ORIGINAL ARTICLE

Outcomes of open abdomen versus primary closure following

emergent laparotomy for suspected secondary peritonitis:
A propensity-matched analysis
Angela M. Kao, MD, Lawrence N. Cetrulo, MD, Maria R. Baimas-George, MD, Tanushree Prasad, MA,
Brant Todd Heniford, MD, Bradley R. Davis, MD, and Kevin R. Kasten, MD, Charlotte, North Carolina

BACKGROUND: Optimal management following index laparotomy is poorly defined in secondary peritonitis patients. Although “open abdomen”
(OA), or temporary abdominal closure with planned relaparotomy, is used to reassess bowel viability or severity of contamination,
recent studies demonstrate comparable morbidity and mortality with primary abdominal closure (PC). This study evaluates differences
between OA and PC following emergent laparotomy.
METHODS: Using the Premier database at a quaternary care center (2012–2016), nontrauma patients with secondary peritonitis requiring
emergent laparotomy were identified (N = 534). Propensity matching for PC (n = 331; 62%) or OA (n = 203; 38%) was performed
using variables: Mannheim Peritonitis Index, lactate, and vasopressor requirement. One hundred eleven closely matched pairs (PC:
OA) were compared.
RESULTS: Five hundred thirty-four patients (55.0% female; mean age, 59.6 ± 15.5 years) underwent emergent laparotomy. Of the OA patients,
136 (67.0%) had one relaparotomy, while 67 (33.0%) underwent multiple reoperations. Compared to daytime cases, laparotomies
performed overnight (6 pm–6 am) had more temporary closures with OA (42.8% OA vs. 57.2% PC, p = 0.04). When assessing by
surgeon type, PC was performed in 78.7% of laparotomies by surgical subspecialties compared to 56.7% (p < 0.0001) of
acute care surgeons. After propensity matching, OA patients had increased postoperative complications (71.2% vs. 41.4%,
p < 0.0001), mortality (22.5% vs. 11.7%, p = 0.006), and longer median length of stay (13 vs. 9 days, p = 0.0001).
CONCLUSION: Open abdomen was performed in 38.0% of patients, with one-third of those requiring multiple reoperations. Complications, mortality
rates, and costs associated with OAwere significantly increased when compared to PC. Given these findings, future studies are needed
to determine appropriate indications for OA. (J Trauma Acute Care Surg. 2019;87: 623–629. Copyright © 2019 Wolters Kluwer
Health, Inc. All rights reserved.)
LEVEL OF EVIDENCE: Therapeutic/care management, level IV.
KEY WORDS: Open abdomen; vacuum-assisted closure; on-demand laparotomy; emergency; peritonitis.

T he open abdomen (OA) technique, or temporary abdominal

closure with planned relaparotomy, was first discussed by
Ogilvie in World War II.1 Open abdomen was resurrected in
the 1970s as a life-saving damage control strategy with its role
in a trauma surgeon's armamentarium well-described.2 The
indications and outcomes related to the use of OA in nontrauma
patients requiring emergent laparotomy, however, has not been
clearly delineated.3 Its use in secondary peritonitis has been sug-
gested owing to the substantial associated morbidity, mortality,
and subsequent burden imposed on the health care system.4 With
the availability and convenience of temporary abdominal closure
methods such as Abthera vacuum-assisted closure (VAC), pa-
tients increasingly undergo OA in the setting of mesenteric
Submitted: November 5, 2018, Revised: March 6, 2019, Accepted: March 31, 2019,
Published online: April 25, 2019.
From the Division of Gastrointestinal and Minimally Invasive Surgery, Department of
Surgery, Carolinas Medical Center, Charlotte, NC (A.M.K., L.N.C., M.B-G., T.P.,
B.T.H., B.R.D., K.R.K.).
Podium presentation at American College of Surgeons 104th Annual Clinical Con-
gress, Scientific Forum, Boston, MA, October 2018.
Address for reprints: Kevin R. Kasten, MD, 1025 Morehead Medical Drive, Suite 300
Charlotte, NC 28203; email: kevin.kasten@atriumhealth.org.
DOI: 10.1097/TA.0000000000002345
J Trauma Acute Care Surg
Volume 87, Number 3
ischemia with questionable bowel viability, gross contamination
from bowel perforation or diverticulitis, and/or in situations of he-
modynamic instability due to concern for anastomosis construc-
tion.5 Unfortunately, with the adoption of temporary abdominal
closure techniques in nontraumatic situations, surgeons find
themselves without consensus guidelines regarding appropriate
indications and technical recommendations.6
Before widespread adoption of a new surgical technique
occurs, it is important to determine the appropriate indications
and technical recommendations, especially when there is a high
potential for complications. Recent studies demonstrate an in-
creased risk of complications with OA, most notably the forma-
tion of enteroatmospheric fistulas.7,8 A randomized controlled
trial failed to show a survival benefit with planned relaparotomy
compared to an “on-demand” approach when secondary perito-
nitis patients underwent primary fascial closure at initial opera-
tion.9 Even with the pendulum shifting away from routine OA
utilization,10,11 it continues to be performed in a subgroup of
critically ill patients with physiologic or anatomical derangements
the operative surgeon believes precludes abdominal closure.5
However, evidence to support the role of OA in these situations
is contradictory and inconclusive. While several studies have
shown comparable outcomes between OA and primary closure
623

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.


Kao et al.
(PC), others have demonstrated reduced in-hospital mortality
using OA with VAC after adjusting for Acute Physiologic Assess-
ment and Chronic Health Evaluation (APACHE) scores.12–14 In
addition to disease severity, the initial operative approach often
depends on the source and degree of contamination, patient's
physiologic stability, and surgeon/institution comfort using
OA.14 Few studies have evaluated patient and surgical conditions
to determine when and why surgeons may be more inclined to
temporarily close the abdomen. This paucity of data contributes
to nonstandard protocols used throughout multiple medical cen-
ters and indicates a need for robust study.
The efficacy and optimal scenarios for use of OA are dif-
ficult to assess owing to significant patient heterogeneity, physi-
cian comfort with OA, and lack of strict indications to guide the
unfamiliar surgeon. This institutional analysis of patients under-
going emergent laparotomy seeks to evaluate morbidity and
mortality associated with OA management, using propensity-
matching adjustment for inherent selection differences associ-
ated with each treatment group. Based on literature review, we
hypothesize that the use of OA for surgical management of
suspected secondary peritonitis is associated with higher in-
hospital complication and mortality rates when compared with
PC. Here, we aim to describe surgeon use of OA at a quaternary
care center including patient demographics, clinical indications,
and subsequent outcomes. This study adds much-needed data
regarding use of OA in nontrauma patients.
METHODS
Study Design
Following institutional board review approval, an external
administrative database was retrospectively queried. The study
Figure 1. International Classification of Diseases, Ninth Revision / International Classification of Diseases, Ninth Revision diagnosis codes.
624
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg
Volume 87, Number 3
took place at Carolinas Medical Center in Charlotte, NC,
which serves as a regional quaternary referral center for North
and South Carolina. The study group included all patients
with secondary peritonitis who underwent an emergent
laparotomy between 2012 and 2016. Secondary peritonitis
was defined as active or impending peritoneal infection
secondary to intraabdominal or intrapelvic source of hollow
viscus perforation, bowel ischemia/necrosis, or nonbacterial
peritonitis.15 Patients demonstrating physical examination
findings consistent with localized or generalized peritonitis
and diagnoses including closed loop obstruction, volvulus,
and strangulated or internal hernias were also selected for
inclusion given the natural progression to bowel ischemia
and perforation in the absence of surgical treatment. Patients
were identified from the database based on International
Classification of Diseases (ICD), Ninth and Tenth versions,
and Clinical Modification (ICD-9-CM and ICD-10-CM).
Patients with specific diagnoses (Fig. 1) who underwent an
emergent laparotomy were selected from the database.
Emergent laparotomy was defined as an operative procedure
performed within 48 hours of surgery consultation. Patient’s
records were reviewed by two independent sources (A.K.
and L.C.) to confirm the emergent nature of the procedure.
Exclusion criteria included age younger than 18 years, index
laparotomy at an outside institution, and patients with
diagnosis of acute or necrotizing pancreatitis given its
propensity for reoperative management.
Data Collection
Data were collected retrospectively from electronic medi-
cal records and operative notes. Complications and cost data
were obtained using the Premier database. Patients were
© 2019 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg
Volume 87, Number 3
separated into two treatment groups based on approach at index
laparotomy (primary abdominal closure and open abdomen),
with the clinical definitions as follows:
Open abdomen (OA)—At the termination of the index
laparotomy, the fascia is left open with an Abthera VAC device
placed intra-abdominally and connected to negative pressure
suction with plan for repeat laparotomy before ultimate closure.
Patients were subsequently returned to the operating room for
reassessment of the peritoneal cavity and all other indicated pro-
cedures. Timing of relaparotomy was based on clinical stability
and the discretion of the managing surgical team, typically oc-
curring within 24 to 48 hours. Based on relaparotomy findings,
patients either underwent fascial closure or repeat OA with sub-
sequent laparotomies as necessary.
Primary abdominal closure (PC)—During the index lapa-
rotomy, patient's fascia is reapproximated primarily as definitive
closure and additional reoperation is performed on-demand
based on clinical suspicion by the surgical team. Indications
for reoperation included clinical deterioration or failure of im-
provement due to suspected intraabdominal cause.
Perioperative data were recorded and included patients’
demographics, clinical presentation, surgical indication, postop-
erative complications, and hospital cost. Indication for emergent
laparotomy as defined in this study was based on retrospective
review of intraoperative findings in conjunction with operative
notes and preoperative variables, while surgeon preoperative
indication for emergent laparotomy was based on physical
examination findings and clinical signs/symptoms of peritonitis.
Surgical indications were categorized into obstruction, ischemia/
necrosis, contamination/sepsis, and perforation. Patients with
presence of multiple findings, such as closed loop obstruction
leading to perforation, were categorized into the latter group ac-
cording to their most emergent indication for surgery. Disease se-
verity was measured using the Mannheim Peritonitis Index (MPI)
and the APACHE-II score. The MPI includes patient's age, pres-
ence of malignancy, duration of preoperative symptoms, source of
sepsis, quality of intraabdominal exudate, and chronic organ fail-
ure.16 The APACHE-II score integrates patient's age, physiologic
parameters, and laboratory data within the first 24 hours of inten-
sive care unit admission.17 Operative details included time of day
surgery performed, operative duration, source and etiology of
peritonitis, and date of abdominal closure. Intraoperative details
collected from anesthesia records also included presence of hypo-
tension (defined as mean arterial pressure <65) and need for
blood transfusions or vasopressors. In patients managed with
OA, the number of subsequent operations (after index laparot-
omy) and the procedures performed (i.e., bowel resection,
ileostomy creation) during relaparotomies were recorded. When
available from the operative note, the reason for OA technique
was also noted. Postoperative data were obtained from the Pre-
mier database and included total hospital length of stay (LOS),
hospital billed charges (costs), in-hospital complications, in-
hospital mortality, discharge destination, and 30-day readmission.
Reoperation performed after abdominal fascial closure was de-
fined as an unplanned laparotomy.
Primary outcomes included in-hospital complications and
in-hospital mortality. Secondary outcomes included unplanned
relaparotomy and readmission. Hospital system measures of
interest included LOS and hospital costs. All outcomes were
© 2019 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
Kao et al.
reported as number and frequency, mean with standard deviation,
or median with interquartile range if non-normally distributed.
Statistical Analysis
Descriptive statistics and univariate analysis were performed
to describe patients’ demographics and operative variables. Both
unplanned relaparotomy and death occurring within 48 hours of
initial laparotomy occurred in each treatment group; therefore,
comparisons were made using an intention-to-treat analysis based
on the operating surgeon's planned approach at the index laparot-
omy. Differences in the study population by method of abdominal
closure during index laparotomy were reported using the χ2 test for
comparing the proportions and the Wilcoxon rank-sum test for
comparing the medians, with statistical significance set at two-
tailed p < 0.05. Propensity score nearest neighbor matching
was performed for PC or OA based on logistic regression model,
generating 1:1 closely matched pairs for comparison. Con-
founding variables used for propensity matching included pre-
operative MPI score, vasopressor requirements during index
laparotomy, and lactate level of 3 or greater. Selection of
propensity-matching variables was based on statistical signifi-
cance as well as clinical applicability and relevance to intraoper-
ative decision to close the abdomen during initial laparotomy.
While other variables also reached statistical significance on
univariate analysis, many were institution specific and therefore
were not included to limit the influence of factors outside the
surgeon's control. The combination of MPI score, lactate level,
and vasopressor requirement is not only accessible in the preop-
erative and intraoperative setting, but also captures the severity
of disease and patient’s physiology and thus can be readily eval-
uated by any surgeon to aid in decision making. Multivariate re-
gression analysis was also performed on the propensity-matched
cohort to evaluate the strength of association between other po-
tentially confounding variables that also reached statistical sig-
nificance on univariate analysis, including surgical indication
(perforation, contamination/sepsis, ischemia/necrosis, and
obstruction/other), patient’s point of origin (health care facility
vs. home), and operative length (minutes). All data were ana-
lyzed using Statistical Analysis Software, version 9.4 (SAS In-
stitute, Inc., Cary, NC).
RESULTS
Five hundred thirty-four patients [55.0% female with median
age of 61 years (interquartile range, 50–71 years)] underwent
emergent laparotomy for secondary peritonitis. Following the
index laparotomy, 203 patients (38.0%) were managed with
OA, while the remaining 331 patients (62.0%) underwent PC.
Of the OA patients, 136 (67.0%) underwent fascial closure dur-
ing the first relaparotomy (second operative procedure), while
67 (33.0%) had multiple subsequent planned reoperations
(mean, 1.6 ± 1.5).
Patients’ demographics and surgical characteristics of
each group are shown in Table 1. On unmatched univariate anal-
ysis, OA patients were older (median age, 63 vs. 59 years,
p = 0.004) and more often required transfer to our institution
from an outside facility owing to complexity of care (66.2%
vs. 28.8%, p < 0.0001). When assessing by surgeon type, OA
was performed in 21.3% of laparotomies by surgical subspecialties
625
 J Trauma Acute Care Surg
Kao et al. Volume 87, Number 3

MPI scores (median MPI, 23 vs. 18; p < 0.0001) and peak lactate
TABLE 1. Baseline Characteristics in Emergent
(median lactate level, 3 vs. 1.6; p < 0.0001). When comparing in-
Laparotomy Patients
dications for laparotomy, 75.2% of patients with ischemia or necro-
Open Abdomen Primary Closure sis had OA performed compared to 24.8% undergoing PC
n = 203 n = 331 p (p < 0.0001). Open abdomen was performed following index lap-
Age, years 63 (54–72) 59 (47–70) 0.004
arotomy in 39.6% of perforations and 18.4% of obstructions
Female s 109 (54.2) 179 (55.4) 0.79
(p < 0.0001). Open abdomen was performed more often in patients
Mannheim Peritonitis Index 23 (16–28) 18 (13–24) <0.0001
undergoing surgery at night (52.7% vs. 43.7%, p = 0.044), with in-
Apache score 18 (13–24) 14 (9–19) <0.0001
traoperative hypotension (58.2% vs. 46.5%, p = 0.018), and those
Lactate (mmol/L) 3 (1.8–5) 1.6 (1.1–2.5) <0.0001
requiring vasopressors (68.5% vs. 23.6%, p < 0.0001) or blood
ED admission 94 (46.8) 216 (66.9) <0.0001
transfusions (30.1% vs. 9.0%, p < 0.0001) during index
laparotomy.
Point of Origin <0.0001
Home 68 (33.8) 230 (71.2)
Unmatched analysis clearly demonstrated worsened
Healthcare facility 133 (66.2) 93 (28.8)
outcomes in the OA group when compared to PC. Mortality
Overnight case (6 pm–6 am) 107 (52.7) 143 (43.7) 0.0441
(24.9% vs. 6.5%, p < 0.0001), any complication (72.6% vs.
Surgeon type <0.0001
39.0%, p < 0.0001), and all subcategories of complication
Acute care surgery 171 (85.5) 224 (67.7)
rates were significantly higher in the OA group (Table 2).
Gynycologyic oncology 2 (1.0) 23 (7.0)
As expected, hospital LOS (13 vs. 8 days, p < 0.0001) and
Hepatopancreatobiliary 10 (5.0) 11 (3.3)
hospital cost ($41,759 vs. $18,591, p < 0.0001) were signifi-
Minimally invasive surgery 4 (2.0) 40 (12.1)
cantly higher in the OA group. Unplanned laparotomy oc-
Surgical oncology 11 (5.5) 18 (5.4)
curred at a significantly higher rate in PC compared to OA
Transplant surgery 1 (0.5) 12 (3.6)
patients (3.0% vs. 7.0%, p = 0.048). There was no difference
Indication for laparotomy <0.0001
in readmission.
Perforation 61 (30.5) 93 (29.5)
Following propensity matching, 111 closely matched
Contamination/sepsis 20 (10.0) 18 (5.7)
pairs (OA:PC) were compared. Importantly, differences in age,
Ischemia 79 (39.5) 26 (8.3)
MPI score, APACHE score, lactate level, and overnight surgery
Obstruction 40 (20.0) 178 (56.5)
became nonsignificant. Conversely, point of origin, surgeon spe-
Cloudy/Purulent exudate 74 (36.8) 99 (30.0) 0.104
cialty, and procedure indication remained significantly different
Fecal exudate 46 (22.9) 39 (11.8) 0.0007
(Table 3). Open abdomen patients demonstrated increased rates
Intraoperative hypotension 89 (58.2) 151 (46.5) 0.018
of postoperative complications (71.2% vs. 41.4%, p < 0.0001)
Vasopressor(s) during 122 (68.5) 77 (23.6) <0.0001
(Table 4). Specifically, there were higher rates of gastrointestinal
index laparotomy (15.3% vs. 5.4%, p = 0.015) and infectious complications
Blood transfusion 46 (30.1) 29 (9.0) <0.0001 (23.4% vs. 14.4%, p = 0.015). Open abdomen patients also
Operative duration, mins 85 (55–126) 107 (76–157) <0.0001 had longer hospital LOS (median, 13 days vs. 9 days;
p = 0.0001) and were more likely to be discharged to a facility
n (%) or median (interquartile range).
instead of home (58.6 vs. 45.9%, p = 0.006). There was no lon-
ger a significant difference in unplanned relaparotomy (2.7% vs.
compared to 43.3% (p < 0.0001) of acute care surgeons. Preopera- 8.1%, p = 0.07), and readmissions among patients surviving to
tively, OA patients had increased severity of illness as indicated by discharge (20.7% vs. 17.7%, p = 0.608) were similar between

TABLE 2. Postoperative Outcomes in the Aggregate Cohort

Open Abdomen Primary Closure
n = 203 n = 331 p
Cost $41,759 ($26,616–$65,918) $18,591 ($13,546–$31,078) <0.0001
LOS, days 13 (8–20) 8 (5–13) <0.0001
30-day readmission 32 (21.2) 50 (16.6) 0.23
In-hospital mortality 50 (24.9) 21 (6.5) <0.0001
Unplanned laparotomy 6 (3.0) 23 (7.0) 0.048
Any complication 146 (72.6) 126 (39.0) <0.0001
Cardiovascular complications 64 (31.5) 45 (13.6) <0.0001
Skin/Soft tissue complications 20 (9.9) 14 (4.2) 0.009
Respiratory complications 58 (28.6) 43 (13.0) <0.0001
GI complications 26 (12.8) 21 (6.3) 0.01
Infectious complications 49 (24.1) 39 (11.8) 0.0002
Neurologic complications 28 (13.8) 13 (3.9) <0.0001
Renal complications 64 (31.5) 47 (14.2) <0.0001

626 © 2019 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.

J Trauma Acute Care Surg
Volume 87, Number 3 Kao et al.

cohorts. In-hospital mortality occurred in 22.5% of OA patients

TABLE 3. Patient/Surgical Characteristics After
and 11.7% of PC patients (p = 0.006). There were also signifi-
Propensity-Score Matching
cantly higher costs in OA patients ($38,588 vs. $20,353,
Open Abdomen Primary Closure p < 0.0001).
n = 111 n = 111 p Multivariate analysis of the propensity-matched cohorts
Age, years 63 (54–72) 61 (48–72) 0.487
(111 patients per group) was performed evaluating association
Female sex 56 (50.9) 56 (51.9) 0.889
of surgical indication (perforation, contamination/sepsis, ischemia/
Mannheim Peritonitis Index 22 (15–26) 21 (14–26) 0.515
necrosis, obstruction/other), patient’s point of origin (health care
Apache score 16 (11–21) 16 (12–26) 0.683
facility vs. home), and operative length (minutes) with (1) in-
Lactate, mmol/L) 2.5 (1.6–4.2) 2.3 (1.3–3.6) 0.239
hospital mortality (area under the curve [AUC], 0.69); 95%
ED admission 48 (43.6) 74 (68.5) 0.0002
(confidence interval [CI], 0.599–0.781); (Hosmer-Lemeshow
[HL] goodness-of-fit p = 0.801) and (2) any in-hospital com-
Point of origin <0.0001
Home 33 (30.0) 74 (68.5)
plication (AUC, 0.671; 95% CI, 0.598–0.744; HE goodness
Healthcare facility 77 (70.0) 34 (31.5)
of fit p = 0.21). None of these variables was found indepen-
Overnight case (6 pm-6 am) 65 (58.6) 54 (48.7) 0.138
dently associated with in-hospital mortality. Patient’s point
Surgeon Ttype 0.033
of origin (transfer from health care facility), however, was
Acute care surgery 93 (84.6) 87 (78.4)
associated with any complication (odds ratio, 1.95 (95% CI,
Gynecologic oncology 1 (0.9) 3 (2.7)
1.05–3.63), p = 0.035), while the other two variables were not
Hepatopancreatobiliary 6 (5.4) 2 (1.8)
associated with occurrence of any complication.
Minimally invasive surgery 3 (2.7) 9 (8.1)
Evaluation of bowel viability was the most common rea-
Surgical oncology 7 (6.4) 5 (4.5)
son (33.4%) for surgeons to select OA during the index laparot-
Transplant surgery 0 (0.0) 5 (4.5)
omy. Of the patients with concern for bowel viability, only
Indication for laparotomy <0.0001
16.2% underwent further bowel resection during the first or sub-
Perforation 32 (29.1) 36 (35.6)
sequent relaparotomies, and 69.1% had their abdomen closed
Contamination/sepsis 10 (9.1) 6 (5.9)
during the first relaparotomy. Other surgeon-dictated indications
Ischemia 47 (43.6) 12 (11.9)
(Table 5) included vasopressor requirement (25.1%) and degree
Obstruction 20 (18.2) 47 (46.5)
of intra-abdominal contamination (19.2%). Only 2.5% of pa-
Cloudy/Purulent exudate 37 (33.3) 35 (31.5) 0.774
tients were managed with OA owing to concern for abdominal
Fecal exudate 22 (19.8) 18 (16.2) 0.484
compartment syndrome, while 14.8% had no reason recorded.
Intraoperative hypotension 44 (46.8) 57 (51.8) 0.476
Of the 136 OA patients, 24 did not survive to subsequent
Vasopressor(s) during 56 (50.5) 54 (48.7) 0.788
procedure. Of the remaining patients undergoing only one
index laparotomy planned relaparotomy, 79 (58.1%) had no new operative findings
Blood transfusion 25 (26.3) 14 (12.8) 0.01 during relaparotomy, and 33 (24.3%) underwent only abdominal
Operative duration, mins 92 (53–130) 112 (75–165) 0.01 washout before abdominal closure. An additional 32 patients
(23.5%) without positive findings during the first relaparotomy
n (%) or median (interquartile range).
underwent delayed anastomosis, while 12 (8.8%) had creation
of ostomy and 2 (1.5%) underwent both anastomosis and ostomy.
Only 33 (24.3%) patients had new or persistent intraabdominal

TABLE 4. Postoperative Outcomes in the Matched Cohort

Open Abdomen Primary Closure
n = 111 n = 111 p
Cost $38,588 ($25,405–$63,561) $20,353 ($14,466–$33,674) <0.0001
LOS, days 13 (8.5–20.5) 9 (5.0–13.0) 0.0001
30-day readmission 18 (20.7) 17 (17.7) 0.608
In-hospital mortality 25 (22.5) 13 (11.7) 0.006
Unplanned laparotomy 3 (2.7) 9 (8.1) 0.074
Any complication 79 (71.2) 46 (41.4) <0.0001
Cardiovascular complications 29 (26.1) 18 (16.2) 0.071
Skin/Soft tissue complications 12 (10.8) 5 (4.5) 0.077
Respiratory complications 30 (27.0) 21 (18.9) 0.151
GI complications 17 (15.3) 6 (5.4) 0.015
Infectious complications 26 (23.4) 16 (14.4) 0.015
Neurologic complications 12 (10.8) 8 (7.2) 0.348
Renal complications 31 (27.9) 19 (17.1) 0.054

© 2019 Wolters Kluwer Health, Inc. All rights reserved. 627

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.


Kao et al.
TABLE 5. Surgeon Reasons for Open Abdomen
Bowel Viability 68 (33.4)
Vasopressors 51 (25.1)
Contamination 39 (19.2)
No reason stated 30 (14.8)
Other 10 (4.9)
Intra-abdominal hypertension 5 (2.5)
findings during the first relaparotomy and required further intes-
tinal resection or other indicated procedure that warranted a
planned operation.
DISCUSSION
Emergent surgical management of patients with suspected
secondary peritonitis remains a challenge, particularly in situa-
tions where inadequate source control, hemodynamic instability,
and physiologic derangements may preclude definitive manage-
ment. In this propensity-matched analysis of emergent laparoto-
mies performed for suspected secondary peritonitis, differences
between in-hospital mortality rates demonstrate a benefit for PC
during the index laparotomy over OA management. Our study
showed a positive association between OA management and
rates of postoperative gastrointestinal and infectious complica-
tions, while no difference in rate of unplanned relaparotomy
was observed. Interestingly, the differences in morbidity and
mortality rates remained significant in the matched cohort anal-
ysis, suggesting OA may cause more harm than good when used
in a nontargeted fashion.
Our findings are consistent with literature demonstrating
no added survival benefit with OA. In fact, our study demon-
strated significantly higher mortality and morbidity following
OA, with no difference in rates of unplanned laparotomy or read-
mission. Van Ruler et al.9 demonstrated in a randomized trial of
patients with severe peritonitis that following initial emergent
laparotomy and abdominal closure, there was no significant dif-
ference in mortality or major morbidity with on-demand laparot-
omy versus planned re-laparotomy. Importantly, the authors
noted significant heterogeneity within the secondary peritonitis
patient population including operative surgeon, etiology, source
of infection, and disease severity. Attempts were made to control
for the latter using APACHE-2 scores, ultimately demonstrating
no difference in relative treatment effects; however, limitations
exist when trying to control for other variables. Robledo et al.14
similarly failed to show any statistically significant differences
in morbidity and mortality between OA and PC. However, sev-
eral issues exist with this study that limit generalizability and
practical application by surgeons. First, it used an outdated OA
method involving bridged polypropylene mesh sutured to fascial
edges with betadine-soaked gauze on top. Furthermore, the
study was terminated after accrual of only 40 patients despite
power calculations requiring 250 patients; therefore, conclusions
regarding outcomes between the groups cannot be made. While
advances in temporary abdominal closure devices limit the gen-
eralizability of prior conclusions to modern OA management for
abdominal sepsis, previous studies provide valuable insight and
additional support to refute the rationale for routine use of OA.18
628
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg
Volume 87, Number 3
Despite, or because of, these flawed studies, surgeon prac-
tices reflect continued use of selective OA approach in a cohort
of patients with physiologic derangements or damage control
settings. In this current study, surgeons’ reasons for OA included
need for evaluation of bowel viability, deferred anastomoses due
to vasopressor support, and level of contamination in 79.4% of
OA patients. Interestingly, our unmatched analysis also observed
a higher frequency of OA in cases performed overnight between
6 pm and 6 am, suggesting yet another heterogeneous variable
that may contribute to use of OA. A key strength of the current
analysis is the adjustment for clinical factors that predispose sur-
geons to select OA, such as preoperative clinical status, intraoper-
ative vasopressors, and degree of intra-abdominal contamination.
While these unfavorable factors frequently influence intraopera-
tive decision making and patient survival, few studies have ad-
justed for these variables or used them as basis for exclusion.
Our data demonstrate that surgeons are more likely to perform
temporary abdominal closure in these settings without any added
survival benefit. After controlling for these factors that increased
surgeons' propensity for OA, patients still demonstrated worse
outcomes compared to the PC cohort.
Primary closure results in significantly fewer repeat lapa-
rotomies and intensive care unit time, logically reducing health
care utilization and medical costs.9 Even with clearer indications
in traumatic situations, the trauma literature has also investigated
overuse of OA approach. Higa et al.19 found that with more se-
lective use of the OA technique, there was 10% reduction in
mortality, 33% reduction in reopen laparotomies, and projected
savings in health care costs of over five million dollars. This is
likely magnified in the nontraumatic population, as literature
has demonstrated worse outcomes in patients with peritonitis
versus trauma patients, including higher rates of intra-abdominal
abscess and delayed primary fascial closure rates.10 Previous cost
analyses have shown significantly greater costs and resource uti-
lization associated with planned relaparotomy compared to the
on-demand approach.20 Similarly, in our matched analysis, com-
parison of hospital use outcomes showed OA was associated
with significantly increased LOS and hospital costs. Patients
in the OA cohort were also more likely to require skilled nursing
care following discharge, further adding to the patient and health
care system burden.
The limitations of this analysis include those inherent to a
retrospective study. To reduce inbuilt selection bias, a propensity
score model was used to create well-matched cohorts for com-
parison. While there remains a possibility that other patient
and surgical factors may also influence outcomes, the combined
covariates selected for propensity matching adequately ad-
dressed 80% of surgeon-reported indications for selecting OA
approach. This study was also limited to a single tertiary referral
center and therefore may not accurately represent widespread
surgical practices and may not be generalizable. However, the
study did include a very large sample size of more than 500 lap-
arotomies performed by 57 surgeons from various subspe-
cialties. Finally, this analysis compared in-hospital mortality
and morbidity; therefore, it may underestimate the long-term
consequences associated with severe secondary peritonitis and
OA management. Patients presenting to different health care fa-
cilities after discharge may similarly underestimate readmission
and long-term complication rates; however, as the tertiary
© 2019 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg
Volume 87, Number 3
referral center, many patients requiring readmission are rou-
tinely transferred back to our facility from regional hospitals par-
ticularly within 30 days of a major surgery. Future studies such
as the multi-institutional COOL Trial, an active randomized con-
trolled trial led by the World Society of Emergency Surgery,
could provide further insight.21,22 Comparisons of long-term
outcomes in subgroups with specific indications, such as relook
for bowel ischemia, may also be of additional value.
CONCLUSIONS
The findings of this study demonstrate significant poten-
tial for adverse outcomes and overuse of health care resources
with OA, even when using a selective approach. The implica-
tions of our results and previous studies further highlight the
need for determining appropriate use of OA and planned
relaparotomy approach. With data to suggest OA may not pro-
vide the benefit that operating surgeons perceive, a closer look
at the added value of relaparotomy is warranted. Future studies
directed at identifying predictors of ongoing intra-abdominal
sepsis during relaparotomy and evaluating the incidence of addi-
tional procedures performed during second look laparotomy
may help pinpoint a subset of patients that benefit from
relaparotomy.
AUTHORSHIP
A.K., M.B.G., and L.C. reviewed the literature. A.K., B.T.H., B.D., and K.K.
designed and conceptualized the study. A.K., L.C., M.B.G., and T.P. col-
lected the data collection. A.K., T.P., L.C., and K.K. analyzed and interpreted
the data. A.K., M.B.G., and K.K. drafted the manuscript, A.K., B.T.H., B.D.,
and K.K. made critical revision to the manuscript.
DISCLOSURE
Dr Heniford has prior grants and/or honoraria from W.L. Gore, Allergan,
and Stryker. All the remaining authors declare no conflicts of interest.
This research did not receive any specific grant from funding agencies in
the public, commercial, or not-for-profit sectors.
REFERENCES
1. Ogilvie W. The late complications of abdominal war wounds. Lancet. 1940;
2:253–256.
2. Rotondo MF, Schwab CW, McGonigal MD, et al. 'Damage control': an ap-
proach for improved survival in exsanguinating penetrating abdominal in-
jury. J Trauma. 1993;35(3):375–383.
3. Coccolini F, Montori G, Ceresoli M, et al. The role of open abdomen in non-
trauma patient: WSES Consensus Paper. World J Emerg Surg. 2017;12(1):39.
4. Wittmann DH, Schein M, Condon RE. Management of secondary peritoni-
tis. Ann Surg. 1996;224(1):10–18.
© 2019 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
Kao et al.
5. Rezende-Neto J, Rice T, Abreu ES, Rotstein O, Rizoli S. Anatomical, phys-
iological, and logistical indications for the open abdomen: a proposal for a
new classification system. World J Emerg Surg. 2016;11:28.
6. Diaz JJ, Cullinane DC, Dutton WD, et al. The management of the open ab-
domen in trauma and emergency general surgery: part 1-damage control.
J Trauma. 2010;68(6):1425–1438.
7. Miller RS, Morris JA, Diaz JJ, Herring MB, May AK. Complications after
344 damage-control open celiotomies. J Trauma. 2005;59(6):1365–1374.
8. Yuan Y, Ren J, He Y. Current status of the open abdomen treatment for intra-
abdominal infection. Gastroenterol Res Pract. 2013;2013:532013.
9. van Ruler O, Mahler CW, Boer KR, et al. Comparison of on-demand vs
planned relaparotomy strategy in patients with severe peritonitis: a random-
ized trial. JAMA. 2007;298(8):865–872.
10. Sartelli M, Abu-Zidan FM, Ansaloni L, et al. The role of the open abdomen
procedure in managing severe abdominal sepsis: WSES position paper.
World J Emerg Surg. 2015;10:35.
11. Schein M. Surgical management of intra-abdominal infection: is there any
evidence? Langenbecks Arch Surg. 2002;387(1):1–7.
12. Bleszynski MS, Chan T, Buczkowski AK. Open abdomen with negative
pressure device vs primary abdominal closure for the management of surgi-
cal abdominal sepsis: a retrospective review. Am J Surg. 2016;211(5):
926–932.
13. Rausei S, Dionigi G, Boni L, et al. Open abdomen management of intra-
abdominal infections: analysis of a twenty-year experience. Surg Infect
(Larchmt). 2014;15(3):200–206.
14. Robledo FA, Luque-de-León E, Suárez R, et al. Open versus closed manage-
ment of the abdomen in the surgical treatment of severe secondary peritoni-
tis: a randomized clinical trial. Surg Infect (Larchmt). 2007;8(1):63–72.
15. Holzheimer R. Management of secondary peritonitis. In: Mannick JA, ed.
Surgical Treatment: Evidence-Based and Problem-Oriented. Munich,
Germany: Zuckschwerdt; 2001.
16. Muralidhar VA, Madhu CP, Sudhir S, Srinivasarangan M. Efficacy of
Mannheim Peritonitis Index (MPI) score in patients with secondary peritoni-
tis. J Clin Diagnostic Res. 2014;8(12):NC01–NC03.
17. Viehl CT, Kraus R, Zürcher M, Ernst T, Oertli D, Kettelhack C. The acute
physiology and chronic health evaluation II score is helpful in predicting
the need of relaparotomies in patients with secondary peritonitis of colorectal
origin. Swiss Med Wkly. 2012;142:w13640.
18. Kreis BE, de Mol van Otterloo AJ, Kreis RW. Open abdomen management: a
review of its history and a proposed management algorithm. Med Sci Monit.
2013;19:524–533.
19. Higa G, Friese R, O'Keeffe T, Wynne J, Bowlby P, Ziemba M, Latifi R,
Kulvatunyou N, Rhee P. Damage control laparotomy: a vital tool once
overused. J Trauma Inj Infect Crit Care. 2010;69(1):53–59.
20. Opmeer BC, Boer KR, van Ruler O, et al. Costs of relaparotomy on-demand
versus planned relaparotomy in patients with severe peritonitis: an economic
evaluation within a randomized controlled trial. Crit Care. 2010;14(3):R97.
21. Kirkpatrick AW, Coccolini F, Ansaloni L, et al. Closed Or Open after Source
Control Laparotomy for Severe Complicated Intra-Abdominal Sepsis (the
COOL trial): study protocol for a randomized controlled trial. World J Emerg
Surg. 2018;13:26.
22. Coccolini F, Catena F, Montori G, et al. IROA: the international register of
open abdomen: an international effort to better understand the open abdomen:
Call for participants. World J Emerg Surg. 2015;10:37.
629