Journal of Atherosclerosis and Thrombosis 　Vol. 21, No.

3 261

Original Article

Validation of a Venous Thromboembolism Risk Assessment Model

in Hospitalized Chinese Patients: A Case-Control Study

Haixia Zhou 1, Lan Wang 1, Xiaoling Wu 1, Yongjiang Tang 1, Jing Yang 1, Bo Wang 1, Yu Yan 2, Binmiao Liang 1,
Ke Wang 1, Xuemei Ou 1, Maoyun Wang 1, Yulin Feng 1 and Qun Yi 1

1
Department of Respiratory Medicine, West China Hospital, Sichuan University, Chengdu, China
2
Department of Pharmacy, West China Hospital, Sichuan University, Chengdu, China

Aim: The appropriate selection of hospitalized patients for venous thromboembolism (VTE) prophy-
laxis is an important unresolved issue. We sought to validate the Caprini model, a famous individual
VTE risk assessment model (RAM), in hospitalized Chinese patients.
Methods: We performed a retrospective case-control study among unselected hospitalized patients
admitted to a comprehensive hospital in China. A total of 347 patients were confirmed to have VTE
during hospitalization, and 651 controls were randomly selected to match the patients according to
medical service. Both the patients and controls were retrospectively assessed for the risk of VTE using
the Caprini RAM.
Results: The average Caprini cumulative risk score in the patients was significantly higher than that
observed in the controls (4.69±2.58 vs 3.16±1.82, p ＜ 0.0001). Compared with that observed in the
low-risk group, a classification of high-risk according to the Caprini model was associated with a 1.65-
fold increased risk of VTE (95%CI 1.05-2.61), while that of highest-risk was associated with a 4.84-
fold increased risk of VTE (95%CI 3.06-7.64). After further stratifying the highest risk level with a
cumulative risk score of ≥ 5 into scores of 5-6, 7-8 and ≥ 9, the patients with a score of 5-6 were
found to exhibit a 3.33-fold increased risk of VTE (95%CI 2.06-5.40), those with a score 7-8 exhib-
ited a 9.41-fold increased risk of VTE (95%CI 4.90-18.08) and those with a score of ≥ 9 exhibited a
24.69-fold (95%CI 7.98-76.40) increased risk of VTE compared with their low-risk counterparts.
Conclusions: Our study suggests that the Caprini RAM can be used to effectively stratify hospitalized
Chinese patients into VTE risk categories based on individual risk factors. The classification of the
highest risk level with a cumulative risk score of ≥ 5 provided significantly more clinical information,
and further stratification of this group of patients is needed.

J Atheroscler Thromb, 2014; 21:261-272.

Key words: Caprini risk assessment model, Deep venous thrombosis, Pulmonary embolism,
Venous thromboembolism, Padua prediction score

patients. In the absence of appropriate VTE prophy-

Introduction
Venous thromboembolism (VTE), including
pulmonary embolism (PE) and deep venous thrombo-
sis (DVT), is a serious and potentially fatal complica-
tion that is common in both medical and surgical
Address for correspondence: Qun Yi, Respiratory Department,
West China Hospital, Sichuan University, Guo-xue-xiang 37#,
Wuhou district, Chengdu 610041, Sichuan Province, China
E-mail: yiqun925@126.com
Received: August 5, 2013
Accepted for publication: September 18, 2013
laxis, the reported incidence of VTE ranges from 10%
to 40% in unselected hospitalized patients 1). Not only
is VTE the third most common cause of all hospital-
related deaths 2), it is also the most common prevent-
able cause of hospital death 3). Despite substantial evi-
dence that VTE prophylaxis can reduce the incidence
of VTE, prophylactic treatment remains largely unde-
rused in hospitalized patients 4-7). Accurately assessing
the risk of VTE is critical for improving this situation
and increasing compliance with prophylaxis guidelines.
Several VTE risk assessment models (RAMs)
262
have been proposed and evaluated clinically 8-13). Gen-
erally, these models can be classified as involving two
types of strategies: group risk assessment and individ-
ual risk assessment. Most recent publications have
concluded that it may be more appropriate to use the
individual risk assessment strategy to identify and
evaluate all possible risk factors in order to determine
the true extent of risk for a given patient and provide
appropriate suggestions for prophylactic therapies
according to the risk level 8, 14-16). Among these individ-
ual VTE RAMs, the Caprini model is the most nota-
ble. It has been adopted by many individuals and
organizations 17-19) and has already been validated in
several studies 18, 20-24).
However, most of these validation studies were
conducted in surgical patients, despite the fact that
the Caprini RAM was designed for use in both medi-
cal and surgical patients. There are limited data regard-
ing unselected hospitalized patients who are admitted
for either medical or surgical reasons. In addition, all
of the previous validation studies were conducted in
non-Asian populations. Given the racial differences
and differences in medical environments, conducting
validation studies in Chinese patients is warranted
before this model is adopted in clinical practice in
China. Our previous study preliminarily suggested
that the Caprini RAM is a practical and effective tool
for assessing the risk of VTE among unselected Chi-
nese inpatients when compared with two other VTE
RAMs, the Kucher model and the Padua Prediction
Score. However, the lack of a control group without
VTE in that study made it difficult to draw firm con-
clusions 25).
The primary aim of the present study was to vali-
date the Caprini risk assessment model in unselected
hospitalized Chinese patients using a large retrospec-
tive study of VTE patients versus adequate controls.
In addition, we sought to gain insight into the risk
factors of VTE in the same population.
Patients and Methods
Study Design and Patients
We performed a case-control study of unselected
hospitalized patients admitted to a 4,300-bed compre-
hensive teaching hospital in China over a two-year
period. The study was approved by the institutional
review board of the hospital.
We identified all cases of VTE diagnosed after
admission from all medical and surgical departments
of the hospital between January 2010 and December
2011. The methods and criteria used to include VTE
patients were reported in a previous study 25). There
Zhou et al .
were two sources for enrolling VTE patients: the VTE
Registration Center and the Information Center of
the hospital. The VTE Registration Center was estab-
lished in 2009 by the hospital in order to include VTE
patients in studies, and we also searched information
obtained from the hospital’s Information Center in
order to include VTE patients who may have been
missed by the former source. Hospital ICD-10 codes
for VTE (I26, I80, I82) were used to select cases for
review. The inclusion criteria were as follows: con-
firmed VTE (DVT and/or PE), an age of ≥ 18 years
and a ≥ two-day duration of hospitalization. The exclu-
sion criteria were as follows: VTE on admission, throm-
bosis in a location other than the deep veins of the legs
or arms or a coding error. The diagnosis of DVT was
validated based on positive compression ultrasonogra-
phy, contrast venography or autopsy findings. The
diagnosis of PE was validated based on positive pul-
monary angiogram, spiral computed tomography, high
probability ventilation/perfusion scanning or autopsy
results.
The controls were randomly selected from
among all hospitalized adults (age ≥ 18 years) admitted
to the same departments during the same period as
the patients, without an ICD-10 code for thrombosis.
The controls were frequency matched to the patients
at a ratio of 2:1. The exclusion criteria for the controls
were as follows: a coding error (the occurrence of
DVT and/or PE), a ＜ two-day duration of hospitaliza-
tion or data unavailability.
Risk Assessment Model (RAM)
Caprini and colleagues have used a detailed indi-
vidual risk assessment model in medical and surgical
patients since the late 1980s 11). We adopted the recently
modified version of the Caprini RAM 18), which is also
the version adopted by the most recent update of the
American College of Chest Physicians (ACCP) guide-
lines for VTE prophylaxis 19). The model used a point-
scoring system, and approximately 40 risk factors were
listed with weights of 1 to 5 points each. The relative
scores for individual risk factors were summed to pro-
duce a cumulative risk score that defined the patient’s
risk level: low-risk (0-1), moderate-risk (2), high-risk
(3-4) or highest-risk (≥ 5), each with a recommended
prophylactic regimen. Each patient recruited was ret-
rospectively assessed for the risk of VTE using this
model. Since the ACCP guidelines 26) also adopted
another RAM, the Padua Prediction Score, to estimate
the baseline risk of VTE in hospitalized medical
patients, we additionally compared the Caprini model
with the Padua Prediction Score in this subgroup of
patients. In the Padua Prediction Score assessment,
 Caprini Model in Chinese Inpatients
the VTE risk profile was calculated using 11 common
risk factors for VTE in hospitalized medical patients.
Each risk factor was weighted according to a point
scale, and a high risk of VTE was defined as a cumula-
tive score of ≥ 4 points, while a low risk of VTE was
defined as a cumulative score of ＜ 4 points. Each hos-
pitalized medical patient recruited in this study was
additionally retrospectively assessed for the risk of
VTE using this RAM. The risk factors identified in
both RAMs and the assigned points are shown in Sup-
plementary Table 1.
Standardized case report forms were used to
record patient demographics, diagnoses, risk factors
for VTE in the Caprini model, surgical information,
the length of hospital stay, mortality, results of relevant
laboratory tests and imaging examinations and infor-
mation regarding VTE prophylaxis, etc. by physicians
who received in-depth training to ensure data reliabil-
ity.
Prophylaxis was defined as the administration of
any mechanical or pharmacological prophylaxis for
preventive purposes (not treatment).
Data Analysis
Continuous variables with a normal distribution
are presented as means with standard deviations, and
group comparisons of such variables were made using
the t -test. Continuous variables with a skewed distri-
bution are presented as median values with interquar-
tile ranges, and group comparisons of such variables
were made using the non-parametric test. Discrete
variables are presented as frequencies and percentages,
and group comparisons of such variables were made
using the chi-square or rank-sum test.
A logistic regression analysis was used to calcu-
lated crude and adjusted odds ratios (ORs) for VTE
according to the risk levels and individual risk factors.
ORs and the 95% confident intervals (95% CIs) are
reported. We selected variables for inclusion in the
final multivariate model that had both an OR of ≥ 2.0
and a p-value of ≤ 0.1 in a univariate crude analysis.
The use of VTE prophylaxis was included in the final
model to adjust for potential confounding introduced
by its use.
All reported p-values are two-tailed. A value of
p ＜ 0.05 was considered to be statistically significant.
The data were analyzed using the SPSS software pack-
age version 18.0 (SPSS Inc., Chicago, IL, USA).
263
Results
Characteristics of the VTE Patients and Control
Subjects
A total of 500 patients were initially identified as
exhibiting VTE; 153 patients were excluded due to
having VTE on admission (89), an age of ＜ 18 years
(3), a hospital stay of ＜ two days (45), thrombosis in
a location other than the deep veins of the legs or arms
(14) or a coding error (2). Ultimately, 347 patients
were confirmed to have VTE during hospitalization
and were included in our study. A total of 214
(61.7%) of these VTE patients were hospitalized on
the medical service and 133 (39.3%) were hospitalized
on the surgical service (Supplementary Table 2).
Among the 347 VET patients, 92 (26.5%) had PE,
173 (49.9%) had DVT and 82 (23.6%) were diag-
nosed with both. A total of 688 control subjects were
randomly selected and frequency matched to the
patients at a ratio of 2:1 by the admitting services
(Supplementary Table 2). Of the 688 controls
selected, information for 651 individuals was com-
plete and available for review.
There were no significant differences between the
patients and controls in terms of age, gender, smoking
history, hematocrit, white blood count, platelet count
on admission or body mass index (BMI). However,
the percentage of patients with a BMI of ≥ 25 was sig-
nificantly greater than that observed in the control
subjects (p = 0.001). The patients were 1.61-fold more
likely to die than the controls; however, the difference
was not statistically significant (95% CI, 0.93-2.78;
p = 0.087). The patients exhibited a longer length of
hospital stay and reduced rate of prophylaxis against
VTE compared to the controls (p = 0.001 and 0.012,
respectively) (Table 1).
Associations between the Caprini Risk Levels and
the Risk of VTE
The distribution of the hospitalized patients
according to the Caprini risk level among the patients
and controls is shown in Fig. 1. The average Caprini
cumulative risk score in the patients was significantly
higher than that observed in the controls (4.69±2.58
vs 3.16±1.82, p ＜ 0.0001). According to the Wil-
coxon rank-sum test, the difference in the predictive
effect of the Caprini RAM between the patients and
controls was statistically significant (p ＜0.0001), which
means that the Caprini RAM can be used to more eas-
ily classify individuals into much higher risk levels
among patients than among control subjects without
VTE.
Compared with that observed in the low-risk
264 Zhou et al .

Table 1. Characteristics of the VTE patients and controls

Characteristic Cases (n = 347) Controls (n = 651) p value
Male gender 192 (55.3) 382 (58.7) 0.308
Age (years) 56.5±17.1 56.2±18.1 0.322
BMI (kg/m2)＊ 22.9±3.3 22.1±3.3 0.861
BMI ＞ 25 63 (27.3) 70 (16.4) 0.001
BMI ＞ 30 3 (1.3) 9 (2.1) 0.555
Smoking history
Nonsmokers 216 (62.2) 413 (63.4)
Former smokers 53 (15.3) 106 (16.3) 0.699
Active smokers 78 (22.5) 132 (20.3)
Hematocrit (%)† 36.2±7.2 36.3±7.3 0.475
White blood count (×103 mm−3)† 7.8±3.7 7.7±5.0 0.460
Platelet count (×103 mm−3)† 196.4±100.6 185.8±94.4 0.104
Mortality during hospitalization 25 (7.2) 30 (4.6) 0.087
Length of hospital stay (days) 12 (8-21) 10 (6-17) 0.001
Patients receiving prophylaxis 38 (11.0) 110 (16.9) 0.012
The data are presented as the number of patients (%), mean±SD, median (interquartile range)
BMI = body mass index
＊
Available in 231 patients, 428 controls.
†
Available in 329 patients, 607 controls.

group, a classification of moderate risk was associated

with a 1.43-fold increased risk of VTE; however, the
difference was not statistically significant (95% CI
0.84-2.41). Meanwhile, a high risk was associated
with a 1.65-fold increased risk of VTE (95% CI 1.05-
2.61) and the highest risk was associated with a 4.84-
fold increased risk of VTE (95% CI 3.06-7.64) after
adjusting for VTE prophylaxis use. Within the highest
risk level with a cumulative risk score of 5 or more,
the risk of VTE increased almost linearly in associa-
tion with an increase in the cumulative risk score
(Table 2, Fig. 2). Compared with the low-risk group,
the patients with a Caprini score of 5-6 exhibited a
3.33-fold increased risk of VTE (95% CI 2.06-5.40),
those with a Caprini score of 7-8 exhibited a 9.41-fold
increased risk of VTE (95% CI 4.90-18.08) and those
with a Caprini score of ≥ 9 exhibited a 24.69-fold
increased risk of VTE (95% CI 7.98-76.40) after
adjusting for VTE prophylaxis use. With an increase
of 1 risk score within the highest risk level, the OR of
VTE increased 1.64 fold (95% CI 1.35-2.01, p ＜
0.0001, data not shown).
Fig. 1. Percentage of subjects classified as having different risk
levels according to the Caprini RAM among the
Comparison of the Caprini RAM and Padua Predic- patients and controls.
tion Score Results Among the Hospitalized Medi-
cal Patients
Table 3 compares the Caprini RAM and Padua cumulative risk score in the patients was significantly
Prediction Score results among the hospitalized medi- higher than that observed in the controls (average
cal patients. In both RAMs, the average or median score: 4.92 vs 3.28, p ＜ 0.0001 for Caprini RAM;
 Caprini Model in Chinese Inpatients 265

Table 2. Associations between the Caprini risk levels and risk of VTE
Cases Controls Adjusted OR
OR for VTE
Risk level (n = 347), (n = 651), p value for VTE
(95% CI)
n (%) n (%) (95% CI)‡
＊
Low risk (0-1) 31 (8.9) 120 (18.4) ＜ 0.0001 1.00 (Reference) 1.00 (Reference)
Moderate risk (2) 44 (12.7) 125 (19.2) 1.36 (0.81-2.30) 1.43 (0.84-2.41)
High risk (3-4) 105 (30.3) 263 (40.4) 1.55 (0.98-2.44) 1.65 (1.05-2.61)
Highest risk (≥ 5) 167 (48.1) 143 (22.0) 4.55 (2.89-7.17) 4.84 (3.06-7.64)
Highest risk with Score 5-6 95 (27.4) 117 (18.0) − 3.14 (1.95-5.07) 3.33 (2.06-5.40)
Highest risk with Score 7-8 47 (13.5) 22 (3.4) − 8.27 (4.35-15.72) 9.41 (4.90-18.08)
Highest risk with score ≥ 9 25 (7.2) 4 (0.6) − 24.19 (7.84-74.67) 24.69 (7.98-76.40)
†
Average Caprini cumulative risk score, mean±SD 4.69±2.58 3.16±1.82 ＜ 0.0001 − −

VTE = venous thromboembolism, OR = odds ratio, CI = confidence interval, SD = standard deviation

＊
Wilcoxon rank-sum test
†
Independent-samples t -test,
‡
Adjusted for the use of VTE prophylaxis

median score: 2 vs 1, p ＜ 0.0001 for the Padua Predic-

tion Score). After adjusting for the use of VTE pro-
phylaxis, a classification of high-highest risk according
to the two RAMs was associated with similar ORs for
VTE (3.01 (95% CI 2.03-4.48) for the Caprini RAM
vs 2.90 (95% CI 1.93-4.37) for the Padua Prediction
Score) compared with the low-moderate risk classifica-
tion. However, the Caprini RAM identified 82.3% of
VTE patients as having a high or highest risk, while
the Padua Prediction Score identified only 30.1% of
VTE patients as being at high risk.

Effects of Caprini Risk Factors on the Risk of VTE

in the Study Population
Of the risk factors listed in the Caprini RAM,
heart failure (OR = 3.43; 95% CI 1.67-7.06, p = 0.001),
obesity (BMI ＞ 25) (1.98; 1.34-2.91, p = 0.001), severe
lung disease (2.25; 1.62-3.12, p ＜ 0.0001), swollen
legs (current) (12.12; 8.64-17.00, p ＜ 0.0001), central
venous access (5.71; 1.15-28.44, p = 0.033), being
confined to a bed (＞ 72 h) (6.22; 3.85-10.03, p ＜
0.0001), a history of DVT/PE (17.75; 4.10-76.98, p ＜ Fig. 2. Percentage of subjects with different Caprini risk scores
0.0001) and hip, pelvis or leg fractures ( ＜1 mo) (3.52; among the patients and controls.
1.29-9.60, p = 0.014) were found to be significantly
associated with an increased risk of in-hospital VTE in
a univariate analysis (Table 4). The occurrence of (OR = 0.77; 95% CI 0.56-1.07, p = 0.115, data not
major surgery ( ＞ 45 min) during the current hospital- shown).
ization was associated with a decreased risk of VTE In the multivariate logistic regression analysis, we
(0.45; 0.28-0.72, p = 0.001). However, when we ana- included risk factors with an OR of ≥ 2.0 and a
lyzed all patients who underwent major surgery within p-value of ≤ 0.1 in the univariate analysis. A control
one month, including those who underwent major variable for VTE prophylaxis was also included in this
surgery during the current hospitalization and those regression model. The multivariate analysis identified
with a history of prior major surgery, there were no acute myocardial infarction (OR = 16.11; 95% CI
associations between this factor and the risk of VTE 1.45-179.22, p = 0.024), obesity (BMI ＞ 25) (2.02;
266
Table 3. Comparison of the Caprini RAM and Padua Prediction Score among the hospitalized medical patients
Caprini RAM＊
RAMs
Patients with
high-highest risk, %
Cases (n = 226) 82.3%
Controls (n = 402) 60.4%
† ‡
p value ＜ 0.0001
Adjusted OR for VTE (95%CI)¶ 3.01 (2.03-4.48)
RAM = risk assessment model, SD = standard deviation, OR = odds ratio, VTE = venous thromboembolism, CI = confidence interval
＊
According to the Caprini RAM, a high-highest risk means a cumulative risk score of ≥ 3; according to the Padua Prediction Score, a high risk
means a cumulative risk score of ≥ 4.
†
Chi-square
‡
Independent-samples t -test
§
Non-parametric test
¶
Adjusted for the use of VTE prophylaxis
1.27-3.20, p = 0.003), severe lung disease (2.71; 1.82-
4.03, p ＜ 0.0001), swollen legs (current) (12.64; 8.76-
18.24, p ＜ 0.0001), being confined to a bed ( ＞ 72 h)
(7.37; 4.19-12.96, p ＜ 0.0001), a history of DVT/PE
(20.72; 3.88-110.51, p ＜ 0.0001) and hip, pelvis or
leg fractures ( ＜ 1 mo) (3.24; 1.01-10.43, p = 0.049) to
be significantly associated with VTE, while heart fail-
ure (2.42; 0.96-6.14, p = 0.063) and central venous
access (5.28; 0.81-34.53, p = 0.083) were only margin-
ally significant (Table 4).
Discussion
VTE is a common and severe complication among
hospitalized patients, including medical and surgery
patients, with many risk factors contributing to the
disease. However, knowledge regarding which patients
are at a high risk of in-hospital VTE and the adminis-
tration of prophylaxis in at-risk patients are both sub-
optimal. The use of a validated risk assessment model
(RAM) for assessing hospitalized patients may improve
the current situation. In this study, we validated that
the Caprini RAM can be used to stratify the risk of
VTE in unselected hospitalized patients based on indi-
vidual risk factors. We found that increases in the
Caprini risk level and cumulative risk score were asso-
ciated with an increased risk of in-hospital VTE; out
of the four risk levels determined using the RAM, the
classification of the highest risk level provided signifi-
cantly more clinical information for identifying at-risk
hospitalized patients. Compared with the low-risk
classification, a classification of highest risk was associ-
ated with a 4.84-fold increased risk of VTE. After fur-
ther stratifying the patients with the highest risk (a
Zhou et al .
The Padua Prediction Score＊
Cumulative risk score,
Cumulative risk score, Patients with
median (interquartile
mean±SD high risk, %
range)
4.92±2.64 30.1% 2 (1-4)
3.28±1.95 12.7% 1 (0-2)
† §
＜ 0.0001 ＜ 0.0001 ＜ 0.0001
− 2.90 (1.93-4.37) −
cumulative risk score of ≥ 5) into groups with 5-6, 7-8
and ≥ 9 risk levels, the increase in the odds ratio for
VTE in these patients accelerated as the cumulative
risk score increased. Compared with the low-risk
group, the patients with a Caprini score of 5-6 exhib-
ited a 3.33-fold increased risk of VTE (95% CI 2.06-
5.40), the patients with a Caprini score of 7-8 exhib-
ited a 9.41-fold increased risk of VTE (95% CI 4.90-
18.08) and the patients with a Caprini score of ≥ 9
exhibited a 24.69-fold (95% CI 7.98-76.40) increased
risk of VTE after adjusting for VTE prophylaxis use.
These findings are consistent with the results of a
previous validation study of the Caprini RAM in sur-
gical hospitalized patients 18) that demonstrated a link
between the individual patient risk assigned by the
Caprini RAM and the VTE outcome observed within
30 days after surgery. The incidence of VTE was asso-
ciated with an increased risk level: highest risk, 1.94%;
high risk, 0.97%; moderate, 0.70%; low, 0%. The
authors found that the difference in the incidence of
VTE between the high and highest risk levels was sta-
tistically significant (p ＜ 0.001), while the differences
between the low and moderate risk levels and the
moderate and high risk levels were not statistically sig-
nificant. After further disaggregating the patients in
the highest risk category, the authors observed that the
differences in the incidence of VTE among the patients
with a Caprini score of 5 to 6 (1.3%), 7 to 8 (2.6%)
and ≥ 9 (6.5%) were statistically significant and that
the increase in the incidence of acquired VTE appeared
to accelerate in accordance with the cumulative risk
score. Similarly, Pannucci et al.23) validated the Cap-
rini model in plastic surgery patients with a Caprini
score of more than 3 and found that, compared with
 Caprini Model in Chinese Inpatients 267

Table 4. Distribution of VTE risk factors in the Caprini model in the patients and controls, with crude and adjusted ORs for VTE
Relative Cases Controls Crude OR Adjusted OR＊
Risk factor p value p value
Risk Score (n = 347) (n = 651) (95% CI) (95% CI)

Age, ＜ 41 (not a risk factor) − 18.7% 23.2% 0.77 (0.56-1.06) 0.113

Age, 41-60 1 36.0% 32.3% 1.20 (0.91-1.57) 0.197
Age, 61-74 2 30.3% 26.7% 1.19 (0.89-1.59) 0.237
Age, 75＋ 3 15.0% 17.8% 0.81 (0.57-1.16) 0.255
Acute myocardial infarction 1 1.2% 0.2% 7.58 (0.84-68.08) 0.071 16.11 (1.45-179.22) 0.024
Heart failure 1 6.1% 1.8% 3.43 (1.67-7.06) 0.001 2.42 (0.96-6.14) 0.063
Varicose veins 1 2.9% 5.4% 0.52 (0.26-1.07) 0.075
Obesity (BMI ＞ 25)† 1 27.3% 16.1% 2.00 (1.34-2.91) 0.001 2.02 (1.27-3.20) 0.003
Inflammatory bowel disease 1 1.2% 0.8% 1.51 (0.40-5.65) 0.543
Sepsis (＜ 1 mo) 1 0.9% 0.5% 1.89 (0.38-9.38) 0.440
COPD or abnormal pulmonary function 1 9.5% 6.9% 1.42 (0.89-2.26) 0.147
Severe lung disease, including pneumonia (＜ 1 mo) 1 26.5% 13.8% 2.25 (1.62-3.12) ＜ 0.0001 2.71 (1.82-4.03) ＜ 0.0001
Oral contraceptives or hormone replacement therapy 1 0.6% 0.6% 0.94 (0.17-5.15) 0.941
Pregnancy or postpartum (＜ 1 mo) 1 1.4% 0.6% 2.37 (0.63-8.86) 0.202
History of unexpected stillborn infant, recurrent
spontaneous abortion (≥ 3), premature birth with 1 0.3% 0.5% 0.64 (0.07-6.02) 0.684
toxemia or growth-restricted infant
Medical patient currently at bed rest 1 11.2% 15.1% 0.72 (0.48-1.06) 0.096
Minor surgery planned 1 5.5% 8.8% 0.60 (0.35-1.03) 0.065
History of prior major surgery (＜ 1 mo) 1 12.1% 9.2% 1.36 (0.89-2.06) 0.153
Swollen legs (current) 1 55.6% 9.4% 12.12 (8.64-17.0) ＜ 0.0001 12.64 (8.76-18.24) ＜ 0.0001
Central venous access 2 1.7% 0.3% 5.71 (1.15-28.44) 0.033 5.28 (0.81-34.53) 0.083
Arthroscopic surgery 2 0.0% 0.0% − −
Major surgery ( ＞ 45 min) 2 6.9% 14.1% 0.45 (0.28-0.72) 0.001
Malignancy (present or previous) 2 13.5% 13.1% 1.04 (0.71-1.53) 0.828
Laparoscopic procedure ＞ 45 min 2 0.3% 0.2% 1.88 (0.12-30.13) 0.656
Patient confined to bed ( ＞ 72 h) 2 19.9% 3.8% 6.22 (3.85-10.03) ＜ 0.0001 7.37 (4.19-12.96) ＜ 0.0001
Immobilizing plaster cast (＜ 1 mo) 2 0.3% 0.2% 1.88 (0.12-30.13) 0.656
History of DVT/PE 3 5.2% 0.3% 17.75 (4.10-76.98) ＜ 0.0001 20.72 (3.88-110.51) ＜ 0.0001
Positive Factor V Leiden; positive prothrombin
3 − − − −
G20210A; elevated serum homocysteine‡
§
Positive Lupus anticoagulant 3 1.7% 1.8% 0.97 (0.06-15.82) 0.980
Heparin-induced thrombocytopenia (HIT) 3 0.9% 0.0% − 0.042
Family history of VTE 3 1.2% 0.2% 7.58 (0.84-68.08) 0.071 6.93 (0.51-93.62) 0.145
Elevated anticardiolipin antibodies¶ 3 20% 15.4% 1.38 (0.26-7.34) 0.709
Stroke (＜ 1 mo) 5 1.4% 0.5% 3.16 (0.75-13.29) 0.117
Multiple trauma (＜ 1 mo) 5 1.2% 0.5% 2.52 (0.56-11.32) 0.228
Elective major lower extremity arthroplasty 5 0.0% 0.3% − 0.546
Hip, pelvis, or leg fracture (＜ 1 mo) 5 1.2% 0.9% 3.52 (1.29-9.60) 0.014 3.24 (1.01-10.43) 0.049
Acute spinal cord injury (paralysis) (＜ 1 mo) 5 0.6% 0.2% 3.77 (0.34-41.70) 0.279
VTE prophylaxis − 11.0% 16.9% 0.61 (0.41-0.90) 0.012 0.50 (0.30-0.85) 0.010

OR = odds ratio, VTE = venous thromboembolism, CI = confidence interval, BMI = body mass index, mo = month
＊
Variables with both an OR of ≥ 2.0 and a p-value of ≤ 0.1 in the univariate crude analysis were selected for inclusion in the final multivariate model. VTE prophylaxis was also
included.
†
Available in 231 patients, 428 controls.
‡
These risk factors could not be tested in the hospital.
§
Available in 59 patients, 57 controls.
¶
Available in 45 patients, 13 controls.

patients with a Caprini score of 3 to 4, patients with a unselected hospitalized patients are fully consistent
Caprini score of 7 to 8 or ≥ 9 were significantly more with those observed in validation studies carried out
likely to develop VTE (OR 20.9, p ＜ 0.001; OR 4.5, among surgical inpatients and suggest that the Caprini
p = 0.04; respectively). Our findings obtained in RAM can be used to effectively stratify patients with
268 Zhou et al .
respect to the risk of VTE based on each patient’s
unique risk factors and that it is necessary to further
stratify patients previously lumped into the same
“highest risk” category in order to determine the true
extent of risk.
In this study, more than half (61.7%) of the VTE
events occurred in the patients hospitalized on the
medical service, which is in keeping with the previous
literature 27, 28). Recently published ACCP guidelines
for administering VTE prophylaxis among hospital-
ized medical patients adopted the Padua Prediction
Score to assess the baseline risk in this group of
patients with a low or high risk of VTE 26); hence, we
additionally compared the Caprini RAM with this
scoring system among a subgroup of patients. Although
the risk for VTE associated with the high/highest risk
classification determined according to both RAMs was
similar, the Caprini model exhibited greater sensitivity
in identifying high-risk individuals among the hospi-
talized medical patients evaluated in this study (82.3%
of the VTE patients were classified as having a high/
highest risk according to the Caprini RAM versus
30.1% according to the Padua Prediction Score). This
difference may be attributed to the characteristics of
the Padua Prediction Score, which is based on 11 risk
factors, compared with the more than 30 risk factors
used in the Caprini model. Patients identified as being
at low risk may have one or more risk factors that
failed to be identified in this model. In addition, the
classification of the four risk levels determined using
the Caprini RAM may provide more information for
clinicians to guide decision-making for VTE prophy-
laxis based on the patient’s risk level, compared with
the use of a binary low/high risk classification.
Of the risk factors listed in the Caprini RAM, we
found that acute myocardial infarction, obesity (BMI
＞ 25), severe lung disease, swollen legs (current), being
confined to a bed ( ＞ 72 h), a history of DVT/PE and
hip, pelvis or leg fractures ( ＜ 1 mo) were associated
with an increased risk of VTE among unselected hos-
pitalized patients after controlling for the use of VTE
prophylaxis. Heart failure and central venous access
were marginally significantly associated with the risk
of VTE. All of these factors are well-recognized risk
factors associated with the development of VTE 29-31).
However, as pointed out by Anderson and cowork-
ers 31), these risk factors are not of equal weight. We
observed that the strongest risk factor, with an odds
ratio of ＞ 20, was a history of VTE, followed by acute
myocardial infarction and swollen legs (current), with
ORs of ＞ 10, and obesity (BMI ＞ 25), severe lung dis-
ease, being confined to a bed ( ＞ 72 h), heart failure,
central venous access and hip, pelvis or leg fractures
( ＜ 1 mo), with ORs of 2-9. The odds ratios of some
factors, notably a history of VTE, myocardial infarc-
tion and swollen legs (current), were higher than their
assigned relative weight in the Caprini RAM. While
the relative weights in the Caprini RAM were estab-
lished and revised based on results obtained from clin-
ical trials in Western populations 18), our findings
reflect the VTE-related characteristics of hospitalized
Chinese patients.
Some established risk factors for VTE included
in the Caprini RAM, such as major surgery, multiple
episodes of trauma, age, malignancy, varicose veins,
anticardiolipin antibodies, etc., were not found to be
statistically significantly associated with VTE in the
present study. One possible explanation for this find-
ing may be that some of these factors are not impor-
tant determinants of VTE in the Chinese population.
Insufficient power is another possible reason, espe-
cially for thrombophilia-related factors, data for which
were only available for a small number of patients. It
is surprising to note that major surgery, a well-estab-
lished strong risk factor for VTE, was not found to be
associated with VTE in the present study. However, in
a further analysis, we found that major surgery was sig-
nificantly associated with the use of VTE prophylaxis
(OR = 5.72; 3.75-8.73, p ＜ 0.0001, data not shown).
While VTE prophylaxis has been proven to reduce the
risk of VTE, the association between major surgery
and VTE prophylaxis inevitably dilutes the association
between VTE prophylaxis and the risk of VTE.
This study is associated with several limitations.
First, the risk factors were identified in a retrospective
manner that was unable to identify all patient risk fac-
tors, and screening hospitalized patients for asymp-
tomatic VTE was not routinely performed in our hos-
pital, which may have resulted in the failure to detect
the incidence of VTE in some control subjects, espe-
cially those classified as being at high or highest risk.
Second, matching the controls to the patients accord-
ing to the medical service added validity to the study
in some respects, although it made it difficult to evalu-
ate various risk factors, such as the presence of cancer
(Oncology Department) and varicose veins (Vascular
Surgery Department). Third, as mentioned previ-
ously 25), although some patients in the postoperative
period following obstetric or gynecological surgery
were included in our study, there is no obstetrics or
gynecology department at our hospital. Therefore, the
VTE population evaluated in this study may be less
representative of all VTE hospitalized patients, and
our study results must be interpreted with caution in
these patient categories.
To our knowledge, this study is the first to vali-
 Caprini Model in Chinese Inpatients
date the Caprini VTE risk assessment model in
unselected hospitalized patients using a large retro-
spective case-control study. We found that stratifying
individuals according to the Caprini risk level can be
used to effectively discriminate the risk of VTE in
hospitalized patients, whether using an “intro-com-
parison” among the different risk levels within the
model or an “inter-comparison” with other RAMs. In
this study, the classification of the highest risk level
with a cumulative risk score of ＞ 5 provided signifi-
cantly more clinical information for identifying
unselected hospitalized patients who were at risk of
VTE and would benefit most from VTE prophylaxis,
compared with those at any other distinct risk level.
Our results showed that further stratifying this group
of patients to determine the true extent of risk is nec-
essary in order to provide appropriate prophylaxis.
However, future multicenter and prospective valida-
tion studies of the use of this model in hospitalized
Chinese patients are needed.
Acknowledgments
This study was supported by the National Science
and Technology Pillar Program of China in the Elev-
enth Five-year Plan Period (grant 2006BAI01A06).
We greatly appreciate the collaboration received
from the staff of the Respiratory Department, VTE
Registration Center and Information Center of West
China Hospital.
We also wish to thank Xiao-Lei Chen, Da Luo
and Juan Li for their work on the data collection.
Conflicts of Interest
None.
References
1) Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama
CM, Lassen MR, Colwell CW: Prevention of venous
thromboembolism: American College of Chest Physicians
Evidence-Based Clinical Practice Guidelines (8th Edi-
tion). Chest, 2008; 133: 381S-453S
2) Michota FA: Bridging the gap between evidence and prac-
tice in venous thromboembolism prophylaxis: the quality
improvement process. J Gen Intern Med, 2007; 22: 1762-
1770
3) Lilienfeld DE: Decreasing mortality from pulmonary
embolism in the United States, 1979-1996. Int J Epide-
miol, 2000; 29: 465-469
4) Chopard P, Dorffler-Melly J, Hess U, Wuillemin WA,
Hayoz D, Gallino A, Bachli EB, Canova CR, Isenegger J,
Rubino R, Bounameaux H: Venous thromboembolism
prophylaxis in acutely ill medical patients: definite need
269
for improvement. J Intern Med, 2005; 257: 352-357
5) Kucher N, Spirk D, Kalka C, Mazzolai L, Nobel D, Ban-
yai M, Frauchiger B, Bounameaux H: Clinical predictors
of prophylaxis use prior to the onset of acute venous
thromboembolism in hospitalized patients SWIss Venous
ThromboEmbolism Registry (SWIVTER). J Thromb
Haemost, 2008; 6: 2082-2087
6) Tapson VF, Decousus H, Pini M, Chong BH, Froehlich
JB, Monreal M, Spyropoulos AC, Merli GJ, Zotz RB,
Bergmann JF, Pavanello R, Turpie AG, Nakamura M,
Piovella F, Kakkar AK, Spencer FA, Fitzgerald G, Ander-
son FA Jr, Investigators I: Venous thromboembolism pro-
phylaxis in acutely ill hospitalized medical patients: find-
ings from the International Medical Prevention Registry
on Venous Thromboembolism. Chest, 2007; 132: 936-
945
7) Yu HT, Dylan ML, Lin J, Dubois RW: Hospitals’ compli-
ance with prophylaxis guidelines for venous thromboem-
bolism. Am J Health Syst Pharm, 2007; 64: 69-76
8) Cohen AT, Alikhan R, Arcelus JI, Bergmann JF, Haas S,
Merli GJ, Spyropoulos AC, Tapson VF, Turpie AG: Assess-
ment of venous thromboembolism risk and the benefits of
thromboprophylaxis in medical patients. Thromb Hae-
most, 2005; 94: 750-759
9) Samama MM, Combe S, Conard J, Horellou MH: Risk
assessment models for thromboprophylaxis of medical
patients. Thromb Res, 2012; 129: 127-132
10) Barbar S, Noventa F, Rossetto V, Ferrari A, Brandolin B,
Perlati M, De Bon E, Tormene D, Pagnan A, Prandoni P:
A risk assessment model for the identification of hospital-
ized medical patients at risk for venous thromboembo-
lism: the Padua Prediction Score. J Thromb Haemost,
2010; 8: 2450-2457
11) Caprini JA: Thrombosis risk assessment as a guide to
quality patient care. Dis Mon, 2005; 51: 70-78
12) Kucher N, Koo S, Quiroz R, Cooper JM, Paterno MD,
Soukonnikov B, Goldhaber SZ: Electronic alerts to prevent
venous thromboembolism among hospitalized patients. N
Engl J Med, 2005; 352: 969-977
13) Spyropoulos AC, Anderson FA Jr, Fitzgerald G, Decousus
H, Pini M, Chong BH, Zotz RB, Bergmann JF, Tapson V,
Froehlich JB, Monreal M, Merli GJ, Pavanello R, Turpie
AG, Nakamura M, Piovella F, Kakkar AK, Spencer FA,
Investigators I: Predictive and associative models to iden-
tify hospitalized medical patients at risk for VTE. Chest,
2011; 140: 706-714
14) Caprini JA: Risk assessment as a guide for the prevention
of the many faces of venous thromboembolism. Am J
Surg, 2010; 199: S3-10
15) Caprini JA: Individual risk assessment is the best strategy
for thromboembolic prophylaxis. Dis Mon, 2010; 56: 552-
559
16) Samama MM, Dahl OE, Mismetti P, Quinlan DJ, Ros-
encher N, Cornelis M, de Vries H, van Beusekom I,
Kahan JP: An electronic tool for venous thromboembo-
lism prevention in medical and surgical patients. Haema-
tologica, 2006; 91: 64-70
17) Murphy RX Jr, Alderman A, Gutowski K, Kerrigan C,
Rosolowski K, Schechter L, Schmitz D, Wilkins E: Evi-
dence-based practices for thromboembolism prevention:
270 Zhou et al .
summary of the ASPS Venous Thromboembolism Task
Force Report. Plast Reconstr Surg, 2012; 130: 168e-175e
18) 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: 344-350
19) Gould MK, Garcia DA, Wren SM, Karanicolas PJ, Arce-
lus JI, Heit JA, Samama CM: Prevention of VTE in non-
orthopedic surgical patients: Antithrombotic Therapy and
Prevention of Thrombosis, 9th ed: American College of
Chest Physicians Evidence-Based Clinical Practice Guide-
lines. Chest, 2012; 141: e227S-277S
20) Caprini JA, Arcelus JI, Reyna JJ: Effective risk stratifica-
tion of surgical and nonsurgical patients for venous throm-
boembolic disease. Semin Hematol, 2001; 38: 12-19
21) Arcelus JI, Candocia S, Traverso CI, Fabrega F, Caprini
JA, Hasty JH: Venous thromboembolism prophylaxis and
risk assessment in medical patients. Semin Thromb
Hemost, 1991; 17 Suppl 3: 313-318
22) Hatef DA, Kenkel JM, Nguyen MQ, Farkas JP, Abtahi F,
Rohrich RJ, Brown SA: Thromboembolic risk assessment
and the efficacy of enoxaparin prophylaxis in excisional
body contouring surgery. Plast Reconstr Surg, 2008; 122:
269-279
23) Pannucci CJ, Bailey SH, Dreszer G, Fisher Wachtman C,
Zumsteg JW, Jaber RM, Hamill JB, Hume KM, Rubin
JP, Neligan PC, Kalliainen LK, Hoxworth RE, Pusic AL,
Wilkins EG: Validation of the Caprini risk assessment
model in plastic and reconstructive surgery patients. J Am
Coll Surg, 2011; 212: 105-112
24) Zakai NA, Wright J, Cushman M: Risk factors for venous
thrombosis in medical inpatients: validation of a throm-
bosis risk score. J Thromb Haemost, 2004; 2: 2156-2161
25) Zhou HX, Peng LQ, Yan Y, Yi Q, Tang YJ, Shen YC,
Feng YL, Wen FQ: Validation of the Caprini risk assess-
ment model in Chinese hospitalized patients with venous
thromboembolism. Thromb Res, 2012; 130: 735-740
26) Kahn SR, Lim W, Dunn AS, Cushman M, Dentali F, Akl
EA, Cook DJ, Balekian AA, Klein RC, Le H, Schulman S,
Murad MH: 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:
e195S-226S
27) Heit JA, Silverstein MD, Mohr DN, Petterson TM,
O’Fallon WM, Melton LJ 3rd: Risk factors for deep vein
thrombosis and pulmonary embolism: a population-based
case-control study. Arch Intern Med, 2000; 160: 809-815
28) Goldhaber SZ, Dunn K, MacDougall RC: New onset of
venous thromboembolism among hospitalized patients at
Brigham and Women’s Hospital is caused more often by
prophylaxis failure than by withholding treatment. Chest,
2000; 118: 1680-1684
29) Heit JA, Silverstein MD, Mohr DN, Petterson TM, Lohse
CM, O’Fallon WM, Melton LJ 3rd: The epidemiology of
venous thromboembolism in the community. Thromb
Haemost, 2001; 86: 452-463
30) Heit JA, O’Fallon WM, Petterson TM, Lohse CM, Silver-
stein MD, Mohr DN, Melton LJ 3rd: Relative impact of
risk factors for deep vein thrombosis and pulmonary
embolism: a population-based study. Arch Intern Med,
2002; 162: 1245-1248
31) Anderson FA Jr, Spencer FA: Risk factors for venous
thromboembolism. Circulation, 2003; 107: I9-I16
 Caprini Model in Chinese Inpatients 271

Supplementary Table 1. Risk factors and definition of the risk levels in the Caprini RAM and Padua Prediction Score models
RAM Risk Factors in the RAMs Definition of risk levels
Caprini model Score 5: Stroke; Multiple trauma; Elective major lower extremity arthroplasty; Low risk
(2009) Hip, pelvis or leg fracture; Acute spinal cord injury (paralysis) (cumulative risk score 0-1)
Score 3: Age ( ≥ 75); History of VTE; Positive Factor V Leiden; Positive Moderate risk
prothrombin G20210A; Elevated serum homocysteine; Positive Lupus (cumulative risk score 2)
anticoagulant; Other congenital or acquired thrombophilia; Heparin- High risk
induced thrombocytopenia (HIT); Family history of VTE; Elevated (cumulative risk score 3-4)
anticardiolipin antibodies Highest risk
Score 2: Age (61-74); Central venous access; Arthroscopic surgery; Major (cumulative risk score ≥ 5)
surgery; Malignancy; Laparoscopic procedure ＞ 45 min; Patient confined
to bed; Immobilizing plaster cast
Score 1: Age (41-60); Acute myocardial infarction; Heart failure; Varicose
veins; Obesity (BMI ＞ 25); Inflammatory bowel disease; Sepsis; COPD or
abnormal pulmonary function; Severe lung disease; Oral contraceptives or
HRT; Pregnancy or postpartum; History of unexpected stillborn infant,
recurrent spontaneous abortion ( ≥ 3), premature birth with toxemia or
growth-restricted infant; Medical patient currently at bed rest; Minor
surgery planned; History of prior major surgery; Swollen legs
The Padua Prediction Score 3: Active cancer; Previous VTE (with the exclusion of superficial vein Low risk
Score (2010) thrombosis); Reduced mobility; Already known thrombophilic condition (Cumulative score ＜ 4)
Score 2: Recent ( ≤ 1 month) trauma and/or surgery High risk
Score 1: Elderly age ( ≥ 70 years); Heart and/or respiratory failure; Acute (Cumulative score ≥ 4)
myocardial infarction or ischemic stroke; Acute infection and/or
rheumatologic disorder; Obesity (BMI ≥ 30); Ongoing hormonal treatment
RAM = risk assessment model, BMI = body mass index, HRT = hormone replacement therapy, COPD = chronic obstructive pulmonary disease,
VTE = venous thromboembolism
272 Zhou et al .

Supplementary Table 2. Number of patients and control hospitalized subjects according to the medical service
Cases (%) Controls (%)
Medical service
(n = 347) (n = 651)
Internal Medical department
Respiratory Medicine 73 (21.0) 136 (20.9)
General internal medicine 27 (7.8) 49 (7.5)
Cardiology 20 (5.8) 38 (5.8)
Geriatrics 18 (5.2) 34 (5.2)
ICU 7 (2.0) 14 (2.2)
Hematology 10 (2.9) 20 (3.1)
Neurology 10 (2.9) 19 (2.9)
Oncology 15 (4.3) 26 (4.0)
Infectious Diseases 7 (2.0) 13 (2.0)
Nephrology 6 (1.7) 12 (1.8)
Gastroenterology 10 (2.9) 17 (2.6)
Integrated TCM and Western Medicine 5 (1.4) 10 (1.5)
Rheumatology 3 (0.9) 6 (0.9)
Other medical departments＊ 3 (0.9) −
Surgical department
Vascular Surgery 99 (28.5) 193 (29.6)
Orthopedics 7 (2.0) 13 (2.0)
Cardio-Thoracic Surgery 6 (1.7) 12 (1.8)
General Surgery 16 (4.6) 29 (4.5)
Neurosurgery 3 (0.9) 6 (0.9)
Urology Surgery 2 (0.6) 4 (0.6)
ICU = intensive care unit, TCM = traditional Chinese medicine
＊
Other medical departments include dermatovenereology, pain management and psychiatry. VTE patients from these depart-
ments were not matched with controls.