Risk of venous thromboembolism in patients with splenic injury and
splenectomy
A nationwide cohort study
Jiun-Nong Lin1–3; Hsuan-Ju Chen4,5; Ming-Chia Lin6; Chung-Hsu Lai3; Hsi-Hsun Lin3; Chih-Hui Yang7; Chia-Hung Kao8,9
Introduction
The spleen, located in the upper-left portion of the abdomen
underneath the rib cage, is vulnerable to blunt and penetrating abdominal
trauma. The spleen plays critical haematologic and immunologic
roles in humans. For example, it removes old red blood
cells and pools a reserve of blood, including red blood cells, white
blood cells, and platelets (1). The stored blood cells can be released
into circulation in an emergency, after a splenectomy, or when
otherwise required (1, 2).
Approximately 22,000 all-cause splenectomies were performed
in the United States in 2005 (3, 4). Haematologic disorders are the
most common indications of splenectomy, including haemolytic
anaemia, autoimmune haemolytic anaemia, sickle cell disease,
_-thalassaemia, and immune thrombocytopenic purpura (3). Currently,
splenectomy is not recommended for managing splenic
trauma. Instead, spleen-preserving nonoperative procedures, such
as transcatheter arterial embolisation, have replaced surgical splenectomy
(5–7). However, emergent splenectomy is still recommended
for life-threatening haemorrhage and severe splenic injury
(grade III-V) because of the high failure rate of nonoperative management
(6, 8, 9).
Using splenectomy to treat haematologic diseases creates several
complications, including bacterial infections, cancer, ischaemic
heart disease, stroke, and venous thromboembolism
(VTE) (10–14). However, haematologic disorders can be associated
with these vascular complications (3). Little is known regarding
the association of VTE with splenic injury and splenectomy intrauma patients (15–17). Thus, we conducted a nationwide retrospective
cohort study to determine the risk of VTE in patients with
splenic injury and splenectomy by analysing the National Health
Insurance Research Database (NHIRD) of Taiwan.
Materials and methods
Data source
Taiwan launched a single-payer National Health Insurance (NHI)
program on March 1, 1995. The NHI database covers nearly the
entire population in Taiwan (> 23 million). The NHIRD, a claims
database, is managed and maintained by the National Health Research
Institutes of Taiwan. To protect patient privacy, all data are
anonymised through cryptographic scrambling before release to
the public for research. In this study, the hospitalisation claims
data of all enrollees in Taiwan were analysed, including information
on sex, date of birth, dates of admission and discharge, and
expenditures by admission. The diagnoses and procedures were
based on the International Classification of Diseases, Ninth Revision,
Clinical Modification (ICD-9-CM) codes. This study was approved
by the Institutional Review Board of China Medical University
and Hospital (CMUH104-REC2-115).
Study population
All patients included in this study were aged _ 20 years with newly
diagnosed splenic injury (ICD-9-CM code 865.xx), as recorded in
the inpatient registry file of the NHIRD from 2000–2006. Patients
with diagnoses of VTE or undergoing splenectomy (ICD-9-CM
codes for procedure 41.2x, 41.43, and 41.5x) before the date of
diagnosed splenic injury were excluded from this study. Among
the splenic injury cohort, patients who underwent splenectomy
were classified as the splenectomised group, whereas those who received
no splenectomy were classified as the nonsplenectomised
group. The first diagnosis date of splenic injury was regarded as
the index date. Patients without splenic injury, splenectomy, and
VTE were randomly selected from the registry of inpatients and
frequency-matched with splenic injury patients in a 4:1 ratio according
to sex, five-year age interval, and the year of index date.
Outcomes and comorbidities
The primary outcome was VTE, including portal vein thrombosis
(ICD-9-CM code 452), pulmonary embolism (ICD-9-CM codes
415.1 and 415.19; excluding 415.11), and other venous embolism
and thrombosis (ICD-9-CM code 453.xx). All enrolled patients
were followed from the index date to the occurrence of primary
outcome, December 31, 2011, death or withdrawal from NHI. The
risk of VTE after splenic injury and splenectomy was stratified by
the follow-up duration after the index date (_ 90, 91–365, and
> 365 days) and length of hospital stay for the splenic injury and
splenectomy (_ 8, 9–13, and _ 14 days). According to each inpatient
diagnosis, we calculated the Charlson Comorbidity Index
(CCI) as the comorbidity measure. CCI score is the sum of the
weighted score of 17 comorbidities. A weight is assigned in each
indicated diagnosis and added together to provide a total CCI
score. Comorbidities, including myocardial infarction, congestive
heart failure, peripheral vascular disease, cerebrovascular disease,
dementia, chronic pulmonary disease, rheumatologic disease, peptic
ulcer disease, mild liver disease, and diabetes mellitus, were calculated
as weighted 1. Moderate to severe renal disease, diabetes
mellitus with chronic complications, hemiplegia or paraplegia, leukaemia,
tumour of any type, and malignant lymphoma were
weighted 2. Moderate to severe liver disease was weighted 3. Acquired
immune deficiency syndrome and metastatic solid tumour
were weighted 6 (18, 19).
Statistical analysis
Differences among cohorts were examined using the Chi-square
test for categorical variables and Student’s t-test for continuous
variables. Cumulative incidences of VTE in the splenectomised,
nonsplenectomised, and comparison cohorts were explored using
the Kaplan-Meier method, and the differences were determined
using log-rank tests. Cox proportional hazards regression models
were used to assess the association of VTE with splenic injury and
splenectomy after adjustments for sex, age, and CCI score. Hazard
ratios (HRs) and 95 % confidence intervals (CIs) were calculated to
quantify the risk of VTE. All analyses were performed using SAS
software Version 9.3 (SAS Institute, Cary, NC, USA). Differences
were considered statistically significant if the two-tailed p-value
was < 0.05.
Results
During the study period, 6162 splenic injury patients and 24,648
comparison patients were enrolled in this study (_ Table 1). In the
splenic injury cohort, 3,033 patients received splenectomy, and
3,129 patients did not receive splenectomy. The age of the patients
with splenic injury was 41.93 ± 16.44 years (mean ± standard deviation,
SD), and 83.33 % of patients were younger than 60 years.
Men accounted for approximately 71 % of the patients in each cohort.
A higher CCI score was found in the splenic injury cohort
than in the comparison cohort (0.32 ± 0.91 vs 0.16 ± 0.65, p
< 0.001). The mean ± SD follow-up periods were 7.08 ± 3.18 and
7.7 ± 2.64 years in the splenic injury and comparison cohorts, respectively
(p < 0.001). Splenectomised patients had a shorter follow-
up duration than did nonsplenectomised patients (6.98 ± 3.45
vs 7.18 ± 2.90 years; p = 0.01).
Among the 6,162 splenic injury patients, 44 patients, including
19 nonsplenectomised and 25 splenectomised patients, developed
VTE (incidence density rate, 10.08 per 10,000 person-years). Additionally,
99 patients had VTE among the 24,648 comparison patients
(incidence density rate, 5.21 per 10,000 person-years)
(_ Table 2). Compared with patients in the comparison cohort,
splenic injury patients with and without splenectomy exhibited a
2.21-fold and 1.71-fold increased risk of VTE (95 % CI, 1.43–3.43;
95 % CI, 1.05–2.80, respectively). The risk of VTE was 1.97-foldhigher in the splenic injury cohort than in the comparison cohort,
irrespective of splenectomy status (95 % CI, 1.38–2.81). Among
patients with splenic injury, splenectomised patients had a
1.35-fold greater risk of VTE compared with nonsplenectomised
patients; however, the difference was not significant (95 % CI,
0.74–2.45).
The Kaplan-Meier analysis revealed that splenic injury patients
had a significantly higher cumulative incidence of VTE than did
the patients in the comparison cohort (log-rank test, p < 0.001;
_ Figure 1 A). Significant differences in the cumulative incidences
of VTE were detected among these three cohorts (_ Figure 1 B; p
< 0.001), and the cumulative incidence of VTE was the highest in
splenic injury patients with splenectomy.
We analysed the effect of sex, age, and CCI score on the risk of
VTE after splenic injury and splenectomy (_ Table 3). The incidence
density rates of VTE in female and male splenic injury patients
were 6.30 and 11.63 per 10,000 person-years, respectively;
both incidence density rates were higher than that in the comparison
cohort (3.34 and 5.96 per 10,000 person-years, respectively).
Among the splenic injury patients, men, but not women, exhibited
a significant 1.97-fold (95 % CI, 1.33–2.92) higher risk of developing
VTE than did male patients in the comparison cohort. Furtheranalysis showed that the risk of VTE was only significant in male
splenectomised patients (adjusted HR, 2.24; 95 % CI, 1.39–3.63).
Age-specific analysis revealed that the incidence density rates of
VTE increased with age in each cohort. The risk of VTE was significant
in splenic injury patients aged < 60 years compared with in
patients in the nonsplenic injury cohort. Among the splenic injury
patients, only those aged < 60 years who underwent splenectomy
had significantly increased risks of VTE. Splenic patients in all age
groups who did not receive splenectomy had no increased risk for
VTE. With regard to CCI score, we observed that both splenectomised
and nonsplenectomised splenic injury patients with a CCI
score of 0 had a significantly higher risk of VTE compared with
the nonsplenic injury cohort. However, the risk of VTE after
splenic injury did not increase in patients with a CCI score _ 1.
The simultaneous comorbidities of patients with VTE events during
the follow-up periods were analysed (Suppl. Table 1, available online
at Among the 44 VTE patients in
the splenic injury cohort, liver cirrhosis (52.27 %) is the most prevalent
comorbidity, followed by cancer (40.91 %) and diabetes mellitus
(31.82 %). Hypertension (38.38 %), cancer (33.33 %), and liver cirrhosis
(27.27 %) were the leading three comorbidities among the 99 VTE
patients in the nonsplenic injury cohort. Liver cirrhosis was the only
significantly more prevalent comorbidity in the splenic injury cohort
than in the nonsplenic injury cohort (p = 0.006).
_ Table 4 shows the risk of VTE in different vessels. The risk of
portal vein thrombosis was significantly elevated in patients with
splenic injury (adjusted HR, 3.05; 95 % CI, 1.70–5.45), either in
splenectomised patients (adjusted HR, 2.53; 95 % CI, 1.15–5.54)
and in nonsplenectomised patients (adjusted HR, 3.58; 95 % CI,
1.78–7.20). There was no significant difference of portal vein
thrombosis between splenectomised and nonsplenectomised cohorts.
The risk of pulmonary embolism was not significantly different
following splenic injury (adjusted HR, 1.04; 95 % CI,
0.39–2.77), irrespective of splenectomy status. The risk of VTE
other than portal vein thrombosis and pulmonary embolism was
elevated in the splenic injury cohort compared with the nonsplenic
injury cohort (adjusted HR, 1.75; 95 % CI, 1.04–2.96). This VTE
risk was significantly increased in the splenectomised patients (adjusted
HR, 2.25; 95 % CI, 1.22–4.16), but not in the nonsplenectomised
patients (adjusted HR, 1.24; 95 % CI, 0.56–2.75).
We analysed the relationship between VTE risk and length of
hospital stay for splenic injury and splenectomy (_ Table 5). The
hospitalisation days were classified into _ 8, 9–13, and _ 14 days
according to tertiles. Compared with the comparison cohort, the
HRs of VTE increased as the length of hospital stay increased (p
for trend < 0.001). Among those who were hospitalised for more
than 14 days, the risk of VTE significantly increased following
splenic injury (adjusted HR, 2.43; 95 % CI, 1.52–3.59).
When stratified by follow-up periods, the risk of VTE was significantly
increased in the splenic injury cohort within 90 days
(adjusted HR, 24.94; 95 % CI, 2.99–208) and more than 365 days
after the index date (adjusted HR 1.72; 95 % CI 1.16–2.56)
(_ Table 6). The VTE risk was substantially increased within 90
days after the index date in the splenectomised patients (adjusted
HR, 43.35; 95 % CI, 5.05–372). However, the VTE risk did not increase
during 91–365 days after splenic injury. When the followup
duration was more than 365 days after the index date, both
splenectomised (adjusted HR, 1.70; 95 % CI, 1.01–2.86) and nonsplenectomised
patients (adjusted HR, 1.75; 95 % CI, 1.04–2.95)
remained at a higher risk of VTE.Discussion
Recently, a large retrospective study reported the long-term complications
of all-cause splenectomy, revealing that splenectomy increases
the risks of pulmonary embolism and deep-vein thrombosis
(11). However, an older study showed no correlation between
the occurrence of deep vein thrombosis and splenectomy in patients
with Hodgkin disease (12). The present study, which included
only trauma patients, revealed that splenic injury elevated
the risk of VTE (adjusted HR, 1.97; 95 % CI, 1.38–2.81), irrespective
of whether the patients underwent splenectomy. A large Australian
study reported a VTE rate of 130 per 10,000 person-years
after all-cause splenectomy (20). The incidence density rate of
VTE following splenectomy in splenic trauma patients was 11.81
per 10,000 person-years in this study. Our VTE rate is much lower
than that in the Australian study (20). The prevalence of VTE
varies among different races and ethnicities. Asians have a lower
prevalence of VTE than do African Americans, Caucasians, and
Hispanics (21–23).
Although the incidence of VTE in women of childbearing age is
higher than that in men, the incidence in postmenopausal women
is lower than that in men (10, 21). Studies have revealed an association
between VTE and increasing age (10, 21). Comorbidities
such as cancer, respiratory failure, myocardial infarction, and congestive
heart failure were also shown to increase the risk of VTE
(10). Among the VTE patients in our study, there was no significant
difference in the prevalence of comorbidities between the
splenic injury cohort and nonsplenic injury cohort except liver cirrhosis.
Epidemiological studies inspecting the association between
liver cirrhosis and VTE demonstrated conflicting results (24). The
role of liver cirrhosis in the pathogenesis of VTE in splenic injury
patients remains unclear.
Portal vein thrombosis is a well-known complication after splenectomy
indicated for haematologic diseases (25, 26). The process
of haematologic diseases and the presence of splenomegaly were
identified as the most important risk factors for portal vein thrombosis
after splenectomy (25). In our study, we observed that the
risk of portal vein thrombosis was significantly elevated in thesplenic injury cohort, either in splenectomised and nonsplenectomised
patients. However, no statistical difference of portal vein
thrombosis was identified between splenectomised and nonsplenectomised
patients. Our findings suggest that splenic injury per
se might increase the risk of portal vein thrombosis beyond the
splenectomy.
Van Haren et al. (27) assessed the risk of VTE in trauma patients
who hospitalised in intensive care unit. They found that patients
with more surgeries, longer intensive care unit stay, and abdominal
injury with a higher Abbreviated Injury Scale score were
at a higher risk for VTE. Brakenridge et al. (28) also reported a
similar result. In our study, the information of injury severity is not
available in the NHIRD. We used length of hospital stay as a proxy
for severity of splenic injury. Our results showed that the VTE risk
increased along with hospitalisation days. When the length of hospital
stay was more than 14 days, splenic injury patients exhibited
a 2.43-fold increased risk of VTE.
Long-term risks of VTE have been rarely described in the literature
(11, 29). Thomsen et al. (29) reported a 32.6-, 7.1-, and
3.4-fold higher risks of VTE within 90 days, 91–365 days, and
> 365 days after all-cause splenectomy compared with the general
population. Among the splenic injury patients, the VTE risk > 365
days post-splenectomy was 3.1-fold higher (95 % CI, 1.8–5.4) than
in the normal comparisons (29). Our results showed an increased
risk of VTE within 90 days and > 365 days after splenic injury and
splenectomy. But the VTE risk did not increase during 91–365
days following splenic injury. The exact reason for the partial inconsistency
between our results and the study by Thomsen et al. is
unclear. Race and ethnicity differences may play a role in the variation
of VTE risk.
Our study revealed increased short-term and long-term risks of
VTE in splenic injury patients than in the control patients. Severe
splenic injury could result in hyposplenism or splenic dysfunction.
Surgery has been shown to increase the risk of VTE within 30 days
of operation (30–32). The mechanisms for VTE after surgery are
multifactorial, including blood stasis resulting from immobility,vessel endothelial damage, and upregulation of P-selectin and local
prothrombotic microparticles (30). However, the postprocedural
or postoperative risks for VTE are short-term. The increased longterm
risk of VTE was less likely caused by only the surgical procedures.
In other words, our results suggested that the increased
risk for VTE was due to the splenic injury or splenectomy. Some
mechanisms have been proposed for VTE after splenic injury and
splenectomy, including reactive thrombocytosis, platelet activation,
hypercoagulability, activation of the endothelium, decreased
levels of protein C and protein S, elevated thrombin generation,
altered lipid pro_les, leukocytosis, and increased C-reactive
protein levels (33–36). Further studies to understand the pathogenesis
of VTE after splenic injury and splenectomy are warranted.
The major strengths of this study include the large number of
cases (N = 6,162), single indication for splenectomy, and long duration
of follow-up. However, several limitations should be noted.
First, a paucity of information was identified regarding various
risk factors for VTE, such as smoking, obesity, and socioeconomic
status. In addition, no haematologic information, such as platelet
counts, prothrombin time, and partial thromboplastin time, was
contained in the NHIRD. These factors were possible confounding
factors in this study. Second, the present study is a retrospectively
observational cohort study. Despite the meticulous design and
analysis of comorbidity, we cannot prove the real etiology of VTE.
Third, the severity of splenic injury is not included in the NHIRD.
We cannot evaluate the risk of VTE according to the severity of
splenic trauma. However, we used length of hospital stay as a
proxy for severity of splenic injury. Fourth, the NHIRD lack information
on pharmacologic prophylaxis and inferior vena cava
filters. These factors could influence the occurrence of VTE. Fifth,
the diagnoses of VTE were based on the ICD-9-CM diagnostic