAMs) Sensitivity Specificity Positive predictive value Negative predictive value Caprini RAM 70.9% 73.4% 72.8% 71.7% Padua RAM 23.4% 85.6% 62.0% 52.8% χ P-value P < 0.05 was considered statistically significant. ORIGINAL ARTICLE VTE Non-VTE P-value Caprini RAM Cumulative risk scores 3.92 ± ± High superhigh risk, n (%) 227 (70.9) 85 (26.6) Padua RAM Cumulative risk scores 1.64 ± ± High risk, n (%) 75 (23.4) 46 (14.4) P < 0.05 was considered statistically significant. VTE: venous thromboembolism. Figure 1: Receiver operating characteristic (ROC) curve of the Caprini and Padua risk assessment models.

5 542 X. Liu et al. / Interactive CardioVascular and Thoracic Surgery higher than those of the Padua RAM (AUC: ± vs ± 0.022; CI (95%): vs , P = 0.001), which indicated a higher forecast accuracy of the Caprini RAM. In addition, the Youden index was in the Caprini RAM when score 3 was considered to be the critical point of VTE-positive and -negative. However, the Youden index in the Padua RAM was only when the critical point was set at 4, and it was significantly lower than 0.14 at a critical point of 3. DISCUSSION VTE (DVT and PE) is one of the most common diseases in medical and surgical patients and is associated with high morbidity and mortality. Although a high risk of VTE complications was exhibited in a substantial proportion of hospitalized patients [3, 16], the administration of thromboprophylaxis was still not fully performed [4, 17, 18]. Therefore, appropriate selection of effective and accurate prophylaxis for VTE has become an important unresolved issue. In this study, the Caprini and Padua RAMs recommended by American College of Chest Physicians were implemented both on VTE and on non-vte patients. As a result, the Caprini RAM could distinguish more high-risk patients than the Padua RAM. The sensitivity, positive and negative predictive value, and AUC of the Caprini RAM were all higher than those of the Padua RAM. All these phenomena indicated higher prediction accuracy in the Caprini RAM than the Padua RAM, which was consistent with previous studies [19, 20]. As was reported previously, a total of 103 (62%) patients with DVT were considered to be at high risk for DVT by the Caprini RAM, while the number of high DVT risk patients analysed by the Padua RAM were 93 (54.7%) [19]. According to the comparison of partial VTE risk factors between the VTE and the non-vte group, VTE patients were exposed to many more risk factors than non-vte patients, such as patients confined to bed (>72 h), central venous access or varicose veins, all of which were not included in the Padua RAM yet. Hence, it seemed that the high accuracy of the Caprini RAM can be explained by more assessment factors in this model compared with the Padua model, and the low sensitivity of the Padua RAM for VTE may also be attributed to the fact that many patients with multiple high risks were ignored. In addition, although the Caprini RAM was usually recommended in the evaluation of surgical patients [9, 15, 21], it was also proved to be suitable for medical patients in this study. This may be because more effective evaluation was validated in surgical patients; however, this model was originally designed for the study of medical and surgical patients. On the other hand, the specificity of the Caprini RAM was shown to be lower than the Padua RAM. This result may also be related to fewer assessment factors in the Padua RAM, and illustrate a high accuracy in the evaluation of low-risk patients by the Padua RAM [22]. In this study, the Youden index at the critical point 4 was lower than 3 in the Padua RAM. This phenomenon was consistent with previous studies that PE could also be found in many patients with the Padua RAM scores of 3 [23]. However, it also has been reported that the critical point 4 in the Padua RAM could be effective for stratification of low and high risk of VTE in either prospective or retrospective studies [8, 24, 25]. These differences may be explained by the insufficient sample size in this study, and further research studies on the critical point (3 or 4) are still needed. Several limitations in our study must be addressed. First, only partial VTE risks were compared between VTE patients and non-vte patients. Second, subgroup analysis between DVT and PE patients was not conducted. Third, non-vte patients had no CT scan or angiography but only ultrasonography and D-dimer examination. In addition, a retrospective study on this issue is of low priority at this point. In conclusion, the Caprini RAM based on individual risk factors had more validity than the Padua RAM in evaluation of VTE among hospitalized medical patients. The implementation of Caprini RAM may improve the efficiency of prophylaxis, and decrease the incidence of VTE in the clinical setting. However, further verifications on the Caprini RAM by large samples are still needed. Funding Conflict of interest: none declared. REFERENCES [1] Huang W, Anderson FA, Spencer FA, Gallus A, Goldberg RJ. Riskassessment models for predicting venous thromboembolism among hospitalized non-surgical patients: a systematic review. J Thromb Thrombolysis 2013;35: [2] Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR et al. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest 2008;133:381S 453S. [3] Heit JA, O Fallon WM, Petterson TM, Lohse CM, Silverstein MD, Mohr DN et al. Relative impact of risk factors for deep vein thrombosis and pulmonary embolism: a population-based study. Arch Intern Med 2002;162: [4] Cohen AT, Tapson VF, Bergmann JF, Goldhaber SZ, Kakkar AK, Deslandes B et al. Venous thromboembolism risk and prophylaxis in the acute hospital care setting (endorse study): a multinational cross-sectional study. Lancet 2008;371: [5] Caprini JA, Hyers TM. Compliance with antithrombotic guidelines. Manag Care 2006;15:49 50., 53 60, 66. [6] Pannucci CJ, Shanks A, Moote MJ, Bahl V, Cederna PS, Naughton NN et al. Identifying patients at high risk for venous thromboembolism requiring treatment after outpatient surgery. Ann Surg 2012;255: [7] Pannucci CJ, Barta RJ, Portschy PR, Dreszer G, Hoxworth RE, Kalliainen LK et al. Assessment of postoperative venous thromboembolism risk in plastic surgery patients using the 2005 and 2010 Caprini Risk score. Plast Reconstr Surg 2012;130: [8] Barbar S, Noventa F, Rossetto V, Ferrari A, Brandolin B, Perlati M et al. A risk assessment model for the identification of hospitalized medical patients at risk for venous thromboembolism: the Padua Prediction Score. J Thromb Haemost 2010;8: [9] Caprini JA. Thrombosis risk assessment as a guide to quality patient care. Dis Mon 2005;51:70 8. [10] Eldor A. Applying risk assessment models in non-surgical patients: effective risk stratification. Blood Coagul Fibrinolysis 1999;10:S91 7. [11] Woller SC, Stevens SM, Jones JP, Lloyd JF, Evans RS, Aston VT et al. Derivation and validation of a simple model to identify venous thromboembolism risk in medical patients. Am J Med 2011;124: e2. [12] Yale SH, Medlin SC, Liang H, Peters T, Glurich I, Mazza JJ. Risk assessment model for venothromboembolism in post-hospitalized patients. Int Angiol 2005;24: [13] McCaffrey R, Bishop M, Adonis-Rizzo M, Williamson E, McPherson M, Cruikshank A et al. Development and testing of a DVT risk assessment tool: providing evidence of validity and reliability. Worldviews Evid Based Nurs 2007;4: [14] Samama MM, Dahl OE, Mismetti P, Quinlan DJ, Rosencher N, Cornelis M et al. An electronic tool for venous thromboembolism prevention in medical and surgical patients. Haematologica 2006;91: [15] Bahl V, Hu HM, Henke PK, Wakefield TW, Campbell DA Jr, Caprini JA. A validation study of a retrospective venous thromboembolism risk scoring method. Ann Surg 2010;251: [16] Anderson FA Jr, Spencer FA. Risk factors for venous thromboembolism. Circulation 2003;107:I9 16.

6 X. Liu et al. / Interactive CardioVascular and Thoracic Surgery 543 [17] Chopard P, Dorffler-Melly J, Hess U, Wuillemin WA, Hayoz D, Gallino A et al. Venous thromboembolism prophylaxis in acutely ill medical patients: definite need for improvement. J Intern Med 2005;257: [18] Kucher N, Spirk D, Baumgartner I, Mazzolai L, Korte W, Nobel D et al. Lack of prophylaxis before the onset of acute venous thromboembolism among hospitalized cancer patients: the SWIss Venous Thromboembolism Registry (SWIVTER). Ann Oncol 2010;21: [19] Pop TR, Vesa SC, Trifa AP, Crisan S, Buzoianu AD. PAI-1 4G/5G and MTHFR C677T polymorphisms increased the accuracy of two prediction scores for the risk of acute lower extremity deep vein thrombosis. Rom J Morphol Embryol 2014;55: [20] Zhou H, Wang L, Wu X, Tang Y, Yang J, Wang B et al. Validation of a venous thromboembolism risk assessment model in hospitalized Chinese patients: a case-control study. J Atheroscler Thromb 2014;21: [21] Pannucci CJ, Bailey SH, Dreszer G, Fisher Wachtman C, Zumsteg JW, Jaber RM et al. Validation of the Caprini risk assessment model in plastic and reconstructive surgery patients. J Am Coll Surg 2011;212: [22] Nendaz M, Spirk D, Kucher N, Aujesky D, Hayoz D, Beer JH et al. Multicentre validation of the Geneva Risk Score for hospitalised medical patients at risk of venous thromboembolism. Explicit ASsessment of Thromboembolic Risk and Prophylaxis for Medical PATients in SwitzErland (ESTIMATE). Thromb Haemost 2014;111: [23] Maynard G, Jenkins IH, Merli GJ. Venous thromboembolism prevention guidelines for medical inpatients: mind the (implementation) gap. J Hosp Med 2013;8: [24] Bogari H, Patanwala AE, Cosgrove R, Katz M. Risk-assessment and pharmacological prophylaxis of venous thromboembolism in hospitalized patients with chronic liver disease. Thromb Res 2014;134: [25] Vardi M, Ghanem-Zoubi NO, Zidan R, Yurin V, Bitterman H. Venous thromboembolism and the utility of the Padua Prediction Score in patients with sepsis admitted to internal medicine departments. J Thromb Haemost 2013;11: ORIGINAL ARTICLE