- Email: [email protected]
Liver injuries—Improved results with a formal protocol including angiography
Injury, Int. J. Care Injured (2007) 38, 1075—1083
www.elsevier.com/locate/injury
Liver injuries–—Improved results with a formal protocol including angiography Christine Gaarder a,*, Pa ˚l Aksel Naess b, Torsten Eken c, Nils Oddvar Skaga d, Johan Pillgram-Larsen e, Nils Einar Klow e, Trond Buanes f a
Trauma Unit, Emergency Division, Ullevaal University Hospital, N-0407 Oslo, Norway Department of Paediatric Surgery, Ullevaal University Hospital, Oslo, Norway c Department of Anaesthesia, Aker University Hospital, Oslo, Norway d Department of Anaesthesia, Ullevaal University Hospital, Oslo, Norway e Department of Radiology, Ullevaal University Hospital, Oslo, Norway f Department of Gastrointestinal Surgery, Ullevaal University Hospital, Oslo, University of Oslo, Norway b
Accepted 5 February 2007
KEYWORDS Non-operative management; Damage control surgery; Liver injury; Angiography; Embolisation
Summary Objective: We hypothesised that a formal treatment protocol for liver injuries including angiography would increase the non-operative management (NOM) rate and would be efficient as an adjunct to damage control surgery. Methods: During the 4-year period from 1 August 2000, a total of 138 adult patients with liver injuries were admitted to the largest trauma centre in Norway and prospectively included in the institutional trauma registry. On 1 August 2002, a protocol mandating angiography in all NOM patients with OIS grades 3—5 liver injuries and after packing of the liver was implemented. All patients admitted during the subsequent 2-year period (group 2) were compared with the previous 2 years as historic controls (group 1). Results: Fifty-five patients were included in group 1 and 59 in group 2. The groups were statistically comparable, both with a mean ISS of 31. Patients selected for NOM increased from 28 (51%) to 45 (76%) ( p < 0.05), without increasing failure rate, liverrelated complications, mortality or transfusion rate. Angiography was performed in 26 patients in group 2 (44%). Only nine patients underwent embolisation (35%), and five of these were in the NOM group. Angiography was negative in the eight NOM stable patients with OIS grade 3 injury. Conclusion: The implementation of a formal NOM protocol decreased total laparotomy rate and seemed to improve patient outcome without jeopardising patient
* Corresponding author. Tel.: +47 41318992. E-mail address: [email protected] (C. Gaarder). 0020–1383/$ — see front matter # 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.injury.2007.02.001
1076
C. Gaarder et al. safety. Surprisingly few of the patients undergoing angiography required embolisation. Angiography is not indicated in stable OIS grade 3 liver injuries, and the protocol in our institution has been adjusted accordingly. AE seems to be a valuable adjunct to DCS with packing of liver injuries. # 2007 Elsevier Ltd. All rights reserved.
Introduction Non-operative management (NOM) of blunt liver injuries in haemodynamically stable or stabilised patients has become standard practice during the last decade.13,18—20 Success rates for NOM are usually reported to exceed 80%.6,13,14,19,20,23 However, the reported overall percentage of patients selected for NOM varies between 47% and 85%.4,9,13,14,19,22,23 Many studies have tried to define independent predictors of outcome and improve the way we treat liver injuries.12,21 The introduction of angiographic embolisation (AE) has been reported to increase the success rate of NOM.3,7,10,16,20,22,23 However, most studies describing NOM and the use of AE are retrospective and compare different groups in the same population, treated differently. Furthermore, reported failure rates, mortality, complications and transfusion rates vary according to definitions, population, and criteria for NOM and angiography. Operative treatment of liver injuries, even in experienced hands, still carries a high mortality and morbidity risk.2,4,7,9,14,18 However, just as NOM is the treatment of choice in haemodynamically stable patients, unstable patients still need an operation. Irrespective of criteria for NOM, operative intervention remains necessary in a significant number of patients. Operative treatment options are depending on available resources and expertise. AE seems to be a valuable adjunct to operative management.1,2,11 since most patients are haemodynamically abnormal at the end of a damage control laparotomy, and ongoing arterial bleeding is difficult to rule out clinically.11 The risk of bleeding in OIS grades 4 and 5 liver injuries has been recognised by several authors.3,9,19 An indication for angiography in these patients offers no controversy, whereas the indication for grade 3 is debated.2,4,5,10,11 However, in a prospective study, Hagiwara et al. reported on 51 patients with liver injuries grades 3—5 undergoing angiography.10 All 14 patients with a CT scan grade 4 or 5 injury were actively bleeding, but also 18 of the 37 patients with grade 3 injuries. On 1 August 2002, a new protocol was implemented in our institution mandating angiography in all patients after damage control surgery (DCS) with packing of the liver and in patients managed non-
operatively with organ injury scale (OIS) grades 3—5 injuries. The purpose of the present study was to evaluate the new protocol, its effect on NOM rate and the need for AE after DCS with packing of the liver.
Patients and methods During the 4-year period from 1 August 2000 to 31 July 2004, a total of 138 adult patients (14 years) with liver injuries were admitted to Ullevaal University Hospital, the largest trauma centre in Norway, and prospectively included in the institutional trauma registry. The trauma registry includes all patients with injury severity score (ISS) > 9 admitted within 24 h after injury as well as all patients admitted for whom the trauma team is activated. Additionally, all penetrating injuries proximal to elbow or knee are included. Of the 138 admitted patients, 24 were categorised as dead on arrival or died in the emergency department and were excluded from further analysis. All prospectively included patients with liver injuries admitted between 1 August 2000 and 31 July 2002, served as historic controls (group 1). During this period, patients who were haemodynamically unstable and had intraabdominal haemorrhage were operated on immediately. All haemodynamically stable patients with suspected intraabdominal injuries underwent computed tomographic (CT) scanning of the abdomen and pelvis. Although NOM was the treatment of choice in the absence of findings suggestive of associated injuries mandating laparotomy, the treatment decision was left to the surgeon in charge. Angiography was not available during this period. From 1 August 2002, the existing treatment policy was formalised in a protocol introducing angiography in severe liver injuries (Fig. 1). All patients with liver injuries prospectively included during the subsequent 2-year period constituted group 2. Patients remaining haemodynamically unstable despite fluid resuscitation were still not eligible for NOM, and underwent immediate laparotomy. All haemodynamically stable or stabilised patients were considered for NOM and had their liver injury diagnosed on CT scan and graded according to the organ injury scale (OIS) as described by the
Liver injuries–—Improved results
1077
Figure 1 Algorithm for treatment of liver injuries. DPL, diagnostic peritoneal lavage; FAST, focused assessment with sonography in trauma; OIS, Organ Injury Scale; DCS, Damage Control Surgery.
American Association for the Surgery of Trauma (AAST) (Table 1).17 The decision to treat the patient non-operatively was based entirely on the patient’s haemodynamic status and the absence of associated injuries mandating laparotomy, and was not influenced by liver injury grade, grade of haemoperitoneum, age, or the presence of head injury. The abdominal CT scans were interpreted by the trauma team leader and the resident radiologist on call.
Angiography and arterial embolisation The new protocol (Fig. 1) mandated angiography in all adult patients with liver injury OIS grades 3—5
and in any patient with clinical suspicion of ongoing bleeding, defined as falling haemoglobin and tachycardia, or any transfusion requirement where the liver could not be ruled out as a significant bleeding source. Acute angiography was performed in haemodynamically stable or stabilised patients when CT demonstrated extravasation or the patient showed clinical signs of ongoing bleeding. In addition, acute angiography was performed in all patients who had undergone DCS with packing of the liver. The other patients underwent angiography the next morning. Our institution is fortunate to have a highly competent angiography service available 24/7/365, on 30 min notice.
1078
C. Gaarder et al.
Table 1 AAST organ injury scale–—liver injury Grade
Injury description
I
Haematoma Laceration
Subcapsular <10% surface area Capsular tear <1 cm parenchymal depth
II
Haematoma
Subcapsular, 10—50% surface area Intraparenchymal <10 cm diameter 1—3 cm parenchymal depth, <10 cm length
Laceration III
Haematoma
Subcapsular >50% surface area or expanding Ruptured subcapsular or parenchymal haematoma Intraparenchymal haematoma >10 cm or expanding >3 cm parenchymal depth
Laceration IV
Laceration
Parenchymal disruption involving 25—75% of hepatic lobe or 1—3 Coinaud’s segments in a single lobe
V
Laceration
Parenchymal disruption involving >75% of hepatic lobe or >3 Coinaud’s segments in a single lobe Juxtahepatic venous injury, i.e. retrohepatic vena cava/ central major hepatic veins
Vascular VI
Vascular
Hepatic avulsion
Angiography was performed via femoral artery puncture. A 4—5 Fr diagnostic catheter was placed in the celiac trunk and advanced into the hepatic artery. According to the protocol, embolisation was performed only when angiography revealed ongoing bleeding, vessel truncation seen as an unsecured bleeding site, or pseudoaneurysm (PSA). Embolisation was performed as peripherally as possible by the placement of microcoils. A completion angiogram was performed to confirm haemostasis of the embolised vessel.
Outcome measures and statistics Data retrieved from the hospital-based trauma registry described demographic characteristics, injury type and severity, as well as haemodynamic status. Data on diagnostic tests, management strategies, and outcome were also collected. Groups 1 and 2 were compared, and outcome is reported as laparotomy rates, failure of NOM, complication and
transfusion rates, and mortality. Failure of NOM was defined as the need for laparotomy after initial non-operative treatment. Failures of NOM were further categorised as being caused by the liver injury or not. Successful AE was defined as achievement of haemostasis primarily, no repeat angiography required, and no delayed bleeding from the liver injury. Complications and the number of transfusions were recorded for the period of hospitalisation in our institution. Angiography related complications were defined as unintentional coil migration, puncture site haematoma or PSA, intimal tears, gallbladder necrosis or allergic contrast reaction. Liver-related complications were defined as delayed liver haemorrhage, intra- and perihepatic abscess, hepatic necrosis, bile leak and biloma. Survival was defined as the patient being alive 30 days after admission. Categorical variables underwent x2 analysis or the Fisher exact test, and continuous variables were subjected to the Mann—Whitney U-test. p < 0.05
Table 2 Patient characteristics Age (range) Male gender, n (%) Blunt injury, n (%) ISS Ps RTS score BE Liver OIS
Group 1 (n = 55)
Group 2 (n = 59)
p-Value
32.4 (17—81) 44 (80) 47 (85) 30.7 18.2 0.74 0.31 6.53 1.43 3.5 4.4 2.5 1.0
33.7 (15—80) 38 (64) 52 (88) 30.6 14.8 0.77 0.26 6.67 1.45 3.6 4.2 2.6 1.2
0.78 0.06 0.89 0.83 > 0.99 0.43 0.67 0.69
ISS, injury severity score; Ps, probability of survival; RTS, revised trauma score; BE, base excess; OIS, organ injury scale. Values are given as mean S.D. where not stated otherwise.
Liver injuries–—Improved results
1079
Table 3 Management strategies and outcome Attempted NOM (%) Failure rate (%) Total laparotomy rate (%) Mortality rate (%) Laparotomy per patient PRBCs first 24 h Total number of PRBCs LOS ICU per patient (days) Complications per patient
Group 1 (n = 55)
Group 2 (n = 59)
p-Value
95% CI for the difference
28 (51) 5 (18) 32 (58) 5 (9) 1.2 (0—7) 7.8 (0—67) 12.1 (0—67) 9.2 (0—75) 1.5 (0—9)
45 (76) 6 (13) 20 (34) 6 (10) 0.7 (0—4) 6.2 (0—60) 9.7 (0—80) 7.3 (0—51) 0.9 (0—8)
0.005 0.85 0.009 0.82 0.02 0.49 0.41 0.41 0.03
0.08; 0.42 0.08; 0.18 0.06; 0.42 0.10; 0.12 1.00; 0.08 5.90; 2.82 8.19; 3.39 6.49; 2.66 1.26; 0.03
NOM, non-operative management; PRBC, packed red blood cells; LOS, length of stay; ICU, intensive care unit; CI, confidence interval. Values are given as mean and total range where not stated otherwise.
was chosen to indicate statistical significance. Statistical analyses were performed using The Statistical Package for the Social Sciences, Version 13.0 (SPSS Inc., Chicago, IL), and Statview 5.0 (SAS Institute Inc.). Data are presented as mean S.D. or mean with total range. The study was approved by the Regional Committee for Medical Research Ethics.
Results Group 1 consisted of 55, and group 2 of 59 patients. The groups were comparable with respect to demographic characteristics, mechanism of injury and injury severity, haemodynamic status on arrival, and had virtually identical mean liver OIS (Table 2). Management strategies and outcome for groups 1 and 2 are listed in Table 3. The emergency lapar-
otomy rate decreased from 49% to 24% ( p = 0.005) without a concomitant increase in failure rate, transfusion rate or LOS in the ICU. Mortality rates were similar in the two groups. There were no liver-related deaths. In group 1, the principal cause of death was head injury in four patients and sepsis and multiple organ failure (MOF) in one patient. In group 2, head injury was the main cause in three patients, and adult respiratory distress syndrome (ARDS) combined with MOF in three patients. The total number of complications (Table 3) per patient decreased significantly ( p = 0.03). Significant complications (Table 4) were registered for 32 patients (58%) in group 1 and 26 patients (44%) in group 2. Liver-related complications were registered in eight patients (15%) in group 1 and three patients (5%) in group 2 ( p = 0.08). Bile leak
Table 4 Significant complications Group 1 (83 complications in 32 of 55 patients)
Group 2 (52 complications in 26 of 59 patients)
Liver-related Delayed hemorrhage Liver pseudoaneurysm/biloma Abscess Liver necrosis Bile leak
4 — 5 1 1
2 — 1 2 3
Other abdominal Abdominal compartment syndrome Wound infection Wound dehiscence/hernia Intestinal fistula Urinary tract infection
2 5 4 3 4
3 2 — — 6
Thoracic Pleural effusion Pneumonia
14 24
4 20
Sepsis/multiple organ failure Sepsis Multiple organ failure
10 6
6 3
1080
C. Gaarder et al. Table 6
Patients undergoing angiography n = 26
Age (range) Male gender (%) ISS RTS score Ps BE Liver OIS
31.5 (17—64) 17 (65) 34.2 13.9 6.5 1.4 0.74 0.27 5.1 4.5 3.4 0.9
ISS, injury severity score; RTS, revised trauma score; Ps, probability of survival; BE, base excess; OIS, organ injury scale. Values are given as mean S.D. where not stated otherwise.
Patients undergoing angiography
Figure 2 Organ injury scale (OIS) grade distribution. The light bars represent group 1, the dark bars group 2. The checked area of each bar represents the number of patients with attempted non-operative management (NOM).
was recorded in three patients and necrosis in two patients in group 2, while only one patient with bile leak was registered in group 1. The mean liver OIS grade in the two groups was similar. The number of patients with OIS grades 3—5 injury was 23 (43%) in group 1 and 32 (54%) in group 2 (Fig. 2). Procedures performed in patients undergoing emergency laparotomy are listed in Table 5. In group 1, 11 laparotomies (41%) were non-therapeutic regarding the liver injury. The corresponding number for group 2 was four (29%). In group 1, 5 of the 28 patients selected for NOM required laparotomy, resulting in a NOM failure rate of 18%. The liver was the cause of the delayed operation in two of these patients, resulting in a liver-specific failure rate of 7%. In group 2, NOM was attempted in 45 patients. Among the six patients (13%) who failed NOM, the liver was the cause in one patient, resulting in a liver-specific failure rate of 2%. The difference was not significant. There were no missed injuries in the patients selected for NOM for known liver injuries.
A total of 26 (44%) of the 59 patients in group 2 underwent angiography (Table 6). Treatment strategies in group 2 are presented in Fig. 3. In 19 patients angiography was performed as an adjunct to NOM, whereas 7 patients underwent angiography after laparotomy with packing. These two subgroups are described separately. NOM patients Of the 19 NOM patients undergoing angiography, 6 patients presented with grade 4 or 5 injuries. Abdominal CT scan demonstrated extravasation in one patient with grade 3 and one with grade 5 injury. Angiography showed ongoing bleeding in three patients, including the two with contrast extravasation on CT scan, and vessel truncation in two patients. Of these five patients, two had grade 3, two had grade 4, and one had grade 5 injury. All five patients were embolised. Angiography was negative in the three patients with OIS grade 2 injury and in the eight NOM stable patients with OIS grade 3 injury. None of these patients had extravasation seen on CT scan. Thus, of the remaining eight NOM patients undergoing angiography, embolisation was performed in five (63%). DCS patients Angiography was performed in seven patients after laparotomy with packing. Four of these patients
Table 5 Emergency laparotomies Group 1 (27 of 55 patients)
Group 2 (14 of 59 patients)
Emergency laparotomy–—liver specific treatment Liver packing Hepatorrhaphy Combination hepatorrhaphy/packing
16 11 3 2
10 8 1 1
Emergency laparotomy–—non-therapeutic for liver
11
4
Liver injuries–—Improved results
Figure 3
1081
Flowchart showing treatment strategies followed in group 2.
presented with OIS grade 3 injury, one with a grade 4 injury, and two with grade 5 injury. Angiography revealed ongoing bleeding in three patients and vessel truncation in one. These four patients (57%) underwent embolisation. In the remaining three patients angiography was negative.
possible intimal tear in the celiac trunk, which was without clinical or therapeutic consequences. There were three deaths in the angiography group. None of them had been embolised. The cause of death was unrelated to the liver injury in all patients, the main cause being head injury in one, and ARDS combined with MOF in two patients.
Success and complications after angiography
Discussion Clinical and angiographical success was obtained in eight of nine embolised patients (89%). The only failure to AE had a grade 5 injury, was taken to angiography, was embolised, but remained unstable and was subsequently operated with packing of the liver. The abdomen was unpacked after 48 h, and closed on day 5. One patient underwent laparotomy because of sepsis and had a nephrectomy performed, unrelated to the angiography. One patient with a grade 5 injury developed necrosis after successful angiographic embolisation following DCS with packing of the liver, and tolerated a formal right hemihepatectomy on day 5. The only complication registered directly associated with the angiographic instrumentation was a
The two most important parameters to describe a study on NOM are the overall percentage of patients selected for NOM and the failure rate. As the first percentage increases to include more patients, the failure rate is also likely to increase. In our population, we show that the new protocol resulted in a significant increase in patients selected for NOM from 51% to 76% (Table 3), without a concomitant increase in failure rate or mortality. Hence, total laparotomy rate also decreased significantly, from 58% to 34%, and there were no liver-related deaths. Although angiography was performed in 26 patients in group 2, only 9 patients were embolised. Furthermore, only five of the embolised patients
1082 were in the NOM group (26%) and one was subsequently operated for bleeding. Obviously, the decrease in laparotomy rate from 58% to 34% cannot be ascribed the effect of AE. During period 1, although NOM was the treatment of choice in haemodynamically stable patients with no associated injuries mandating laparotomy, the treatment decision was left to the surgeon in charge, resulting in a 49% emergency laparotomy rate. By current standards, 51% selected for NOM is relatively low, and the implementation of a standardised protocol with close follow-up might in itself have been the most important factor leading to the increased NOM rate. Although angiography was part of the new protocol it may have been incidental to the reduced laparotomy rate. However, introducing AE offered an additional reliable alternative to operation, thus contributing to the surgeons in charge allowing more aggressive resuscitation before deciding on laparotomy. When surgery is indicated, the goal should be haemorrhage control.8,20 In our institution surgeons have limited exposure to complex liver injuries. A haemodynamically unstable patient undergoing damage control laparotomy for a liver injury will most likely get the liver packed. The rationale for including mandatory angiography after DCS of the liver was that most patients are haemodynamically abnormal at the end of a damage control laparotomy, and ongoing arterial bleeding is difficult to rule out clinically.11 Of the seven DCS patients subjected to angiography in the present material, four underwent embolisation (57%). The risk of bleeding with NOM in OIS grades 4 and 5 liver injuries is significant,3,9,19 and mandatory angiography in these patients offers no controversy.2,4,5,10,11 Some studies show the benefit of angiography in OIS grade 3 injuries.10 Other studies question the value of such a low threshold for angiography and conclude with contrast extravasation on CT scan or clinical signs of ongoing bleeding as selection criteria.22,23 In our population, only 5 of the 19 NOM patients underwent embolisation. All 11 patients with grade 2 and haemodynamically stable patients with grade 3 liver injury had no extravasation seen on CT scan and negative angiographies, and were not embolised. For various reasons, eight NOM patients with grade 3 injury were not subjected to angiography. Two patients died before angiography could be performed. The other six were successfully observed. Although demonstrating some compliance problems with the protocol, they also add to the fact that haemodynamically stable patients with grade 3 injury and no extravasation on CT scan are well off without
C. Gaarder et al. mandatory angiography. Excluding the patients with grade 2 injury and assuming we had excluded haemodynamically stable grade 3 patients from the protocol, our positive angiography rate in the study population would be acceptable to us, with five of eight patients being embolised (63%). After the study period, our protocol has been adjusted accordingly. Worrying complication rates associated with NOM and AE have been presented by others.7,12,16,22,23 In our population, the number of liver-related complications was 11 in group 1 and 8 in group 2, with a higher rate of necrosis and bile leaks registered in group 2. In the angiography group, two patients developed necrosis after embolisation of the liver. One was in the NOM group and recovered uneventfully after observation alone. The other patient underwent DCS with packing of the liver and subsequent successful embolisation of ongoing arterial bleeding. A right-sided hemihepatectomy was performed on day 5 when the patient was physiologically corrected. Bile leak was registered in three patients in group 2 and only one in group 1. Some drainage of bile postoperatively would be expected and might have been inadequately documented in the patient charts, thus contributing to underrating complications in group 1. The same could be argued for other liver-related complications and illustrates one weakness associated with the use of historic controls. In spite of this, the total number of significant complications per patient decreased, probably due to the lowered laparotomy rate. Although the rate of missed associated hollow viscus injuries in patients with liver injuries is low, reported at 2.3% by Miller et al.,15 the risk still worries us as the percentage of NOM patients increases. We saw no missed injuries in patients selected for NOM for known liver injury. Both increased and decreased transfusion requirements with NOM and AE have been reported.14,16,21,23 In the present material, a significant increase in the number of patients successfully treated non-operatively did not increase the total number of transfusions required. LOS in the ICU is obviously influenced by a multitude of factors. However, in spite of a tendency to keep NOM patients with severe liver injuries longer in the ICU for safety reasons during the study period, their LOS in the ICU did not increase. In the absence of associated injuries requiring intensive care, the patients with grades 3—5 liver injuries who have undergone angiography are now moved out of ICU whenever the haemoglobin has remained stable for a 24 h period. The patients may be discharged after 1 week.
Liver injuries–—Improved results
Conclusion The implementation of a formal NOM protocol decreased total laparotomy rate and seemed to improve patient outcome without jeopardising patient safety. Surprisingly few of the patients undergoing angiography required embolisation. Angiography is not indicated in stable OIS grade 3 liver injuries with no clinical or radiological signs of bleeding, and the protocol in our institution has been adjusted accordingly. AE seems to be a valuable adjunct to DCS with packing of liver injuries.
Conflict of interest None of the authors have any financial and personal relationships with other people, or organisations, that could inappropriately influence (bias) their work, all within 3 years of the beginning the work submitted.
References 1. Asensio JA, Demetriades D, Chahwan. et al. Approach to the management of complex hepatic injuries. J Trauma 2000;48: 66—9. 2. Asensio JA, Roldan G, Petrone. et al. Operative management and outcomes in 103 AAST-OIS grades IV and V complex hepatic injuries: trauma surgeons still need to operate, but angioembolisation helps. J Trauma 2003;54:647—53. 3. Carrillo EH, Spain DA, Wohltmann CD, et al. Interventional techniques are useful adjuncts in non-operative management of hepatic injuries. J Trauma 1999;46:619—22. 4. Christmas AB, Wilson AK, Manning B, et al. Selective management of blunt hepatic injuries including non-operative management is a safe and effective strategy. Surgery 2005;138: 606—10. 5. Ciraulo DL, Luk S, Palter M, et al. Selective hepatic arterial embolisation of grade IV and V blunt hepatic injuries: an extension of resuscitation in the non-operative management of traumatic hepatic injuries. J Trauma 1998;45:353—8. 6. Croce MA, Fabian TC, Menke. et al. Non-operative management of blunt hepatic trauma is the treatment of choice for haemodynamically stable patients. Results of a prospective trial. Ann Surg 1995;221:744—53.
1083 7. Duane TM, Como JJ, Bochicchio GV, Scalea TM. Reevaluating the management and outcomes of severe blunt liver injury. J Trauma 2004;57:494—500. 8. Fang JF, Chen RJ, Lin BC, et al. Blunt hepatic injury: minimal intervention is the policy of treatment. J Trauma 2000;49: 722—8. 9. Goldman R, Zilkoski M, Mullins R, et al. Delayed celiotomy for the treatment of bile leak, compartment syndrome, and other hazards of non-operative management of blunt liver injury. Am J Surg 2003;185:492—7. 10. Hagiwara A, Murata A, Matsuda T, et al. The efficacy and limitations of transarterial embolisation for severe hepatic injury. J Trauma 2002;52:1091—6. 11. Johnson JW, Gracias VH, Gupta R, et al. Hepatic angiography in patients undergoing damage control laparotomy. J Trauma 2002;52:1102—6. 12. Kozar RA, Moore FA, Cothren CC, et al. Risk factors for hepatic morbidity following non-operative management: multicenter study. Arch Surg 2006;141:451—8. 13. Malhotra AK, Fabian TC, Croce MA, et al. Blunt hepatic injury: a paradigm shift from operative to non-operative management in the 1990s. Ann Surg 2000;231:804—13. 14. Meredith JW, Young JS, Bowling J, Roboussin D. Non-operative management of blunt hepatic trauma: the exception or the rule? J Trauma 1994;36:529—34. 15. Miller PR, Croce MA, Bee TK, et al. Associated injuries in blunt solid organ trauma: implications for missed injury in nonoperative management.. J Trauma 2002;53:238—42. 16. Mohr AM, Lavery RF, Barone A, et al. Angiographic embolisation for liver injuries: low mortality, high morbidity. J Trauma 2003;55:1077—81. 17. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury scaling: spleen and liver (1994 revision). J Trauma 1995;38:323—4. 18. Pachter HL, Hofstetter SR. The current status of non-operative management of adult blunt hepatic injuries. Am J Surg 1995;169:442—54. 19. Pachter HL, Knudson MM, Esrig B, et al. Status of non-operative management of blunt hepatic injuries in 1995: a multicenter experience with 404 patients. J Trauma 1996;40:31—8. 20. Richardson JD, Franklin GA, Lukan JK, et al. Evolution in the management of hepatic trauma: a 25-year perspective. Ann Surg 2000;232:324—30. 21. Robinson III WP, Ahn J, Stiffler A, et al. Blood transfusion is an independent predictor of increased mortality in non-operatively managed blunt hepatic and splenic injuries. J Trauma 2005;58:437—44. 22. Velmahos GC, Toutouzas K, Radin R, et al. High success with non-operative management of blunt hepatic trauma: the liver is a sturdy organ. Arch Surg 2003;138:475—80. 23. Wahl WL, Ahrns KS, Brandt MM, et al. The need for early angiographic embolisation in blunt liver injuries. J Trauma 2002;52:1097—101.
www.elsevier.com/locate/injury
Liver injuries–—Improved results with a formal protocol including angiography Christine Gaarder a,*, Pa ˚l Aksel Naess b, Torsten Eken c, Nils Oddvar Skaga d, Johan Pillgram-Larsen e, Nils Einar Klow e, Trond Buanes f a
Trauma Unit, Emergency Division, Ullevaal University Hospital, N-0407 Oslo, Norway Department of Paediatric Surgery, Ullevaal University Hospital, Oslo, Norway c Department of Anaesthesia, Aker University Hospital, Oslo, Norway d Department of Anaesthesia, Ullevaal University Hospital, Oslo, Norway e Department of Radiology, Ullevaal University Hospital, Oslo, Norway f Department of Gastrointestinal Surgery, Ullevaal University Hospital, Oslo, University of Oslo, Norway b
Accepted 5 February 2007
KEYWORDS Non-operative management; Damage control surgery; Liver injury; Angiography; Embolisation
Summary Objective: We hypothesised that a formal treatment protocol for liver injuries including angiography would increase the non-operative management (NOM) rate and would be efficient as an adjunct to damage control surgery. Methods: During the 4-year period from 1 August 2000, a total of 138 adult patients with liver injuries were admitted to the largest trauma centre in Norway and prospectively included in the institutional trauma registry. On 1 August 2002, a protocol mandating angiography in all NOM patients with OIS grades 3—5 liver injuries and after packing of the liver was implemented. All patients admitted during the subsequent 2-year period (group 2) were compared with the previous 2 years as historic controls (group 1). Results: Fifty-five patients were included in group 1 and 59 in group 2. The groups were statistically comparable, both with a mean ISS of 31. Patients selected for NOM increased from 28 (51%) to 45 (76%) ( p < 0.05), without increasing failure rate, liverrelated complications, mortality or transfusion rate. Angiography was performed in 26 patients in group 2 (44%). Only nine patients underwent embolisation (35%), and five of these were in the NOM group. Angiography was negative in the eight NOM stable patients with OIS grade 3 injury. Conclusion: The implementation of a formal NOM protocol decreased total laparotomy rate and seemed to improve patient outcome without jeopardising patient
* Corresponding author. Tel.: +47 41318992. E-mail address: [email protected] (C. Gaarder). 0020–1383/$ — see front matter # 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.injury.2007.02.001
1076
C. Gaarder et al. safety. Surprisingly few of the patients undergoing angiography required embolisation. Angiography is not indicated in stable OIS grade 3 liver injuries, and the protocol in our institution has been adjusted accordingly. AE seems to be a valuable adjunct to DCS with packing of liver injuries. # 2007 Elsevier Ltd. All rights reserved.
Introduction Non-operative management (NOM) of blunt liver injuries in haemodynamically stable or stabilised patients has become standard practice during the last decade.13,18—20 Success rates for NOM are usually reported to exceed 80%.6,13,14,19,20,23 However, the reported overall percentage of patients selected for NOM varies between 47% and 85%.4,9,13,14,19,22,23 Many studies have tried to define independent predictors of outcome and improve the way we treat liver injuries.12,21 The introduction of angiographic embolisation (AE) has been reported to increase the success rate of NOM.3,7,10,16,20,22,23 However, most studies describing NOM and the use of AE are retrospective and compare different groups in the same population, treated differently. Furthermore, reported failure rates, mortality, complications and transfusion rates vary according to definitions, population, and criteria for NOM and angiography. Operative treatment of liver injuries, even in experienced hands, still carries a high mortality and morbidity risk.2,4,7,9,14,18 However, just as NOM is the treatment of choice in haemodynamically stable patients, unstable patients still need an operation. Irrespective of criteria for NOM, operative intervention remains necessary in a significant number of patients. Operative treatment options are depending on available resources and expertise. AE seems to be a valuable adjunct to operative management.1,2,11 since most patients are haemodynamically abnormal at the end of a damage control laparotomy, and ongoing arterial bleeding is difficult to rule out clinically.11 The risk of bleeding in OIS grades 4 and 5 liver injuries has been recognised by several authors.3,9,19 An indication for angiography in these patients offers no controversy, whereas the indication for grade 3 is debated.2,4,5,10,11 However, in a prospective study, Hagiwara et al. reported on 51 patients with liver injuries grades 3—5 undergoing angiography.10 All 14 patients with a CT scan grade 4 or 5 injury were actively bleeding, but also 18 of the 37 patients with grade 3 injuries. On 1 August 2002, a new protocol was implemented in our institution mandating angiography in all patients after damage control surgery (DCS) with packing of the liver and in patients managed non-
operatively with organ injury scale (OIS) grades 3—5 injuries. The purpose of the present study was to evaluate the new protocol, its effect on NOM rate and the need for AE after DCS with packing of the liver.
Patients and methods During the 4-year period from 1 August 2000 to 31 July 2004, a total of 138 adult patients (14 years) with liver injuries were admitted to Ullevaal University Hospital, the largest trauma centre in Norway, and prospectively included in the institutional trauma registry. The trauma registry includes all patients with injury severity score (ISS) > 9 admitted within 24 h after injury as well as all patients admitted for whom the trauma team is activated. Additionally, all penetrating injuries proximal to elbow or knee are included. Of the 138 admitted patients, 24 were categorised as dead on arrival or died in the emergency department and were excluded from further analysis. All prospectively included patients with liver injuries admitted between 1 August 2000 and 31 July 2002, served as historic controls (group 1). During this period, patients who were haemodynamically unstable and had intraabdominal haemorrhage were operated on immediately. All haemodynamically stable patients with suspected intraabdominal injuries underwent computed tomographic (CT) scanning of the abdomen and pelvis. Although NOM was the treatment of choice in the absence of findings suggestive of associated injuries mandating laparotomy, the treatment decision was left to the surgeon in charge. Angiography was not available during this period. From 1 August 2002, the existing treatment policy was formalised in a protocol introducing angiography in severe liver injuries (Fig. 1). All patients with liver injuries prospectively included during the subsequent 2-year period constituted group 2. Patients remaining haemodynamically unstable despite fluid resuscitation were still not eligible for NOM, and underwent immediate laparotomy. All haemodynamically stable or stabilised patients were considered for NOM and had their liver injury diagnosed on CT scan and graded according to the organ injury scale (OIS) as described by the
Liver injuries–—Improved results
1077
Figure 1 Algorithm for treatment of liver injuries. DPL, diagnostic peritoneal lavage; FAST, focused assessment with sonography in trauma; OIS, Organ Injury Scale; DCS, Damage Control Surgery.
American Association for the Surgery of Trauma (AAST) (Table 1).17 The decision to treat the patient non-operatively was based entirely on the patient’s haemodynamic status and the absence of associated injuries mandating laparotomy, and was not influenced by liver injury grade, grade of haemoperitoneum, age, or the presence of head injury. The abdominal CT scans were interpreted by the trauma team leader and the resident radiologist on call.
Angiography and arterial embolisation The new protocol (Fig. 1) mandated angiography in all adult patients with liver injury OIS grades 3—5
and in any patient with clinical suspicion of ongoing bleeding, defined as falling haemoglobin and tachycardia, or any transfusion requirement where the liver could not be ruled out as a significant bleeding source. Acute angiography was performed in haemodynamically stable or stabilised patients when CT demonstrated extravasation or the patient showed clinical signs of ongoing bleeding. In addition, acute angiography was performed in all patients who had undergone DCS with packing of the liver. The other patients underwent angiography the next morning. Our institution is fortunate to have a highly competent angiography service available 24/7/365, on 30 min notice.
1078
C. Gaarder et al.
Table 1 AAST organ injury scale–—liver injury Grade
Injury description
I
Haematoma Laceration
Subcapsular <10% surface area Capsular tear <1 cm parenchymal depth
II
Haematoma
Subcapsular, 10—50% surface area Intraparenchymal <10 cm diameter 1—3 cm parenchymal depth, <10 cm length
Laceration III
Haematoma
Subcapsular >50% surface area or expanding Ruptured subcapsular or parenchymal haematoma Intraparenchymal haematoma >10 cm or expanding >3 cm parenchymal depth
Laceration IV
Laceration
Parenchymal disruption involving 25—75% of hepatic lobe or 1—3 Coinaud’s segments in a single lobe
V
Laceration
Parenchymal disruption involving >75% of hepatic lobe or >3 Coinaud’s segments in a single lobe Juxtahepatic venous injury, i.e. retrohepatic vena cava/ central major hepatic veins
Vascular VI
Vascular
Hepatic avulsion
Angiography was performed via femoral artery puncture. A 4—5 Fr diagnostic catheter was placed in the celiac trunk and advanced into the hepatic artery. According to the protocol, embolisation was performed only when angiography revealed ongoing bleeding, vessel truncation seen as an unsecured bleeding site, or pseudoaneurysm (PSA). Embolisation was performed as peripherally as possible by the placement of microcoils. A completion angiogram was performed to confirm haemostasis of the embolised vessel.
Outcome measures and statistics Data retrieved from the hospital-based trauma registry described demographic characteristics, injury type and severity, as well as haemodynamic status. Data on diagnostic tests, management strategies, and outcome were also collected. Groups 1 and 2 were compared, and outcome is reported as laparotomy rates, failure of NOM, complication and
transfusion rates, and mortality. Failure of NOM was defined as the need for laparotomy after initial non-operative treatment. Failures of NOM were further categorised as being caused by the liver injury or not. Successful AE was defined as achievement of haemostasis primarily, no repeat angiography required, and no delayed bleeding from the liver injury. Complications and the number of transfusions were recorded for the period of hospitalisation in our institution. Angiography related complications were defined as unintentional coil migration, puncture site haematoma or PSA, intimal tears, gallbladder necrosis or allergic contrast reaction. Liver-related complications were defined as delayed liver haemorrhage, intra- and perihepatic abscess, hepatic necrosis, bile leak and biloma. Survival was defined as the patient being alive 30 days after admission. Categorical variables underwent x2 analysis or the Fisher exact test, and continuous variables were subjected to the Mann—Whitney U-test. p < 0.05
Table 2 Patient characteristics Age (range) Male gender, n (%) Blunt injury, n (%) ISS Ps RTS score BE Liver OIS
Group 1 (n = 55)
Group 2 (n = 59)
p-Value
32.4 (17—81) 44 (80) 47 (85) 30.7 18.2 0.74 0.31 6.53 1.43 3.5 4.4 2.5 1.0
33.7 (15—80) 38 (64) 52 (88) 30.6 14.8 0.77 0.26 6.67 1.45 3.6 4.2 2.6 1.2
0.78 0.06 0.89 0.83 > 0.99 0.43 0.67 0.69
ISS, injury severity score; Ps, probability of survival; RTS, revised trauma score; BE, base excess; OIS, organ injury scale. Values are given as mean S.D. where not stated otherwise.
Liver injuries–—Improved results
1079
Table 3 Management strategies and outcome Attempted NOM (%) Failure rate (%) Total laparotomy rate (%) Mortality rate (%) Laparotomy per patient PRBCs first 24 h Total number of PRBCs LOS ICU per patient (days) Complications per patient
Group 1 (n = 55)
Group 2 (n = 59)
p-Value
95% CI for the difference
28 (51) 5 (18) 32 (58) 5 (9) 1.2 (0—7) 7.8 (0—67) 12.1 (0—67) 9.2 (0—75) 1.5 (0—9)
45 (76) 6 (13) 20 (34) 6 (10) 0.7 (0—4) 6.2 (0—60) 9.7 (0—80) 7.3 (0—51) 0.9 (0—8)
0.005 0.85 0.009 0.82 0.02 0.49 0.41 0.41 0.03
0.08; 0.42 0.08; 0.18 0.06; 0.42 0.10; 0.12 1.00; 0.08 5.90; 2.82 8.19; 3.39 6.49; 2.66 1.26; 0.03
NOM, non-operative management; PRBC, packed red blood cells; LOS, length of stay; ICU, intensive care unit; CI, confidence interval. Values are given as mean and total range where not stated otherwise.
was chosen to indicate statistical significance. Statistical analyses were performed using The Statistical Package for the Social Sciences, Version 13.0 (SPSS Inc., Chicago, IL), and Statview 5.0 (SAS Institute Inc.). Data are presented as mean S.D. or mean with total range. The study was approved by the Regional Committee for Medical Research Ethics.
Results Group 1 consisted of 55, and group 2 of 59 patients. The groups were comparable with respect to demographic characteristics, mechanism of injury and injury severity, haemodynamic status on arrival, and had virtually identical mean liver OIS (Table 2). Management strategies and outcome for groups 1 and 2 are listed in Table 3. The emergency lapar-
otomy rate decreased from 49% to 24% ( p = 0.005) without a concomitant increase in failure rate, transfusion rate or LOS in the ICU. Mortality rates were similar in the two groups. There were no liver-related deaths. In group 1, the principal cause of death was head injury in four patients and sepsis and multiple organ failure (MOF) in one patient. In group 2, head injury was the main cause in three patients, and adult respiratory distress syndrome (ARDS) combined with MOF in three patients. The total number of complications (Table 3) per patient decreased significantly ( p = 0.03). Significant complications (Table 4) were registered for 32 patients (58%) in group 1 and 26 patients (44%) in group 2. Liver-related complications were registered in eight patients (15%) in group 1 and three patients (5%) in group 2 ( p = 0.08). Bile leak
Table 4 Significant complications Group 1 (83 complications in 32 of 55 patients)
Group 2 (52 complications in 26 of 59 patients)
Liver-related Delayed hemorrhage Liver pseudoaneurysm/biloma Abscess Liver necrosis Bile leak
4 — 5 1 1
2 — 1 2 3
Other abdominal Abdominal compartment syndrome Wound infection Wound dehiscence/hernia Intestinal fistula Urinary tract infection
2 5 4 3 4
3 2 — — 6
Thoracic Pleural effusion Pneumonia
14 24
4 20
Sepsis/multiple organ failure Sepsis Multiple organ failure
10 6
6 3
1080
C. Gaarder et al. Table 6
Patients undergoing angiography n = 26
Age (range) Male gender (%) ISS RTS score Ps BE Liver OIS
31.5 (17—64) 17 (65) 34.2 13.9 6.5 1.4 0.74 0.27 5.1 4.5 3.4 0.9
ISS, injury severity score; RTS, revised trauma score; Ps, probability of survival; BE, base excess; OIS, organ injury scale. Values are given as mean S.D. where not stated otherwise.
Patients undergoing angiography
Figure 2 Organ injury scale (OIS) grade distribution. The light bars represent group 1, the dark bars group 2. The checked area of each bar represents the number of patients with attempted non-operative management (NOM).
was recorded in three patients and necrosis in two patients in group 2, while only one patient with bile leak was registered in group 1. The mean liver OIS grade in the two groups was similar. The number of patients with OIS grades 3—5 injury was 23 (43%) in group 1 and 32 (54%) in group 2 (Fig. 2). Procedures performed in patients undergoing emergency laparotomy are listed in Table 5. In group 1, 11 laparotomies (41%) were non-therapeutic regarding the liver injury. The corresponding number for group 2 was four (29%). In group 1, 5 of the 28 patients selected for NOM required laparotomy, resulting in a NOM failure rate of 18%. The liver was the cause of the delayed operation in two of these patients, resulting in a liver-specific failure rate of 7%. In group 2, NOM was attempted in 45 patients. Among the six patients (13%) who failed NOM, the liver was the cause in one patient, resulting in a liver-specific failure rate of 2%. The difference was not significant. There were no missed injuries in the patients selected for NOM for known liver injuries.
A total of 26 (44%) of the 59 patients in group 2 underwent angiography (Table 6). Treatment strategies in group 2 are presented in Fig. 3. In 19 patients angiography was performed as an adjunct to NOM, whereas 7 patients underwent angiography after laparotomy with packing. These two subgroups are described separately. NOM patients Of the 19 NOM patients undergoing angiography, 6 patients presented with grade 4 or 5 injuries. Abdominal CT scan demonstrated extravasation in one patient with grade 3 and one with grade 5 injury. Angiography showed ongoing bleeding in three patients, including the two with contrast extravasation on CT scan, and vessel truncation in two patients. Of these five patients, two had grade 3, two had grade 4, and one had grade 5 injury. All five patients were embolised. Angiography was negative in the three patients with OIS grade 2 injury and in the eight NOM stable patients with OIS grade 3 injury. None of these patients had extravasation seen on CT scan. Thus, of the remaining eight NOM patients undergoing angiography, embolisation was performed in five (63%). DCS patients Angiography was performed in seven patients after laparotomy with packing. Four of these patients
Table 5 Emergency laparotomies Group 1 (27 of 55 patients)
Group 2 (14 of 59 patients)
Emergency laparotomy–—liver specific treatment Liver packing Hepatorrhaphy Combination hepatorrhaphy/packing
16 11 3 2
10 8 1 1
Emergency laparotomy–—non-therapeutic for liver
11
4
Liver injuries–—Improved results
Figure 3
1081
Flowchart showing treatment strategies followed in group 2.
presented with OIS grade 3 injury, one with a grade 4 injury, and two with grade 5 injury. Angiography revealed ongoing bleeding in three patients and vessel truncation in one. These four patients (57%) underwent embolisation. In the remaining three patients angiography was negative.
possible intimal tear in the celiac trunk, which was without clinical or therapeutic consequences. There were three deaths in the angiography group. None of them had been embolised. The cause of death was unrelated to the liver injury in all patients, the main cause being head injury in one, and ARDS combined with MOF in two patients.
Success and complications after angiography
Discussion Clinical and angiographical success was obtained in eight of nine embolised patients (89%). The only failure to AE had a grade 5 injury, was taken to angiography, was embolised, but remained unstable and was subsequently operated with packing of the liver. The abdomen was unpacked after 48 h, and closed on day 5. One patient underwent laparotomy because of sepsis and had a nephrectomy performed, unrelated to the angiography. One patient with a grade 5 injury developed necrosis after successful angiographic embolisation following DCS with packing of the liver, and tolerated a formal right hemihepatectomy on day 5. The only complication registered directly associated with the angiographic instrumentation was a
The two most important parameters to describe a study on NOM are the overall percentage of patients selected for NOM and the failure rate. As the first percentage increases to include more patients, the failure rate is also likely to increase. In our population, we show that the new protocol resulted in a significant increase in patients selected for NOM from 51% to 76% (Table 3), without a concomitant increase in failure rate or mortality. Hence, total laparotomy rate also decreased significantly, from 58% to 34%, and there were no liver-related deaths. Although angiography was performed in 26 patients in group 2, only 9 patients were embolised. Furthermore, only five of the embolised patients
1082 were in the NOM group (26%) and one was subsequently operated for bleeding. Obviously, the decrease in laparotomy rate from 58% to 34% cannot be ascribed the effect of AE. During period 1, although NOM was the treatment of choice in haemodynamically stable patients with no associated injuries mandating laparotomy, the treatment decision was left to the surgeon in charge, resulting in a 49% emergency laparotomy rate. By current standards, 51% selected for NOM is relatively low, and the implementation of a standardised protocol with close follow-up might in itself have been the most important factor leading to the increased NOM rate. Although angiography was part of the new protocol it may have been incidental to the reduced laparotomy rate. However, introducing AE offered an additional reliable alternative to operation, thus contributing to the surgeons in charge allowing more aggressive resuscitation before deciding on laparotomy. When surgery is indicated, the goal should be haemorrhage control.8,20 In our institution surgeons have limited exposure to complex liver injuries. A haemodynamically unstable patient undergoing damage control laparotomy for a liver injury will most likely get the liver packed. The rationale for including mandatory angiography after DCS of the liver was that most patients are haemodynamically abnormal at the end of a damage control laparotomy, and ongoing arterial bleeding is difficult to rule out clinically.11 Of the seven DCS patients subjected to angiography in the present material, four underwent embolisation (57%). The risk of bleeding with NOM in OIS grades 4 and 5 liver injuries is significant,3,9,19 and mandatory angiography in these patients offers no controversy.2,4,5,10,11 Some studies show the benefit of angiography in OIS grade 3 injuries.10 Other studies question the value of such a low threshold for angiography and conclude with contrast extravasation on CT scan or clinical signs of ongoing bleeding as selection criteria.22,23 In our population, only 5 of the 19 NOM patients underwent embolisation. All 11 patients with grade 2 and haemodynamically stable patients with grade 3 liver injury had no extravasation seen on CT scan and negative angiographies, and were not embolised. For various reasons, eight NOM patients with grade 3 injury were not subjected to angiography. Two patients died before angiography could be performed. The other six were successfully observed. Although demonstrating some compliance problems with the protocol, they also add to the fact that haemodynamically stable patients with grade 3 injury and no extravasation on CT scan are well off without
C. Gaarder et al. mandatory angiography. Excluding the patients with grade 2 injury and assuming we had excluded haemodynamically stable grade 3 patients from the protocol, our positive angiography rate in the study population would be acceptable to us, with five of eight patients being embolised (63%). After the study period, our protocol has been adjusted accordingly. Worrying complication rates associated with NOM and AE have been presented by others.7,12,16,22,23 In our population, the number of liver-related complications was 11 in group 1 and 8 in group 2, with a higher rate of necrosis and bile leaks registered in group 2. In the angiography group, two patients developed necrosis after embolisation of the liver. One was in the NOM group and recovered uneventfully after observation alone. The other patient underwent DCS with packing of the liver and subsequent successful embolisation of ongoing arterial bleeding. A right-sided hemihepatectomy was performed on day 5 when the patient was physiologically corrected. Bile leak was registered in three patients in group 2 and only one in group 1. Some drainage of bile postoperatively would be expected and might have been inadequately documented in the patient charts, thus contributing to underrating complications in group 1. The same could be argued for other liver-related complications and illustrates one weakness associated with the use of historic controls. In spite of this, the total number of significant complications per patient decreased, probably due to the lowered laparotomy rate. Although the rate of missed associated hollow viscus injuries in patients with liver injuries is low, reported at 2.3% by Miller et al.,15 the risk still worries us as the percentage of NOM patients increases. We saw no missed injuries in patients selected for NOM for known liver injury. Both increased and decreased transfusion requirements with NOM and AE have been reported.14,16,21,23 In the present material, a significant increase in the number of patients successfully treated non-operatively did not increase the total number of transfusions required. LOS in the ICU is obviously influenced by a multitude of factors. However, in spite of a tendency to keep NOM patients with severe liver injuries longer in the ICU for safety reasons during the study period, their LOS in the ICU did not increase. In the absence of associated injuries requiring intensive care, the patients with grades 3—5 liver injuries who have undergone angiography are now moved out of ICU whenever the haemoglobin has remained stable for a 24 h period. The patients may be discharged after 1 week.
Liver injuries–—Improved results
Conclusion The implementation of a formal NOM protocol decreased total laparotomy rate and seemed to improve patient outcome without jeopardising patient safety. Surprisingly few of the patients undergoing angiography required embolisation. Angiography is not indicated in stable OIS grade 3 liver injuries with no clinical or radiological signs of bleeding, and the protocol in our institution has been adjusted accordingly. AE seems to be a valuable adjunct to DCS with packing of liver injuries.
Conflict of interest None of the authors have any financial and personal relationships with other people, or organisations, that could inappropriately influence (bias) their work, all within 3 years of the beginning the work submitted.
References 1. Asensio JA, Demetriades D, Chahwan. et al. Approach to the management of complex hepatic injuries. J Trauma 2000;48: 66—9. 2. Asensio JA, Roldan G, Petrone. et al. Operative management and outcomes in 103 AAST-OIS grades IV and V complex hepatic injuries: trauma surgeons still need to operate, but angioembolisation helps. J Trauma 2003;54:647—53. 3. Carrillo EH, Spain DA, Wohltmann CD, et al. Interventional techniques are useful adjuncts in non-operative management of hepatic injuries. J Trauma 1999;46:619—22. 4. Christmas AB, Wilson AK, Manning B, et al. Selective management of blunt hepatic injuries including non-operative management is a safe and effective strategy. Surgery 2005;138: 606—10. 5. Ciraulo DL, Luk S, Palter M, et al. Selective hepatic arterial embolisation of grade IV and V blunt hepatic injuries: an extension of resuscitation in the non-operative management of traumatic hepatic injuries. J Trauma 1998;45:353—8. 6. Croce MA, Fabian TC, Menke. et al. Non-operative management of blunt hepatic trauma is the treatment of choice for haemodynamically stable patients. Results of a prospective trial. Ann Surg 1995;221:744—53.
1083 7. Duane TM, Como JJ, Bochicchio GV, Scalea TM. Reevaluating the management and outcomes of severe blunt liver injury. J Trauma 2004;57:494—500. 8. Fang JF, Chen RJ, Lin BC, et al. Blunt hepatic injury: minimal intervention is the policy of treatment. J Trauma 2000;49: 722—8. 9. Goldman R, Zilkoski M, Mullins R, et al. Delayed celiotomy for the treatment of bile leak, compartment syndrome, and other hazards of non-operative management of blunt liver injury. Am J Surg 2003;185:492—7. 10. Hagiwara A, Murata A, Matsuda T, et al. The efficacy and limitations of transarterial embolisation for severe hepatic injury. J Trauma 2002;52:1091—6. 11. Johnson JW, Gracias VH, Gupta R, et al. Hepatic angiography in patients undergoing damage control laparotomy. J Trauma 2002;52:1102—6. 12. Kozar RA, Moore FA, Cothren CC, et al. Risk factors for hepatic morbidity following non-operative management: multicenter study. Arch Surg 2006;141:451—8. 13. Malhotra AK, Fabian TC, Croce MA, et al. Blunt hepatic injury: a paradigm shift from operative to non-operative management in the 1990s. Ann Surg 2000;231:804—13. 14. Meredith JW, Young JS, Bowling J, Roboussin D. Non-operative management of blunt hepatic trauma: the exception or the rule? J Trauma 1994;36:529—34. 15. Miller PR, Croce MA, Bee TK, et al. Associated injuries in blunt solid organ trauma: implications for missed injury in nonoperative management.. J Trauma 2002;53:238—42. 16. Mohr AM, Lavery RF, Barone A, et al. Angiographic embolisation for liver injuries: low mortality, high morbidity. J Trauma 2003;55:1077—81. 17. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury scaling: spleen and liver (1994 revision). J Trauma 1995;38:323—4. 18. Pachter HL, Hofstetter SR. The current status of non-operative management of adult blunt hepatic injuries. Am J Surg 1995;169:442—54. 19. Pachter HL, Knudson MM, Esrig B, et al. Status of non-operative management of blunt hepatic injuries in 1995: a multicenter experience with 404 patients. J Trauma 1996;40:31—8. 20. Richardson JD, Franklin GA, Lukan JK, et al. Evolution in the management of hepatic trauma: a 25-year perspective. Ann Surg 2000;232:324—30. 21. Robinson III WP, Ahn J, Stiffler A, et al. Blood transfusion is an independent predictor of increased mortality in non-operatively managed blunt hepatic and splenic injuries. J Trauma 2005;58:437—44. 22. Velmahos GC, Toutouzas K, Radin R, et al. High success with non-operative management of blunt hepatic trauma: the liver is a sturdy organ. Arch Surg 2003;138:475—80. 23. Wahl WL, Ahrns KS, Brandt MM, et al. The need for early angiographic embolisation in blunt liver injuries. J Trauma 2002;52:1097—101.