ORIGINAL ARTICLE
The Effects of Splenic Artery Embolization on Nonoperative
Management of Blunt Splenic Injury: A 16-Year Experience
Ashraf A. Sabe, BA, Jeffrey A. Claridge, MD, David I. Rosenblum, DO, Kevin Lie, MD,
and Mark A. Malangoni, MD
Introduction: Nonoperative management (NOM) of blunt splenic injury has
become the preferred treatment for hemodynamically stable patients. The
application of splenic artery embolization (SAE) in NOM has been contro-
versial. We hypothesized that incorporation of initial use of SAE into a practice
protocol for patients at high risk for NOM failure (contrast extravasation or
pseudoaneurysm on computed tomography, grade 3 injury with large hemoperi-
toneum, grade 4 injuries) would improve patient outcomes.
Methods: A retrospective analysis of three continuums of practice was
performed: group I (January 1991–June 1998), SAE not part of routine
NOM; group II (July 1998 –December 2001), introduction and discretionary
use of SAE; and group III (January 2002–June 2007), standardized use of
initial SAE for patients considered at high risk of nonoperative failure. The
primary outcome measure was the success of NOM. Failure of NOM was
defined as the need for abdominal operation. Secondary outcomes were
mortality, length of stay, and splenic salvage.
Results: Over 16 years, 815 patients with blunt splenic injury were treated
at our level 1 trauma center. There were 222 patients in group I, 195 in group
II, and 398 in group III. There was an increase in the use of SAE over time
with a significant improvement in the utilization of NOM (61% in group I;
82% in group II; 88% in group III; p ⬍ 0.05). This was associated with an
increase in successful NOM (77%, group I; 94%, group II; 97%, group III;
p ⬍ 0.0001 group I vs. group II and III). Mortality, length of stay, and splenic
salvage were similar in groups II and III but significantly improved when
compared with group I.
Conclusions: The increased use of initial SAE in high-risk patients expanded
the successful use of NOM but was not associated with other incremental
improvements.
Key Words: Splenic artery embolization, Blunt spleen injury.
(J Trauma. 2009;67: 565–572)
Submitted for publication December 4, 2008.
Accepted for publication June 4, 2009.
Copyright © 2009 by Lippincott Williams & Wilkins
From the Departments of Surgery (A.A.S., J.A.C., M.A.M.), and Radiology
(D.I.R., K.L.), Case Western Reserve University School of Medicine, Metro-
Health Medical Center, Cleveland, Ohio.
Supported by National Center for Research Resources (NCRR), a component of
the National Institutes of Health (NIH) and NIH Roadmap for Medical
Research, grant 1KL2RR024990 (to J.A.C.).
The contents of this manuscript are solely the responsibility of the authors and do
not necessarily represent the official view of NCRR or NIH.
Presented at the 67th Annual Meeting of the American Association for the Surgery
of Trauma, September 24 –27, 2008, Maui, Hawaii.
Address for reprints: Jeffrey A. Claridge, MD, MS, FACS, MetroHealth Medical
Center, Room H939, Hamann Bldg, 2500 MetroHealth Drive, Cleveland, OH
44109-1998; email: jclaridge@metrohealth.org.
DOI: 10.1097/TA.0b013e3181b17010
The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
O ver the last decade, nonoperative management (NOM)
has become the preferred treatment for hemodynami-
cally stable patients with blunt splenic injury. Early reports
established the use of bed rest, frequent monitoring of red
blood cell count, and serial abdominal exams as standard
principles of NOM.1–3 Computed tomography (CT) has evolved
as the preferred method to identify which patients are appropri-
ate for NOM and to select when patients should undergo further
investigation. Advancements in CT technology have improved
our ability to define the degree of splenic injury and more
accurately to identify patients who are more likely to fail
NOM.4 –12 Splenic artery embolization (SAE) has been dem-
onstrated to effectively stop bleeding in selected patients,
thus avoiding the need for operation.10 –13
Splenic injury grades ⬎3, derived from The American
Association for the Surgery of Trauma Splenic Injury Scale
(SIS), have been associated with a greater likelihood for
failure of NOM.5,9,11,14,15 This grading system alone has not
been sufficient to define which patients with splenic injury
can be safely managed nonoperatively. Other findings on CT,
including active extravasation, traumatic pseudoaneurysm, or
a large hemoperitoneum have been associated with the need for
operation.10,11,13 The incorporation of these parameters along
with the grade of injury has improved the ability to predict which
patients are likely to have successful NOM.8,9,11,13,16
In a previous study,13 we compared a cohort of patients
from 1991 to 1998 who had standard NOM to a later group
in which SAE was selectively used to improve the success of
NOM. That analysis demonstrated that the increased use of
SAE correlated with increased success of NOM and was
associated with decreased mortality.13
We hypothesized that a practice protocol, which incor-
porated the initial use of SAE for patients identified to be at
high risk for NOM failure (contrast extravasation or pseudo-
aneurysm on CT, grade 3 injury with large hemoperitoneum,
grade 4 injury), would further improve the success of NOM
for appropriate patients with blunt splenic injury.
PATIENTS AND METHODS
Patients for this retrospective analysis were derived
from our hospital trauma registry between the years 1991 and
2007 and consisted of all patients older than 16 years with
blunt splenic injury admitted to MetroHealth Medical Center,
an American College of Surgeons verified Level I trauma center.
The following data were collected from patient records: demo-
graphics, initial vital signs, hematocrit, and Injury Severity Score
565
Sabe et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009

(ISS). Abdominal CT scans were concurrently reviewed by
radiologists and attending surgeons, and splenic injuries were
graded according to the SIS of the American Association for
the Surgery of Trauma.5 Relevant operative notes and pathol-
ogy reports were reviewed to confirm the SIS grade. For
patients who underwent SAE, angiographic procedure notes
were evaluated for documentation of contrast extravasation
and the presence of a pseudoaneurysm. Technical failure of
SAE was defined as the inability to successfully cannulate
and embolize the splenic artery. A SAE failure was defined as
a patient who needed subsequent operation after SAE.
Patients were divided into three continuums of practice:
group I (January 1991–June 1998), SAE not part of routine
NOM; group II (July 1998 –December 2001), introduction
and discretionary use of SAE; and group III (January 2002–
June 2007), standardized use of initial SAE for patients
determined to be at high risk of failure. The following
protocol was followed on arrival of patients to the emergency
department with suspected intra-abdominal injury: if hemo-
dynamically unstable, patients had immediate exploratory
laparotomy; those who were hemodynamically stable under-
went abdominal CT with intravenous contrast unless transferred
from another hospital with an adequate CT scan for assessment
of injury. Prospectively defined indications for initial SAE in
group III included contrast extravasation or pseudoaneurysm
on CT; grade 3 injuries with a large hemoperitoneum; or
grade 4 injuries. The protocol also suggested that all patients
with grade 5 injury should have an immediate operation.
Those who did not fulfill these criteria underwent standard
NOM without SAE. Initial SAE was performed within 12
hours of presentation and involved occlusion of the proximal
splenic artery with coils in 99% of patients.
The aforementioned groups were compared based on
initial treatment, ultimate treatment, and outcomes. The pri-
mary outcome variable was successful NOM, which was
defined as the percent of patients not requiring abdominal
surgery. Secondary outcome variables included mortality,
length of stay (LOS), and overall splenic salvage (any out-
come other than splenectomy). Failure of NOM is defined as
patients who initially managed without operation and subse-
quently required an operation for hemorrhage.
Statistical analysis was performed using SPSS software
(SPSS Inc, Chicago, IL). Student’s t tests were used to
compare continuous data, which are expressed as mean ⫾
SEM. GraphPad software (GraphPad Software, Inc, San Di-
ego, CA) was used for ANOVA of multiple groups with
Tukey post hoc testing while categorical data were compared
by Fisher’s exact test or Pearson ␹2 analysis as appropriate.
When categorical data among the three groups were com-
pared, further subgroup comparisons were made only if the p
value was ⬍0.05. This study was approved by the Metro-
Health System Institutional Review Board.
RESULTS
A total of 815 adults with blunt splenic injury were
identified. There were 222 patients in group I, 195 in group II,
and 398 in group III. Characteristics, treatments, and out-
comes of all patients are presented in Table 1. Patients were
predominantly men. ISS was significantly greater in group I
(27.2 ⫾ 0.9) compared with groups II and III (22.9 ⫾ 0.9 and
21.2 ⫾ 0.6, respectively, p ⬍ 0.0001). The mean SIS scores
and the percentage of each grade of splenic injury were
similar among the groups.

TABLE 1. Characteristics, Treatments, and Outcomes of Patients With Blunt Spleen Injuries
Group I (n ⴝ 222) Group II (n ⴝ 195) Group III (n ⴝ 398) p
Characteristics
Age (yr) 35.0 ⫾ 1.2 37.9 ⫾ 1.4 37.4 ⫾ 0.9 NS
Male (%) 142 (64) 122 (62.6) 251 (63.1) NS
Injury severity score 27.2 ⫾ 0.9† 22.9 ⫾ 0.9 21.2 ⫾ 0.6 ⬍0.0001
Splenic injury score 2.8 ⫾ 0.1 2.7 ⫾ 0.1 2.6 ⫾ 0.1 NS
Initial systolic blood pressure (mm Hg) 130.2 ⫾ 2.1 128.3 ⫾ 1.9 126.3 ⫾ 1.3 NS
Initial pulse (beats/min) 100.9 ⫾ 1.7 100.0 ⫾ 1.5 99.4 ⫾ 1.2 NS
Hematocrit (%) 34.3 ⫾ 0.5 34.8 ⫾ 0.6 35.5 ⫾ 0.4 NS
Treatments
Initial operation (%) 86 (38.7)† 35 (17.9)† 47 (11.8)† ⬍0.05
Initial nonoperative management (%) 136 (61.3)† 160 (82.1)† 351 (88.2)† ⬍0.05
Initial splenic artery embolization (%) 0† 17 (8.7)† 92 (23.1)† ⬍0.0001
Overall splenic artery embolization (%) 6 (2.7)† 34 (17.4)† 118 (29.6)† ⬍0.01
Outcomes
Successful nonoperative management (% of 105 (77.2)† 151 (94.4) 340 (96.9) ⬍0.0001
initial nonoperative management)
Mortality (%) 26 (11.7)† 12 (6.2)† 20 (5.0)† ⬍0.05
Length of stay (d) 14.7 ⫾ 1† 10.3 ⫾ 0.8 9.8 ⫾ 0.5 ⬍0.0001
Splenic salvage (%) 125 (56.3)† 166 (85.1) 351 (88.2) ⬍0.0001
Data presented as mean ⫾ standard error of the mean.
* Significant compared with group I.
† Significant compared with all other groups.
NS, not significant.

566 © 2009 Lippincott Williams & Wilkins

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009 A 16-Year Experience With Splenic Trauma

There was a significant increase in the use of NOM
(p ⬍ 0.05) and concurrent increase in the success rate of
NOM over time. This was associated with a significant
increase in the initial and overall use of SAE (p ⬍ 0.01).
Mortality decreased significantly over time (group I vs.
groups II and III; p ⬍ 0.0001), and the LOS was shorter in
groups II and III compared with group I (p ⬍ 0.0001).
Furthermore, the splenic salvage rate increased in groups II
and III compared with group I (p ⬍ 0.0001). As reflected in
Figure 1, these findings correlated with an increase in proto-
col compliance with the use of SAE (protocol described in the
methods) across groups (p ⬍ 0.004).
Patients Managed With Initial Operation
Table 2 demonstrates the characteristics and outcome
parameters of patients who had immediate operation. Six
patients had urgent exploratory laparotomy (5 for diaphrag-
matic rupture and 1 for emergent cesarean section), and
minor splenic injuries were discovered incidentally at oper-
ation and did not require treatment. Two of these patients
subsequently required SAE for delayed splenic rupture. Com-
pared with the other groups, there was a significant decrease
in the incidence of splenorrhaphy (p ⬍ 0.05) in group III.
There were no statistically significant differences in the
mortality, LOS, or splenic salvage among these three groups
of patients.
Patients Initially Treated With NOM
Six hundred forty-seven patients (79.4%) had planned
NOM. The characteristics, treatments, and outcomes of these
patients are presented in Table 3. The rate of successful NOM
increased over time (p ⬍ 0.0001), as the use of initial SAE
increased (0%, group I; 10.6%, group II; 26.2%, group III;
p ⬍ 0.0001). Both ISS and SIS are significantly greater in
patients managed nonoperatively compared with patients who
had an operation (p ⬍ 0.0001; Table 4). The mean SIS grade
was lower in patients who had successful NOM (p ⬍ 0.001;
Table 5). There was also a significant increase in protocol
compliance in patients who had successful NOM compared
with those who failed NOM (p ⬍ 0.05; Table 5).
Mortality decreased from 5.8% in group I to 1.2% in
groups II and III combined (p ⬍ 0.01). Splenic salvage
improved from 81% to 97% over time (p ⬍ 0.0001, group I
vs. groups II and III). There was no significant difference in
the LOS between the groups.
Patients Managed With Initial SAE
Patients who underwent initial SAE had a higher SIS
than those who were initially observed in groups II and III
(⬍0.0001). In group I, no patients were managed with initial
SAE. The average SIS in patients who had initial SAE (3.5 ⫾
0.1) was similar those who had initial operation (3.7 ⫾ 0.1)
in groups II and III. Five of the 109 patients (4.6%) who were

77
*
NOM Success % 94

97

81 *

Splenic Salvage % 97

97

4 Group I

SAE % 18 *** n = 136

Group II
29 n = 160
Group III
n = 351

Protocol Compliance % 76 **

89

0 10 20 30 40 50 60 70 80 90 100
Figure 1. A comparison of patients managed nonoperatively. Protocol compliance is defined in the methods. *p ⬍ 0.05,
group I versus group II and III, **p ⬍ 0.05, group II versus III, and ***p ⬍ 0.05, between all the three groups.

© 2009 Lippincott Williams & Wilkins 567

Sabe et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009

TABLE 2. Characteristics, Treatments, and Outcomes of Patients Who Had Initial Operation
Group I (n ⴝ 86) Group II (n ⴝ 35) Group III (n ⴝ 47) p
Characteristics
Age (yr) 36.1 ⫾ 1.9 42.4 ⫾ 4.0 39.6 ⫾ 2.9 NS
Injury severity score 34.1 ⫾ 1.5 32.5 ⫾ 2.1 29.7 ⫾ 2.1 NS
Splenic injury score 3.5 ⫾ 0.1 3.7 ⫾ 0.2 3.6 ⫾ 0.2 NS
Initial systolic blood pressure (mm Hg) 121.8 ⫾ 3.4 118.5 ⫾ 6.12 113.8 ⫾ 5.1 NS
Initial pulse (beats/min) 106.6 ⫾ 2.5 112.2 ⫾ 3.6* 106.9 ⫾ 4.9 ⬍0.0001
Hematocrit (%) 31.5 ⫾ 1.0 27.0 ⫾ 1.0* 31.1 ⫾ 1.1 0.022
Treatments
Splenectomy (%) 69 (80.2) 24 (68.6) 38 (80.9) NS
Splenorrhaphy (%) 17 (19.8) 10 (28.6) 4 (8.5)† 0.017
Operation for other indication (%) 0 1 (2.9) 5 (10.6)* 0.002
Outcomes
Mortality (%) 18 (20.9) 11 (31.4) 15 (31.9) NS
Length of stay (d) 20.0 ⫾ 2.4‡ 14.0 ⫾ 3.2 12 ⫾ 1.7 0.05
Splenic salvage (%) 17 (19.8) 11 (31.4) 9 (19.2) NS
Data presented as mean ⫾ standard error of the mean.
* Significant compared with group I.
† Significant compared with group II.
‡ Significant compared with all the other groups.
NS, not significant.

TABLE 3. Characteristics, Treatments and Outcomes in Patients Initially Managed Nonoperatively

Group I (n ⴝ 136) Group II (n ⴝ 160) Group III (n ⴝ 351) p
Characteristics
Age (yr) 34.3 ⫾ 1.5 36.9 ⫾ 1.5 37.1 ⫾ 0.9 NS
Injury severity score 22.8 ⫾ 1.0 20.5 ⫾ 0.9 19.9 ⫾ 0.6* ⬍0.0001
Splenic injury score 2.4 ⫾ 0.1 2.5 ⫾ 0.1 2.5 ⫾ 0.1 NS
Initial systolic blood pressure (mm Hg) 135 ⫾ 2.6 130.5 ⫾ 1.9 128.0 ⫾ 1.2* ⬍0.0001
Initial pulse (beats/min) 97.6 ⫾ 2.1 97.2 ⫾ 1.5 98.4 ⫾ 1.1 NS
Hematocrit (%) 35.9 ⫾ 0.5 36.5 ⫾ 0.6 36.1 ⫾ 0.4 NS
Treatments
Initial splenic artery embolization (%) 0 (0)† 17 (10.6)† 92 (26.2)† ⬍0.0001
Late splenic artery embolization (%) 6 (4.4) 17 (10.6)* 24 (6.8) 0.036
Outcomes
Successful nonoperative managment (%) 105 (77.2)† 151 (94.4) 340 (96.9) ⬍0.0001
Mortality (%) 8 (5.8)† 1 (0.6) 5 (1.4) ⬍0.01
Length of stay (d) 11.4 ⫾ 0.8 9.5 ⫾ 0.7 9.5 ⫾ 0.6 NS
Splenic salvage (%) 110 (80.9)† 155 (96.9) 342 (97.4) ⬍0.0001
Data presented as ⫾ standard error of the mean.
* Significant compared with group I.
† Significant compared with all the other groups.
NS, not significant.

designated to have initial SAE and one of the 47 patients
(2.3%) who had late SAE had a technical failure. Ultimately,
two of these patients required splenectomy, three had splen-
orrhaphy, and one was observed successfully. Seven patients
who had initially successful SAE had persistent bleeding. All
of these patients required splenectomy.
Discretionary (Group II) Versus Protocol-Based
(Group III) Use of Initial SAE
As demonstrated in Figure 1, when retrospectively
applying our protocol to group II where the use of initial
SAE was discretionary, the overall compliance rate was 76%.
Compliance improved to 89% when this protocol was prospec-
tively applied (p ⫽ 0.011). The success of NOM increased
between these two groups from 94.4% to 96.9%; however, this
was not statistically significant. The increased use of SAE was
not associated with a difference in mortality, LOS, or splenic
salvage. Outcomes stratified by SIS grade are demonstrated in
Table 6. NOM was most successful in patients with lower grade
injuries and decreased as SIS increased. Splenic salvage also
decreased with increasing SIS. Six patients who had NOM in
groups II and III died. None died of persistent bleeding or a
missed intra-abdominal injury.

568 © 2009 Lippincott Williams & Wilkins

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009 A 16-Year Experience With Splenic Trauma

initially observed (Fig. 2). Patients selected for initial SAE

TABLE 4. Comparison of Patients by Initial Treatment in
Groups II and III had a higher SIS grade of injury than those who did not have
SAE. The mean grade of splenic injury for patients who had
Initial Initial
Variable Operation NOM p
SAE was similar to patients who presented to the trauma
center with hemodynamic instability and had immediate op-
Age (yr) 40.8 ⫾ 2.4 37.0 ⫾ 0.8 NS eration. Other reports have demonstrated a higher grade of
Injury severity score 30.9 ⫾ 1.5 20.1 ⫾ 0.5 ⬍0.0001 splenic injuries among patients requiring SAE.1,12,14,17 This
Splenic injury score 3.7 ⫾ 0.1 2.5 ⫾ 0.1 ⬍0.0001 suggests that the use of SAE, whether done initially or for
Initial systolic blood 115.8 ⫾ 3.9 128.8 ⫾ 1.0 ⬍0.0001 those who are failing NOM, can reduce the need for operation
pressure (mm Hg)
Initial pulse (beats/min) 109.2 ⫾ 3.2 98.0 ⫾ 0.9 ⬍0.0001
and increase the rate of overall splenic salvage, despite the
Hematocrit 29.4 ⫾ 0.8 36.2 ⫾ 0.3 ⬍0.0001
greater need for splenectomy at operation. Patients who failed
Total 82 511 —
NOM were significantly older and had a greater ISS and SIS,
which is consistent with our previous study as well as
Data presented as ⫾ standard error of the mean. observations of others.10,14,18
Our protocol identified patients at high risk for failure
if they had vascular blush or pseudoaneurysm on CT, grade 3
TABLE 5. Comparison of Successful vs. Failed injury with large hemoperitoneum, or grade 4 injury. These
Nonoperative Management (NOM) in Groups II and III parameters have been demonstrated to be associated with an
Successful Failed increased failure of NOM in previous reports.8,10,14 Haan et
Variable NOM NOM p al.19 found a significant association with NOM failure and the
Age (yr) 36.7 ⫾ 0.8 45.3 ⫾ 4.3 ⬍0.05 presence of arteriovenous fistula but did not report increased
Injury severity score (ISS) 19.8 ⫾ 0.5 28.6 ⫾ 3.6 0.001 failure rates with large hemoperitoneum or active contrast
Splenic injury score (SIS) 2.5 ⫾ 0.1 3.4 ⫾ 0.2 ⬍0.001 extravasation. Marmery et al.11 recently proposed a new
Initial systolic blood 129.2 ⫾ 1.0 120.5 ⫾ 7.3 NS CT-based grading system in which active bleeding, arterio-
pressure (mm Hg) venous fistula, pseudoaneurysm, and vascular injury are the
Initial pulse (beats/min) 97.9 ⫾ 0.9 103.5 ⫾ 3.6 NS main parameters used to determine the grade of splenic
Hematocrit 36.3 ⫾ 0.3 34.5 ⫾ 1.9 NS injury. Using criteria similar to theirs in the present protocol,
Mortality (%) 6 (1.2) 0 (0) NS we have demonstrated a 97% success rate among patients
Protocol compliance (%) 409 (83.3) 13 (65) ⬍0.05 selected for NOM and 88% rate of overall splenic salvage.
Total 491 20 — These results equal or surpass previous single institution and
Data presented as ⫾ standard error of the mean. multicenter reports.11,14,19,20
Although SAE adds to the cost of care, the benefits of
a reduced LOS and avoidance of operation will most likely
DISCUSSION offset this expense. Although our data does not show a signifi-
This study demonstrates that implementation of a pro- cant improvement in primary and secondary outcome measures,
tocol that incorporates SAE for patients with blunt splenic the initial use of SAE may avoid some of the potential compli-
injury has expanded the use of NOM. The use of SAE was cations of hemodynamic instability that invariably occur if SAE
also associated with greater success of NOM, a decline in is reserved only for those patients who are failing NOM. Addi-
mortality, an increase in overall splenic salvage, and a shorter tionally, the improvement in overall splenic salvage rate that
LOS. However, the prospective application of a structured occurred when SAE was used more frequently may provide a
protocol using defined criteria for initial SAE did not further potential benefit of avoiding the occurrence of overwhelming
improve these outcomes. However, we achieved a 97% postsplenectomy infection. This is an uncommon complication
splenic salvage rate in ⬎350 patients who had NOM with use that has a high mortality rate.21
of this protocol. There are a number of limitations to this study. Fore-
The rate of splenectomy and splenorrhaphy decreased most, this is a retrospective analysis which compares data
while the use of SAE and the success of NOM increased over over the span of 16 years. Thus, differences in outcomes can
time. The criteria for use of SAE may have shifted patients be multifactorial and may in part reflect changes in medical
who previously would have had splenorrhaphy or splenec- practice rather than the effects of SAE. There has been a
tomy to have successful NOM, thus increasing the relative national trend to use NOM more often for the treatment of
rate of splenectomy among operated patients. Although the blunt splenic injuries and its success has improved over time.
grade of splenic injury was similar across all the three groups, This may be due to recognition that patients previously
we did notice a decrease in ISS in more recent years, which thought to need operation are good candidates for NOM, a
may reflect an increase in the overall number of patients with greater threshold of the trauma surgeon to not associate the
splenic injury treated at our trauma center.13 need for a low volume transfusion with NOM failure, and an
The use of initial NOM increased during the 16-year increased application of NOM in patients with serious head
period of this study. Of the 109 patients who had initial SAE, injury and multisystem injuries.2,3,18 A recent report has
90% were successfully managed without operation nearly suggested that improvements in the success of NOM may be
matching the NOM success rate of 93% in patients who were due to the increased number of patients treated at designated

© 2009 Lippincott Williams & Wilkins 569

Sabe et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
TABLE 6. Outcomes by AAST Splenic Injury Score of Patients With Blunt Splenic Injury in Groups II and III
SIS Grade n Initial NOM (%) Splenic Artery Embolization (%)
1 129 123 (95.3) 1 (0.8)
2 123 117 (95.1) 9 (7.3)
3 159 146 (91.8) 60 (37.7)
4 145 109 (75.2) 77 (53.1)
5 20 3 (15.0) 3 (15.0)
Total 576 498 150
Figure 2. Initial treatment plan and the outcome for all patients. (A) 93% successful NOM and (B) 90% successful NOM. OR:
operating room.
trauma centers in recent years, which has resulted in a
decrease in ISS and more splenic injuries that are less
severe.22 The small number of failures of NOM and SAE and
the low mortality rate affects the statistical power to draw
conclusions regarding specific parameters associated with
failure of NOM and death. Experience of the interventional
radiologists performing SAE may have affected technical
failure. We did not assess splenic function after SAE, thus are
unable to comment on whether SAE will protect against later
overwhelming infection. Also, we did not evaluate the use of
blood products and transfusion trigger points. The subjectiv-
ity of radiologists, surgeons, and pathologists who graded
these injuries is another limitation. The use of a prospectively
applied protocol may also have had some beneficial effects
due to heightened awareness of treatment of these patients.
Our findings suggest that the use of SAE based on
CT-defined parameters associated with ongoing bleeding was
associated with significant improved success of NOM in
patients with blunt splenic injury who are hemodynamically
stable at presentation to the trauma center. SAE appears to
increase the utility of NOM in patients with a higher degree
of splenic injury, thus avoiding operation and potential sple-
nectomy in these patients. Thus, we conclude that incorpo-
rating a protocol using SAE in selected patients has increased
the use of NOM with no significant deterioration in out-
comes. We acknowledge that as implementation and success
of NOM continue to increase, so does the need for well-
570
Successful NOM (% of Initial NOM) Splenic Salvage (%)
123 (100) 127 (98.4)
115 (98.3) 117 (95.1)
137 (93.8) 141 (88.7)
101 (92.7) 113 (77.9)
2 (66.7) 2 (10.0)
478 500
coordinated multicenter studies to more clearly define factors
associated with failure of NOM. Additionally, further studies
are needed to assess the long-term outcomes of splenic
function after SAE.
REFERENCES
1. Konstantakos AK, Barnoski AL, Plaisier BR, Yowler CJ, Fallon WF
Jr, Malangoni MA. Optimizing the management of blunt splenic
injury in adults and children. Surgery. 1999;126:805– 812; discussion
812– 813.
2. Pachter HL, Guth AA, Hofstetter SR, Spencer FC. Changing patterns in
the management of splenic trauma: the impact of nonoperative manage-
ment. Ann Surg. 1998;227:708 –717; discussion 717–719.
3. Bain IM, Kirby RM. 10 year experience of splenic injury: an increasing
place for conservative management after blunt trauma. Injury. 1998;29:
177–182.
4. Malangoni MA, Cue JI, Fallat ME, Willing SJ, Richardson JD. Evalu-
ation of splenic injury by computed tomography and its impact on
treatment. Ann Surg. 1990;211:592–597; discussion 597–599.
5. Moore EE, Cogbill TH, Jurkovich GJ, Shackford SR, Malangoni MA,
Champion HR. Organ injury scaling: spleen and liver (1994 revision).
J Trauma. 1995;38:323–324.
6. Federle MP, Courcoulas AP, Powell M, Ferris JV, Peitzman AB. Blunt
splenic injury in adults: clinical and CT criteria for management, with
emphasis on active extravasation. Radiology. 1998;206:137–142.
7. Starnes S, Klein P, Magagna L, Pomerantz R. Computed tomographic
grading is useful in the selection of patients for nonoperative management
of blunt injury to the spleen. Am Surg. 1998;64:743–748; discussion
748 –749.
© 2009 Lippincott Williams & Wilkins
The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
8. Thompson BE, Munera F, Cohn SM, et al. Novel computed tomography
scan scoring system predicts the need for intervention after splenic
injury. J Trauma. 2006;60:1083–1086.
9. Anderson SW, Varghese JC, Lucey BC, Burke PA, Hirsch EF, Soto JA.
Blunt splenic trauma: delayed-phase CT for differentiation of active
hemorrhage from contained vascular injury in patients. Radiology.
2007;243:88 –95.
10. Haan JM, Marmery H, Shanmuganathan K, Mirvis SE, Scalea TM.
Experience with splenic main coil embolization and significance of new
or persistent pseudoaneurym: reembolize, operate, or observe. J Trauma.
2007;63:615– 619.
11. Marmery H, Shanmuganathan K, Alexander MT, Mirvis SE. Optimiza-
tion of selection for nonoperative management of blunt splenic injury:
comparison of MDCT grading systems. AJR Am J Roentgenol. 2007;
189:1421–1427.
12. Wei B, Hemmila MR, Arbabi S, Taheri PA, Wahl WL. Angioemboli-
zation reduces operative intervention for blunt splenic injury. J Trauma.
2008;64:1472–1477.
13. Rajani RR, Claridge JA, Yowler CJ, et al. Improved outcome of adult
blunt splenic injury: a cohort analysis. Surgery. 2006;140:625– 631;
discussion 631– 632.
14. Peitzman AB, Heil B, Rivera L, et al. Blunt splenic injury in adults:
Multi-institutional Study of the Eastern Association for the Surgery of
Trauma. J Trauma. 2000;49:177–187; discussion 187–189.
15. Haan J, Scott J, Boyd-Kranis RL, Ho S, Kramer M, Scalea TM.
Admission angiography for blunt splenic injury: advantages and pitfalls.
J Trauma. 2001;51:1161–1165.
16. Harbrecht BG, Ko SH, Watson GA, Forsythe RM, Rosengart MR, Peitzman
AB. Angiography for blunt splenic trauma does not improve the success rate
of nonoperative management. J Trauma. 2007;63:44 – 49.
17. Haan JM, Bochicchio GV, Kramer N, Scalea TM. Nonoperative man-
agement of blunt splenic injury: a 5-year experience. J Trauma. 2005;
58:492– 498.
18. Cadeddu M, Garnett A, Al-Anezi K, Farrokhyar F. Management of
spleen injuries in the adult trauma population: a ten-year experience.
Can J Surg. 2006;49:386 –390.
19. Haan JM, Biffl W, Knudson MM, et al. Splenic embolization revisited:
a multicenter review. J Trauma. 2004;56:542–547.
20. Davis KA, Fabian TC, Croce MA, et al. Improved success in nonoperative
management of blunt splenic injuries: embolization of splenic artery
pseudoaneurysms. J Trauma. 1998;44:1008 –1013; discussion 1013–1015.
21. Malangoni MA, Dillon LD, Klamer TW, Condon RE. Factors influenc-
ing the risk of early and late serious infection in adults after splenectomy
for trauma. Surgery. 1984;96:775–783.
22. Harbrecht BG, Zenati MS, Ochoa JB, Puyana JC, Alarcon LH, Peitzman
AB. Evaluation of a 15-year experience with splenic injuries in a state
trauma system. Surgery. 2007;141:229 –238.
DISCUSSION
Dr. Naoyuki Kaneko (Japan): Dr. Fabian, Britt, AAST
members and guests, I would like to thank you for giving me
the opportunity to discuss Dr. Sabe’s paper here.
I would like to discuss the paper from a surgical
viewpoint. Although the rate of splenic artery embolization
was significantly higher in Group 3 than in Group 2 in your
study, the splenic salvage rate was similar between them.
Furthermore, in patients initially managed non-opera-
tively, the mortality rate was a little higher among the patients
in Group 3 than in Group 2.
I am aware that many factors such as ISS and surgical
technique affect these results. However, that you might have
tried to over-manage the patients with embolization is debatable.
Thus, my first question is, was there not some inappro-
priate decisions at the time of patients selection for emboli-
zation? The title of the article by Smith and Cioffi in 2006,
namely Splenic Artery Embolization, Have We Gone too
Far? is quite impressive for me.
© 2009 Lippincott Williams & Wilkins
A 16-Year Experience With Splenic Trauma
The splenic injury score determined by CT is also
debatable. The CT and operative grade did not agree in 84
percent of patients according to paper from hepatic injury by
Croce and Fabian.
Although the procedure on CT has remarkably improved
over the past ten years, scanning for not only hepatic but also splenic
injuries still seems to result in considerable inconsistency.
My second question is, did you investigate the discrep-
ancy? I would like to know this especially when surgery was
performed after failed non-operative management.
Smith and Cioffi recommended a low threshold for
surgery in their 2006 paper. Furthermore, they were con-
cerned about the ramifications of splenic artery embolization.
I always have two such concerns. One is the inference of
an operation based on the surgical feasibility of spleen preser-
vation. In my experience splenectomy was performed in approx-
imately 65 percent of the initially operated group but in 85
percent of the delayed operation group. Smith et al described
that splenic injuries are best treated by early operation to achieve
maximum splenic preservation. I agree with this.
Another consequence that concerns me is the training
of surgeons with trauma experience. A decrease in the num-
ber of operations leads to a shortage of surgical experience
among the younger surgeons.
In Group 3 you operated on 58 patients over a period of
five years. If you have two younger doctors to educate, each of
them can only operate on an average of five patients per year.
I do have high regard for the development of angioem-
bolization. However, I do not believe that this strategy will
benefit traumatized patients 20 or 30 years down the road for
whom emergency surgery is inevitable.
Surgeons see less trauma cases in Japan than you see so
this issue is more critical. If an emergency doctor to under-
stands embolization well but not surgical techniques makes a
decision of primary management, embolization tends to be
overused with the occasional unfavorable consequences.
Thus, I think we must establish clear borderlines be-
tween recommendations for surgery and the embolization as
soon as possible by lowering the surgical threshold.
My last question is what is your opinion about the
indications for embolization in the light of such two ramifi-
cations? Regardless, I think very highly of your effort to treat
more severe splenic injuries with embolization. Thank you.
Dr. Frederick A. Moore (Houston, Texas): A hundred
and fifteen or 33 percent of your non-operative patients in the
third group underwent angioembolization.
That incidence of angioembolization I think only the
University of Maryland exceeds that. And the University of
Maryland has popularized the concept of screening angiography
where a lot of people go from the CT scan to the angio suite.
And you listed some indications. One of the things
you didn’t mention in the indication is hemodynamic
stability. If you read the MIMS papers they take, tend to
take people who are hemodynamically unstable to angiog-
raphy. Do you do that?
Second, do all patients who undergo the initial angio-
gram get embolized? You had 25 delayed angiograms so my
571
Sabe et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 67, Number 3, September 2009
question is do you routinely get CT scans before hospital
discharge as recommended by President Fabian?
And the last is one about technique. There is a contro-
versy about doing main splenic artery embolization versus
polar embolization. And which technique do you do?
And what is you complication of angioembolization in
regards to failure to stop bleeding and local complications of
abscess and infarcted spleens? Thank you very much.
Dr. Richard J. Mullins (Portland, Oregon): Congrat-
ulations on an excellent presentation. Following some of
these non-operative spleen injured patients in my experience
is challenging.They have fever, pain, pleural effusions, and
return to the ER after hospital discharge with abdominal pain
and bloating. Is that your experience? Did you have to work
pretty hard to get some of these patients through the non-
operative management?
The majority of splenectomies done in this country for
injury are done in smaller hospitals, not comparable to your
Level I trauma center where you can have experts perform an
angiogram within minutes. Do you recommend a surgeon in
a smaller hospital in rural community outside Cleveland who
is called to evaluate a patient who presents with a bleeding
spleen injury, transfer that patient to Metro Hospital so that
the patient can get their splenic artery embolization? Or is it
all right for the surgeon in a smaller hospital to just take out
the spleen and get the job done?
Dr. Ronald J. Simon (New York, New York): Enjoyed
your paper very much. You showed a reduction in mortality
in your first group versus your second and third groups.
Did you look at whether or not morbidity in your and
mortality in your general trauma population was also reduced
during these time periods because you guys just got better in
taking care of patients and that the difference between Groups
1, Group 2 and Group 3 actually had nothing to do with
splenic artery embolization but just had to do with improved
quality of care over the last ten years? Thank you.
Dr. Joseph P. Minei (Dallas, Texas): I have a question
regarding discretionary use of A gram and protocol-driven
use of A gram and the question is, is really how good are your
surgeons at predicting this outcome?
That is, what are the A gram results of discretionary use
versus protocol-driven use? Are the discretionary use A
grams have a lot higher positive rate versus the protocol use.
And the second question is what are the complication
rate? Did you notice a higher complication in the protocol-
driven group with so many more A grams done? Thank you.
Dr. Mark A. Malangoni (Cleveland, Ohio): Thank
you all for your questions. And I will try and address them all.
For Dr. Kaneko, 7 of 109 patients that had initial splenic
artery embolization failed and that compares to 1 of 46
patients that had late splenic artery embolization.
So you can see our failure rate was higher in that group or
in the group that had initial SAE. And it may be because we are
not quite as precise as we wish to be in selecting patients for splenic
artery embolization who will have that done successfully.
It’s very clear that what we’ve done by using this
protocol is we’ve moved more patients that we would operate
on initially into the non-operative management group.
572
And we think that in addition to that we have selected
out those patients that would more likely have qualified
previously for splenic preservation and now moved those into
the group that never gets operated.
So splenectomy for those patients who do require
operation is going to be more common because the degree of
injury is high.
We are fortunate to have 24/7 interventional radiology
coverage. And that clearly allows us to do this. And there are
not all hospitals that have that sort of resource-intensive
radiology support available.
Clearly Dr. Kaneko, the skills of surgeons who will be
doing splenorrhaphy will be challenged in our environment.
However, I think as you can see from some of the
questions that were asked that this is not common at all
institutions so I think there still will be plenty of opportunities
for people to experience splenic preservation in training.
Dr. Moore, as I mentioned, this is very resource-
intensive. Overall our failure rate for splenic artery emboli-
zation is about 5 to 6 percent.
Some of those are technical failures in that the radiologist
is unable to do the cannulation to even get an angiogram. But
most of these are failures after the angiogram has been done.
We typically take unstable patients directly to the
operating room and do not advocate that they should have
angiography or, for that matter, even CT scan. We do not do
routine follow-up CTs on these patients, however.
Dr. Mullins asked about the variability in practice in
hospitals. And, Richard, you are expert in pointing out to us the
differences that occur between Level I trauma centers and
trauma centers that may not have the same capabilities or
surgeons.
And we would agree with your observations previously
published that this doesn’t work for everyone. And I don’t
believe that there is anything wrong with a surgeon, regard-
less of where they practice, taking a patient to operation in
order to do what is right for the patient.
There was a question asked about whether or not our
improvements were due to the protocol or perhaps due to
other factors that may affect our overall mortality.
We did not examine this specifically but I would just tell you
that in my opinion I don’t believe that our mortality decreased by 50
percent in our patients overall. But we did clearly, clearly did not
examine it so I don’t have that data for you.
Dr. Minei, the one thing I would tell you is that for our
protocol-driven patients there was no real difference in com-
plications in the two groups.
We had thought initially that by using splenic artery
embolization at the onset in these high risk patients that we
might be able to improve overall splenic salvage. That was not
the case.
We had a small but incremental improvement in pa-
tients that had non-operative management.
And I think it goes back to our overall goal is that if we
can get this right we should be able to decrease our rate of
failures from 3 percent to even lower. And that’s what we’re
striving to do. Thank you all very much.
© 2009 Lippincott Williams & Wilkins