Abstract
Objective
Venous thromboembolism (VTE) is a preventable complication of hospitalization. Risk-stratification is the cornerstone of prevention. The Caprini and Padua are two of the most commonly used risk-assessment models (RAMs) to quantify VTE risk. Both models perform well in select, high-risk cohorts. Although VTE RAMs were designed for use in all hospital admissions, they are mostly tested in select, high-risk cohorts. We aim to evaluate the two RAMs in a large, unselected cohort of patients.
Methods
We analyzed consecutive first hospital admissions of 1,252,460 unique surgical and non-surgical patients to 1298 Veterans Affairs facilities nationwide between January 2016 and December 2021. Caprini and Padua scores were generated using the Veterans Affairs’ national data repository. We determined the ability of the two RAMs to predict VTE within 90 days of admission. In secondary analyses, we evaluated prediction at 30 and 60 days, in surgical vs non-surgical patients, after excluding patients with upper extremity deep vein thrombosis, in patients hospitalized ≥72 hours, after including all-cause mortality in a composite outcome, and after accounting for prophylaxis in the predictive model. We used area under the receiver operating characteristic curves (AUCs) as the metric of prediction.
Results
A total of 330,388 (26.4%) surgical and 922,072 (73.6%) non-surgical consecutively hospitalized patients (total N = 1,252,460) were analyzed. Caprini scores ranged from 0 to 28 (median, 4; interquartile range [IQR], 3-6); Padua scores ranged from 0-13 (median, 1; IQR, 1-3). The RAMs showed good calibration and higher scores were associated with higher VTE rates. VTE developed in 35,557 patients (2.8%) within 90 days of admission. The ability of both models to predict 90-day VTE was low (AUCs: Caprini, 0.56; 95% confidence interval [CI], 0.56-0.56; Padua, 0.59; 95% CI, 0.58-0.59). Prediction remained low for surgical (Caprini, 0.54; 95% CI, 0.53-0.54; Padua, 0.56; 95% CI, 0.56-0.57) and non-surgical patients (Caprini, 0.59; 95% CI, 0.58-0.59; Padua, 0.59; 95% CI, 0.59-0.60). There was no clinically meaningful change in predictive performance in any of the sensitivity analyses.
Conclusions
Caprini and Padua RAM scores have low ability to predict VTE events in a cohort of unselected consecutive hospitalizations. Improved VTE RAMs must be developed before they can be applied to a general hospital population.
Keywords: Deep vein thrombosis, Prevention, Pulmonary embolism, Risk assessment, Venous thromboembolism
Article Highlights.
-
•
Type of Research: Multicenter retrospective cohort study
-
•
Key Findings: In a cohort of over 1 million unselected consecutive hospitalizations across the United States, the widely used Caprini and Padua risk-assessment models demonstrate limited predictive ability for venous thromboembolism (VTE) within 90 days of admission (area under the receiver operating characteristic curve: Caprini, 0.56; Padua, 0.59). The predictive ability is similarly low when evaluating both models in surgical and non-surgical patients.
-
•
Take Home Message: The Caprini and Padua risk-assessment models have limited ability to predict VTE events in unselected general hospital admissions. Further development and improvement of VTE risk-assessment models are essential before widespread implementation in the general hospital population for guiding VTE preventive strategies.
Venous thromboembolism (VTE), encompassing deep vein thrombosis (DVT) and pulmonary embolism (PE), is a preventable sequala of hospitalization. Over 900,000 VTE events, associated with over 100,000 deaths, and an economic burden of $2 to $10 billion are reported each year in the United States (U.S.).1,2 Summarizing evidence-based guidelines, the U.S. Surgeon General emphasized “the need to screen hospitalized patients for risk of DVT/PE and to provide appropriate prophylaxis to those at risk.”3 The Joint Commission on the Accreditation of Healthcare Organizations, Agency for Healthcare Research and Quality, and Centers for Disease Control and Prevention have each identified risk-assessment and risk-stratification as the keys to improving VTE prevention in all general hospitalized patients.4
At least 23 VTE risk-assessment models (RAMs) have been developed that quantify a patient’s risk for VTE by summing the weights assigned to clinical risk-factors.5 By combining ranges of scores into ordered groups, the RAMs attempt to accomplish risk-stratification. The Caprini and Padua RAMs are two of the most widely used RAMs.5 They were designed to be used in the general population of hospital patients, not just high-risk patient groups.5 Risk factors and weights in both RAMs were selected based on clinical expertise and information from the literature.6 The Caprini RAM was designed for general surgery patients but was mostly evaluated in high-risk subgroups of surgical patients.7,8 The Padua RAM was designed for general medical patients but was mostly evaluated in acutely ill inpatients.9, 10, 11 Although both RAMs have been validated primarily in high-risk subsets, guidelines suggest using the RAMs in the general population of hospitalized patients; the Caprini RAM in hospitalized surgical patients8,12 and the Padua RAM in hospitalized non-surgical patients.9,10
Increased awareness about the clinical and cost of hospital-associated VTE, and the push to implement risk stratification by health care organizations, has led to an increased use of the Caprini and Padua RAMs in all hospital admissions. However, neither RAM has been well-studied in a large, unselected cohort of non-surgical and surgical patients.6,13
We evaluated the ability of the Caprini and Padua RAMs to predict VTE 0 to 90 days post-admission in patients hospitalized at Veterans Affairs (VA) health care facilities nationwide over a 6-year period.
Methods
Study design and participants
We performed an analysis of the first hospital admission of patients to any VA facility from January 1, 2016, through December 1, 2021. We followed STROBE guidelines for cohort studies in writing this report.14 Patients with a VTE diagnosed in the 90 days prior to admission or those admitted with a diagnosis of VTE were excluded. Patients who underwent a surgical procedure (planned or unplanned) during hospitalization, regardless of the admitting service, were defined as surgical patients. Non-surgical patients were those who did not undergo a surgical procedure during hospitalization. The protocol was approved by the institutional review board of the University of Maryland and the Baltimore VA Research and Development Committee.
Data source
Data were obtained from the Veterans Affairs Informatics and Computing Infrastructure (VINCI), a database of the approximately 9 million patients receiving care at 1298 VA health care facilities nationwide.15,16 VINCI contains all data entered into the VA’s electronic medical record from in- and out-patient encounters.
Outcome and variables
Our main outcome, VTE both during or after admission, was defined as a new International Classification of Diseases (ICD-10) code diagnostic for PE, DVT, or both DVT and PE (Supplementary Table I, online only). VINCI maintains complete records of all inpatient and outpatient encounters of veterans. The 2013 version of the Caprini RAM, as used by Cronin et al, provides a composite score based on the sum of weighted scores of 30 risk factors.8 The Padua RAM provides a score based on 11 risk factors.11 We computed both scores for each patient. Risk factors were identified using ICD-10 diagnostic and procedural codes, Current Procedural Terminology (CPT) codes, demographic data, clinical and nursing orders, laboratory data, medications, prosthetics consultations, and operating room data. Data obtained included length of surgery, duration of bedrest, and date of VTE diagnosis, for example. We carefully interpreted each risk factor in the two RAMs and translated them into appropriate method of data extraction (eg, set of ICD-10 codes). For example, we translated “insertion of central venous catheter” to the CPT codes for Insertion of Central Venous Access Device (36,569-36,573) and for Complete Replacement of Central Venous Access Device Through Same Venous Access Site (36,580-36,585). If Caprini and Padua variables were similarly worded, we used the same variable in final calculation (eg, history of VTE). When height or weight were not recorded at admission, we imputed the body mass index (BMI) using the average of heights and weights obtained within the year before and after admission, after which BMI was available in 96% of patients. We were unable to extract information about family history; therefore, we did not include family history of VTE or family history of a disorder that increases risk of VTE, both of which are part of the Caprini RAM. Furthermore, it should be noted that certain risk factors, including BMI >40 kg/m2, smoking, insulin-treated diabetes, chemotherapy, and transfusions, which were suggested as additions to the Caprini RAM, have, according to those authors, not yet undergone rigorous validation studies to establish their precise influence. Therefore, these risk factors were excluded from our study.
We categorized patients as receiving pharmacologic VTE prophylaxis based on medication orders for unfractionated heparin, low molecular-weight heparins, direct-acting oral anticoagulants, or warfarin as well as CPT codes for inferior vena cava filter placement. We determined if patients received mechanical prophylaxis based on physician orders for intermittent or sequential compression devices.
Statistical methods
We compared demographic and clinical features of our cohort and distribution of Caprini and Padua risk-factors in patients with and without VTE, using the Pearson χ2 test (categorical variables) and Student t-test (continuous variables). We used histograms to visualize the distribution of RAM scores in patients who did and did not develop VTE within 90 days of admission, and to visualize the relationship between RAM scores and VTE. Logistic regressions, outcome of VTE (yes vs no), and predictor variable Caprini or Padua scores, were used to determine the ability of the RAMs to predict outcomes of interest. Predictive ability was quantified by the area under the receiver operating characteristic curves (AUC) obtained from the logistic regressions. The AUCs of different models were compared using Delong-Delong tests.17 P-value < .01 was considered statistically significant. Analyses were performed using R version 4.0.1 (R Core Team) and SAS software version 9.4 (SAS Institute Inc).
We evaluated the ability of each RAM to predict VTE in the entire cohort at 0 to 30, 31 to 60, 61 to 90, and 0 to 90 days (four logistic regressions for each RAM). The models were adjusted only for the patients’ RAM score. To maximize the number of VTE events, all subsequent secondary analyses examined outcomes at a single follow-up, 0 to 90 days.
In secondary analyses, we evaluated whether each RAM had better predictive ability (higher AUC) in sub-populations of the cohort, and for different outcomes of interest. We first evaluated the prediction of 90-day VTE for each RAM in subgroups of the overall cohort: (1) surgical patients, (2) non-surgical patients, and (3) patients admitted for ≥72 hours (a surrogate for increased immobility). Next, we determined the prediction of the 0- to 90-day outcome of (4) VTE excluding upper extremity DVTs (upper extremity DVTs have lower morbidity and mortality) and (5) the composite outcome of VTE or mortality (acute PE may remain undetected as a cause of sudden death). Finally, we examined prediction of 0- to 90-day VTE by the two RAMs in (6) patients who received VTE prophylaxis (mechanical or pharmacologic), (7) patients who did not receive VTE prophylaxis, and (8) in the entire cohort after adjusting for any prophylaxis received (prophylaxis may alter the predictive ability of the RAMs; however, prophylaxis is not accounted for in the Caprini or Padua RAMs).
Results
Participants
A total of 1,282,014 patients were hospitalized from January 1, 2016, to December 1, 2021. We excluded 21,974 patients admitted with a diagnosis of VTE and 7580 patients with a VTE within 0 to 90 days before admission. Our study population included 1,252,460 patients, of whom 26.4% (n = 330,388) were surgical and 73.6% (n = 922,072) were non-surgical patients (Table I). The mean age of the patients was 65.9 years; the cohort was predominantly male (93.0%). A substantial minority were non-White (29.8%). Almost 41% received no VTE prophylaxis.
Table I.
Clinical characteristics of the study cohort
| Variables | All N = 1,252,460 |
VTE group n = 35,557 (2.8%) |
Non-VTE group n = 1,216,903 (97.2%) |
P valuea | U.S. hospitalized populationb |
|---|---|---|---|---|---|
| Age, years | 65.9 (13.8) | 68.1 (12.0) | 65.8 (13.8) | <.001 | 49.9 (481.7) |
| Sex | <.001 | ||||
| Male | 1,164,380 (93.0) | 33,846 (95.2) | 1,130,534 (92.9) | 44.8% | |
| Female | 88,079 (7.0) | 1711 (4.8) | 86,368 (7.1) | 55.2% | |
| Race | <.001 | ||||
| White | 891,379 (71.2) | 24,112 (67.8) | 867,267 (71.3) | 74.1% | |
| Black | 261,905 (20.9) | 8656 (24.3) | 253,249 (20.8) | 15.2% | |
| All others | 82,256 (6.6) | 2250 (6.3) | 82,170 (6.7) | 10.7% | |
| Body mass index, kg/m2 | 29.3 (6.8) | 29.2 (7.1) | 29.3 (6.7) | .018 | N/A |
| Hospital length of stay, days | 4.1 (11.3) | 8.7 (23.5) | 3.9 (10.7) | <.001 | 4.9 |
| Patient classification | <.001 | ||||
| Surgical | 330,388 (26.4) | 8064 (22.7) | 322,324 (26.5) | 25.8% | |
| Non-surgical | 922,072 (73.6) | 27,493 (77.3) | 894,579 (73.5) | 74.2% | |
| Prophylaxis type | |||||
| Prophylaxis (mechanical and/or pharmacologic) | 740,632 (59.1) | 24,899 (70.0) | 715,733 (58.8) | <.001 | N/A |
| None | 511,828 (40.9) | 10,658 (30.0) | 501,170 (41.2) | N/A |
VTE, Venous thromboembolism.
Data are presented as number (%) or mean (standard deviation).
P-values computed with Pearson χ2 test comparing VTE group vs non-VTE group; threshold for statistical significance set at < .01.
Outcome data
Within 90 days after admission, 35,557 patients (2.8%) developed a VTE, of whom 15,056 (42%) were PEs with or without concurrent DVT, and 20,501 (58%) were DVTs alone. The median time to VTE occurrence was 11 days (IQR, 4-35). Although most VTEs developed within the first 30 days of admission, 28% developed at 31 to 90 days (Table II).
Table II.
Distribution of venous thromboembolism (VTE) events and area under the receiver operating characteristic curves (AUC) for Caprini and Padua scores to predict a VTE after 1,252,460 unique consecutive unselected surgical and non-surgical hospital admissions nationwide
| Time from hospital admission | VTE type |
Total VTE events |
Caprini RAM |
Padua RAM |
||||
|---|---|---|---|---|---|---|---|---|
| PE only |
PE and DVT |
DVT only |
N (%) | AUC (95% CI) | P valuea | AUC (95% CI) | P valuea | |
| N | ||||||||
| 0-30 days | 8093 | 3144 | 14,371 | 25,608 (72) | 0.56 (0.55-0.56) | .07 | 0.58 (0.58-0.59) | .58 |
| 31-60 days | 1709 | 620 | 3622 | 5951 (17) | 0.58 (0.57-0.58) | <.001 | 0.59 (0.58-0.60) | .15 |
| 61-90 days | 1138 | 352 | 2508 | 3998 (11) | 0.56 (0.55-0.57) | .47 | 0.58 (0.57-0.59) | .69 |
| 0-90 days | 10,940 | 4116 | 20,501 | 35,557 (100) | 0.56 (0.56-0.56) | 1.00 (ref) | 0.59 (0.58-0.59) | 1.00 (ref) |
CI, Confidence interval; PE, pulmonary embolism; RAM, risk assessment model; ref, reference.
0-30, 31-60, and 61-90-day AUCs are compared with the 0-90 day AUCs (reference group) within each risk assessment model. Comparisons were made within each RAM using the Delong-Delong test.
Descriptive data
Patients who suffered a VTE within 90 days of admission were on average older than those who did not suffer a VTE (68.1 vs 65.8 years; P < .001) (Table I). The VTE group exhibited a BMI of 29.2 kg/m2 compared with 29.3 kg/m2 in the non-VTE group (P = .018). The fraction who were male was slightly higher in the VTE group than in the non-VTE group (95.2% vs 92.9%; P < .001). The fraction who were Black was higher in the VTE group than in the non-VTE group (24.3% vs 20.8%; P < .001).
Sixteen of the 30 risk factors in the Caprini RAM were associated with an increased risk of VTE (Supplementary Table II, online only). Nine of the 11 risk factors in the Padua RAM were associated with an increased risk of VTE (Supplementary Table II, online only). Some risk factors in the Caprini RAM (age 41-60 years, planned minor surgery, inflammatory bowel disease, BMI >25 kg/m2, hormone therapy, planned major surgery, and elective arthroplasty) and in the Padua RAM (hormonal treatment and BMI >30 kg/m2) were associated with a reduced risk of VTE.
The Caprini scores ranged from 0 to 28 (median, 4; IQR, 3-6) (Fig 1). VTE rate increased as the score increased, a trend that held until a score of 15 (Fig 2). Padua scores ranged from 0 to 13 (median, 1; IQR, 1-3) (Fig 1). VTE rate increased as the score increased (Fig 2).
Fig 1.
Distribution of Caprini scores (A) and Padua scores (B) at the time of admission from the first admission of 1,252,460 consecutively hospitalized surgical and non-surgical patients. Patient who did not develop a venous thromboembolism (VTE) event within 90 days of hospital admission on the left and patients that developed a VTE event within 90 days of hospital admission on the right. IQR, inter-quartile range; SD, standard deviation.
Fig 2.
Distribution of venous thromboembolism (VTE) rates by risk-assessment model (RAM) score in 1,252,460 consecutively hospitalized surgical and non-surgical patients at the time of their admission. A, Caprini score; B, Padua score. The Error bar represents the upper bar of 95% confidence interval (CI).
Main results
The Caprini-RAM AUC predicting VTE at 0 to 90 days in the entire cohort was 0.56 (95% confidence interval [CI], 0.56-0.56) (Table II). The AUCs at 0 to 30 and 61 to 90 days were not different from those at 0 to 90 days. The AUC at 31 to 60 days was slightly higher than for 0 to 90 days; the difference was not clinically important (difference in AUC, 0.02) (Table II).
The Padua RAM AUC predicting VTE at 0 to 90 days in the entire cohort was 0.59 (95% CI, 0.58-0.59), higher than the Caprini AUC (P < .001); the difference was of no clinical importance (difference in AUC, 0.03) (Table II). The AUCs at 0 to 30, 31 to 60, and 61 to 90 days were not different from the 0- to 90-day AUC (Table II).
Secondary results
Compared with the 0- to 90-day prediction of VTE in the entire cohort (model 0), Caprini and Padua VTE prediction was (1) better in non-surgical patients and (2) worse in surgical patients, P < .001 for all four comparisons (Table III). Prediction in non-surgical patients was better than in surgical patients; the difference was of no clinical importance (difference of AUCs: Caprini, 0.05; Padua, 0.03; P < .001) (Fig 3). Compared with the reference model, both RAMs performed slightly worse among (3) patients hospitalized for ≥72 hours (Caprini P < .001; Padua P < .001).
Table III.
Secondary analyses of the Caprini and Padua risk assessment models (RAMs) for venous thromboembolism (VTE) outcomes within 0 to 90 days of hospital admission
| Characteristic examined | Model | Analytic approach | Patients (N) | Events (N) | Caprini RAM |
Padua RAM |
||
|---|---|---|---|---|---|---|---|---|
| AUC (95% CI) | P valuea | AUC (95% CI) | P valuea | |||||
| VTE in entire cohort (reference for subsequent analyses) | 0 | Full cohort analysis for all VTE events | 1,252,460 | 35,557 | 0.56 (0.56-0.56) | 1.00 (ref) | 0.59 (0.58-0.59) | 1.00 (ref) |
| VTE by type of patient | 1 | Subgroup analysis of non-surgical patients | 922,072 | 27,493 | 0.59 (0.58-0.59) | <.001 | 0.59 (0.59-0.60) | <.001 |
| 2 | Subgroup analysis of surgical patients | 330,388 | 8064 | 0.54 (0.53-0.54) | <.001 | 0.56 (0.56-0.57) | <.001 | |
| 3 | Subgroup analysis of patients admitted for ≥72 hours | 442,644 | 21,280 | 0.53 (0.52-0.53) | <.001 | 0.56 (0.56-0.57) | <.001 | |
| Different outcomes in entire cohort | 4 | Full cohort analysis for excluding upper extremity DVTs | 1,252,460 | 31,602 | 0.56 (0.56-0.57) | .12 | 0.59 (0.58-0.59) | .57 |
| 5 | Full cohort analysis for VTE and/or all-cause mortality | 1,252,460 | 107,684 | 0.58 (0.58-0.58) | <.001 | 0.60 (0.60-0.60) | <.001 | |
| VTE by prophylaxis status | 6 | Subgroup analysis of patients receiving VTE prophylaxis | 740,632 | 24,899 | 0.55 (0.54-0.55) | <.001 | 0.58 (0.57-0.58) | <.001 |
| 7 | Subgroup analysis of patients not receiving VTE prophylaxis | 511,828 | 10,658 | 0.58 (0.57-0.58) | <.001 | 0.59 (0.59-0.60) | .02 | |
| 8 | Full cohort adjusted for prophylaxis status | 1,252,460 | 35,557 | 0.59 (0.58-0.59) | <.001 | 0.61 (0.60-0.61) | <.001 | |
AUC, Area under the receiver operating characteristic curve; CI, confidence interval; DVT, deep vein thrombosis.
The AUC of models 1 through 8 were compared with baseline model 0. Comparisons were made within each RAM using the Delong-Delong test.
Fig 3.
Receiver operating characteristic curves demonstrating the ability of the Caprini (A) and Padua (B) risk assessment models (RAMS) to predict a venous thromboembolism (VTE) event within 0 to 90 days of hospital admission. Each graph demonstrates the receiver operating characteristic curve for surgical (blue) and non-surgical (red) patients computed separately. AUC, Area under the curve. The confidence intervals (CIs) are provided within brackets.
Prediction by type of outcome
The Caprini and Padua AUCs for prediction of (4) VTE excluding upper extremity were similar to the AUC for all VTEs (Caprini P = .12; Padua P = .57). Both RAMs were slightly better at predicting the (5) composite outcome of VTE or all-cause mortality, vs VTE alone (Caprini P < .001; Padua P < .001). The differences were of no clinical importance (difference of AUCs: Caprini 0.02; Padua 0.01).
Prediction by prophylaxis status
A total of 740,632 patients (59%) received VTE prophylaxis (mechanical or pharmacologic) and 511,828 (41%) did not; 3.4% of those who received prophylaxis developed a VTE 0 to 90 days after admission, whereas 2.1% of those who did not receive prophylaxis developed a VTE (P < .001). For the VTEs occurring outside the first 30 days post-admission (ie, from 31-90 days post-admission), 0.7% of those who received prophylaxis developed a VTE, whereas 0.9% of those who did not receive prophylaxis developed a VTE (P < .001). In (6) patients who receive VTE prophylaxis, the ability to predict VTE at 0 to 90 days for both RAMs was worse than for the entire cohort (AUCs: Caprini, 0.55; 95% CI, 0.54-0.55; P < .001 and Padua, 0.58; 95% CI, 0.57-0.58; P < .001). (7) In patients who did not receive VTE prophylaxis, the ability to predict VTE at 0 to 90 days for both RAMs was better than for the entire cohort (AUCs: Caprini, 0.58; 95% CI, 0.57-0.58; P < .001 and Padua, 0.59; 95% CI, 0.59-0.60; P = .02). (8) The prediction of VTE at 0 to 90 days improved when prophylaxis was included as an independent variable in the models for both RAMs (AUCs: Caprini, 0.59; 95% CI, 0.58-0.59 and Padua, 0.61; 95% CI, 0.60-0.61; P < .001) (Table III). The differences were of no clinical importance (differences of AUCs: Caprini, 0.03; Padua, 0.02).
Discussion
The incidence of VTE within 90 days after hospitalization in our nationwide cohort of 1,252,460 unselected consecutive surgical and non-surgical patients was 2.8%. Approximately 28% of VTEs occurred between 31 and 90 days after admission, highlighting the importance of tracking events beyond the traditional 30 days post-discharge. The low AUCs of the Caprini (AUC, 0.56) and Padua (AUC, 0.59) RAMS reflect their limited utility for VTE risk-stratification. The performance of both RAMs was better in non-surgical compared with surgical patients, although the improvements were not clinically important. The RAMs did not demonstrate clinically important improvement when prediction was limited to patients admitted for ≥72 hours, after excluding upper-extremity DVT from the outcome, after including all-cause mortality in a composite outcome, or after accounting for VTE prophylaxis. These results do not support use of the RAMs in general hospital admissions to predict VTE without improvements in the models, though there may still be a role for them in high-risk subsets.
Our results confirm that hospital admission confers a risk for VTE. In our study, 2.8% of patients developed a VTE within 90 days of hospitalization. Of these, 42% had a PE, an important cause of VTE-related mortality. With several million patients hospitalized each year, many patients are at risk for long-term sequelae of VTE, post-thrombotic syndrome, and post-PE syndrome. These iatrogenic complications can be reduced by prophylactic measures. However, mechanical prophylaxis (eg, compression) may not be tolerated or feasible, and pharmacoprophylaxis increases the risk for bleeding. These risks must be weighed against the risk for VTE; therefore, reliable stratification of VTE risk is a prerequisite to appropriate prophylaxis.
Mixed results have been reported from the assessment of predictive performance for the Caprini and Padua RAMs. Although the Caprini RAM has been evaluated in over 200 studies worldwide, most sample-sizes have been modest (<10,000 patients) with few VTE events (<100).21 The Padua RAM has been evaluated primarily in small cohorts of medical patients.22,23 The best performance (AUCs) for the RAMs were obtained in studies with modest numbers of high-risk patients, often fewer than 1000.24, 25, 26 The two RAMs have been compared previously.5,23,26, 27, 28, 29, 30 The Caprini RAM shows better prediction of mortality than Padua,23 and better prediction (AUC) for VTE (AUC 0.77 vs 0.62 for Padua; P < .05).27 An analysis of acutely ill medical patients reported low predictive abilities from both, Caprini (AUC, 0.60) and Padua (AUC, 0.64) RAMs.30 In our larger, nationwide, mixed surgical and non-surgical cohort, with a large number of VTEs (n = 35,557), both RAMs had limited predictive ability for VTE. This may explain why, even though the tools have been available for several years, there are limited data on whether broad clinical use of the RAMs reduces VTE rates.31 A 43-hospital Michigan Hospital Medicine Safety Consortium-instituted systematic VTE risk assessment was performed in hospitalized patients (primarily Caprini RAM). Even though risk-assessment and the use of prophylaxis increased, the 90-day post-admission VTE rate was not reduced.32
In our study, most VTEs occurred within 0 to 30 days post-admission; however, one-quarter occurred between days 31 and 90 (Table II). Our results may underestimate incidence, as we are relying on administrative data and as PE is an often-overlooked cause of death when it occurs remote from a hospitalization. When we included all-cause mortality in a composite outcome, prediction of both RAMs improved, which could suggest an underestimation of PE incidence. We found significant VTE incidence up to 90 days post-admission. Future studies should assess whether risk extends beyond 90 days, and compare VTE incidence in age-, sex- and risk factor-matched non-hospitalized patients to define the true additional risk for VTE conferred by hospitalization.
Both RAMs showed good calibration. Increasing Caprini RAM scores up to 15 were associated with an increased rate of VTE, at which point the paucity of events (and resulting larger CIs) likely contributed to a less linear association (Fig 2). Grant et al reported increasing VTE rates up to Caprini scores of 7, after which the rate stayed elevated but without a consistent rise.33 Higher Padua scores were associated with increased VTE rates, consistent with prior reports.23
There may be several reasons why RAM scores did not translate into better prediction of VTE. Not all risk factors in the RAMs had a strong relationship with VTE (Supplementary Table II, online only). Surprisingly, some risk factors were protective. One reason for the modest predictive power of the RAMs may be that both RAMs were derived empirically.29 The Caprini RAM was developed using data from 538 surgical patients; the Padua RAM from 1180 non-surgical patients.7,11 The prevalence of 13 of 30 Caprini risk factors was not different in patients with vs those without VTE. Five factors had a higher prevalence in the no-VTE vs VTE group (Supplementary Table II, online only).
Obesity is known to increase risk for VTE two- to three-fold.34, 35, 36 We found that obesity was not associated with VTE risk (Supplementary Table II, online only). This finding could be related to our study population, older, mostly male patients, and with many comorbidities, reducing the overall effect of obesity. Exogenous hormone use, particularly estrogen, is a known risk factor with a 1.5- to three-fold higher risk of VTE.34,37 In our study, hormone use was more prevalent in the non-VTE group. This may be a function of the low overall prevalence of hormone use in our population; only 1% of the cohort (n = 11,793 patients) had recent (within 30 days) exogenous hormone use. Future studies should rigorously evaluate the factors that are not correlated with higher VTE risk, and versions of the RAM should reconsider including these in the model.
The Caprini RAM has been tested (and is possibly used) more often in surgical patients, whereas the Padua RAM has been tested (and possibly used) more often in non-surgical patients. When we assessed for potential differential performance of the RAMs by patient sub-type, we found that both RAMs performed slightly better in non-surgical (AUCs: Caprini RAM, 0.59; Padua RAM, 0.60) compared with surgical patients. The differences were not clinically important. It is possible that some risk factors for VTE in surgical patients are not included in the RAMs (eg, duration of surgery, type of anesthesia, or organ system being operated on). An evaluation for inclusion in the RAMs of a more exhaustive set of risk factors is needed to improve RAM performance.
No widely used RAM includes VTE prophylaxis in its assessment of VTE risk.6 Similarly, studies evaluating the predictive ability of the RAMs have not routinely accounted for VTE prophylaxis.26, 27, 28 We found the predictive ability of both RAMs improved when we accounted for prophylaxis (AUCs: Caprini RAM, 0.59; Padua RAM, 0.61), and when we examined only those patients that did not receive prophylaxis (AUCs: Caprini RAM, 0.58; Padua RAM, 0.59) (Table III). Although the improvements did not make either RAM clinically useful, they indicate the possibility of incorporating prophylaxis in future testing of RAMs and in validation studies. We found that VTE prophylaxis was associated with a higher risk of VTE. This finding was noted in a study of 14,660 hospitalized medical patients; 57% of patients who developed a VTE had received prophylaxis vs 46% of those that had not (P < .001),30 and this was confirmed in a study of surgical patients.38 This counterintuitive finding may reflect the fact that clinicians, either by calculation of a RAM score, or by clinical intuition, identify patients at high risk for VTE and more often prescribe prophylaxis for these patients. It could also be related to surveillance bias in studies. If true, it suggests that in high-risk patients, the prescribed prophylaxis is not as effective as needed to allow VTE prophylaxis to overcome possible bias by indication.
Limitations
Our study is the largest evaluation of VTE RAMs in an unselected population of general VA hospital admissions. Our reliance on administrative data to define scores of several risk factors means that Caprini and Padua RAM scores could be under- or overestimated, given inaccurate clinical information. Some risk factors are not precisely defined in the RAMs, such as “serious trauma.” As a result, their extraction from the medical record was imprecise. Despite this, collection of these risk factors directly from patients would be equally imprecise because of the lack of a clear definition of the risk factor. Although it may be argued that using VA data decreases the generalizability of our results to female patients, despite the small fraction of women in our cohort, it is the largest cohort (total number, n = 88,079) of women studied for VTE risk-stratification. Although VA clinicians obtain data about events that occur between VA encounters, clinical data about events that occurred, or did not occur, at non-VA facilities after hospital discharge may have been missed. There is no clear reason why more events would be missed than non-events, and so it is not clear that these missing events (or non-events) would bias our results. We were able to evaluate multiple follow-up periods, different outcome measures, and perform secondary analyses. We were not able to include family history in computing the Caprini score as these data were available for only a small fraction of patients. We were missing BMI data on 4% of patients. Although we had data indicating if mechanical prophylaxis was ordered for patients, we could not confirm patient adherence to the therapy. Possible confounding by indication exists, particularly in the analysis of prediction of VTE in patients receiving and not receiving prophylaxis.
Conclusions
The Caprini and Padua RAMs have low predictive ability for VTE among general hospital admissions, for both surgical and non-surgical patients. Implementing and ensuring compliance for widespread VTE risk assessment and risk stratification using standardized risk-assessment models across large health care systems is effort- and resource-intensive. The currently available risk models may need to be readdressed, and rigorously assessed and modified prior to universal adoption and use in the general hospital population.
Author Contributions
Conception and design: BL
Analysis and interpretation: HH, SS, PN, MMC, TS, BE, JS, CM, YY, JS, BL
Data collection: HH, TS
Writing the article: HH, JS, BL
Critical revision of the article: HH, SS, PN, MMC, TS, BE, JS, CM, YY, JS, BL
Final approval of the article: HH, SS, PN, MMC, TS, BE, JS, CM, YY, JS, BL
Statistical analysis: HH, SS, JS
Obtained funding: HH, BL
Overall responsibility: BL
Disclosures
C.D.M. has served as a consultant to Bayer, Incyte, Merck, Pfizer, and Takeda.
Footnotes
Funding: This study was funded by the Institute for Clinical & Translational Research (ICTR) Award 1UL1TR003098, National Institutes of Health (NIH) Awards NS080168, NS097876 and AG000513 and Veterans Affairs Awards HSRD C19-20-407, RRD RX000995 and CSRD CX001621 (B.K.L.); NIH Award AG028747 and Baltimore VA Medical Centre GRECC (J.D.S.); and NIHT32 AG00262 and American Venous Forum Award JF2021 (H.H.).
The editors and reviewers of this article have no relevant financial relationships to disclose per the Journal policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest.
Additional material for this article may be found online at www.jvsvenous.org.
Appendix
Additional material for this article may be found online at www.jvsvenous.org.
Appendix (online only)
Supplementary Table I (online only).
List of International Classification of Diseases, 10th revision (ICD-10) codes for diagnosis of venous thromboembolism (VTE)
| ICD-10 code | Description |
|---|---|
| I2601 | Septic pulmonary embolism with acute cor pulmonale |
| I2602 | Saddle embolus of pulmonary artery with acute cor pulmonale |
| I2609 | Other pulmonary embolism with acute cor pulmonale |
| I2690 | Septic pulmonary embolism without acute cor pulmonale |
| I2692 | Saddle embolus of pulmonary artery without acute cor pulmonale |
| I2693 | Single subsegmental pulmonary embolism without acute cor pulmonale |
| I2694 | Multiple subsegmental pulmonary emboli without acute cor pulmonale |
| I2699 | Other pulmonary embolism without acute cor pulmonale |
| I81 | Portal vein thrombosis |
| I820 | Budd-Chiari syndrome |
| I82210 | Acute embolism and thrombosis of superior vena cava |
| I82220 | Acute embolism and thrombosis of inferior vena cava |
| I82290 | Acute embolism and thrombosis of other thoracic veins |
| I823 | Embolism and thrombosis of renal vein |
| I82401 | Acute embolism and thrombosis of unspecified deep veins of right lower extremity |
| I82402 | Acute embolism and thrombosis of unspecified deep veins of left lower extremity |
| I82403 | Acute embolism and thrombosis of unspecified deep veins of lower extremity, bilateral |
| I82409 | Acute embolism and thrombosis of unspecified deep veins of unspecified lower extremity |
| I82411 | Acute embolism and thrombosis of right femoral vein |
| I82412 | Acute embolism and thrombosis of left femoral vein |
| I82413 | Acute embolism and thrombosis of femoral vein, bilateral |
| I82419 | Acute embolism and thrombosis of unspecified femoral vein |
| I82421 | Acute embolism and thrombosis of right iliac vein |
| I82422 | Acute embolism and thrombosis of left iliac vein |
| I82423 | Acute embolism and thrombosis of iliac vein, bilateral |
| I82429 | Acute embolism and thrombosis of unspecified iliac vein |
| I82431 | Acute embolism and thrombosis of right popliteal vein |
| I82432 | Acute embolism and thrombosis of left popliteal vein |
| I82433 | Acute embolism and thrombosis of popliteal vein, bilateral |
| I82439 | Acute embolism and thrombosis of unspecified popliteal vein |
| I82441 | Acute embolism and thrombosis of right tibial vein |
| I82442 | Acute embolism and thrombosis of left tibial vein |
| I82443 | Acute embolism and thrombosis of tibial vein, bilateral |
| I82449 | Acute embolism and thrombosis of unspecified tibial vein |
| I82451 | Acute embolism and thrombosis of right peroneal vein |
| I82452 | Acute embolism and thrombosis of left peroneal vein |
| I82453 | Acute embolism and thrombosis of peroneal vein, bilateral |
| I82459 | Acute embolism and thrombosis of unspecified peroneal vein |
| I82461 | Acute embolism and thrombosis of right calf muscular vein |
| I82462 | Acute embolism and thrombosis of left calf muscular vein |
| I82463 | Acute embolism and thrombosis of calf muscular vein, bilateral |
| I82469 | Acute embolism and thrombosis of unspecified calf muscular vein |
| I82491 | Acute embolism and thrombosis of other specified deep vein of right lower extremity |
| I82492 | Acute embolism and thrombosis of other specified deep vein of left lower extremity |
| I82493 | Acute embolism and thrombosis of other specified deep vein of lower extremity, bilateral |
| I82499 | Acute embolism and thrombosis of other specified deep vein of unspecified lower extremity |
| I824Y1 | Acute embolism and thrombosis of unspecified deep veins of right proximal lower extremity |
| I824Y2 | Acute embolism and thrombosis of unspecified deep veins of left proximal lower extremity |
| I824Y3 | Acute embolism and thrombosis of unspecified deep veins of proximal lower extremity, bilateral |
| I824Y9 | Acute embolism and thrombosis of unspecified deep veins of unspecified proximal lower extremity |
| I824Z1 | Acute embolism and thrombosis of unspecified deep veins of right distal lower extremity |
| I824Z2 | Acute embolism and thrombosis of unspecified deep veins of left distal lower extremity |
| I824Z3 | Acute embolism and thrombosis of unspecified deep veins of distal lower extremity, bilateral |
| I824Z9 | Acute embolism and thrombosis of unspecified deep veins of unspecified distal lower extremity |
| I82601 | Acute embolism and thrombosis of unspecified veins of right upper extremity |
| I82602 | Acute embolism and thrombosis of unspecified veins of left upper extremity |
| I82603 | Acute embolism and thrombosis of unspecified veins of upper extremity, bilateral |
| I82609 | Acute embolism and thrombosis of unspecified veins of unspecified upper extremity |
| I82621 | Acute embolism and thrombosis of deep veins of right upper extremity |
| I82622 | Acute embolism and thrombosis of deep veins of left upper extremity |
| I82623 | Acute embolism and thrombosis of deep veins of upper extremity, bilateral |
| I82629 | Acute embolism and thrombosis of deep veins of unspecified upper extremity |
| I82A11 | Acute embolism and thrombosis of right axillary vein |
| I82A12 | Acute embolism and thrombosis of left axillary vein |
| I82A13 | Acute embolism and thrombosis of axillary vein, bilateral |
| I82A19 | Acute embolism and thrombosis of unspecified axillary vein |
| I82B11 | Acute embolism and thrombosis of right subclavian vein |
| I82B12 | Acute embolism and thrombosis of left subclavian vein |
| I82B13 | Acute embolism and thrombosis of subclavian vein, bilateral |
| I82B19 | Acute embolism and thrombosis of unspecified subclavian vein |
| I82C11 | Acute embolism and thrombosis of right internal jugular vein |
| I82C12 | Acute embolism and thrombosis of left internal jugular vein |
| I82C13 | Acute embolism and thrombosis of internal jugular vein, bilateral |
| I82C19 | Acute embolism and thrombosis of unspecified internal jugular vein |
| I82890 | Acute embolism and thrombosis of other specified veins |
| I8290 | Acute embolism and thrombosis of unspecified vein |
Supplementary Table II (online only).
| Risk factor | Caprini points | Padua points | VTE, % (n = 35,557) | No VTE, % (n = 1,216,903 | P-valuec |
|---|---|---|---|---|---|
| Age 41-60 years old | 1 | – | 18.9 | 22.8 | <.001 |
| Planned minor surgery | 1 | – | 1.3 | 2.7 | <.001 |
| Varicose veins | 1 | – | 0.4 | 0.2 | <.001 |
| Recent major surgery | 1 | – | 0.9 | 0.7 | <.001 |
| Inflammatory bowel disease | 1 | – | 3.3 | 3.7 | <.001 |
| Swollen legs | 1 | – | 1.2 | 0.6 | <.001 |
| BMI ≥25 kg/m2 | 1 | – | 68.4 | 70.7 | <.001 |
| Heart attack | 1 | – | 4.0 | 2.9 | <.001 |
| Congestive heart failure | 1 | – | 12.9 | 10.1 | <.001 |
| Serious infection | 1 | – | 22.3 | 15.5 | <.001 |
| Lung disease | 1 | – | 16.8 | 16.2 | .0015 |
| Bed rest <72 hours | 1 | – | 1.3 | 1.4 | .823 |
| Hormone therapy | 1 | 1 | 0.5 | 1.0 | <.001 |
| Pregnancy | 1 | – | 0.0 | 0.0 | .803 |
| History of stillbirth | 1 | – | 0.0 | 0.0 | .530 |
| Age 61-74 years old | 2 | – | 52.5 | 48.0 | <.001 |
| Cancer | 2 | – | 34.7 | 27.7 | <.001 |
| Planned major surgery | 2 | – | 14.3 | 20.4 | <.001 |
| Leg cast or mold | 2 | – | 0.03 | 0.04 | .764 |
| Central venous catheter | 2 | – | 2.3 | 1.1 | <.001 |
| Age ≥75 years old | 3 | – | 25.8 | 22.9 | <.001 |
| Bed rest ≥72 hours | 3 | 3 | 1.4 | 0.6 | <.001 |
| History of VTE | 3 | 3 | 7.4 | 1.5 | <.001 |
| Family history of VTEd | 3 | – | – | – | – |
| History of clotting disorder | 3 | – | 1.7 | 0.6 | <.001 |
| Elective arthroplasty | 5 | – | 3.5 | 3.9 | <.001 |
| Broken hip, pelvis, or leg | 5 | – | 1.6 | 1.1 | <.001 |
| Serious trauma | 5 | – | 0.0 | 0.0 | <.001 |
| Spinal cord injury | 5 | – | 0.2 | 0.1 | .0010 |
| Stroke | 5 | – | 4.1 | 4.1 | .937 |
| BMI ≥30 kg/m2 | – | 1 | 38.7 | 39.5 | .0020 |
| Infection or rheumatologic disorder | – | 1 | 25.7 | 21.1 | <.001 |
| Heart attack and/or stroke | – | 1 | 6.4 | 5.7 | <.001 |
| Heart and/or respiratory failure | – | 1 | 16.8 | 11.5 | <.001 |
| Age ≥70 years | – | 1 | 46.6 | 41.2 | <.001 |
| Recent trauma and/or surgery | – | 2 | 0.9 | 0.7 | <.001 |
| Thrombophilia | – | 3 | 1.6 | 0.4 | <.001 |
| Active cancer | – | 3 | 28.4 | 21.1 | <.001 |
BMI, Body mass index; VTE, venous thromboembolism.
The most recent version of the risk assessment model was used (2013).8
The most recent version of the risk assessment model was used (2010).9
P-values computed with Pearson χ2 test.
Unable to extract this information from the medical record; the item was dropped from the computation of the Caprini score.
References
- 1.Beckman M.G., Hooper W.C., Critchley S.E., Ortel T.L. Venous thromboembolism. Am J Prev Med. 2010;38:S495–S501. doi: 10.1016/j.amepre.2009.12.017. [DOI] [PubMed] [Google Scholar]
- 2.CDC Impact of blood clots on the United States | CDC. Centers for disease Control and prevention. 2018. https://www.cdc.gov/ncbddd/dvt/infographic-impact.html Accessed December 17, 2021.
- 3.Office of the Surgeon General, National Heart Lung and Blood Institute . Office of the Surgeon General; Rockville, MD: 2008. The Surgeon General’s Call to action to prevent deep vein thrombosis and pulmonary embolism. [PubMed] [Google Scholar]
- 4.Woller S. CDC, Joint commission engage in VTE prevention Research. The Joint commission. 2021. https://www.jointcommission.org/resources/news-and-multimedia/blogs/improvement-insights/2021/08/cdc-and-joint-commission-engage-in-vte-prevention-research/ Accessed August 10, 2022.
- 5.Pandor A., Tonkins M., Goodacre S., et al. Risk assessment models for venous thromboembolism in hospitalised adult patients: a systematic review. BMJ Open. 2021;11 doi: 10.1136/bmjopen-2020-045672. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Darzi A.J., Repp A.B., Spencer F.A., et al. Risk-assessment models for VTE and bleeding in hospitalized medical patients: an overview of systematic reviews. Blood Adv. 2020;4:4929–4944. doi: 10.1182/bloodadvances.2020002482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Caprini J.A., Arcelus J.I., Hasty J.H., Tamhane A.C., Fabrega F. Clinical assessment of venous thromboembolic risk in surgical patients. Semin Thromb Hemost. 1991;17:304–312. [PubMed] [Google Scholar]
- 8.Cronin M., Dengler N., Krauss E.S., et al. Completion of the updated Caprini risk assessment model (2013 version) Clin Appl Thromb. 2019;25:1–10. doi: 10.1177/1076029619838052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gerotziafas G.T., Papageorgiou L., Salta S., Nikolopoulou K., Elalamy I. Updated clinical models for VTE prediction in hospitalized medical patients. Thromb Res. 2018;164:S62–S69. doi: 10.1016/j.thromres.2018.02.004. [DOI] [PubMed] [Google Scholar]
- 10.Kahn S.R., Lim W., Dunn A.S., et al. Prevention of VTE in nonsurgical patients. Chest. 2012;141:e195S–e226S. doi: 10.1378/chest.11-2296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Barbar S., Noventa F., Rossetto V., et al. A risk assessment model for the identification of hospitalized medical patients at risk for venous thromboembolism: the Padua Prediction Score: Padua Prediction Score. J Thromb Haemost. 2010;8:2450–2457. doi: 10.1111/j.1538-7836.2010.04044.x. [DOI] [PubMed] [Google Scholar]
- 12.Gould M.K., Garcia D.A., Wren S.M., et al. Prevention of VTE in nonorthopedic surgical patients. Antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2012;141(2 SUPPL):e227S–e277S. doi: 10.1378/chest.11-2297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chamoun N., Matta S., Aderian S.S., et al. A prospective observational cohort of clinical outcomes in medical inpatients prescribed pharmacological thromboprophylaxis using different clinical risk assessment models(COMPT RAMs) Sci Rep. 2019;9 doi: 10.1038/s41598-019-54842-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.von Elm E., Altman D.G., Egger M., et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335:806–808. doi: 10.1136/bmj.39335.541782.AD. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Veterans Affairs Health Administration Providing Health Care for Veterans. US Dep Veterans Aff. 2020. https://www.va.gov/health/ Accessed April 15, 2022.
- 16.U.S. Department of Veterans Affairs VA Informatics and computing infrastructure (VINCI) homepage. 2018. https://www.hsrd.research.va.gov/for_researchers/vinci/ Accessed April 15, 2022.
- 17.DeLong E.R., DeLong D.M., Clarke-Pearson D.L. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–845. [PubMed] [Google Scholar]
- 18.HCUP National Trends in Inpatient Stays . Agency for Healthcare Research and Quality; 2022. Healthcare cost and Utilization Project (HCUP)https://datatools.ahrq.gov/ Accessed August 7, 2022. [PubMed] [Google Scholar]
- 19.HCUP Weighted Summary Statistics Report . Agency for Healthcare Research and Quality; 2022. NIS 2020 Core file means of continuous data elements.https://hcup-us.ahrq.gov/db/nation/nis/tools/stats/MaskedStats_NIS_2020_Core_Weighted_V2.PDF Accessed August 7, 2022. [Google Scholar]
- 20.Loyd C., Blue K., Turner L., et al. National norms for hospitalizations due to ambulatory care sensitive conditions among adults in the US. J Gen Intern Med. 2023:1–7. doi: 10.1007/s11606-023-08161-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hayssen H., Cires-Drouet R., Englum B., et al. Systematic review of venous thromboembolism risk categories derived from Caprini score. J Vasc Surg Venous Lymphat Disord. 2022;10:1401–1409.e7. doi: 10.1016/j.jvsv.2022.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nendaz M., Spirk D., Kucher N., 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;111:531–538. doi: 10.1160/TH13-05-0427. [DOI] [PubMed] [Google Scholar]
- 23.Zhou H., Hu Y., Li X., 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;25:1091–1104. doi: 10.5551/jat.43653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lobastov K., Barinov V., Schastlivtsev I., Laberko L., Rodoman G., Boyarintsev V. 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;4:153–160. doi: 10.1016/j.jvsv.2015.09.004. [DOI] [PubMed] [Google Scholar]
- 25.Shang M.M., Yan R., Wang X.L., Gong W.P., Guo Z.Q. Comparison of 2013 and 2009 versions of Caprini risk assessment models for predicting VTE in Chinese cancer patients: a retrospective study. J Thromb Thrombolysis. 2020;50:446–451. doi: 10.1007/s11239-020-02038-2. [DOI] [PubMed] [Google Scholar]
- 26.Tian B., Li H., Cui S., Song C., Li T., Hu B. A novel risk assessment model for venous thromboembolism after major thoracic surgery: a Chinese single-center study. J Thorac Dis. 2019;11:1903–1910. doi: 10.21037/jtd.2019.05.11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chen X., Pan L., Deng H., et al. Risk assessment in Chinese hospitalized patients comparing the Padua and Caprini scoring algorithms. Clin Appl Thromb. 2018;24(suppl):127S–135S. doi: 10.1177/1076029618797465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.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;23:538–543. doi: 10.1093/icvts/ivw158. [DOI] [PubMed] [Google Scholar]
- 29.Stuck A.K., Spirk D., Schaudt J., Kucher N. Risk assessment models for venous thromboembolism in acutely ill medical patients: a systematic review. Thromb Haemost. 2017;117:801–808. doi: 10.1160/TH16-08-0631. [DOI] [PubMed] [Google Scholar]
- 30.Moumneh T., Riou J., Douillet D., et al. Validation of risk assessment models predicting venous thromboembolism in acutely ill medical inpatients: a cohort study. J Thromb Haemost. 2020;18:1398–1407. doi: 10.1111/jth.14796. [DOI] [PubMed] [Google Scholar]
- 31.Bullock-Palmer R.P., Weiss S., Hyman C. Innovative approaches to increase deep vein thrombosis prophylaxis rate resulting in a decrease in hospital-acquired deep vein thrombosis at a tertiary-care teaching hospital. J Hosp Med. 2008;3:148–155. doi: 10.1002/jhm.291. [DOI] [PubMed] [Google Scholar]
- 32.Flanders S.A., Greene M.T., Grant P., et al. Hospital performance for pharmacologic venous thromboembolism prophylaxis and rate of venous thromboembolism : a cohort study. JAMA Intern Med. 2014;174:1577–1584. doi: 10.1001/jamainternmed.2014.3384. [DOI] [PubMed] [Google Scholar]
- 33.Grant P.J., Greene M.T., Chopra V., Bernstein S.J., Hofer T.P., Flanders S.A. Assessing the Caprini score for risk assessment of venous thromboembolism in hospitalized medical patients. Am J Med. 2016;129:528–535. doi: 10.1016/j.amjmed.2015.10.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Cushman M. Epidemiology and risk factors for venous thrombosis. http://www.nhlbisupport.com/bmi Accessed June 21, 2022. [DOI] [PMC free article] [PubMed]
- 35.Abdollahi M., Cushman M., Rosendaal F.R. Obesity: risk of venous thrombosis and the interaction with coagulation factor levels and oral contraceptive use. Thromb Haemost. 2003;89:493–498. [PubMed] [Google Scholar]
- 36.Stein P.D., Beemath A., Olson R.E. Obesity as a risk factor in venous thromboembolism. Am J Med. 2005;118:978–980. doi: 10.1016/j.amjmed.2005.03.012. [DOI] [PubMed] [Google Scholar]
- 37.Crous-Bou M., Harrington L.B., Kabrhel C. Environmental and genetic risk factors associated with venous thromboembolism. Semin Thromb Hemost. 2016;42:808–820. doi: 10.1055/s-0036-1592333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Sutzko D.C., Obi A.T., Kamdar N., et al. Low to moderate risk non-orthopedic surgical patients do not benefit from VTE chemoprophylaxis. Ann Surg. 2022;276:e691–e697. doi: 10.1097/SLA.0000000000004646. [DOI] [PubMed] [Google Scholar]