ORIGINAL ARTICLE
Nontrauma open abdomens: A prospective observational study Brandon R. Bruns, MD, Sarwat A. Ahmad, MD, Lindsay O'Meara, CRNP, Ronald Tesoriero, MD, Margaret Lauerman, MD, Elena Klyushnenkova, PhD, Rosemary Kozar, MD, PhD, Thomas M. Scalea, MD, and Jose J. Diaz, MD, Baltimore, Maryland
Damage-control surgery with open abdomen (OA) is described for trauma, but little exists regarding use in the emergency general surgery. This study aimed to better define the following: demographics, indications for surgery and OA, fascial and surgical site complications, and in-hospital/long-term mortality. We hypothesize that older patients will have increased mortality, patients will have protracted stays, they will require specialized postdischarge care, and the indications for OA will be varied. METHODS: A prospective observational study of emergency general surgery OA patients from June 2013 to June 2014 was performed. Demographics, clinical/operative variables, comorbidities, indications for procedure and OA, wound/fascial complications, and disposition were collected. Patients were stratified into age groups (≤60, 61–79, and ≥80 years). Six-month and 1-year mortality was determined by query of the Social Security Death Index. RESULTS: A total of 338 laparotomies were performed, of which 96 (28%) were managed with an OA. Median age was 61 years (interquartile range [IQR], 0–68 years), and 51% were male. The median Charlson Comorbidity Index was 2 (IQR, 1.5–5.1), and the median hospital stay was 25 days (IQR, 15–50 days). The most common indications for operation were perforated viscus/free air (20%), mesenteric ischemia (17%), peritonitis (16%), and gastrointestinal hemorrhage (12%). The most common indication for OA was damage control (37%). In the 63 patients with fascial closure, there were 9 (14%) wound infections and 6 (10%) fascial dehiscences. A total of 30% of the patients died in the hospital, and an additional six patients died 6 months after discharge. Patients in the oldest age stratum were more likely to die at 6 months than those in the lower strata. CONCLUSION: Older patients were more likely to die by 6 months, the median hospital stay was 3 weeks, and there were multiple indications for OA management. With a 6-month mortality of 36% and 70% of survivors requiring postdischarge care, this population represents a critically ill population meriting additional study. (J Trauma Acute Care Surg. 2016;80: 631–636. Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.) LEVEL OF EVIDENCE: Prognostic and epidemiologic study, level III. KEY WORDS: Emergency surgery; open abdomen; damage control; acute care surgery. BACKGROUND:
D
amage-control surgery is a well-described technique in which the trauma patient at the extreme of physiologic compromise undergoes abbreviated laparotomy, followed by resuscitation in an intensive care unit environment before definitive treatment of injuries. The use of this technique for severely injured patients in shock has been shown to improve survival.1,2 Perhaps extrapolating from the lessons learned in trauma, the technique of damage-control surgery and management with open abdomen (OA) has been used for critically ill nontrauma patients. Despite relatively wide adoption of this technique in emergency general surgery (EGS) patients, little quality data exist, leading many to call for more investigation into OA use in nontrauma populations.3,4 In fact, a randomized controlled trial by the Dutch Peritonitis Study Group showed equivalent mortalities in patients managed with planned relaparotomy versus on-demand relaparotomy. Furthermore, investigators found decreased resource use and decreased costs in patients managed with as-needed Submitted: June 29, 2015, Revised: October 21, 2015, Accepted: November 9, 2015, Published online: January 21, 2016. From the University of Maryland, R Adams Cowley Shock Trauma Center, Baltimore, Maryland. Address for reprints: Brandon R. Bruns, MD, 22 S Greene St, S4D07, Baltimore, MD 2120; email: [email protected]. DOI: 10.1097/TA.0000000000000958
reexploration.5 Despite this, numerous groups continue to examine the indications and outcomes in EGS OA patients.6–8 Despite small retrospective case series examining EGS patients managed with an OA, large prospective studies are lacking. The aim of the current work was to identify EGS patients managed with an OA at an academic medical center to better define patient demographics, indications for surgery and OA management, fascial and surgical site infectious complications, and both in-hospital and long-term mortality. We hypothesize that (1) older EGS OA patients will have more medical comorbidities and inferior outcomes as compared with younger EGS patients; (2) regardless of age, EGS OA patients will have significant morbidity and mortality, have protracted lengths of stay, require multiple operations, and require postdischarge specialized care; and (3) the indications for OA management in the EGS population will be varied and multifactorial.
PATIENTS AND METHODS After institutional review board approval was obtained, patients managed with an OA from June 1, 2013, to June 1, 2014, at the University of Maryland Medical Center, a large tertiary referral center for the state, were prospectively enrolled in this observational study. All patients were managed by the acute
J Trauma Acute Care Surg Volume 80, Number 4
631
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
care emergency surgery team, which is dedicated to the care of the EGS patient. OA patients were defined as those patients undergoing operative therapy for which the surgeon of record did not pursue primary fascial closure at the index operation but instead proceeded with temporary abdominal closure. The type of temporary abdominal closure was at the discretion of the surgeon but was either a modified Barker9 noncommercial vacuum dressing or the ABThera (KCI) wound closure system. Patients transferred to the medical center with an OA were excluded. Patient demographics and clinical variables were prospectively collected. The Charlson Comorbidity Index (CCI) was calculated in the standard fashion.10 The indication for operation that led to OA was recorded from surgeon documentation. Indications for OA management were also recorded and included damage control, need for a second look, excessive contamination, compartment syndrome, or multiple indications. Damage control was defined as ongoing intraoperative acidosis, hypothermia, or coagulopathy (clinical or laboratory). Second look was chosen when the surgeon elected to leave the abdomen open to reassess bowel viability at a later operation. Excessive contamination was a subjective judgement by the surgeon that too much contamination was present for safe closure of the fascial. Contamination was further categorized into local purulent (purulence limited to one quadrant of the peritoneal cavity), diffuse purulent (purulent material throughout the peritoneal cavity), succus (presence of bilious fluid in peritoneal cavity), and feculent (presence of fecal material in peritoneal cavity). Intraoperative variables for the index operation leading to OA were retrospectively collected from the anesthesia record. Intraoperative use of vasopressor drips, crystalloid infusion amounts, and transfusion requirements were collected. Discharge disposition and in-hospital mortality were recorded. Rehabilitation facilities were defined as lower-acuity facilities requiring active patient participation with rehabilitation activities with the hopes of discharge to an independent living situation. Skilled nursing facilities include ventilator rehabilitation and weaning facilities and do not require active participation with rehabilitation activities; thus, they are considered long-term placement solutions. Hospice facilities serve the primary goal of providing comfort services and medications during the dying process. In addition, the Social Security Death Index was accessed to assess morality within the initial 6-month and 1-year periods after discharge. Surgical site infections (SSIs) were recorded and were classified according to the Centers for Disease Control and Prevention classification: superficial incisional, deep incisional, and organ space infections.11 In determining the presence of woundrelated SSI (nonorgan space SSIs), the deepest level of infection was counted. In other words, if a patient developed a superficial SSI and a deep incisional SSI, only the deep incisional SSI was counted. This was done in an effort to not overestimate the number of SSIs. Continuous variables are presented as median and interquartile range (IQR). Categorical variables are presented as median and percentage. Mann-Whitney U-test or Kruskal-Wallis test (for more than two groups) were used for continuous variables, as appropriate. Nonparametric test for trend (CochranArmitage trend test) was used for the ordered group of age to test for trend in proportions for in-hospital and 6-month mortality. 632
Fisher's exact test was used for categorical variables. In an effort to adjust for individual patient risk factors, a multivariate logistic regression model was created adjusting for the influence of comorbidities (via the CCI) on both in-hospital and 6-month mortality for different age strata (≤60, 61–79, and ≥80 years). Odds ratios (ORs) and confidence intervals (CIs) were reported for the logistic model. Given the prospective nature of the data collection, the only data points missing were laboratory values; these were excluded from the analysis given the random nature of occurrence. Statistical significance was assumed at p < 0.05. SAS 9.4 software (SAS Institute, Cary, NC) was used to complete the statistical analysis.
RESULTS A total of 338 primary laparotomies (excluding reexploration of initial laparotomy) were performed during the 1-year study period, of which 96 patients (28%) were managed with an OA. The median age of the OA patient was 61 years (IQR, 50–68 years), 51% were male, and the median CCI was 2 (IQR, 1–4.5). At the time of the index operation leading to OA management, the median lactate was 2.4 mmol/L (IQR, 1.5–5.1 mmol/L), base deficit was 4.5 (IQR, 0.2–8), pH was 7.34 (IQR, 7.27–7.40), and white blood cell count was 13.6 K/μL (IQR, 8.35–20.51 K/μL). In patients managed with an OA, the median hospital length of stay was 25 days (IQR, 15–50 days), with a median of 16 intensive care unit days (IQR, 8–33 days) and 13 ventilator days (IQR, 5–31 days). In-hospital mortality for the OA patient population was 30% (29 patients) (Table 1). Twenty-six OA patients (27%) required a tracheostomy during hospitalization. In the 67 patients surviving to discharge, 39 (58%) were discharged to rehabilitation facilities, 20 (30%) to home, 7 (10%) to skilled nursing facilities, and 1 (2%) to hospice. The most common preoperative indications for index laparotomy leading to OA were perforated viscus and/or the presence of extraluminal gas on abdominal imaging (20%), mesenteric ischemia (17%), peritonitis and/or septic abdomen (16%), and gastrointestinal hemorrhage (12%) (Table 2). The most common index operation leading to OA was for bowelderived pathology (n = 60, 63%) (Table 3). Damage-control surgery (n = 35, 37%) was the most commonly reported indication for the decision to proceed with an OA. The need for a second look (n = 26, 27%), excessive contamination (n = 10, 10%), TABLE 1. Demographics and Hospital Characteristics of Patients Managed With an OA Sex, n (%) Male Female Age, y Hospital days ICU days Ventilator days In-hospital mortality, n (%)
49 (51) 47 (49) 61 (50–68) 25 (15–50) 16 (8–33) 13 (5–31) 29 (30)
All data presented as median (IQR), unless otherwise noted. ICU, intensive care unit.
© 2016 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
TABLE 2. Indication for Laparotomy That Leads to OA
TABLE 4. Indication for OA Management
Indication for Laparotomy
n (%)
Indication for OA
n (%)
Perforated viscus/extraluminal gas on imaging Mesenteric ischemia Peritonitis/septic abdomen Gastrointestinal hemorrhage Intestinal obstruction Incarcerated hernia Abdominal compartment syndrome Pancreatitis Other Elective operation
19 (20) 16 (17) 15 (16) 11 (12) 9 (9) 7 (7) 6 (6) 5 (5) 4 (4) 4 (4)
Damage control Facilitate early “second look” >1 indication or “other” Excessive contamination Decompression
35 (37) 26 (27) 19 (20) 10 (10) 6 (6)
abdominal compartment syndrome (n = 6, 6%), and multiple indications/“other” (n = 19, 20%) were the remaining OA management indications (Table 4). After exclusion of two missing anesthesia flow sheets, a total of 94 intraoperative anesthesia records from the index operation leading to OA were available for review. Fifty-three patients (57%) required intraoperative vasopressor drips. Median crystalloid requirement was 2,900 mL (IQR, 2,000–4,500 mL). Sixtysix patients required transfusion of at least 1 U of packed red blood cells. Of those requiring transfusion, the median number of units was 3 (IQR, 2–6). Of the 93 patients who survived to return to the operating room, the median time to return to surgery was 36 hours (IQR, 24–48 hours). In the 84 patients who achieved some manner of abdominal closure, the median number of returns to the operating room before closure was 2 (IQR, 1–3). In those that were closed, 61 (73%) had no wound complications and 23 (27%) experienced the following complications: 14 repair failures or dehiscence of the fascia, 3 abdominal sepses, 2 abdominal compartment syndromes, 2 bleedings, and 2 complications were classified as “other.” All 23 patients with wound complications required reoperation. Of the OA patients, 39 were judged to have intraperitoneal contamination during the index operation leading to OA management. The nature of the contamination was local purulent in 7, diffuse purulent in 14, feculent in 8, and succus in 8. Fifty-seven patients were judged, by the operating surgeon, to not have intraperitoneal contamination. Of those without contamination, 39 (68%) achieved primary fascial closure versus 24 (62%) of those who did have intraperitoneal contamination (p = 0.518). TABLE 3. Operative Intervention Performed at Index Operation Leading to OA Operative Intervention
n (%)
Bowel resection/drainage Gastrointestinal hemorrhage cessation Abdominal decompression Pancreatitis Debridement of necrotic tissue Gastric Hepatobiliary Abdominal wall reconstruction “Other”
60 (63) 8 (8) 6 (6) 4 (4) 3 (3) 2 (2) 2 (2) 2 (2) 9 (10)
Among the 63 patients with primary fascial closure, there were 9 (14%) total wound-related SSIs. Fascial dehiscence developed in six patients (10%). In those with fascial closure, 38 had skin closed at the time of fascial closure and 25 had their skin left open. There were seven SSIs (18%) in the skin closure group versus two (8%) in the open group (p = 0.300). Fascial dehiscence occurred in five (13%) of those with skin closure and only one (4%) of those with skin left open (p = 0.338). Fifty-eight patients (60%) received bowel resection at initial operation. Of those patients undergoing bowel resection, 50 (86%) were left in intestinal discontinuity at the index operation. During subsequent return to the operating room, 43 anastomoses were created in 36 patients. Of the 43 anastomoses, 33 were hand sewn and 10 were stapled. Two patients (5%) had clear anastomotic leaks; both had only one anastomosis, one stapled and one hand sewn. There were a total of six patients who developed enteric fistulas in the entire cohort. Three of the fistula patients had bowel resections as primary operative therapy, and three had no bowel resection. Four patients who subsequently developed a fistula were closed on their first return trip to the operating room; two were closed with mesh bridges and two with primary fascial closure. The other two fistula patients had 4 (closed with absorbable mesh) and 10 (primary fascial closure) return trips to the operating room before closure. Two of the six patients who developed fistula subsequently died in the hospital. With regard to contamination, four fistulas formed in patients with contamination present at the index operation, while two formed in patients without contamination (p = 0.397). A total of 29 OA patients (30%) died while in the hospital. Those who died in the hospital were similar to those who survived but notably had a higher CCI, lower hemoglobin, and higher lactate before operation (Table 5). An additional six patients proceeded to die within 6 months of discharge from the index admission. Of those six patients, two were discharged from the hospital to skilled nursing facilities, one was discharged to hospice, and three were discharged to rehabilitation. Stratification by age groups revealed no difference in in-house mortality for patients 80 years or older when compared with those 60 years or younger and those aged 61 years to 79 years. However, there was a significant increase in 6-month mortality in older patients, with a 6-month mortality of 36% (35 patients) (Table 6). There were no significant associations between age and in-hospital mortality when controlling for CCI (p = 0.2). For 6-month mortality, CCI was found to be significant for age stratum (p = 0.02) and was thus used in the model. Patients 80 years or older were more likely to die than those in the stratum of 60 years or younger (OR, 4.1; 95% CI, 1.00–16.96). Comparing patients 61 years to 79 years with those 60 years or younger,
© 2016 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
633
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
TABLE 5. Comparison of Those Patients Who Survived to Discharge and Those Who Died in the Hospital Live Total, n (%) Age, y BMI, kg/m2 CCI Perioperative laboratory values Hemoglobin, g/dL WBC, K/μL pH Base deficit Lactate, mmol/L
Die (In-Hospital)
67 (70) 60 (49–67) 31 (23–38) 2 (1–4) 10.0 (8–13) 13.8 (8.8–22.2) 7.33 (7.27–7.40) 4.7 (0.7–8.0) 2.2 (1.4–3.8)
29 (30) 65 (54–72) 35 (29–39) 3 (2–6) 9.3 (7.4–10.5) 11.9 (7.4–19.0) 7.31 (7.24–7.39) 7.0 (2.4–8.8) 3.2 (1.9–7.2)
p 0.103 0.159 0.038 0.027 0.241 0.507 0.237 0.027
All data are presented as median (IQR), unless otherwise indicated. BMI, body mass index; CCI, Charlson Comorbidity Index; WBC, white blood cell.
there was no difference in mortality (OR, 1.4; 95% CI, 0.53–3.77). In evaluating the Social Security Death Index for the period from 6 months to 1 year after discharge, there were no additional deaths.
DISCUSSION The current work represents a prospective observational cohort of a large number of EGS patients managed with an OA over a 1-year period. EGS OA patients required protracted hospitalizations and multiple trips to the operating room and numerous wound complications. Most patients managed with an OA had bowel-related pathologies. There was no difference in fascial closure rates between those with contamination and those without, and one quarter of closed patients required reoperation for closure-related complications. With an in-hospital mortality of 30% and 6-month mortality of 36%, this population represents a critically ill group at high risk of death. When stratified by age, 6-month mortality was higher among older patients managed with an OA as compared with younger age strata. Of those who do survive, the vast majority (70%) are discharged to rehabilitation and skilled nursing facilities. In the current study, a total of 96 patients, 28% of all patients undergoing primary laparotomy, were managed with an OA. The sheer number of patients managed with an OA is in stark contrast to other investigators but does allow for a more in-depth examination of the use of OA in EGS. Stawicki et al.8 reported a 2-year experience of nontrauma patients managed with an OA consisting of 16 patients, although the overall number of laparotomies performed during the period is not reported. Similarly, Khan et al.6 describe a 3-year retrospective analysis of 42 nontrauma OA patients, again with no clear statement of the overall number of laparotomies. The high percentage of patients managed with an OA and the large overall number of OA patients is likely representative of the nature of the EGS practice at the University of Maryland. During the 1-year study period, the acute care emergency surgery service received 11% of its new consults as outside transfers admitted to critical care beds, 35% were outside-hospital transfers to surgical floors, and only 7% of new patient originated from the hospital's emergency department. Given the high number of transfers, it is likely that the 634
local patient population represents a group with advanced pathology and altered hemodynamics, which likely contributes to the high number of OA patients in the current series. Sixty-six percent of the OA population in the current study achieved primary facial closure, with no difference between those with and without intraperitoneal contamination during the index laparotomy. This is in contradistinction to other investigations, which have shown that the presence of contamination negatively affects the ability to obtain fascial closure.12 Fistula formation occurred in 6% of the EGS OA population in this series, which is lower than the 12% reported in septic abdomens by Bruhin et al.13 in 2014 and the 44% abdominal fistula rate reported by Stawicki et al.8 in 2008. These findings likely represent the accumulation of knowledge and refinement of both surgical and critical care practices, leading to better recognition and avoidance of potential pitfalls. In addition, patients with skin closure had higher rates of SSI and fascial dehiscence in the current study, although not statistically significant. In a population of trauma patients managed with an OA, Pommerening et al.14 showed an association between skin closure and SSI but not fascial dehiscence. Additional study in the EGS population will be necessary to better elucidate optimal management of the skin in this patient population. Specific focus on the octogenarian population revealed a total of 11 patients (11% of OA patients), with a median CCI of 3 (IQR, 1–5). In-hospital mortality was 45%, in comparison with the 37% mortality reported by Arhinful et al.15 at the Mayo Clinic. They performed a 12-year retrospective analysis of 67 octogenarians and concluded that despite the significant comorbidities and high mortality rates, most of these patients managed with an OA will leave the hospital. Similarly, Subramanian et al.16 performed a 10-year chart review of 33 patients older than 65 years managed with an OA and also concluded that the technique of OA should at least be considered in the elderly. In the current series, one third of the six additional deaths in the 6-month period after discharge came in the population of 80 years or older, leading to a 6-month mortality of 64% in this population. In light of this significant mortality, the OA technique should be considered and discussed at length with octogenarian patients and family members. Almost 90% of patients receiving bowel resection in this study were left in intestinal discontinuity at the initial operation, with 2 anastomotic leaks (5% leak rate) in 43 small bowel–to–small
TABLE 6. Age-Stratified Comparison of Comorbidities, In-Hospital, and 6-Month Postdischarge Mortality
Total population CCI, median (IQR) In-hospital Lived Died 6 mo Lived Died
≤60 y
61–79 y
≥80 y
p
43 (45) 1 (0–3)
42 (44) 4 (1–5)
11 (11) 3 (1–5)
Nontrauma open abdomens: A prospective observational study Brandon R. Bruns, MD, Sarwat A. Ahmad, MD, Lindsay O'Meara, CRNP, Ronald Tesoriero, MD, Margaret Lauerman, MD, Elena Klyushnenkova, PhD, Rosemary Kozar, MD, PhD, Thomas M. Scalea, MD, and Jose J. Diaz, MD, Baltimore, Maryland
Damage-control surgery with open abdomen (OA) is described for trauma, but little exists regarding use in the emergency general surgery. This study aimed to better define the following: demographics, indications for surgery and OA, fascial and surgical site complications, and in-hospital/long-term mortality. We hypothesize that older patients will have increased mortality, patients will have protracted stays, they will require specialized postdischarge care, and the indications for OA will be varied. METHODS: A prospective observational study of emergency general surgery OA patients from June 2013 to June 2014 was performed. Demographics, clinical/operative variables, comorbidities, indications for procedure and OA, wound/fascial complications, and disposition were collected. Patients were stratified into age groups (≤60, 61–79, and ≥80 years). Six-month and 1-year mortality was determined by query of the Social Security Death Index. RESULTS: A total of 338 laparotomies were performed, of which 96 (28%) were managed with an OA. Median age was 61 years (interquartile range [IQR], 0–68 years), and 51% were male. The median Charlson Comorbidity Index was 2 (IQR, 1.5–5.1), and the median hospital stay was 25 days (IQR, 15–50 days). The most common indications for operation were perforated viscus/free air (20%), mesenteric ischemia (17%), peritonitis (16%), and gastrointestinal hemorrhage (12%). The most common indication for OA was damage control (37%). In the 63 patients with fascial closure, there were 9 (14%) wound infections and 6 (10%) fascial dehiscences. A total of 30% of the patients died in the hospital, and an additional six patients died 6 months after discharge. Patients in the oldest age stratum were more likely to die at 6 months than those in the lower strata. CONCLUSION: Older patients were more likely to die by 6 months, the median hospital stay was 3 weeks, and there were multiple indications for OA management. With a 6-month mortality of 36% and 70% of survivors requiring postdischarge care, this population represents a critically ill population meriting additional study. (J Trauma Acute Care Surg. 2016;80: 631–636. Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.) LEVEL OF EVIDENCE: Prognostic and epidemiologic study, level III. KEY WORDS: Emergency surgery; open abdomen; damage control; acute care surgery. BACKGROUND:
D
amage-control surgery is a well-described technique in which the trauma patient at the extreme of physiologic compromise undergoes abbreviated laparotomy, followed by resuscitation in an intensive care unit environment before definitive treatment of injuries. The use of this technique for severely injured patients in shock has been shown to improve survival.1,2 Perhaps extrapolating from the lessons learned in trauma, the technique of damage-control surgery and management with open abdomen (OA) has been used for critically ill nontrauma patients. Despite relatively wide adoption of this technique in emergency general surgery (EGS) patients, little quality data exist, leading many to call for more investigation into OA use in nontrauma populations.3,4 In fact, a randomized controlled trial by the Dutch Peritonitis Study Group showed equivalent mortalities in patients managed with planned relaparotomy versus on-demand relaparotomy. Furthermore, investigators found decreased resource use and decreased costs in patients managed with as-needed Submitted: June 29, 2015, Revised: October 21, 2015, Accepted: November 9, 2015, Published online: January 21, 2016. From the University of Maryland, R Adams Cowley Shock Trauma Center, Baltimore, Maryland. Address for reprints: Brandon R. Bruns, MD, 22 S Greene St, S4D07, Baltimore, MD 2120; email: [email protected]. DOI: 10.1097/TA.0000000000000958
reexploration.5 Despite this, numerous groups continue to examine the indications and outcomes in EGS OA patients.6–8 Despite small retrospective case series examining EGS patients managed with an OA, large prospective studies are lacking. The aim of the current work was to identify EGS patients managed with an OA at an academic medical center to better define patient demographics, indications for surgery and OA management, fascial and surgical site infectious complications, and both in-hospital and long-term mortality. We hypothesize that (1) older EGS OA patients will have more medical comorbidities and inferior outcomes as compared with younger EGS patients; (2) regardless of age, EGS OA patients will have significant morbidity and mortality, have protracted lengths of stay, require multiple operations, and require postdischarge specialized care; and (3) the indications for OA management in the EGS population will be varied and multifactorial.
PATIENTS AND METHODS After institutional review board approval was obtained, patients managed with an OA from June 1, 2013, to June 1, 2014, at the University of Maryland Medical Center, a large tertiary referral center for the state, were prospectively enrolled in this observational study. All patients were managed by the acute
J Trauma Acute Care Surg Volume 80, Number 4
631
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
care emergency surgery team, which is dedicated to the care of the EGS patient. OA patients were defined as those patients undergoing operative therapy for which the surgeon of record did not pursue primary fascial closure at the index operation but instead proceeded with temporary abdominal closure. The type of temporary abdominal closure was at the discretion of the surgeon but was either a modified Barker9 noncommercial vacuum dressing or the ABThera (KCI) wound closure system. Patients transferred to the medical center with an OA were excluded. Patient demographics and clinical variables were prospectively collected. The Charlson Comorbidity Index (CCI) was calculated in the standard fashion.10 The indication for operation that led to OA was recorded from surgeon documentation. Indications for OA management were also recorded and included damage control, need for a second look, excessive contamination, compartment syndrome, or multiple indications. Damage control was defined as ongoing intraoperative acidosis, hypothermia, or coagulopathy (clinical or laboratory). Second look was chosen when the surgeon elected to leave the abdomen open to reassess bowel viability at a later operation. Excessive contamination was a subjective judgement by the surgeon that too much contamination was present for safe closure of the fascial. Contamination was further categorized into local purulent (purulence limited to one quadrant of the peritoneal cavity), diffuse purulent (purulent material throughout the peritoneal cavity), succus (presence of bilious fluid in peritoneal cavity), and feculent (presence of fecal material in peritoneal cavity). Intraoperative variables for the index operation leading to OA were retrospectively collected from the anesthesia record. Intraoperative use of vasopressor drips, crystalloid infusion amounts, and transfusion requirements were collected. Discharge disposition and in-hospital mortality were recorded. Rehabilitation facilities were defined as lower-acuity facilities requiring active patient participation with rehabilitation activities with the hopes of discharge to an independent living situation. Skilled nursing facilities include ventilator rehabilitation and weaning facilities and do not require active participation with rehabilitation activities; thus, they are considered long-term placement solutions. Hospice facilities serve the primary goal of providing comfort services and medications during the dying process. In addition, the Social Security Death Index was accessed to assess morality within the initial 6-month and 1-year periods after discharge. Surgical site infections (SSIs) were recorded and were classified according to the Centers for Disease Control and Prevention classification: superficial incisional, deep incisional, and organ space infections.11 In determining the presence of woundrelated SSI (nonorgan space SSIs), the deepest level of infection was counted. In other words, if a patient developed a superficial SSI and a deep incisional SSI, only the deep incisional SSI was counted. This was done in an effort to not overestimate the number of SSIs. Continuous variables are presented as median and interquartile range (IQR). Categorical variables are presented as median and percentage. Mann-Whitney U-test or Kruskal-Wallis test (for more than two groups) were used for continuous variables, as appropriate. Nonparametric test for trend (CochranArmitage trend test) was used for the ordered group of age to test for trend in proportions for in-hospital and 6-month mortality. 632
Fisher's exact test was used for categorical variables. In an effort to adjust for individual patient risk factors, a multivariate logistic regression model was created adjusting for the influence of comorbidities (via the CCI) on both in-hospital and 6-month mortality for different age strata (≤60, 61–79, and ≥80 years). Odds ratios (ORs) and confidence intervals (CIs) were reported for the logistic model. Given the prospective nature of the data collection, the only data points missing were laboratory values; these were excluded from the analysis given the random nature of occurrence. Statistical significance was assumed at p < 0.05. SAS 9.4 software (SAS Institute, Cary, NC) was used to complete the statistical analysis.
RESULTS A total of 338 primary laparotomies (excluding reexploration of initial laparotomy) were performed during the 1-year study period, of which 96 patients (28%) were managed with an OA. The median age of the OA patient was 61 years (IQR, 50–68 years), 51% were male, and the median CCI was 2 (IQR, 1–4.5). At the time of the index operation leading to OA management, the median lactate was 2.4 mmol/L (IQR, 1.5–5.1 mmol/L), base deficit was 4.5 (IQR, 0.2–8), pH was 7.34 (IQR, 7.27–7.40), and white blood cell count was 13.6 K/μL (IQR, 8.35–20.51 K/μL). In patients managed with an OA, the median hospital length of stay was 25 days (IQR, 15–50 days), with a median of 16 intensive care unit days (IQR, 8–33 days) and 13 ventilator days (IQR, 5–31 days). In-hospital mortality for the OA patient population was 30% (29 patients) (Table 1). Twenty-six OA patients (27%) required a tracheostomy during hospitalization. In the 67 patients surviving to discharge, 39 (58%) were discharged to rehabilitation facilities, 20 (30%) to home, 7 (10%) to skilled nursing facilities, and 1 (2%) to hospice. The most common preoperative indications for index laparotomy leading to OA were perforated viscus and/or the presence of extraluminal gas on abdominal imaging (20%), mesenteric ischemia (17%), peritonitis and/or septic abdomen (16%), and gastrointestinal hemorrhage (12%) (Table 2). The most common index operation leading to OA was for bowelderived pathology (n = 60, 63%) (Table 3). Damage-control surgery (n = 35, 37%) was the most commonly reported indication for the decision to proceed with an OA. The need for a second look (n = 26, 27%), excessive contamination (n = 10, 10%), TABLE 1. Demographics and Hospital Characteristics of Patients Managed With an OA Sex, n (%) Male Female Age, y Hospital days ICU days Ventilator days In-hospital mortality, n (%)
49 (51) 47 (49) 61 (50–68) 25 (15–50) 16 (8–33) 13 (5–31) 29 (30)
All data presented as median (IQR), unless otherwise noted. ICU, intensive care unit.
© 2016 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
TABLE 2. Indication for Laparotomy That Leads to OA
TABLE 4. Indication for OA Management
Indication for Laparotomy
n (%)
Indication for OA
n (%)
Perforated viscus/extraluminal gas on imaging Mesenteric ischemia Peritonitis/septic abdomen Gastrointestinal hemorrhage Intestinal obstruction Incarcerated hernia Abdominal compartment syndrome Pancreatitis Other Elective operation
19 (20) 16 (17) 15 (16) 11 (12) 9 (9) 7 (7) 6 (6) 5 (5) 4 (4) 4 (4)
Damage control Facilitate early “second look” >1 indication or “other” Excessive contamination Decompression
35 (37) 26 (27) 19 (20) 10 (10) 6 (6)
abdominal compartment syndrome (n = 6, 6%), and multiple indications/“other” (n = 19, 20%) were the remaining OA management indications (Table 4). After exclusion of two missing anesthesia flow sheets, a total of 94 intraoperative anesthesia records from the index operation leading to OA were available for review. Fifty-three patients (57%) required intraoperative vasopressor drips. Median crystalloid requirement was 2,900 mL (IQR, 2,000–4,500 mL). Sixtysix patients required transfusion of at least 1 U of packed red blood cells. Of those requiring transfusion, the median number of units was 3 (IQR, 2–6). Of the 93 patients who survived to return to the operating room, the median time to return to surgery was 36 hours (IQR, 24–48 hours). In the 84 patients who achieved some manner of abdominal closure, the median number of returns to the operating room before closure was 2 (IQR, 1–3). In those that were closed, 61 (73%) had no wound complications and 23 (27%) experienced the following complications: 14 repair failures or dehiscence of the fascia, 3 abdominal sepses, 2 abdominal compartment syndromes, 2 bleedings, and 2 complications were classified as “other.” All 23 patients with wound complications required reoperation. Of the OA patients, 39 were judged to have intraperitoneal contamination during the index operation leading to OA management. The nature of the contamination was local purulent in 7, diffuse purulent in 14, feculent in 8, and succus in 8. Fifty-seven patients were judged, by the operating surgeon, to not have intraperitoneal contamination. Of those without contamination, 39 (68%) achieved primary fascial closure versus 24 (62%) of those who did have intraperitoneal contamination (p = 0.518). TABLE 3. Operative Intervention Performed at Index Operation Leading to OA Operative Intervention
n (%)
Bowel resection/drainage Gastrointestinal hemorrhage cessation Abdominal decompression Pancreatitis Debridement of necrotic tissue Gastric Hepatobiliary Abdominal wall reconstruction “Other”
60 (63) 8 (8) 6 (6) 4 (4) 3 (3) 2 (2) 2 (2) 2 (2) 9 (10)
Among the 63 patients with primary fascial closure, there were 9 (14%) total wound-related SSIs. Fascial dehiscence developed in six patients (10%). In those with fascial closure, 38 had skin closed at the time of fascial closure and 25 had their skin left open. There were seven SSIs (18%) in the skin closure group versus two (8%) in the open group (p = 0.300). Fascial dehiscence occurred in five (13%) of those with skin closure and only one (4%) of those with skin left open (p = 0.338). Fifty-eight patients (60%) received bowel resection at initial operation. Of those patients undergoing bowel resection, 50 (86%) were left in intestinal discontinuity at the index operation. During subsequent return to the operating room, 43 anastomoses were created in 36 patients. Of the 43 anastomoses, 33 were hand sewn and 10 were stapled. Two patients (5%) had clear anastomotic leaks; both had only one anastomosis, one stapled and one hand sewn. There were a total of six patients who developed enteric fistulas in the entire cohort. Three of the fistula patients had bowel resections as primary operative therapy, and three had no bowel resection. Four patients who subsequently developed a fistula were closed on their first return trip to the operating room; two were closed with mesh bridges and two with primary fascial closure. The other two fistula patients had 4 (closed with absorbable mesh) and 10 (primary fascial closure) return trips to the operating room before closure. Two of the six patients who developed fistula subsequently died in the hospital. With regard to contamination, four fistulas formed in patients with contamination present at the index operation, while two formed in patients without contamination (p = 0.397). A total of 29 OA patients (30%) died while in the hospital. Those who died in the hospital were similar to those who survived but notably had a higher CCI, lower hemoglobin, and higher lactate before operation (Table 5). An additional six patients proceeded to die within 6 months of discharge from the index admission. Of those six patients, two were discharged from the hospital to skilled nursing facilities, one was discharged to hospice, and three were discharged to rehabilitation. Stratification by age groups revealed no difference in in-house mortality for patients 80 years or older when compared with those 60 years or younger and those aged 61 years to 79 years. However, there was a significant increase in 6-month mortality in older patients, with a 6-month mortality of 36% (35 patients) (Table 6). There were no significant associations between age and in-hospital mortality when controlling for CCI (p = 0.2). For 6-month mortality, CCI was found to be significant for age stratum (p = 0.02) and was thus used in the model. Patients 80 years or older were more likely to die than those in the stratum of 60 years or younger (OR, 4.1; 95% CI, 1.00–16.96). Comparing patients 61 years to 79 years with those 60 years or younger,
© 2016 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
633
J Trauma Acute Care Surg Volume 80, Number 4
Bruns et al.
TABLE 5. Comparison of Those Patients Who Survived to Discharge and Those Who Died in the Hospital Live Total, n (%) Age, y BMI, kg/m2 CCI Perioperative laboratory values Hemoglobin, g/dL WBC, K/μL pH Base deficit Lactate, mmol/L
Die (In-Hospital)
67 (70) 60 (49–67) 31 (23–38) 2 (1–4) 10.0 (8–13) 13.8 (8.8–22.2) 7.33 (7.27–7.40) 4.7 (0.7–8.0) 2.2 (1.4–3.8)
29 (30) 65 (54–72) 35 (29–39) 3 (2–6) 9.3 (7.4–10.5) 11.9 (7.4–19.0) 7.31 (7.24–7.39) 7.0 (2.4–8.8) 3.2 (1.9–7.2)
p 0.103 0.159 0.038 0.027 0.241 0.507 0.237 0.027
All data are presented as median (IQR), unless otherwise indicated. BMI, body mass index; CCI, Charlson Comorbidity Index; WBC, white blood cell.
there was no difference in mortality (OR, 1.4; 95% CI, 0.53–3.77). In evaluating the Social Security Death Index for the period from 6 months to 1 year after discharge, there were no additional deaths.
DISCUSSION The current work represents a prospective observational cohort of a large number of EGS patients managed with an OA over a 1-year period. EGS OA patients required protracted hospitalizations and multiple trips to the operating room and numerous wound complications. Most patients managed with an OA had bowel-related pathologies. There was no difference in fascial closure rates between those with contamination and those without, and one quarter of closed patients required reoperation for closure-related complications. With an in-hospital mortality of 30% and 6-month mortality of 36%, this population represents a critically ill group at high risk of death. When stratified by age, 6-month mortality was higher among older patients managed with an OA as compared with younger age strata. Of those who do survive, the vast majority (70%) are discharged to rehabilitation and skilled nursing facilities. In the current study, a total of 96 patients, 28% of all patients undergoing primary laparotomy, were managed with an OA. The sheer number of patients managed with an OA is in stark contrast to other investigators but does allow for a more in-depth examination of the use of OA in EGS. Stawicki et al.8 reported a 2-year experience of nontrauma patients managed with an OA consisting of 16 patients, although the overall number of laparotomies performed during the period is not reported. Similarly, Khan et al.6 describe a 3-year retrospective analysis of 42 nontrauma OA patients, again with no clear statement of the overall number of laparotomies. The high percentage of patients managed with an OA and the large overall number of OA patients is likely representative of the nature of the EGS practice at the University of Maryland. During the 1-year study period, the acute care emergency surgery service received 11% of its new consults as outside transfers admitted to critical care beds, 35% were outside-hospital transfers to surgical floors, and only 7% of new patient originated from the hospital's emergency department. Given the high number of transfers, it is likely that the 634
local patient population represents a group with advanced pathology and altered hemodynamics, which likely contributes to the high number of OA patients in the current series. Sixty-six percent of the OA population in the current study achieved primary facial closure, with no difference between those with and without intraperitoneal contamination during the index laparotomy. This is in contradistinction to other investigations, which have shown that the presence of contamination negatively affects the ability to obtain fascial closure.12 Fistula formation occurred in 6% of the EGS OA population in this series, which is lower than the 12% reported in septic abdomens by Bruhin et al.13 in 2014 and the 44% abdominal fistula rate reported by Stawicki et al.8 in 2008. These findings likely represent the accumulation of knowledge and refinement of both surgical and critical care practices, leading to better recognition and avoidance of potential pitfalls. In addition, patients with skin closure had higher rates of SSI and fascial dehiscence in the current study, although not statistically significant. In a population of trauma patients managed with an OA, Pommerening et al.14 showed an association between skin closure and SSI but not fascial dehiscence. Additional study in the EGS population will be necessary to better elucidate optimal management of the skin in this patient population. Specific focus on the octogenarian population revealed a total of 11 patients (11% of OA patients), with a median CCI of 3 (IQR, 1–5). In-hospital mortality was 45%, in comparison with the 37% mortality reported by Arhinful et al.15 at the Mayo Clinic. They performed a 12-year retrospective analysis of 67 octogenarians and concluded that despite the significant comorbidities and high mortality rates, most of these patients managed with an OA will leave the hospital. Similarly, Subramanian et al.16 performed a 10-year chart review of 33 patients older than 65 years managed with an OA and also concluded that the technique of OA should at least be considered in the elderly. In the current series, one third of the six additional deaths in the 6-month period after discharge came in the population of 80 years or older, leading to a 6-month mortality of 64% in this population. In light of this significant mortality, the OA technique should be considered and discussed at length with octogenarian patients and family members. Almost 90% of patients receiving bowel resection in this study were left in intestinal discontinuity at the initial operation, with 2 anastomotic leaks (5% leak rate) in 43 small bowel–to–small
TABLE 6. Age-Stratified Comparison of Comorbidities, In-Hospital, and 6-Month Postdischarge Mortality
Total population CCI, median (IQR) In-hospital Lived Died 6 mo Lived Died
≤60 y
61–79 y
≥80 y
p
43 (45) 1 (0–3)
42 (44) 4 (1–5)
11 (11) 3 (1–5)
