j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.JournalofSurgicalResearch.com

Readmission With Major Abdominal

Complications After Penetrating Abdominal
Trauma

Kamil Hanna, MD, Samer Asmar, MD, Michael Ditillo, DO,

Mohamad Chehab, MD, Muhammad Khurrum, MD, Letitia Bible, MD,
Molly Douglas, MD, and Bellal Joseph, MD*
Division of Trauma, Critical Care, Emergency Surgery, and Burns, Department of Surgery, College of Medicine,
University of Arizona, Tucson, Arizona

article info abstract

Article history: Background: Despite improvements in operative techniques, major abdominal complica-
Received 10 March 2020 tions (MACs) continue to occur after penetrating abdominal trauma (PAT). This study
Received in revised form aimed to evaluate the burden of MAC after PAT.
13 July 2020 Methods: The (2012-2015) National Readmission Database was queried for all adult (age
Accepted 18 July 2020
18 y) trauma patients with penetrating injuries who underwent exploratory laparotomy
Available online xxx and were readmitted within 6 mo of index hospitalization discharge. Patients were strat-
ified by firearm injuries (FIs) and stab injuries (SIs). Primary outcomes were rates of MAC:
Keywords: intra-abdominal abscesses (IAAs), superficial surgical site infection (SSI), and fascial
Major abdominal complications dehiscence within 6 mo after discharge. Secondary outcomes were both nonabdominal
Readmission complications and mortality, postdischarge, and 6-mo readmission. Regression analysis
Penetrating abdominal trauma was performed.
Damage control laparotomy Results: A total of 4473 patients (FI, 2326; SI, 2147) were included in the study; the mean age
was 32  14 y, the Injury Severity Score was 19 (15-25), and 23% underwent damage control
laparotomy (DCL). The rate of MAC within 6 mo was 22% (IAA 19%, SSI 7%, and fascial
dehiscence 4%). Patients with FIs had a higher rate of IAA (27% versus 10%; P < 0.01), SSI
(11% versus 3%; P < 0.01), fascial dehiscence (5% versus 3%; P ¼ 0.03), nonabdominal com-
plications (54% versus 24%; P < 0.01), and postdischarge mortality (8% versus 6%; P < 0.01)
compared with patients with SIs . On regression analysis, DCL (P < 0.01), large bowel
perforation (P < 0.01), biliary-pancreatic injury (P < 0.01), hepatic injury (P < 0.01), and blood
transfusion (P ¼ 0.02) were predictors of MAC.
Conclusions: MAC developed in one in five patients after PAT. FIs have a higher potential for
hollow viscus injury and peritoneal contamination, and are more predictive of MAC and
nonabdominal complications, especially after DCL.
Level of Evidence: Level III Prognostic.
ª 2020 Elsevier Inc. All rights reserved.

Quickshot Presentation at the American College of Surgeons Clinical Congress, San Francisco, California, October 27-31, 2019.
* Corresponding author. Division of Trauma, Critical Care, and Emergency Surgery, Department of Surgery, University of Arizona, 1501 N.
Campbell Avenue, Room 5411 P.O. Box 245063 Tucson, AZ 85724. Tel.: þ1 520 626 5056; fax: þ1 520 626 5016.
E-mail address: [email protected] (B. Joseph).
0022-4804/$ e see front matter ª 2020 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.jss.2020.07.060
70 j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

Introduction States. Institutional review board approval was exempted

because the NRD only contains deidentified data.
Penetrating trauma is one of the major causes of death glob-
ally, and it is a significant portion of the global trauma burden. Study population
According to estimates, penetrating trauma accounts for 15%-
45% of injuries incurred in the civilian urban population.1,2 We identified all adult (age
18 y) trauma patients with
Penetrating abdominal trauma (PAT) is usually caused by a penetrating injuries who underwent exploratory laparotomy
stab or gunshot wound, and it entails injuries to the small and were readmitted within 6 mo of index hospitalization
bowel (50%), large bowel (40%), solid organs such as the liver discharge. In the NRD, patients cannot be followed from year
(30%), and the vascular system (25%).2 Although the optimal to year. We selected patients who were admitted in the first
management of patients with PAT has been debated for de- 6 mo of the year to ensure a follow-up period of at least 6-
cades and varies based on institutional protocols,3 exploratory 12 mo. To maintain consistency across patients, the follow-up
laparotomy remains the standard of care as a safe approach period was then limited to 6 mo for all patients. Patients who
for the timely identification and treatment of solid organ and were discharged from the emergency department are not
hollow viscus injuries.4 However, open abdominal exploration included in the database. Survey weights were applied for
is associated with significant postoperative morbidity, espe- national estimates according to the Healthcare Cost and Uti-
cially in the setting of damage control laparotomy (DCL), lization Project standards.8 We excluded pediatric patients
which is usually performed in severely injured and physio- and patients who did not undergo an exploratory laparotomy
logically deranged patients.5 (nonoperative management) and those who underwent lapa-
The rising number of nonfatal injuries6 places a significant roscopy. To reduce heterogeneity in the patient sample and
burden on the health care system in terms of the spectrum of wide variations in injury patterns and severity, we decided to
long-term sequelae that often require readmission. Despite be restrictive in our inclusion and exclusion criteria and select
improvements in operative techniques, resuscitation, and adult patients with operatively managed PAT using the open
peritoneal irrigation, major abdominal complications (MACs) approach (i.e., patients who had indications for exploratory
such as intra-abdominal abscesses (IAAs) continue to occur laparotomy).
after PAT, and they account for much of the late morbidity
associated with PAT.7 Although there are multiple reasons Patient stratification
why PAT patients with MACs are likely to be readmitted, the
identification of patients with a higher risk injury pattern Patients were stratified into two groups based on the cause of
would raise the index of suspicion for early detection and their PAT: firearm injuries (FIs) and stab injuries (SIs).
intervention. In the setting of PAT, fecal contamination,
foreign bodies, multiple visceral injuries, extensive peritoneal Data points and definitions
contamination, anastomotic leaks, and continued bleeding
predispose patients to MAC.7 Because there is a paucity of For each patient, we abstracted the following data points from
recent data evaluating the long-term occurrence of MAC after the database: demographic characteristics (age, gender, pri-
PAT, our aim was to evaluate the long-term burden of MAC mary payer, and household income), comorbidities (Charlson
and the time of occurrence after PAT. We hypothesized that Comorbidity Index), injury parameters (median Injury
the 6-mo incidence of MAC is high in operatively managed Severity Score [ISS] and interquartile range, body regions
patients with PAT and is associated with a unique set of pre- Abbreviated Injury Scale, small and large bowel injury, hepatic
dictors that allow the identification of patients at risk. injury, biliary-pancreatic injury, gastric injury, splenic injury,
as well as kidney, ureter, and bladder injuries). Injury patterns
and physiological derangements were decoded using corre-
sponding International Classification of Diseases (ICD)-9
Methods diagnostic codes as shown in Appendix 1. We also decoded
physiological derangements (hypothermia, coagulopathy, and
Data sources acidosis) and management approach (DCL, definitive lapa-
rotomy [DL]). In addition, data were collected on the index
We performed a 4-y analysis of the (2012-2015) Healthcare admission length of stay, readmission, and 6-mo incidence of
Cost and Utilization Project Nationwide Readmission Data- nonabdominal complications (defined as the occurrence of
base (NRD), which is provided by the Agency for Healthcare any of the following complications: acute respiratory distress
Research and Quality. It contains 15 million annual admis- syndrome, pneumonia, myocardial infarction, pulmonary
sions and provides longitudinal information about a nation- embolism, deep vein thrombosis, and acute renal failure).
ally representative sample of patients after index hospital MAC was defined as the occurrence of IAA, superficial surgical
admission and subsequent readmissions throughout 1 y site infection (SSI), or fascial dehiscence within 6 mo post-
within the same state. Sample weights can be used to derive operatively. The list of nonabdominal complications was
national estimates.8 The database uses an identifier to track included based on the American College of Surgeons Trauma
patients across admissions to different hospitals. These fea- Quality Improvement Program data abstraction. Quality
tures make the NRD the largest, most accurate, and compre- improvement programs have specified a number of pertinent
hensive source of hospital readmission data in the United nonabdominal complications to be tracked and recorded in
 hanna et al  abdominal complications after trauma 71

patients with trauma from multiple centers for quality A multivariable logistic regression model was used to
improvement and benchmarking.9 We reported the 6-mo perform the analysis to determine the predictors of MAC
incidence of superficial SSI, which is a superficial infection within 6 mo postoperatively, while adjusting for measurable
located at the surgical incision. We also collected data on confounding variables. In this regression analysis, we
hospital charges throughout the patients’ index hospitaliza- adjusted for demographic characteristics (age, gender, pri-
tion along with hospital-related variables, such as bed size mary expected payer, and median household income), indi-
(categorized into small, medium, or large based on variable vidual comorbidities (hypertension, diabetes mellitus, chronic
cutoff points depending on the urban-rural designation of the obstructive pulmonary disease, congestive heart failure, pe-
hospital), ownership, location, and teaching status. ripheral vascular disease, liver failure, chronic kidney disease,
obesity, alcohol use disorder, autoimmune disease, and being
Damage control versus definitive management underweight), injury parameters (ISS, body regions Abbrevi-
ated Injury Scale, bowel, hepatic, biliary-pancreatic, gastric,
Temporary abdominal closure, as used in the setting of splenic, and kidney, ureter, and bladder injuries), manage-
damage control surgery, is not defined by a single ICD-9 code. ment approach (DCL versus DL), admission physiological de-
Patients who underwent an exploratory laparotomy were rangements (hypothermia, coagulopathy, and acidosis), and
identified using the two ICD-9 codes: 54.11 and 54.19 (explor- hospital parameters.
atory laparotomy and other laparotomy, respectively). DCL was To evaluate the association between each potential pre-
defined as a second laparotomy based on the recurrence of the dictor variable and the development of MAC, we performed a
same two ICD-9 codes or the additional presence of the ICD-9 univariate analysis. Potential predictors with a P value <0.2
code 54.12 (reopening of recent laparotomy site). The number of were then included in the multivariable logistic regression
code recurrences was also used to decode the number of model. Variables were considered significant predictors of
abdominal re-explorations. DL was defined as those without a MAC at a P value <0.05. We then performed the Hosmer-
second laparotomy or a reopening of a recent laparotomy site. Lemeshow test to assess the fitness of our overall model. In
our logistic regression model, the Hosmer-Lemeshow test
exceeded 0.05 and the tolerance was >0.1 for all independent
Transferred patients
variables with a variance inflation factor of <10.0. A Kaplan-
Meier survival analysis was performed to ascertain the me-
On the transfer of patients to another institution, the NRD
dian time to developing MAC in both groups. The median time
does not allow the hospitalization at the receiving hospital to
to developing MAC was compared using the log-rank test. A
be counted as a readmission. The database collapses the pairs
subanalysis comparing outcomes among patients who un-
of records representing a transfer into a single “combined”
derwent DCL was performed to evaluate the effect of the
record and removes the original separate discharge records.
number of subsequent laparotomies on outcomes. We
considered a P value of <0.05 as statistically significant. All
Outcome measures statistical analyses were performed using the Statistical
Package for the Social Sciences (SPSS, version 23; SPSS, Inc,
Primary outcome measures were the 6-mo incidence of MAC: Armonk, NY).
IAA, superficial SSI, and fascial dehiscence. Secondary out-
comes were the 6-mo incidence of nonabdominal complica-
tions (acute respiratory distress syndrome, pneumonia, Results
myocardial infarction, pulmonary embolism, deep vein
thrombosis, and acute renal failure), mortality, and read- A total of 4799 patients with trauma met the inclusion and
mission. Both primary and secondary outcomes were exclusion criteria of the study, of which 4473 survived their
collected in the postdischarge period. index admission and were followed longitudinally. Figure 1
represents the patient flow diagram. The mean age was
Statistical analysis 33  14 y and 87% were male. Most patients belonged to the
lowest income quartile (48%) and were covered by Medicaid
To decide on the optimal statistical test to compare variables (32%), followed by self-pay (24%) and private insurance (20%).
across the two groups, we performed normality testing In addition, 37% of patients had associated comorbidities
including the Shapiro-Wilk test, the Kolmogorov-Smirnov (Charlson Comorbidity Index >0). In terms of their injuries,
test, visual inspection of histograms for continuous vari- most patients sustained moderate injuries, as the median ISS
ables, and Q-Q plots to assess alignment with normal theo- was 10 (4-18), and 35% sustained serious injuries (ISS >15).
retical quantiles. Continuous parametric data were reported The most commonly injured organ was the small bowel
as means (with standard deviations), continuous nonpara- (15%), followed by the liver (12%), large bowel (10%), genito-
metric data were reported as medians (with interquartile urinary organs (8%), and biliary-pancreatic organs (5%).
ranges), and categorical data were reported as proportions. To Moreover, 17% of the sample had at least one element of
analyze the differences between the two groups on a univar- hypotension, coagulopathy, or hypothermia. A total of 1031
iate level, we used the chi-square tests for categorical vari- (23%) patients underwent DCL, and the remaining patients
ables, the Mann-Whitney U-tests for continuous underwent a DL. Of those who underwent DCL, the median
nonparametric data, and the independent Student t-tests for number of subsequent laparotomies was 1 (1-2). The median
continuous parametric data. index hospital length of stay was 7 (3-13) d, and the median
72 j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

Fig. 1 e Patient flow diagram.

index hospitalization health care charges were $72,200 mortality (8% versus 6%; P < 0.01). Table 2 outlines the uni-
($36,569-$157,650). Most patients were admitted to hospitals variate analysis of outcomes. On Kaplan-Meier survival anal-
of large bed size (69%), which were private nonprofit owned ysis, the overall median time to developing MAC was 8 (6-13)
(63%) and were located in large metropolitan areas (60%). d. Patients in the FI group had a significantly shorter time to
Overall, 82% of patients were admitted to teaching hospitals. developing MAC relative to patients with SI (7 [6-12] versus 10
Table 1 outlines the baseline characteristics of the study [8-14] d; P < 0.01). Figure 2 outlines the development of MAC
sample. over time for the two study groups.
In comparison to patients with SIs, patients with FIs were On multivariable regression analysis, FIs were an inde-
younger (P < 0.01), more likely to be male (P < 0.01), in the pendent predictor of MAC (odds ratio [OR] 1.4 [1.11-1.77];
lowest income quartile (P < 0.01), and covered by Medicare P < 0.01) relative to SIs. Other significant predictors were
(P < 0.01). However, patients with FIs were less likely to have DCL (OR 1.63 [1.19-2.24]; P < 0.01) versus DL, large bowel
comorbid conditions (P < 0.01) relative to those in the SI group. penetration (OR 2.56 [1.98-3.3]; P < 0.01), biliary-pancreatic
In terms of their injuries, victims of FIs had a higher ISS injury (OR 1.64 [1.13-2.36]; P < 0.01), hepatic injury (OR 1.57
(P < 0.01) and a greater proportion of serious injuries (P < 0.01) [1.21-2.04]; P < 0.01), age
65 y (OR 1.13 [1.09-1.27]; P < 0.01),
relative to those in the SI group. They were also more likely to packed red blood cell (PRBC) transfusion (OR 1.27 [1.04-1.56];
sustain small bowel (P < 0.01), large bowel (P < 0.01), hepatic P ¼ 0.02). Of all the comorbidities included in the model, the
(P < 0.01), biliary-pancreatic (P < 0.01), and genitourinary in- presence of alcohol use disorder (OR 2.69 [2.14-3.33]; P < 0.01),
juries (P < 0.01). A higher proportion of patients in the FI group diabetes mellitus (OR 3.31 [2.13-5.14]; P < 0.01), autoimmune
underwent DCL (P < 0.01) relative to those in the SI group. diseases (OR 5.21 [3.05-6.17]; P < 0.01), and being underweight
The usage of a cell saver transfusion was restricted to 1% of (OR 1.51 [1.12-2.03]; P < 0.01) were associated with the occur-
the patient population. For these patients, 67% had hepatic rence of MAC. However, cell saver transfusion was not asso-
injuries, 48% had biliary-pancreatic injuries, 20% had kidney- ciated with the occurrence of MAC (OR 0.68 [0.52-1.27];
ureter-bladder injuries, 10% had omental injuries, and 4.1% P ¼ 0.65). Table 3 outlines the independent predictors of
had small and large bowel injuries. MAC obtained from multivariable regression analysis.
The overall 6-mo incidence of MAC was 22% (IAA 19%, SSI On subanalysis of patients who underwent DCL, 1031 pa-
7%, and fascial dehiscence 4%), postdischarge mortality was tients were identified and restratified into those who under-
7%, and 19% of the patients were readmitted within 6 mo. On went a single subsequent laparotomy (SL) (824 [80%]) and
univariate analysis, patients with FIs were more likely to those who underwent more than one SL (207 [20%]). Relative to
develop an IAA (27% versus 10%; P < 0.01), SSI (11% versus 3%; those who underwent one SL, those who underwent more
P < 0.01), fascial dehiscence (5% versus 3%; P ¼ 0.03), and than one SL had higher rates of IAA (34% versus 26%; P < 0.01),
nonabdominal complications (54% versus 24%; P < 0.01) within SSI (26% versus 15%; P < 0.01), fascial dehiscence (87% versus
6 mo postoperatively than patients with SI. In addition, pa- 77%; P < 0.01), and 6-mo postdischarge mortality (6% versus
tients with FI had a higher rate of 6-mo postdischarge 2%; P ¼ 0.004). However, no difference was found in the rate of
 hanna et al  abdominal complications after trauma 73

Table 1 e Baseline characteristics of the study sample.

Variables Overall Firearm injury Stab injury P
(n ¼ 4473) (n ¼ 2326) (n ¼ 2147) value
Demographic characteristics
Age (y), mean  SD 33  13 31  13 35  14 <0.01
Male, n (%) 3869 (87) 2126 (90) 1743 (82) <0.01
Primary expected payer, n (%)
Medicare 285 (6) 70 (3) 215 (10) <0.01
Medicaid 1441 (32) 861 (37) 580 (27) <0.01
Private insurance 891 (20) 419 (18) 472 (22) <0.01
Self-pay 1072 (24) 535 (23) 537 (25) <0.01
Other 786 (18) 442 (19) 344 (16) <0.01
Median household income, n (%)
$1-$37,999 2162 (48) 1325 (56) 837 (40) <0.01
$38,000e$47,999 893 (20) 475 (20) 418 (20) <0.01
$48,000e$63,999 848 (19) 320 (14) 528 (25) <0.01

$64,000 425 (10) 162 (7) 263 (12) <0.01
Comorbidities
CCI >0, n (%) 1653 (37) 683 (29) 970 (45) <0.01
Injury parameters
ISS, median (IQR) 15 (6-19) 19 (8-22) 10 (3-16) <0.01
ISS
15, n (%) 1795 (40) 1280 (54) 515 (24) <0.01
AIS, median (IQR)
Head and neck 2 (1-3) 2 (1-2) 2 (1-3) <0.01
Thorax 1 (1-3) 1 (1-2) 2 (1-3) <0.01
Abdomen 3 (2-3) 3 (3-3) 3 (2-3) <0.01
Extremity 1 (1-1) 1 (1-1) 1 (1-1) <0.01
Hypothermia, n (%) 78 (2) 65 (3) 13 (0.6) 0.001
Coagulopathy, n (%) 38 (1) 22 (0.9) 16 (0.7) 0.001
Acidosis, n (%) 543 (12) 323 (14) 220 (10) <0.01
Intraoperative intervention
Gastric injury, n (%) 476 (11) 326 (14) 150 (7) <0.01
Small bowel injury, n (%) 645 (14) 544 (23) 101 (5) <0.01
Large bowel injury, n (%) 452 (10) 390 (17) 62 (3) <0.01
Hepatic injury, n (%) 539 (12) 418 (18) 121 (6) <0.01
Splenic injury, n (%) 428 (10) 256 (11) 172 (8) <0.01
Biliary-pancreatic injury, n (%) 203 (5) 148 (6) 55 (3) <0.01
Omental injury, n (%) 210 (5) 97 (4) 113 (5) 0.054
KUB injury, n (%) 342 (8) 295 (13) 47 (2) <0.01
Anastomosis, n (%) 237 (5) 194 (8) 43 (2) <0.01
Ostomy, n (%) 162 (4) 129 (6) 33 (2) <0.01
Damage control laparotomy, n (%) 1031 (23) 830 (35) 201 (10) <0.01
Number of relaparotomies, median (IQR) 1 (1-2) 1 (1-2) 1 (1-2) 0.074
Packed RBC transfusion, n (%) 1352 (30) 837 (36) 515 (24) <0.01
Cell saver transfusion, n (%) 49 (1) 42 (2) 7 (0.3) <0.01
Total parenteral nutrition 95 (2) 77 (3) 18 (1) <0.01
LOS (d), median (IQR) 7 (5-18) 9 (5-20) 5 (3-9) <0.01
Index hospitalization health care charges, $1000, 80 (41-190) 109 (50-240) 50 (31-94) <0.01
median (IQR)
Hospital variables
Bed size
Small 447 (10) 140 (6) 279 (13) <0.01
(continued)
74 j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

Table 1 e (continued )
Variables Overall Firearm injury Stab injury P
(n ¼ 4473) (n ¼ 2326) (n ¼ 2147) value
Medium 939 (21) 582 (25) 386 (18) <0.01
Large 3086 (69) 1605 (69) 1481 (69) 0.13
Ownership
Government 1163 (26) 628 (27) 537 (25) 0.14
Private nonprofit 2818 (63) 1465 (63) 1374 (64) 0.38
Private investor owned 492 (11) 233 (10) 258 (12) 0.17
Location
Large metropolitan 2684 (60) 1535 (66) 1181 (55) <0.01
Small metropolitan 1655 (37) 768 (33) 859 (40) <0.01
Micropolitan 89 (2) 23 (1) 86 (4) <0.01
Rural 45 (1) 23 (1) 21 (1) 0.88
Teaching hospital 3668 (82) 1977 (85) 1696 (79) <0.01

AIS ¼ Abbreviated Injury Scale; CCI ¼ Charlson Comorbidity Index; IQR ¼ interquartile range; ISS ¼ Injury Severity Score; KUB ¼ kidney-ureter-
bladder; LOS ¼ length of stay; RBC ¼ red blood cell; SD ¼ standard deviation.

6-mo readmission (34% versus 23%; P ¼ 0.116). Table 4 outlines also promotes preventive practices that improve the quality of
the results of the subanalysis of patients who underwent DCL. care and reduce readmission rates.
Despite multiple improvements in operative interventions,
peritoneal irrigation, the range of available antibiotics, and
critical care, the incidence of MAC remains relatively high,
Discussion and it should be taken into consideration when evaluating the
safety and quality of care at trauma centers. There are no
The results of our study indicate that in a cohort of operatively recent reports describing the overall incidence of MAC over a
managed PAT patients with moderate to severe injury who long-term period, and many studies evaluating these com-
underwent exploratory laparotomy on admission, approxi- plications vary in terms of their follow-up period. Morales
mately one in five patients will develop MAC within 6 mo et al.11 conducted a prospective cohort study of 916 abdominal
postoperatively. They also indicate that there is a spectrum of trauma patients who underwent an exploratory laparotomy
MAC (i.e., IAA, fascial dehiscence, and SSIs) that can occur aiming to estimate the incidence of intra-abdominal in-
with significant rates in the postdischarge period. The data fections over 30 d after operative intervention and to identify
demonstrated multiple factors that increase the likelihood of pertinent risk factors that contribute to the overall incidence.
MAC, including firearm involvement, patient age, and man- They reported that the incidence of abdominal infections was
agement approach (DCL versus DL). These complications may 11%, which is lower than the incidence reported in our study.
require the escalation of the acuity of care and lead to multiple Despite a comparable ISS between the two cohorts, there are
organ failure, abdominal wall defects, and possibly gastroin- multiple reasons that could potentially explain this discrep-
testinal tract fistulae.10 Understanding these factors is of ancy. They studied a relatively younger patient population
paramount importance to clinicians who need to risk stratify and used an age cutoff of 12 y. There was also a significantly
patients and gauge their index of suspicion in the setting of a lower rate of utilization of DCL as only 2% of their patients
readmission event. The identification of patients at high risk underwent an abbreviated procedure. We hypothesize that

Table 2 e Univariate analysis results.

Major abdominal complications Firearm injury (n ¼ 2326) Stab injury (n ¼ 2147) P value
IAA, n (%) 628 (27) 214 (10) <0.01
Time to abscess (d), median (IQR) 7 (6-12) 10 (8-14) <0.01
SSI, n (%) 255 (11) 64 (3) <0.01
Fascial dehiscence, n (%) 116 (5) 64 (3) 0.03
Nonabdominal complications, n (%) 1276 (54) 506 (24) <0.01
6-mo readmission, n (%) 418 (18) 429 (20) 0.097
6-mo postdischarge mortality, n (%) 186 (8) 128 (6) <0.01

IAA ¼ intra-abdominal abscess; IQR ¼ interquartile range; Non-abdominal complications ¼ ¼ acute respiratory distress syndrome, pneumonia,
myocardial infarction, pulmonary embolism, deep vein thrombosis, or acute renal failure; SSI ¼ surgical site infection.
 hanna et al  abdominal complications after trauma 75

Fig. 2 e Major abdominal complication development over time. Color version of figure is available online.

the major reason for the observed discrepancy is the differ- but could be related to immunologic alterations and immu-
ences in the mechanism of injury between the two patient nologically active substances, such as cytokines, histamine,
cohorts. Only 10% of their patient sample sustained PAT, and proinflammatory lipids usually found in stored PRBC
whereas the majority sustained a blunt abdominal trauma units.15 Intraoperative cell salvage was not commonly used in
unlike our patient cohort where all patients presented with our patient cohort, and it did not increase the odds of MAC.
PAT. When examining their data for the rate of abdominal However, its usage was restricted to patients with relatively
infections including only patients with SIs or FIs, the inci- lower peritoneal contamination. A randomized controlled
dence would be 18.1%, which is roughly in line with the trial by Bowley et al.16 concluded that the use of cell salvage
findings of our study. did not lead to an increase in the rates of infection and mor-
On multivariable logistic regression analysis after adjust- tality in patients with PAT. Although the effect of cell salvage
ing for measurable imbalances between the two groups, we on infection rates is in line with the findings of that particular
observed that FIs had higher odds of MAC. Although different trial, our results do not completely support this conclusion
weapons have the potential to lead to injuries of varying given the low overall rate of utilization of cell salvage and
severity, stabbing weapons are usually more localized and lower degree of peritoneal contamination.
may have a lower depth of penetration, thus limiting the Patients who underwent DCL had significantly higher rates
insult to the peritoneal cavity.12 FI, however, may traverse the of complications. Apparently, the rates of MAC were
abdomen through a certain trajectory and at a higher velocity augmented when the number of re-exploration events was
of penetration, resulting in the potential to cause more dam- higher. George et al.17 conducted a matched analysis
age.11 Consequently, the degree of peritoneal contamination comparing DL to DCL. Although there were sample size limi-
may differ when considering the weapon causing PAT. All tations in the study that could have led to a type 2 error, pa-
these circumstances may dictate the eventual probability of tients who underwent DCL had clinically significant higher
infection. Not surprisingly, injuries with a higher potential for rates of MAC and fascial dehiscence. This could be attributed
peritoneal contamination were found to be independent pre- to the morbidity of temporary abdominal closure.18-22 A trial is
dictors of MAC, such as large bowel penetration, biliary- being conducted regarding this question to provide definitive
pancreatic injury, and hepatic injuries. Regardless of injury answers on the benefits and risks of DCL.23 Regardless of the
patterns and operative interventions, there are also pertinent impact of DCL, the percentage of patients who underwent DCL
intraoperative variables that played a role in determining the in our patient cohort approaches the upper limit of what was
incidence of MAC. reported in the literature. According to estimates, up to 25% of
Receiving a PRBC transfusion, which could be intra- trauma patients undergoing emergent exploration of the
operative or not intraoperative, was found to be indepen- abdomen may be unsuitable for primary fascial closure at the
dently associated with adverse outcomes. This is in index operation.24
agreement with the existing body of literature.13,14 The Multiple measures can be taken to limit the occurrence of
mechanisms behind this association are not fully understood, MAC. Timely exploration of the abdomen25 to control
76 j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

are also other measures, such as the early initiation of broad

Table 3 e Multivariable logistic regression analysis.
spectrum antibiotics based on community- and hospital-
Major abdominal aOR 95% CI P acquired infections,31 copious irrigation, source control,
complications value
effective control of peritoneal contamination, adequate
Firearm injury versus stab injury 1.40 1.11- <0.01 debridement or drainage during the initial operation,10 and
1.77 stoma creation when the risk of anastomotic leak is high.32
DCL versus DL 1.63 1.19- <0.01 The limitations of this study are attributed to the retro-
2.24 spective nature of the analysis, the effect of unmeasurable
Large bowel penetration 2.56 1.98- <0.01 confounding factors (intraoperative interventions, peritoneal
3.30 irrigation, and antibiotic use), and erroneous database en-
Biliary-pancreatic injury 1.64 1.14- <0.01 tries. Patients with nonoperatively managed PAT were not
2.36 included, and the entire spectrum of MAC after any PAT was
Hepatic injury 1.57 1.21- <0.01 not evaluated based on the selected patient cohort. Patients
2.04 with laparoscopy or nonoperatively managed PAT were
Age
65 y 1.13 1.09- <0.01 excluded, and the findings of the study are only generalizable
1.27 to PAT patients who underwent an exploratory laparotomy.
Packed RBC transfusion 1.27 1.04- 0.02 As a result, we may have underestimated the overall burden
1.56 of MAC. To reduce heterogeneity in the patient sample and
Cell saver transfusion 0.68 0.52- 0.65 wide variations in injury patterns and severity, we decided to
1.27 be restrictive in our inclusion and exclusion criteria and
Alcohol use disorder 2.69 2.14- <0.01 select patients with operatively managed PAT using the open
3.33 approach, that is, patients who had indications for explor-
Diabetes mellitus 3.31 2.13- <0.01 atory laparotomy. The study duration was also restricted to
5.14 2011-2015. However, because practice patterns can change
Autoimmune diseases 5.21 3.05- <0.01 with time, more recent nationwide data are needed. Recent
6.17 versions of the NRD have undergone significant changes in
Underweight 1.51 1.12- <0.01 the way diagnostic and procedure data are coded. To prevent
2.03 variations and nonuniformity in the way the data are
Model was adjusted for demographic characteristics (age, gender, analyzed we could not include data from more recent ver-
primary expected payer, and median household income), individ- sions of the database. Because of the retrospective nature of
ual comorbidities, injury parameters (ISS, body regions’ AIS, bowel, the analysis, the cause of death cannot be discerned using
hepatic, biliary-pancreatic, gastric, splenic, and KUB injuries),
this database. The NRD includes mortality only when
management approach (DCL versus DL), admission physiological
derangements (hypothermia, coagulopathy, and acidosis), and
occurring in the context of a hospital admission (index
hospital parameters. admission or any readmission event postdischarge). In
aOR ¼ adjusted odds ratio; AIS ¼ Abbreviated Injury Scale; CI ¼ addition, the NRD tracks patients only on readmission when
confidence interval; DCL ¼ damage control laparotomy; occurring within the same state. Patients may be missed if
DL ¼ definitive laparotomy; ISS ¼ Injury Severity Score; KUB ¼ they were not readmitted or if they were managed in an
kidney-ureter-bladder; RBC ¼ red blood cell.
outpatient setting. Thus, it is possible that we have under-
estimated the incidence. Because pediatric patients were
excluded, the findings of this study are nongeneralizable to
hemorrhage26,27 could reduce transfusion requirements, this unique patient population. However, this study fills a gap
which is a risk factor for infections. The selective use of DCL is in the literature on the burden of MAC in adult patients with
another strategy.28-30 Aiming for fascial closure during the PAT, which had not been previously well described. It also
first exploratory laparotomy, if the patient’s physiology per- ascertains the contribution of the different perioperative
mits, could avert the morbidity of MAC. Furthermore, there factors to the overall burden.

Table 4 e Subanalysis of the patients who underwent damage control laparotomy.

Major abdominal complications 1 SL (n ¼ 824) >1 SL (n ¼ 207) P value
IAA, n (%) 216 (26) 71 (34) <0.01
SSI, n (%) 123 (15) 53 (26) <0.01
Fascial dehiscence, n (%) 85 (10) 47 (23) <0.01
Nonabdominal complications, n (%) 635 (77) 181 (87) <0.01
6-mo readmission, n (%) 192 (23) 71 (34) 0.116
6-mo postdischarge mortality, n (%) 16 (2) 12 (6)c 0.004

IAA ¼ intra-abdominal abscess; Nonabdominal complications ¼ acute respiratory distress syndrome, pneumonia, myocardial infarction,
pulmonary embolism, deep vein thrombosis, or acute renal failure; SL ¼ subsequent laparotomy; SSI ¼ surgical site infection.
 hanna et al  abdominal complications after trauma 77

HCaU. Rockville, MD: Agency for Healthcare Research and

Conclusions Quality; 2015.
9. Shafi S, Nathens AB, Cryer HG, et al. The trauma quality
One in five patients develops MAC after PAT. In contrast to an improvement program of the American College of Surgeons
SI, an FI has a higher potential for hollow viscus injury and Committee on Trauma. J Am Coll Surg. 2009;209:521e530.
10. West MA. Cripps MW Infections. In: Moore EE, Feliciano DV,
peritoneal contamination, and it is more predictive of both
Mattox KL, eds. Trauma, 8e. New York, NY: McGraw-Hill
MAC and nonabdominal complications, especially in the
Education; 2017.
setting of DCL. The incidence of MAC remains relatively high 11. Morales CH, Villegas MI, Villavicencio R, et al. Intra-
and should be taken into consideration when evaluating the abdominal infection in patients with abdominal trauma. Arch
safety and quality of care at trauma centers. Surg. 2004;139:1278e1285.
12. Jeroukhimov I, Wiser I, Hershkovitz Y, et al. Frequency of
intra-abdominal organ injury is higher in patients with
concomitant stab wounds to other anatomical areas. BMC
Acknowledgment Emerg Med. 2018;18:18.
13. Moore EE, Dunn EL, Moore JB, Thompson JS. Penetrating
This research is the authors own work and did not receive any abdominal trauma index. J Trauma. 1981;21:439e445.
14. Edna T-H, Bjerkeset T. Association between blood transfusion
specific grant from funding agencies in the public, commer-
and infection in injured patients. J Trauma. 1992;33:
cial, or not-for-profit sectors. 659e661.
Author contributions: B.J., K.H., S.A., M.K., and L.B. designed 15. Nielsen HJ, Edvardsen L, Vangsgaardt K, Dybkjaer E. Skov PS
this study. Mic.D., M.K., B.J., K.H., and S.A. searched the liter- time-dependent histamine release from stored human blood
ature. Mic.D., M.C., Mol.D., S.A., K.H., and B.J. collected the data. products. Br J Surg. 1996;83:259e262.
Mol.D., S.A., K.H., and B.J. analyzed the data. All authors 16. Bowley DM, Barker P, Boffard KD. Intraoperative blood
salvage in penetrating abdominal trauma: a randomised,
participated in data interpretation and manuscript
controlled trial. World J Surg. 2006;30:1074e1080.
preparation.
17. George MJ, Adams SD, McNutt MK, et al. The effect of damage
control laparotomy on major abdominal complications: a
Disclosures matched analysis. Am J Surg. 2018;216:56e59.
18. Hatch QM, Osterhout LM, Podbielski J, et al. Impact of closure
The authors reported no proprietary or commercial interest in at the first take back: complication burden and potential
overutilization of damage control laparotomy. J Trauma Acute
any product mentioned or concept discussed in this article.
Care Surg. 2011;71:1503e1511.
There are no identifiable conflicts of interests to report.
19. Montalvo JA, Acosta JA, Pablo R, Kathia A, Andrés S. Surgical
complications and causes of death in trauma patients that
require temporary abdominal closure. Am Surg.
Supplementary data 2005;71:219e224.
20. Shapiro MB, Jenkins DH, Schwab CW, Rotondo MF. Damage
control: collective review. J Trauma Acute Care Surg.
Supplementary data to this article can be found online at
2000;49:969e978.
https://doi.org/10.1016/j.jss.2020.07.060.
21. Smith BP, Adams RC, Doraiswamy VA, et al. Review of
abdominal damage control and open abdomens: focus on
gastrointestinal complications. J Gastrointest Liver Dis.
references 2010;19:425e435.
22. Diaz Jr JJ, Cullinane DC, Dutton WD, et al. The management of
the open abdomen in trauma and emergency general surgery:
1. Brasel KJ. Epidemiology. In: Moore EE, Feliciano DV, part 1ddamage control. J Trauma Acute Care Surg.
Mattox KL, eds. Trauma, 8e. New York, NY: McGraw-Hill 2010;68:1425e1438.
Education; 2017. 23. Harvin JA, Podbielski J, Vincent LE, et al. Damage control
2. Burns AWB. Penetrating Abdominal Trauma. Treasure Island, laparotomy trial: design, rationale and implementation of a
FL: StatPearls; 2019. randomized controlled trial. Trauma Surg Acute Care Open.
3. Biffl WL, Leppaniemi A. Management guidelines for 2017;2:e000083.
penetrating abdominal trauma. World J Surg. 24. Miller RS, Morris Jr JA, Diaz Jr JJ, Herring MB, May AK.
2015;39:1373e1380. Complications after 344 damage-control open celiotomies. J
4. Shaftan GW. Indications for operation in abdominal trauma. Trauma Acute Care Surg. 2005;59:1365e1374.
Am J Surg. 1960;99:657e664. 25. Schein M, Saadia R. Peritonitis: contamination and
5. Harvin JA, Wray CJ, Steward J, et al. Control the damage: infectiondPrinciples of treatment. Schein’s common sense
morbidity and mortality after emergent trauma laparotomy. emergency abdominal surgery. Verlag Berlin Heidelberg:
Am J Surg. 2016;212:34e39. Springer; 2010:99e105.
6. Nonfatal Injury Data. National Center for Injury Prevention and 26. Li J, Sun SL, Tian JH, et al. Cell salvage in emergency trauma
Control, CDC. Hyattsville, MD: Centers for Disease Control and surgery. Cochrane Database Syst Rev. 2015;1:CD007379.
Prevention, National Center for Injury Prevention and 27. Brown CVR, Foulkrod KH, Sadler HT, et al. Autologous blood
Control; 2017. transfusion during emergency trauma operations. Arch Surg.
7. Gibson DM, Feliciano DV, Mattox KL, Gentry LO, Jordan Jr GL. 2010;145:690e694.
Intraabdominal abscess after penetrating abdominal trauma. 28. Higa G, Friese R, O’Keeffe T, et al. Damage control laparotomy:
Am J Surg. 1981;142:699e703. a vital tool once overused. J Trauma. 2010;69:53e59.
8. Project HCaU Cost-to-Charge Ratio Files: User Guide for 29. Watson JJ, Nielsen J, Hart K, et al. Damage control laparotomy
Nationwide Readmissions Database (NRD) CCRs. In: Project utilization rates are highly variable among Level I trauma
78 j o u r n a l o f s u r g i c a l r e s e a r c h  j a n u a r y 2 0 2 1 ( 2 5 7 ) 6 9 e7 8

centers: PROPPR findings. J Trauma Acute Care Surg. 31. Goldberg SR, Anand RJ, Como JJ, et al. Prophylactic antibiotic
2017;82:481. use in penetrating abdominal trauma: an Eastern Association
30. Joseph B, Azim A, Zangbar B, et al. Improving mortality in for the Surgery of Trauma practice management guideline. J
trauma laparotomy through the evolution of damage control Trauma Acute Care Surg. 2012;73:S321eS325.
resuscitation: analysis of 1,030 consecutive trauma 32. Sharpe JP, Magnotti LJ, Fabian TC, Croce MA. Evolution of the
laparotomies. J Trauma Acute Care Surg. 2017;82: operative management of colon trauma. Trauma Surg Acute
328e333. Care open. 2017;2:e000092.