Successful Nonoperative Management of the Most Severe Blunt Liver Injuries:  A Multicenter Study of the Research Consortium of New England Centers for Trauma FREE

During the last decades, nonoperative management (NOM) of blunt liver injuries (BLIs) has become the standard of care for hemodynamically stable patients, who account for approximately 85% of all those with blunt hepatic trauma.^1- 5 With increasing experience, even severe BLIs have been managed by NOM. The splenic paradigm has shown that the literature offered an overenthusiastic picture about NOM success across all grades. Because the few high-grade splenic injuries were often diluted within the many low-grade injuries, the overall high success rates of NOM were misleadingly perceived as applicable to all grades.⁶ A multicenter study⁷ from our Research Consortium of New England Centers for Trauma (ReCONECT) group in 2010 showed that 38% of grade 4 and grade 5 splenic injuries with NOM eventually failed NOM. It also documented that 64% of all high-grade injuries required splenectomy emergently or after failed NOM.

Anecdotal experience shows that severe BLIs are more hemostatic than severe splenic injuries. In a study² of 206 patients, successful NOM of the liver was 17% higher than successful NOM of the spleen. To date, the only study⁸ referring exclusively to high-grade liver injuries reports successful NOM in 41% of patients but is limited by few grade 5 injuries (10% of the entire population).

The ReCONECT group has combined the collective experience of multiple trauma centers to increase the sample size for injuries of low frequency.^9- 10 The objective of the present study was to determine the rates and predictors of failure of NOM (f-NOM) in patients with grade 4 and grade 5 BLIs. We hypothesized that such high-grade BLIs can be safely managed by NOM.

Quiz Ref IDWe retrospectively included all adult patients with a grade 4 or grade 5 BLI who were admitted between January 1, 2000, and January 31, 2010, to 11 trauma centers in New England. Grading was based on computed tomography (CT) findings and according to the American Association for the Surgery of Trauma Organ Injury Scale¹¹ (Table 1). For patients who were taken to the operating room and given a different intraoperative grade than that assigned by CT, we recorded the intraoperative grade. All centers are verified by the American College of Surgeons Committee on Trauma as level I (9 centers) or level II (2 centers) trauma centers. Patients younger than 15 years, patients who received an urgent operation at an outside hospital, and patients who were dead at the scene or on arrival were excluded.ˇ Similarly, patients with grade 6 injury (liver avulsion) were not included because such patients rarely arrive at the hospital alive, and if they do, they usually die within a few hours.

Patients were categorized as receiving NOM or an immediate operation (IO). Hemodynamic instability and signs of peritonitis were indications for IO. Nonoperative management was defined by a clear note in the medical record committing the patient to NOM or by the fact that an operation was booked later than 3 hours after the diagnosis of BLI. The decision to use 3 hours as the threshold for NOM definition was selected by consensus based on the infrastructure of the participating centers, which typically allowed for expeditious transfer to the operating room. We also wanted to account for patients with complex multiple trauma who spent an initial period of active resuscitation and diagnostic evaluation before a final decision was made about NOM vs IO.

Quiz Ref IDFailure of NOM was defined as the need for surgery after a trial of NOM or as death from BLI during NOM. Success of NOM (s-NOM) occurred if a patient did not receive an abdominal operation during the index hospital stay and did not succumb to the liver injury.

We collected data on demographics, Injury Severity Score, liver injury grade (4 vs 5), associated injuries, and mechanism of blunt trauma (motor vehicle–related crash, fall, assault, or other). We also recorded admission hemodynamics, CT findings, and the presence of free abdominal blood on CT (recorded as diffuse or only around the liver). We also noted the following: type of management (NOM vs IO), indication for operative intervention, operative procedure, and operative findings, as well as intensive care unit and hospital stay and morbidity and mortality. The main outcome measure was f-NOM. It was further classified as liver-specific f-NOM, indicating that an operation was performed to treat bleeding, leak, or infection from the liver injury, or as non–liver-specific f-NOM, indicating that an operation was performed to treat other abdominal organ injury (eg, bleeding from the spleen).

Patients who received NOM vs IO were compared. In addition, patients having f-NOM were compared with patients having s-NOM. Selected continuous variables were dichotomized across clinically meaningful values: age was dichotomized at 55 years, Injury Severity Score at 25, systolic blood pressure on admission at 100 mm Hg, heart rate on admission at 100 beats/min, and hematocrit on admission at 30%. Continuous variables were summarized using mean (SD) values and were compared using 2-sample t test or were summarized using median values (interquartile ranges) and compared using the Wilcoxon rank sum test. Categorical variables (reported as counts and proportions) were compared using the χ² test or Fisher exact test. Logistic regression analysis was performed to identify independent predictors of f-NOM significant at .05. Odds ratios (95% CIs) were reported for each predictor. The incidence of f-NOM based on different combinations of independent predictors of f-NOM was examined. P ≤ .05 indicated statistical significance. Commercially available software (SAS version 9.2; SAS Institute, Inc) was used for all analyses. The study was approved by the institutional review boards of all participating centers.

Quiz Ref IDDuring the 10-year study period, 393 adult patients with grade 4 and grade 5 BLI were included at 11 ReCONECT centers that participated in this study. One hundred thirty-one patients (33.3%) underwent an IO, while 262 patients (66.7%) were offered a trial of NOM. One hundred five of 131 patients (80.2%) underwent damage control surgery with packing of the liver.

Quiz Ref IDThe mean (SD) age of the study population was 33 (16) years (median age, 28 years; age range, 15-95 years), and the mean (SD) Injury Severity Score was 32 (14) (median, 29; range, 4-75). Fifty-four percent of patients were male, and 43.8% of patients had other abdominal organ injury, 13.9% had a brain injury, and 31.8% had major fracture. The mean (SD) intensive care unit stay for 324 patients who required critical care was 9 (17) days (median, 3 days; range, 0-164 days), and the mean (SD) hospital stay among the total population was 16 (22) days (median, 8 days; range, 1-204 days). Mortality was 21.4% (84 patients), including 60 patients who died within 24 hours, 11 patients who died between the second and seventh days, and 13 patients who died later.

Except for age, heart rate on admission, major fracture, and other extra-abdominal injury, all recorded variables were significantly different between patients receiving NOM vs IO. Not surprisingly, morbidity and mortality were higher among patients undergoing IO. The intensive care unit stays and hospital stays were similar between the 2 groups when all patients were evaluated but were longer among patients undergoing IO when only survivors were analyzed (Table 2).

Nonoperative management failed in 23 of 262 patients (8.8%) The rates of failure were 8.1% among patients with grade 4 BLI and 14.3% among patients with grade 5 BLI (P = .28). Patients with f-NOM had lower systolic blood pressure on admission, longer intensive care unit stay and hospital stay, and higher rates of other abdominal organ injury, hepatic angiography, and morbidity (Table 3). Of 23 patients with f-NOM, 17 failed NOM because of liver-specific reasons (7 due to recurrent liver bleeding and 10 due to biliary peritonitis). Of 7 patients with recurrent liver bleeding, 5 received packing of the liver, and 2 underwent nonanatomical resections. The remaining 6 patients with f-NOM had non–liver-specific issues (small bowel injury in 3, colon injury in 1, duodenal injury in 1, and gallbladder necrosis in 1). Therefore, liver-specific f-NOM was 6.5%.

Seventeen patients with liver-specific f-NOM underwent surgery a mean (SD) of 6 (7) days (range, 0-26 days) after admission; the mean (SD) values were 2 (2) days (range, 0-17 days) after admission for those with recurrent liver bleeding and 9 (8) days (range, 1-26 days) after admission for those with biliary peritonitis. All but 2 of 7 patients with f-NOM who required surgery for recurrent liver bleeding underwent an exploratory laparotomy within 24 hours of admission. No liver-specific morbidity was recorded as a direct consequence of f-NOM, and none of these patients died.

Quiz Ref IDThe following 2 independent predictors of f-NOM were identified: systolic blood pressure on admission of 100 mm Hg or less (odds ratio, 2.70; 95% CI, 1.07-6.77) and the presence of other abdominal organ injury (odds ratio, 2.92; 95% CI, 1.10-7.76) (P = .03 for both). If both independent predictors were present, 22.6% of individuals had f-NOM, as opposed to 10.6% if 1 independent predictor was present and 3.5% if neither of the 2 independent predictors was present. The negative predictive value for the absence of both independent predictors was 96.5%.

Of 262 patients treated with NOM, 94 (35.9%) received hepatic angiography, and 65 (24.8%) underwent embolization. Among 239 patients with s-NOM, 79 (33.1%) received hepatic angiography, and 55 (23.0%) underwent embolization. Three of 55 patients (5.5%) developed recurrent liver bleeding, which was controlled by reembolization in 2 and ceased spontaneously in 1. Of 7 patients who failed NOM due to recurrent liver bleeding, 5 received hepatic angiography, and 4 underwent embolization before surgery. The success rate of embolization (including reembolization) was 93.2% (55 of 59).

Endoscopic retrograde cholangiopancreatography (ERCP) was performed in 16 patients with NOM for biliary leaks. Three of them failed NOM. An additional 5 patients received ERCP after IO. Percutaneous drainage of abdominal collections was performed in 14 patients with NOM and in 6 patients with IO.

Overall, 39.2% (154 of 393) of the total study population underwent surgery, including 23 patients with f-NOM and 131 patients with IO. Among them, 63.7% had grade 4 BLI, and 36.3% had grade 5 BLI. Two patients with f-NOM underwent laparoscopy, 1 for biliary peritonitis and 1 for small bowel injury; all other patients underwent exploratory laparotomy.

The shift toward NOM of BLI has been profound during the last few decades. From the early articles advocating near-mandatory operation for all liver injuries^12- 13 to recent recommendations of treating most BLIs with NOM,¹⁴ there is clearly a major change in the standard of care. As usually happens with new methods, the rapidly growing enthusiasm for NOM of solid visceral injuries allowed an overstatement of its scope and outcomes. For example, the unchecked optimism about the positive outcomes of NOM on the injured spleen is balanced by recent evidence showing that more than one-third of severe splenic injuries fail NOM.⁷

The liver is known to respond well to NOM.^{1- 2,4,8,14- 19} However, as occurred with the spleen, the data describing high nonoperative success rates included primarily low-grade liver injuries. Severe injuries of the liver were typically treated with an operation. For example, among 128 patients with grade 4 BLIs and 31 patients with grade 5 BLIs described by Kozar et al,¹⁷ only 40% with grade 4 injury and 4% with grade 5 injury were offered NOM. In 1996, a multicenter study²⁰ from the Western Trauma Association demonstrated few NOM failures. Of 404 patients having BLIs treated with NOM, 58 had grade 4 or grade 5 injuries. There were 6 NOM failures (1.5%), and only 3 of them were specific to the liver. The article did not specify the grade of injury in these 6 patients or the number of patients with grade 4 or grade 5 injuries managed with IO. In a study¹⁴ of 55 patients having BLIs with NOM at a hospital in Los Angeles, California, 8 patients failed NOM (14.5%), although none because of liver-specific reasons. Notably, 6 of 8 failures were among patients with liver injury grade 3 or higher. Other groups have reported on high-grade liver injuries but focused on operations only⁵ or did not distinguish between patients who were operated on immediately and those who were operated on after NOM had failed.⁸

Our consortium of 11 trauma centers in New England focused exclusively on grade 4 and grade 5 BLIs. One-third of patients herein underwent IO, and the remaining two-thirds were offered a trial NOM. More than 90% of patients receiving NOM were discharged without a midline laparotomy. The liver was the specific cause of NOM failure in 6.5% of patients. Notably, 7 of 17 patients with liver-specific f-NOM had recurrent liver bleeding, and the remaining 10 had biliary peritonitis. None of 7 patients with recurrent liver bleeding experienced complications that could be attributed to f-NOM, and none of them died.

The high rate of s-NOM herein was in part because of bleeding control via angiographic embolization. One-quarter of patients receiving NOM underwent embolization, with a 93.2% success rate for control of recurrent liver bleeding. Other studies^21- 23 have shown that liver embolization for trauma is safe and effectively stops bleeding. In our population, most angiographies were performed within the first 24 hours of admission, attesting to the fact that embolization was an important element of the NOM strategy.

Additional interventions, such as ERCP, stenting, or percutaneous drainage of abdominal collections, have been recommended as effective tools to control bile leaks following severe liver trauma.²⁴ Laparoscopic drainage of biliary collections has been used to ameliorate severe inflammatory response in selected patients with NOM who remain tachycardic and febrile 3 to 5 days after injury.²⁵ In our study, we used ERCP and percutaneous drainage of abdominal collections in 6% and 5% of the population, respectively.

Our multi-institutional collaboration enhances the sample size of an uncommon injury but affects the ability to collect important details, particularly in view of its retrospective design. Therefore, precise information about decision making is missing. We set an arbitrary number of hours after admission, beyond which an operation was offered because of f-NOM. It is possible that some of these patients were never offered NOM but deteriorated during the period of evaluation. In this context, “failure” of NOM should not be attributed to the inability of trauma surgeons to triage patients appropriately but rather to the natural history of some injuries, which continue to bleed, despite optimal management. We observed no patients who were clearly harmed by f-NOM. Although such patients undoubtedly exist, f-NOM does not seem to subject patients to increased risk of complications and death. Of course, close observation and monitoring of such severe injuries are required before consideration of NOM. Finally, we could make no statements about several issues that are widely debated and continue to remain without answers: How long should these patients remain in the hospital? When should they be allowed to return to strenuous activities? What is the role of routine successive imaging? Our study was not designed to answer these questions.

In summary, grade 4 and grade 5 liver injuries respond well to NOM. In contrast to splenic injuries of similar grades, many of which are destined to fail NOM,⁷ most severe BLIs can be managed without an operation. High-grade liver injuries seem to behave in a dichotomous way: they bleed immediately and dramatically and are in obvious need of surgical intervention, or they do not bleed and can be managed reliably by NOM, with a low likelihood of subsequent bleeding. There is little reason to intervene surgically in those hemodynamically stable patients, no matter how striking the CT image may be. In this multicenter study, 66.7% of patients with grade 4 or grade 5 BLI were offered NOM, which was successful in 91.3% of them. Only 6.5% of patients with f-NOM failed because of their liver injury.