Sage Journals: Discover world-class research

Abstract

Background

Currently, no universally accepted standardized VTE risk assessment model (RAM) is specifically designed for critically ill patients. Although the ICU-venous thromboembolism (ICU-VTE) RAM was initially developed in 2020, it lacks prospective external validation.

Objectives

To evaluate the predictive performance of the ICU-VTE RAM in terms of VTE occurrence in mixed medical-surgical ICU patients.

Methods

We prospectively enrolled adult patients in the ICU. The ICU-VTE score and Caprini or Padua score were calculated at admission, and the incidence of in-hospital VTE was investigated. The performance of the ICU-VTE RAM was evaluated and compared with that of Caprini or Padua RAM using the receiver operating curve.

Results

We included 269 patients (median age: 70 years; 62.5% male). Eighty-three (30.9%) patients experienced inpatient VTE. The AUC of the ICU-VTE RAM was 0.743 (95% CI, 0.682-0.804, P < 0.001) for mixed medical-surgical ICU patients. Comparatively, the performance of the ICU-VTE RAM was superior to that of the Pauda RAM (AUC: 0.727 vs 0.583, P < 0.001) in critically ill medical patients and the Caprini RAM (AUC: 0.774 vs 0.617, P = 0.128) in critically ill surgical patients, although the latter comparison was not statistically significant.

Conclusions

The ICU-VTE RAM may be a practical and valuable tool for identifying and stratifying VTE risk in mixed medical-surgical critically ill patients, aiding in managing and preventing VTE complications.

Keywords

venous thromboembolism inpatient risk assessment model critically ill

Introduction

Venous thromboembolism (VTE), including pulmonary embolism (PE) and deep vein thrombosis (DVT), ranks as the third leading vascular disease following acute myocardial infarction and stroke.¹ It represents a significant global disease burden, with approximately 10 million cases reported annually.² Critically ill patients face a heightened risk of VTE due to a combination of general risk factors and specific risk factors related to the intensive care unit (ICU), such as sedation, immobilization, vasopressor use, or central venous catheters. The incidence of VTE in ICU patients ranges from 25.0% to 37.2%, nearly doubling that of general medical patients.^3–5 In addition, VTE is associated with an increased risk of mortality.^6–8 Therefore, accurate identification of individuals at high risk of VTE in the ICU, coupled with the implementation of multiple preventive measures, may help prevent or delay the progression of venous thrombosis and improve prognosis. Consequently, early assessment of VTE risk in ICU patients appears to be particularly important.

The Caprini and Padua risk assessment models (RAMs) are the most widely used RAMs in medical^9,10 and surgical patients,^11,12 respectively. The ninth edition of the American College of Chest Physicians Antithrombotic Therapy and Prevention of Thrombosis guidelines (AT9) acknowledges the Caprini RAM for nonorthopedic surgical populations and the Padua RAM for medical patients.^13,14 However, to date, there is no universally accepted standardized VTE RAM is designed specifically for critically ill patients. Viarasilpa et al¹⁵ developed the ICU-VTE RAM, a predictive model based on six independent risk factors: central venous catheterization (5 points), immobilization ≥ 4 days (4 points), prior VTE (4 points), mechanical ventilation (2 points), lowest hemoglobin during hospitalization ≥ 90 g/L (2 points), and baseline platelet count > 250,000/μL (1 point). This model was derived from the data of 37,050 critically ill patients during hospitalization at the Henry Ford Health System.¹⁵ It was utilized to predict the risk of venous thrombosis in critically ill patients and categorize them into three grades: low risk (0-8 points), intermediate risk (9-14 points), and high risk (15-18 points).

However, the ICU-VTE RAM has not undergone prospective evaluation, and its clinical significance requires further verification. Notably, the original study focused solely on symptomatic VTE and did not consider asymptomatic VTE. It would be reasonable to assess the accuracy of the ICU-VTE RAM in predicting both symptomatic and asymptomatic VTE. Therefore, we aimed to perform an external validation to assess the predictive performance of the ICU-VTE RAM for VTE occurrence and its ability to stratify VTE risk in mixed medical-surgical ICU patients.

Methods

Study Design and Participants

This single-center prospective study was conducted in the ICU of the Central Hospital of Wuhan. The trial is registered on ClinicalTrials.gov ((NCT05487066). Between May 2022 and October 2023, a total of 332 patients fulfilled the following inclusion criteria: aged 18 years or older, expected to stay in the ICU for more than 48 h, and volunteered and signed an informed consent form. Patients who met any of the following conditions were excluded: (1) had a diagnosis of VTE upon admission or within 24 h of ICU admission; (2) used therapeutic anticoagulation before or within 24 h of ICU admission; or (3) had a diagnosis of VTE without imaging confirmation. Ultimately, 269 patients were included in this study ( Figure 1 ), and they were categorized into the “with VTE” group and the “without VTE” groups. The study was approved by the Ethics Committees of the Central Hospital of Wuhan (Approval Number: 2022-117, date of approval: April 29, 2022).

Figure 1.

Flowchart of this study.

Procedures

The demographic characteristics, causes of ICU admission, comorbidities, and baseline laboratory results of patients who were included in the routine workup at ICU admission were collected. The Acute Physiology and Chronic Health Evaluation II (APACHE II) score, Padua score for medical patients, or Caprini score for surgical patients and the ICU-VTE score for all patients were determined within 24 h of admission. It should be noted that we collected hemoglobin values and evaluated for expected the immobilization ≥ 4 days at ICU admission to calculate the ICU-VTE score in our cohort. Enrolled patients were stratified into different risk levels based on their Caprini RAM score or Padua RAM score and received a nursing protocol for VTE prophylaxis. Patients classified as very low-risk according to the Caprini RAM and low-risk according to the Padua RAM were encouraged to ambulate early. In contrast, low-risk patients, according to the Caprini RAM, received mechanical prophylaxis, including graduated compression stockings (GCS) or intermittent pneumatic compression (IPC). For patients with moderate to high risk according to the Caprini RAM and high risk according to the Padua RAM, the physician determined the appropriate prophylaxis, whether mechanical, pharmacologic, or a combination of both, based on an assessment of bleeding risk using tools published by the National Programme for Prevention and Management of Pulmonary Embolism and Deep Venous Thrombosis.¹⁶ Pharmacological thromboprophylaxis was considered adequate if initiated within 48 h of admission and covered >80% of the ICU stay. Routine duplex ultrasonography was performed for DVT surveillance, and CT pulmonary angiogram (CTPA) was performed to confirm the diagnosis of pulmonary embolism (PE). All VTE events were verified by a team of expert radiologists. We further followed the enrolled patients until they were discharged from the hospital. Additional clinical information on the treatments related to VTE risk was collected. During the follow-up period, the occurrence of VTE (including timing and type) and clinical outcomes were documented. The primary outcome was the occurrence of VTE during hospitalization. The secondary outcomes included all-cause mortality during hospitalization, ICU length of stay, and hospital length of stay. All the data were cross-checked and centrally validated.

Statistics Analysis

Statistical analysis was performed using the SPSS software package (version 26.0). The normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed continuous variables are reported as the means ± standard deviations (SD), and group comparisons were performed using the Student's t-test. Skewed distribution continuous variables are presented as medians and interquartile ranges (IQRs), and group comparisons were conducted using the Mann-Whitney U test. Categorical variables are presented as numbers and percentages and were compared using the Chi-squared test or Fisher's exact test. No variables were missing and no imputation methods were used in the analysis. The rates of in-hospital VTE, stratified by the ICU-VTE RAM, were analyzed. Forward stepwise logistic regression analysis was performed to identify risk factors for in-hospital VTE, and the results were expressed as odds ratios (ORs) with corresponding 95% confidence intervals (95% CIs). Considering that the limited number of VTE patients (n = 83) in our study might have resulted in overfitting in the model, only variables that were not included as factors in the ICU-VTE RAM and had a P value < 0.05 in our univariate regression model were entered into the multivariate logistic regression analysis. Receiver operating characteristic (ROC) curves were constructed to assess sensitivity and specificity, and the area under the ROC curve (AUC) was calculated using the GraphPad Prism statistical software (version 8). The AUCs were compared using the Delong–Delong test. P values < 0.05 (two-tailed) were considered to indicate statistical significance.

Results

Baseline Characteristics of the Study Subjects

The patients’ demographic and clinical characteristics at baseline are presented in Table 1 . A total of 269 individuals were ultimately included in the study, comprising 168 males (62.5%). The median age of the participants was 70 years (IQR 61-79). The primary reason for ICU admission was medical diseases, accounting for 228 patients (84.8%), while surgical diseases accounted for the remaining 41 patients (15.2%). The median APACHE II score was 17 (IQR 13-22). Eighty-three (30.9%) patients experienced inpatient VTE, 19 (22.9%) of whom experienced symptomatic VTE and 64 (77.1%) of whom experienced asymptomatic VTE. Eighty-two patients had DVT, and 1 patient had PE. Among the patients with DVT, 5 (1.9%) had upper extremity DVT, 72 (26.8%) had isolated lower extremity DVT, and 5 (1.9%) had upper extremity and lower extremity DVT. Among the 72 patients with isolated lower extremity DVT, 7 (2.6%) had proximal DVT (PDVT), 58 (21.6%) had isolated distal deep vein thrombosis (IDDVT), which was defined as a thrombus involving any lower extremity vein distal to the popliteal vein at the knee, and 7 (2.6%) had PDVT coexisting with DDVT. The median time from ICU admission to VTE diagnosis was 4 days (IQR 3-7). The in-hospital mortality rate for all patients was 8.1%. The median ICU length of stay was 8 days (IQR 5-12), while the median hospital length of stay was 10 days (IQR 6-19).

Table 1.

Baseline Characteristics of the Study.

Characteristics	N (%)/Median (IQR)
Age, y	70(61-79)
Male, %	168(62.5)
Body Mass Index, kg/m²	23.01(20.43-24.97)
Causes of ICU admission, %
Medical	228(84.8)
Surgical	41(15.2)
Severity of illness scoring
APACHE II	17(13-22)
Invasive mechanical ventilation, %	50(18.6)
Vasopressor use, %	105(39.0)
The primary outcome, %
In-hospital VTE	83(30.9)
Symptomatic VTE	19(22.9)
Asymptomatic VTE	64(77.1)
Characteristics	N (%)/Median (IQR)
Locations and number of VTE
PE	1(0.4)
DVT	82(30.5)
Upper-extremity DVT	5(1.9)
Lower-extremity DVT	72(26.8)
Proximal DVT	7(2.6)
Isolated distal DVT	58(21.6)
Proximal DVT coexisting distal DVT	7(2.6)
Upper-extremity coexisting Lower-extremity DVT	5(1.9)
Mean (SD) time to VTE after admission to the ICU	4(3-7)
The secondary outcome
In-hospital mortality, %	22(8.1)
Length of ICU stay, d	8(5-12)
Length of hospital stay, d	10(6-19)

Abbreviations: DVT, deep vein thrombosis; PE, pulmonary embolus; VTE, venous thromboembolism.

Predictors of in-hospital VTE

According to the univariate analysis ( Table 2 ), no significant differences were observed in the age, sex, or underlying comorbidities between the VTE and non-VTE groups. However, patients with VTE had a greater proportion of patients with surgical disease (21.7% vs 12.4%, P = 0.049), higher APACHE II scores (20 vs 16, P = 0.006), and greater D-dimer levels (4.62 vs 1.98, P < 0.001) than did those without VTE. Regarding treatment, there was a significantly greater proportion of patients with VTE who received invasive mechanical ventilation (31.3% vs 12.9%, P < 0.001), had a longer duration of mechanical ventilation (4 days vs 3 days, P = 0.013), and had a greater proportion of patients who received central venous catheterization (95.2% vs 84.4%, P < 0.001). Furthermore, compared with the non-VTE group, the VTE group exhibited increased lengths of ICU stay (10 days vs 7 days, P = 0.001), hospital stay (13 days vs 9 days, P = 0.002), and in-hospital mortality (13.2% vs 5.9%, P = 0.042). There were no significant differences in the use of vasopressors (usage, duration, or initial dose), transfusion of blood components, use of CRRT, or anticoagulant prophylaxis between the VTE and non-VTE groups (all P > 0.05). The ICU-VTE RAM was used to predict the risk of in-hospital VTE, and the VTE group had notably higher scores (median 11, IQR, 9-12 vs median 9, IQR, 6-11, P < 0.001), with the difference remaining significant even after stratification into low-risk, intermediate-risk, and high-risk groups (P < 0.001). We further performed logistic regression analyses and the key details are summarized in Supplementary Table 1. A higher ICU-VTE score was found to be a significant predictor of in-hospital VTE (OR, 1.459; 95% CI, 1.264-1.685; P < 0.001).

Table 2.

Comparison of Clinical Characteristics Between Patients with VTE and Without VTE.

Variable	N (%)/Median (IQR)		P-value
Variable	With VTE (n = 83)	Without VTE (n = 186)	P-value
Age, y	70(59-81)	70(62-79)	0.807
Male	47(56.6)	121 (65.1)	0.187
Body mass index, kg/m²	22.86(19.60-24.77)	23.15(20.79-25.10)	0.336
Causes of ICU admission, %			0.049
Medical	65(78.3)	163(87.6)
Surgical	18(21.7)	23(12.4)
Comorbidities, %
Chronic lung disease	10(12.0)	36(19.3)	0.142
Diabetes	23(27.7)	63(33.9)	0.317
Hypertension	48(57.8)	116(62.4)	0.481
Cerebrovascular disease	15(18.1)	50(26.9)	0.119
Cardiovascular disease	23(27.7)	42(22.6)	0.364
Chronic kidney disease	22(26.5)	43(23.1)	0.549
Variable	N (%)/Median (IQR)		P-value
Variable	With VTE (n = 83)	Without VTE (n = 186)	P-value
Cancer	12(14.5)	38(20.4)	0.245
Operation history	30(36.1)	60(32.3)	0.709
Personal history of VTE	3(3.61)	2(1.08)	0.173
Severity of illness
APACHE II score	20 (15-25)	16 (12-20)	0.006
Laboratory findings on admission to ICU
Platelet, × 10⁹/L	171(105-228)	159(87-229)	0.340
Hemoglobin, g/L	99(79-119)	94(71-118)	0.067
D-dimer, mg/L	4.62(2.18-8.70)	1.98(0.92-4.57)	<0.001
Prothrombin time, s	12.9(11.6-14.9)	12.8(11.8-14.8)	0.622
International normalized ratio	1.14(1.02-1.32)	1.14(1.04-1.31)	0.929
Activated partial thromboplastin time, s	32.4(26.0-37.1)	30.5(26.8-35.4)	0.364
Creatinine, μmol/L	121.3(59.2-270.8)	107.1(66.8-277.5)	0.924
Variable	N (%)/Median (IQR)		P-value
Variable	With VTE (n = 83)	Without VTE (n = 186)	P-value
In-patient treatment
Invasive mechanical ventilation, %	26(31.3)	24(12.9)	<0.001
Duration of invasive mechanical ventilation, d	4(2-13)	3(1-5)	0.013
Vasopressor use, %	34(41.0)	71(38.2)	0.665
Duration of vasopressor use, d	3(2-5)	3(2-8)	0.583
CRRT use, %	28(33.7)	62(33.3)	0.949
Duration of CRRT, d	5(4-10)	4(3-7)	0.149
Central venous catheterization, %	79(95.2)	157(84.4)	0.013
Transfusion of blood component, %	43(51.8)	95(51.1)	0.912
Pharmacologic prophylaxis, %	20(24.1)	56(30.1)	0.312
LMWH	19(22.9)	52(27.9)
UFH	1(1.2)	4(2.2)
Bleeding risk assessment at admission-high risk	69(83.1)	145(77.9)	0.331
Variable	N (%)/Median (IQR)		P-value
Variable	With VTE (n = 83)	Without VTE (n = 186)	P-value
Outcome
In-hospital mortality, %	11(13.2)	11(5.9)	0.042
Length of ICU stay, d	10(7-14)	7(5-11)	0.001
Length of hospital stay, d	13(8-22)	9(6-17)	0.002
The ICU-VTE score	11(9-12)	9(6-11)	<0.001
ICU-VTE risk score at hospital admission, %			<0.001
Low risk, 0–8 points	8(9.6)	80(43.0)
Intermediate risk, 9–14 points	73(88.0)	105(56.4)
High risk, 15–18 points	2(2.4)	3(1.6)

Abbreviations: ICU, intensive care unit; APACHE II, acute physiology and chronic health evaluation II; VTE, venous thromboembolism; CRRT, continuous renal replacement therapy; UFH, unfractionated heparin; LMWH, low molecular weight heparin; ICU-VTE, ICU-venous thromboembolism; IQR, interquartile range.

Risk Stratification by the ICU-VTE RAM

Patients were classified into different risk categories based on their ICU-VTE scores upon admission to the ICU ( Table 3 ). Among the total cohort, 39 patients had ICU-VTE scores ranging from 0 to 5, and 49 patients had ICU-VTE scores ranging from 6 to 8, placing them in the low-risk group. Additionally, 140 patients had ICU-VTE scores ranging from 9 to 11, and 38 patients had ICU-VTE scores ranging from12 to 14, these patients were classified into the moderate-risk group. Only 3 patients with ICU-VTE scores greater than 15 were classified as being at high risk upon admission. The incidence of VTE significantly increased with increasing ICU-VTE scores. Further categorization of patients revealed that the incidence of acquired VTE accelerated in relation to the cumulative risk score. Compared with the low-risk group with ICU-VTE scores ranging from 0 to 5, the risk was 2.58-fold greater in the low-risk group with scores from ranging 6 to 8, 11.27-fold greater in the moderate-risk group with scores ranging from 9 to 11, 20.56-fold greater in the moderate-risk group with scores ranging from 12 to 14, and 36.99-fold greater in the high-risk group with scores ranging from 15 to 18.

Table 3.

Odds for Inpatient Venous Thromboembolism Stratified by ICU-VTE RAM.

Risk level	N	VTE Incidence	OR	95% CI	P-value
Low risk (0–5)	39	5.41	1
Low risk (6–8)	49	12.2	2.58	0.49–13.57	0.262
Moderate risk (9–11)	140	37.9	11.27	2.61–48.69	0.001
Moderate risk (12–14)	38	52.6	20.56	4.32–97.71	0.001
High risk (15–18)	3	66.67	36.99	2.27–602.68	0.011
P trend					<0.001

Abbreviations: VTE, venous thromboembolism; ICU-VTE, ICU-venous thromboembolism; OR, odds ratio; CI, confidence interval.

ROC Curve Analysis of the Predictive Value of the ICU-VTE RAM for in-hospital VTE

The ROC curve for predicting all VTEs in critically ill patients is showed in Figure 2 . The AUC of the ICU-VTE RAM for all patients, as determined by the ROC curve, was 0.743 (95% CI, 0.682-0.804, P < 0.001) ( Figure 2A ). The ICU-VTE cutoff score was 9.5, with a sensitivity of 0.747 and a specificity of 0.640. ROC curve analyses for medical patients in the ICU revealed that the AUC of the ICU-VTE RAM was significantly greater than that of the Pauda RAM (AUC: 0.727 vs 0.583; P < 0.001) ( Figure 2B ). In critically ill surgical patients, the AUC of the ICU-VTE RAM was greater than that of the Caprini RAM (AUC: 0.774 vs 0.617; P = 0.128) ( Figure 2C ). We further validated the ability of the ICU-VTE RAM to predict symptomatic and asymptomatic VTEs separately. The ROC curves are shown in Figure 3 . The AUC of the ICU-VTE RAM for the prediction of symptomatic VTE was 0.717 (95% CI, 0.597-0.836, P < 0.001) ( Figure 3A ), while the AUC of the ICU-VTE RAM for the prediction of asymptomatic VTE was 0.707 (95% CI, 0.641-0.774, P < 0.001) ( Figure 3B ).

Figure 2.

Receiver operating characteristic (ROC) curve of the ICU-VTE risk assessment model (RAM) for the prediction of venous thromboembolism in critically ill patients. A. ROC curve of the ICU-VTE RAM in mixed medical-surgical critically ill patients. B. ROC curve of the ICU-VTE and Pauda RAM in critically ill medical patients. C. ROC curve of the ICU-VTE and Caprini RAM in critically ill surgical patients.

Figure 3.

Receiver operating characteristic (ROC) curve of the ICU-VTE risk assessment model (RAM) for predicting symptomatic or asymptomatic venous thromboembolism in critically ill patients. A. ROC curve of the ICU-VTE RAM for the prediction of symptomatic venous thromboembolism. B. ROC curve of the ICU-VTE RAM for the prediction of asymptomatic venous thromboembolism.

Discussion

To our knowledge, this is the first study to validate the ICU-VTE RAM externally. We aimed to externally validate its predictive performance for VTE risk in critically ill patients, assess its role in VTE risk stratification, and compare it with that of the Pauda and Caprini RAM. Our findings highlight the importance of adopting this simple and effective ICU-VTE RAM for critical patients. Intensive surveillance and preventive measures should be implemented when patients exhibit a high ICU-VTE score.

Multivariate logistic regression analysis revealed that a higher ICU-VTE score was a reliable independent risk factor for in-hospital VTE. This confirmed the ability of the ICU-VTE RAM to predict VTE risk among critically ill patients. By categorizing risks, we found a statistically significant difference in VTE incidence among the five groups (P < 0.001). There was an evident increase in the odds ratio for VTE in patients with higher cumulative risk scores, indicating the effective risk stratification potential of the ICU-VTE RAM in critically inpatients. In the current study, the AUC of the ICU-VTE RAM was 0.743 for the entire study population. This is comparable to the AUC of 0.714 reported in a retrospective study of critically ill patients with tumors,¹⁷ but lower than the AUC of 0.87 reported in the original study.¹⁵ These differences in AUC values may be attributed to the lack of external validation of the original study, which focused only on symptomatic VTE, as well as variations in sample sizes between studies. Furthermore, similar AUC values of the ICU-VTE RAM for the prediction of symptomatic and asymptomatic venous thromboembolism were found, indicating that the model can more comprehensively reflect all types of VTE risk.

We compared the ICU-VTE RAM with other RAMs. The ICU-VTE RAM demonstrated superior effectiveness compared to the Pauda RAM (AUC 0.727 vs AUC 0.583, P < 0.001) in assessing VTE risk among medically critically ill patients. In our cohort, the AUC for the Pauda RAM was 0.583, consistent with a prospective study involving medical patients with sepsis (AUC 0.58).¹⁸ Other retrospective studies have reported AUC values of 0.716 or 0.64 or lower for the Padua RAM in general medical patients.^19,20 The Padua RAM exhibited unstable predictive performance in internal medicine patients. It was rarely used in critically ill patients, as evidenced by our study's poor predictive performance in critically ill medical patients.

In the case of critically ill surgical patients, the AUC of the ICU-VTE RAM was not significantly greater than that of the Caprini RAM (AUC: 0.774 vs 0.617, P = 0.128), indicating that the ICU-VTE and Caprini RAMs have similar predictive abilities. Lobastov et al reported an AUC for the Caprini RAM was 0.87 when standard prophylaxis was employed.²¹ Hewes et al found an AUC of 0.818 (95% CI, 0.711-0.908) in esophagectomy patients.²² In other studies, the AUC for the Caprini RAM ranged from 0.53 to 0.754.^23–27 The Caprini RAM is complex, and some of the risk factors only pertain to surgical conditions, making it unsuitable for medical patients. Conversely, the ICU-VTE RAM is simple and intuitive, with only six risk factors related to the ICU. In a recent large-scale, nationwide cohort comprising both surgical and non-surgical patients and encompassing a large number of VTE events (n = 35,557), the predictive ability of both RAMs remained low for both surgical patients (Caprini 0.54, Padua 0.56) and non-surgical patients (Caprini 0.59, Padua 0.59).²⁸

The variability in AUC results for the Caprini and Padua models may stem from limitations in implementation heterogeneity. In our patient cohort, the Padua RAM proved to be inefficient in predicting VTE. The Caprini RAM, although comprehensive, lacks specific predictors for critically ill patients, limiting its ability to assess the risk of VTE in this population. Given the observed unstable performance and limited net benefit, the currently used RAMs are insufficient for patient stratification and determining VTE prophylaxis. The higher AUCs of the ICU-VTE RAM (AUC, 0.743) indicate its superior ability to predict VTE risk in critically ill patients. Our ROC analysis showed that a cut-off value of 9.5 for the ICU-VTE RAM aligned closely with the original study conducted by Viarasilpa et al and another study.¹⁷ Therefore, we recommend defining high-risk patients with a score ≥10 points when utilizing the ICU⁃VTE rating scale. An ICU-VTE score of ≥10 can identify a subgroup of patients at a heightened high risk of VTE, necessitating a more robust prophylactic regimen.

Another significant finding of our study was the acquisition of additional data, which were unavailable in the original study, through prospective external validation. In our study, the incidence of VTE was 30.9%, which surpassed the rate reported in the original study that focused solely on symptomatic VTE.¹⁵ Critically ill patients often do not manifest the typical signs and symptoms of DVT.²⁹ Therefore, it is reasonable for our study to assess DVT based on both symptomatic and asymptomatic cases, which, combined with the inclusion and exclusion criteria for patient selection, may have led to an overestimation of the incidence of DVT. Previous studies have demonstrated that the incidence of newly diagnosed DVT in the ICU ranges from 5% to 37%.^3,5,30,31 Despite the high incidence of DVT in our study, only one PE patient was reported because of the lack of routine screening for PE, the unavailability of CTPA for extremely critically ill patients, and the small sample size of the study. Additionally, our univariate logistic regression analysis revealed that a high APACHE II score was a risk factor for VTE, consistent with previous studies.^32,33 We also observed that critically ill patients with VTE had a longer duration of mechanical ventilation than did those who without VTE. However, the causal relationship between the duration of mechanical ventilation and VTE remains unclear.³⁴ Other studies shown that the using vasopressors is an independent risk factor for VTE.^5,35 In our cohort, the use of a vasopressor did not appear to impact the risk of VTE, a finding not obtained in the original study. This discrepancy may be attributed to the low rate of anticoagulant prophylaxis, which potentially reduces the influence of vasopressors on VTE risk, as well as the small sample size of our study and the heterogeneity of the ICU population.

This study has several noteworthy limitations. First, the sample size was limited to a single center, leading to selection bias in the patient population and preventing comprehensive validation of the ICU-VTE RAM. Although this model is the first to use simple factors at admission to predict VTE risk, the findings must be replicated in an independent cohort. Further validation of the model's accuracy at other centers with a large external dataset is needed. Second, due to the observational nature of this study, residual confounding and potential bias cannot be entirely ruled out. Third, despite the extreme heterogeneity of the enrolled patients (surgical and non-surgical), we envisioned that the model will find application in all ICU patients, unlike traditional models that are based on specific medical or surgical disease categories.

Conclusion

The ICU-VTE RAM may be a practical and valuable tool for identifying and stratifying VTE risk in mixed medical-surgical critically ill patients, aiding in the management and prevention of VTE complications.

Supplemental Material

sj-docx-1-cat-10.1177_10760296241271406 - Supplemental material for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients

Supplemental material, sj-docx-1-cat-10.1177_10760296241271406 for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients by Lijuan Zhang, Fuyang Chen, Su Hu, Yanxia Zhong, Bohua Wei, Xiaopin Wang and Ding Long in Clinical and Applied Thrombosis/Hemostasis

Supplemental Material

sj-xlsx-2-cat-10.1177_10760296241271406 - Supplemental material for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients

Supplemental material, sj-xlsx-2-cat-10.1177_10760296241271406 for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients by Lijuan Zhang, Fuyang Chen, Su Hu, Yanxia Zhong, Bohua Wei, Xiaopin Wang and Ding Long in Clinical and Applied Thrombosis/Hemostasis

Supplemental Material

sj-docx-3-cat-10.1177_10760296241271406 - Supplemental material for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients

Supplemental material, sj-docx-3-cat-10.1177_10760296241271406 for External Validation of the ICU-Venous Thromboembolism Risk Assessment Model in Adult Critically Ill Patients by Lijuan Zhang, Fuyang Chen, Su Hu, Yanxia Zhong, Bohua Wei, Xiaopin Wang and Ding Long in Clinical and Applied Thrombosis/Hemostasis

Footnotes

Acknowledgements

We thank all the authors for their meticulous work.

Authors Contributions

LZ, FC and SH contributed equally and share first authorship. Concept and design: LZ, DL. Acquisition, analysis, or interpretation of data: LZ, FC, SH, YZ. Drafting of the manuscript: LZ. Critical revision of the manuscript for important intellectual content: all authors. Final approval of the manuscript: all authors.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This study was supported by the Administration of Traditional Chinese Medicine of Hubei Province, grant number ZY2023F059 and Wuhan Municipal Health Commission, grant number WZ22Q45.

Wuhan Municipal Health Commission, Administration of Traditional Chinese Medicine of Hubei Province, (grant number WZ22Q45, ZY2023F059).

ORCID iD

Ding Long

Supplemental Material

Supplemental material for this article is available online.

References

Raskob

Angchaisuksiri

Blanco

, et al. Thrombosis: A major contributor to global disease burden. Arterioscler Thromb Vasc Biol. 2014;34(11):2363-2371.

Cohen

Agnelli

Anderson

, et al. Venous thromboembolism (VTE) in Europe. The number of VTE events and associated morbidity and mortality. Thromb Haemost. 2007;98(4):756-764.

Kaplan

Casper

Elliott

, et al. VTE Incidence and risk factors in patients with severe sepsis and septic shock. Chest. 2015;148(5):1224-1230.

Boddi

Peris

. Deep vein thrombosis in intensive care. Adv Exp Med Biol. 2017;906:167-181.

Cook

Crowther

Meade

, et al. Deep venous thrombosis in medical-surgical critically ill patients: Prevalence, incidence, and risk factors. Crit Care Med. 2005;33(7):1565-1571.

Søgaard

Schmidt

Pedersen

Horváth-Puhó

Sørensen

. 30-year Mortality after venous thromboembolism: A population-based cohort study. Circulation. 2014;130(10):829-836.

Spencer

Gore

Lessard

Douketis

Emery

Goldberg

. Patient outcomes after deep vein thrombosis and pulmonary embolism: The worcester venous thromboembolism study. Arch Intern Med. 2008;168(4):425-430.

Naess

Christiansen

Romundstad

Cannegieter

Rosendaal

Hammerstrøm

. Incidence and mortality of venous thrombosis: A population-based study. J Thromb Haemost. 2007;5(4):692-699.

Barbar

Noventa

Rossetto

, 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(11):2450-2457.

10.

Gerotziafas

Papageorgiou

Salta

Nikolopoulou

Elalamy

. Updated clinical models for VTE prediction in hospitalized medical patients. Thromb Res. 2018;164:S62-S69.

11.

Cronin

Dengler

Krauss

, et al. Completion of the updated Caprini risk assessment model (2013 version). Clin Appl Thromb Hemost. 2019;25:107602961983805.

12.

Bartlett

Mauck

Stephenson

Ganesh

Daniels

. Perioperative venous thromboembolism prophylaxis. Mayo Clin Proc. 2020;95(12):2775-2798.

13.

Gould

Garcia

Wren

, et al. Prevention of VTE in nonorthopedic surgical patients. Chest. 2012;141(2):e227S-e277S.

14.

Kahn

Lim

Dunn

, et al. Prevention of VTE in nonsurgical patients: Antithrombotic therapy and prevention of thrombosis, 9th ed: American college of chest physicians evidence-based clinical practice guidelines. Chest. 2012;141(2 Suppl):e195S-e226S.

15.

Viarasilpa

Panyavachiraporn

Marashi

Van Harn

Kowalski

Mayer

. Prediction of symptomatic venous thromboembolism in critically ill patients: The ICU-venous thromboembolism score. Crit Care Med. 2020;48(6):e470-e479.

16.

Physician's Workbook for Prevention and Control of Pulmonary Embolism and Deep Vein Thrombosis in Hospitals. Mar 3, 2019. http://www.thrombosischina.cn/hz/?tid=6 (accessed Mar 5, 2024).

17.

Mei

Wang

Chen

Wang

. The ICU-venous thromboembolism score and tumor grade can predict inhospital venous thromboembolism occurrence in critical patients with tumors. World J Surg Oncol. 2022;20(1):245.

18.

Vardi

Ghanem-Zoubi

Zidan

Yurin

Bitterman

. Venous thromboembolism and the utility of the Padua Prediction Score in patients with sepsis admitted to internal medicine departments. J Thromb Haemostasis. 2013;11(3):467-473.

19.

Zhou

, et al. Assessment of the risk of venous thromboembolism in medical inpatients using the Padua Prediction Score and Caprini risk assessment model. J Atheroscler Thromb. 2018;25(11):1091-1104.

20.

Moumneh

Riou

Douillet

, et al. Validation of risk assessment models predicting venous thromboembolism in acutely ill medical inpatients: A cohort study. J Thromb Haemostasis. 2020;18(6):1398-1407.

21.

Lobastov

Barinov

Schastlivtsev

Laberko

Rodoman

Boyarintsev

. Validation of the Caprini risk assessment model for venous thromboembolism in high-risk surgical patients in the background of standard prophylaxis. J Vasc Surg Venous Lymphat Disord. 2016;4(2):153-160.

22.

Hewes

Hachey

Zhang

, et al. Evaluation of the Caprini model for venothromboembolism in esophagectomy patients. Ann Thorac Surg. 2015;100(6):2072-2078.

23.

Bahl

Henke

Wakefield

Campbell

Caprini

. A validation study of a retrospective venous thromboembolism risk scoring method. Ann Surg. 2010;251(2):344-350.

24.

Hachey

Hewes

Porter

, et al. Caprini venous thromboembolism risk assessment permits selection for postdischarge prophylactic anticoagulation in patients with resectable lung cancer. J Thorac Cardiovasc Surg. 2016;151(1):37-44.e31.

25.

Tian

Cui

Song

. A novel risk assessment model for venous thromboembolism after major thoracic surgery: A Chinese single-center study. J Thorac Dis. 2019;11(5):1903-1910.

26.

Cui

Yang

, et al. Validation of a modified Caprini risk assessment model in lung cancer patients undergoing surgery: Results of a multicenter cross-sectional observational study. J Surg Oncol. 2022;125(5):933-942.

27.

Frankel

Belanger

Tortora

McLaughlin

Staff

Wagner

. Caprini score and surgical times linked to the risk for venous thromboembolism after robotic-assisted radical prostatectomy. Turk J Urol. 2020;46(2):108-114.

28.

Hayssen

Sahoo

Nguyen

, et al. Ability of Caprini and Padua risk-assessment models to predict venous thromboembolism in a nationwide VA study. J Vasc Surg Venous Lymphat Disord. 2024;12(2):101693.

29.

Crowther

Cook

Griffith

, et al. Deep venous thrombosis: Clinically silent in the intensive care unit. J Crit Care. 2005;20(4):334-340.

30.

Marik

Andrews

Maini

. The incidence of deep venous thrombosis in ICU patients. Chest. 1997;111(3):661-664.

31.

Shorr

Williams

. Venous thromboembolism in critically ill patients. Observations from a randomized trial in sepsis. Thromb Haemost. 2009;101(1):139-144.

32.

Zhang

, et al. The cumulative venous thromboembolism incidence and risk factors in intensive care patients receiving the guideline-recommended thromboprophylaxis. Medicine (Baltimore). 2019;98(23):e15833.

33.

Guan

Chang

Zhang

. Interpretable machine learning models for predicting venous thromboembolism in the intensive care unit: An analysis based on data from 207 centers. Crit Care. 2023;27(1):406.

34.

Minet

Lugosi

Savoye

, et al. Pulmonary embolism in mechanically ventilated patients requiring computed tomography. Crit Care Med. 2012;40(12):3202-3208.

35.

Tran

Fernando

Rochwerg

, et al. Prognostic factors associated with development of venous thromboembolism in critically ill patients—A systematic review and meta-analysis. Crit Care Med. 2021;50(4):e370-e381.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.05 MB

0.06 MB