Predictors of outcome in patients requiring surgery for liver trauma☆

Introduction 

Despite its relatively protected position, the liver is the most frequently injured solid intra-abdominal organ.4, 9, 14 Associated injuries to other organs, uncontrolled haemorrhage from the liver and subsequent development of septic complications significantly contribute to morbidity and death.1 Mortality rates have fallen from 66% in World War II to current levels of 28%.7, 27, 30 Reducing the morbidity and mortality from haemorrhagic shock and subsequent sepsis remain the main challenges in the management of liver trauma.7 The aim of this study was to document the outcome of the management of liver trauma and to establish factors that influence outcome.

Patients and methods 

The surgical service at King Edward VIII Hospital consists of three surgical wards. This is a prospective study undertaken in a single surgical ward at King Edward VIII Hospital over a 7-year period from 1998 to 2004. Patients underwent resuscitation prior to being submitted to laparotomy. In a proportion of patients the laparotomy formed part of the resuscitation. A prophylactic antibiotic policy was employed. Local policy for the management of patients with blunt trauma to the abdomen with suspected solid visceral injury, but with no clinical features suggestive of hollow visceral injury, is non-operative management (assuming no haemodynamic instability), i.e. strict bed rest, close observation, regular full blood count and CT scan of the abdomen.

Clinical data including demographics, intra-operative findings, operative management and outcome were collected onto a proforma and subsequently entered into a computer database. Delay before surgery included both pre-hospital and in-hospital duration. Hypovolaemic shock was defined as a systolic BP≤90mmHg. Injuries were graded according to the Organ Injury Scale as described by the American Association for the Surgery of Trauma.19 The severity of injury was documented using the Injury Severity Score (ISS)31 and patients were grouped into three categories (group 1, ISS≤9; group 2, ISS=10–20 and group 3, ISS>20).

Liver injuries found at laparotomy were managed on their merit. Our policy is minimal surgery for minor injuries. Drainage is used only for minimal bleeding or bile leak. Management of major injury includes suture of the laceration, non-anatomic resection and packing depending on the extent of the injury.

Data were analysed using the Statistical Package for the Social Sciences (SPSS) version 11.5. The Chi-squared method was used to assess the influence of shock and delay on outcome and, where numbers were very small, Fisher's exact test was used. The one-way ANOVA test was used to assess the influence of injury mechanism, grade of liver injury and Injury Severity Score on outcome. A p-value of <0.05 was taken as statistically significant.

Results 

A total of 478 patients sustained abdominal trauma during this period. There were 105 with liver injury, of whom 98 were males, giving a male to female ratio of 14:1. The mean age was 27.81±10.33 years (range 13–58 years). Firearms caused 70 of the injuries (67%) followed by stabs in 26 patients (25%) and blunt trauma in 9 patients (9%).

All patients presented with varying degrees of peritonism. Disembowelment occurred in 6 of 26 patients with stab wounds. Hypovolaemic shock was present in 19 patients (18%). Four of these patients underwent laparotomy as a form of resuscitation. Diagnosis of hepatic trauma was made at laparotomy in all patients and none were managed non-operatively. Total delay before surgery was 7.98±9.07h; with delay of ≤6h in 58 patients (56%) and >6h in 47 patients (44%).

There were 84 patients (80%) with haemoperitoneum (range 50–5000ml). The Injury Severity Score (ISS) ranged from 4 to 50 with a mean of 14.85±9.14. There were 46 patients in group 1, 43 patients in group 2 and 16 patients in group 3.

There were 31 patients (30%) with isolated liver trauma and 74 with associated injuries (70%). The latter occurred in 13 stab injuries (50%), 44 firearm injuries (77%) and 7 blunt injuries (78%). Associated hollow visceral injury occurred in 63 patients (60%). Common associated organ injuries were colon (36), stomach (35%), diaphragm (27%), small bowel (24), spleen (13) and kidney (12). Management of the liver injury as stratified according to grade of injury at laparotomy is shown in Table 1. Surgical options for management of liver injuries at laparotomy included conservative management, application of absorbable gelatine sponge (Spongostan: Mascia Brunelli Spa, Milano, Italy; Surgicel: Ethicon, Branswick, NJ, USA), suture of the liver injury, peri-hepatic packing and resection. Resection was non-anatomical.

Table 1. Management of liver injuries in 105 patients stratified according to grade of injury

	Grade	Total	Conservative	Gelatine sponge	Suture	Pack	Resection	Mortalitya
	1	28	21	–	7	–	–	3 (11)
	2	37	19	4	11	3	–	6 (16)
	3	27	9	2	13	2	1	6 (22)
	4	8	–	1	5	2	–	2 (22)
	5	5	–	–	–	4	1	4 (80)
	Total	105	49	7	36	11	2	21 (20)

Grades 1–4 vs grade 5; p<0.0001. Values in parentheses are in %.

Postoperative complications arose in 39 patients (37%) and included pulmonary infection (16%), multiple organ dysfunction syndrome (13%), peritonitis (including anastomotic dehiscence and abscess) in 13 patients, wound sepsis (6%) and fistula (6%). These figures overlap as some patients developed more than one complication. There was no difference in complication rate between the different grades of injury. Table 2, Table 3 address the influence of various factors on morbidity and mortality rates. There was no difference in morbidity between the three injury mechanisms. Patients with the lowest ISS (group 1) had the lowest morbidity. Nine out of 31 patients with isolated liver injuries developed complications (30%) compared to 29 out of 74 with associated injuries (39%) (p=0.501). There was no difference in morbidity between patients with associated hollow visceral injury and those without associated hollow visceral injury (p=0.835) but patients with hollow visceral injury had a higher mortality (p=0.045). Shock on admission, delay before surgery and the use of drains had no influence on morbidity.

Table 2. Influence of injury mechanism, injury severity on morbidity and mortality

		Morbidity		Mortality
		n	%	n	%
	Injury mechanism
	Stab (n=26)	6	23	1	4
	Firearm (n=70)	27	39	16	23
	Blunt (n=9)	6	67	4	44
	Injury Severity Score
	Group 1 (ISS≤9; n=46)	7	15	1	2
	Group 2 (ISS=10–20; n=43)	22	51	10	23
	Group 3 (ISS>20; n=16)	10	63	10	63

Injury mechanism: morbidity—stab vs firearm→p=0.484; stab vs blunt→p=0.132; firearm vs blunt→p=0.548; mortality—stabs vs blunt→p=0.013; stab vs firearm→p=0.101; blunt vs firearm→p=0.207. Injury Severity Score: morbidity—group 1 vs group 2→p<0.0001; group 1 vs group 3→p=0.001; group 2 vs group 3→p=1.0; mortality—group 1 vs group 2→p=0.015; group 1 vs group 3→p<0.0001; group 2 vs group 3→p=0.001.

Table 3. Influence of shock, delay, associated injuries and use of drain on morbidity and mortality

	Factor	Morbidity		Mortality
		n	%	n	%
	Shock
	Shock (n=19)	9	50	11	58
	No shock (n=86)	29	34	10	12
	Delay
	Delay ≤6h (n=58)	24	42	16	28
	Delay >6h (n=47)	14	30	4	9
	Associated injury
	Isolated injuries (n=31)	9	29	1	3
	Associated injuries (n=74)	29	39	20	27
	Associated HVI (n=63)	24	39	17	27
	No associated HVI (n=42)	15	36	4	9.5
	Drains
	Drain (n=39)	17	44	8	21
	No drain (n=66)	22	33	13	20

HVI=hollow visceral injury. Morbidity: shock vs no shock→p=0.301; ≤6h vs >6h delay→p=0.305; isolated vs associated injury→p=0.501; associated HVI vs no HVI→p=0.835; drain vs no drain→p=0.403. Mortality: shock vs no shock→p<0.0001; ≤6h vs >6h delay→p=0.008; isolated vs associated injury→p=0.006; associated HVI vs no HVI→p=0.045; drain vs no drain→p=1.0.

Forty patients required intensive care management (38%) and their ICU stay was 6.55±5.65 days. Twelve patients (63%) who presented in shock required ICU compared to 28 (32%) of those who were not in shock (p=0.017). Twenty (19%) patients required a re-laparotomy for various reasons including removal of packs (7), no improvement (6), peritonitis (3), intra-abdominal abscess formation (2) and intestinal obstruction (1). There were no complications directly specific to liver injury such as bile leaks.

Twenty-one patients died (20%); 44% following blunt trauma, 23% following firearm injury and 4% for stab wounds (p=0.013 for stabs versus blunt trauma; p=0.132 for stabs versus firearms and p=0.548 for firearms versus blunt trauma). Mortality rates among patients in groups 1, 2 and 3 were 2, 23 and 63%, respectively (group 1 versus group 2 p=0.015; group 2 versus group 3 p=0.001; and group 1 versus group 3 p<0.0001). Eleven patients (58%) with shock on admission died compared to 10 (12%) without shock (p<0.0001). Among 58 patients with delay of ≤6h 16 (28%) died, compared to 4 (9%) out 47 of those whose delay was >6h (p=0.008). Patients with grade 5 injuries had a significantly higher mortality rate compared to patients with grades 1–4 injuries (p<0.0001). Significantly more patients with associated injuries died compared to those with isolated injury (3% versus 27%; p=0.006). Table 4 shows the mortality rate increases with the number of associated injuries per patient. The average hospital stay was 11.27±12.09 (range 1–62) days.

Table 4. Mortality stratified according to associated intra-abdominal injuries

	Associated organs		Morbidity		Mortality
	Injuries/patient	No. of patients	n	%	n	%
	0	31	9	31	1	3
	1	31	10	32	6	19
	2	17	6	35	4	24
	3	16	9	56	6	38
	4	9	4	44	3	33
	≥5	1	0	0	1	100

Discussion 

The liver is the most commonly injured intra-abdominal solid organ following trauma.4, 14 Due to its anatomical location, severe liver injuries usually lead to exsanguinating haemorrhage which is the most common cause of death.16 Liver injuries accounted for 22% of abdominal injuries in this study, the majority associated with penetrating trauma. This is not in keeping with the world literature in which penetrating trauma is responsible for only 1–11% and blunt trauma for 23–50% of liver injuries.7, 14, 18, 27 The reason for this discrepancy is that firearm injuries are common in this country, an unfortunate continuing trend.21

The average time before surgery was 7.98±9.07h, which is longer than that reported by others.6 In Durban, as in the rest of the country, the period taken for an injured patient to reach hospital tends to be longer than the accepted norms. This is due to occasionally inadequate emergency medical services.

The majority of liver injuries (70–90%) are minor (grades 1 and 2) and require minimal treatment.3, 11 Minor injuries accounted for 62.5% of injuries in this study. The average Injury Severity Score of 15 was much lower than the 28.4+3.18 reported by Gür et al.14 We believe that stab wounds and low velocity firearm injuries tend to cause less anatomical disruption than blunt trauma, which may explain the low ISS in the current study. The dominant injury mechanism in the study by Gür et al.14 was blunt trauma.

The recognition that 50–80% of liver injuries stop bleeding spontaneously, coupled with better imaging of the injured liver by computerised tomography (CT), has led to the acceptance of non-operative management with resultant decrease in mortality rates.25 This is especially more applicable to blunt liver injuries,6 although selective non-operative management of liver injuries is now becoming acceptable for firearm injuries as well.5, 20, 24 Specific criteria have been suggested to aid the surgeon in the selection of patients suitable for conservative management.10, 24, 26 The critical factor is haemodynamic stability.26 The majority of injuries were firearm injuries and this may explain the lack of non-operative management of liver injuries in this study. Firearms, especially high velocity weapons, are known to cause devastating injuries and few can fulfil these stated criteria although this has been described.24

Prompt, expeditious and appropriate operative management of liver injuries will reduce peri-operative morbidity and mortality. The objectives in the operative management of liver trauma are to control the bleeding and bile leak, remove devitalised tissue, control infection in patients with associated hollow visceral injury and to establish adequate drainage of the abdomen.15 Manual compression, Pringle manoeuvre and continued resuscitation with blood component therapy are essential to control the bleeding.15 Damage control in the form of peri-hepatic packing is ideal for patients who are in extremis or those undergoing initial laparotomy at a district hospital.15 There are also a variety of manoeuvres to stop bleeding ranging from suturing, diathermy and laser, fibrin glue and resection.15 The other aspects of management are secondary to control of haemorrhage but are equally crucial.

While some authors have previously advocated an extensive surgical approach such as anatomic resection, performed through conventional anatomic planes,29 recent trends have been toward a more minimal surgical procedure such as non-anatomic resection to remove ischaemic parts of the liver.6, 8, 15

The morbidity rate associated with liver injury varies significantly depending on the mechanism of injury and ranges from 5 to 24%.6, 7 Mechanism of injury, delay before surgery, shock on admission, grade of injury, associated injury and the use of drains had no effect on morbidity in this series. After haemorrhage, sepsis remains the major factor for liver related morbidity.7 The severity of liver injury, the number of associated abdominal injuries and shock have been demonstrated to correlate with an increasing rate of sepsis.7, 12, 23 Other studies,7 including the present series, did not show this trend. Although associated hollow visceral injury might be assumed to be associated with increased bacterial contamination and therefore sepsis,7 it was not shown to be so by the present study. A high ISS was associated with a high morbidity.

The literature quotes liver injury specific complications at 7%.7 There were no liver injury specific complications in this study. The failure of associated injury and the use of drains to influence morbidity in this study is contrary to what the literature suggests.7, 22, 23 There is conclusive evidence that open drainage is not advisable for liver injuries13 and that sump drainage is hazardous and should not be used.7 The use of closed drainage to evacuate blood and bile, however, remains controversial.7 In a prospective randomised study Mullins et al.22 found a liver related septic complication rate of 8% for both drainage and no drainage. Gilmore et al.13 on the other hand found more complications with drainage than without drainage. There is a slightly higher incidence of peri-hepatic abscess with no drainage than with drainage.7 There is now growing consensus that drainage is not necessary for mild injuries (grades 1 and 2) if there is no bleeding or bile leak.2, 7

Mortality from liver injuries is generally in two phases: early deaths arise from haemorrhage and hypovolaemic shock from either the liver or associated major vascular injuries and late deaths from sepsis and multiple organ dysfunction syndrome (MODS).7 The majority of early deaths from blunt trauma are primarily due to brain injury although haemorrhage does play an important role with major liver fracture, and late deaths can be attributed to closed head injury and sepsis with MODS.7

The mortality rate associated with liver injuries varies according to mechanism of injury and ranges between 1 and 40% but this figure may double or even treble with associated intra-abdominal injuries.3, 6, 7, 14, 19, 28 The mortality rate of 20% in this series falls within this range. Although representing only 8% of the patient population, blunt trauma comprised 20% of all deaths in this study and had the highest mortality rate compared to firearm and stab wounds. Blunt injury impacts on a larger surface area of the liver and is therefore more likely to produce a higher grade of injury. Although blunt injury caused the highest mortality the numbers were too small to conclude that it is a predictor of mortality. Furthermore, any associated injury tends to be more severe resulting in higher mortality rate compared to penetrating injury.7

Mortality rate was highest for grade 5 injuries. This is not surprising as grade 5 injury is associated with massive force causing extensive damage. Also not surprising was the high mortality rate associated with the highest category of Injury Severity Score. Patients that were admitted in shock had a higher mortality rate compared to those admitted without shock. This is due to the fact that shock results from severe exsanguinating haemorrhage, resulting in metabolic instability and MODS. This trend has been observed in other series.17

Patients with a delay of 6h or less before surgery had a higher mortality compared to those with delay of more than 6h. This surprising statistic may be explained by the concept of natural triage. Patients who present with little delay will represent a spectrum of all grades of liver injury including severe damage with shock. Those with major trauma who are subjected to long delays will inevitably succumb to their injury without reaching hospital, leaving a higher proportion of less grave injuries in the delayed group and a higher survival rate. While early presentation was associated with a high mortality it cannot be concluded that delay before surgery is a predictor of outcome as the high early mortality is due to other factors discussed above.

We are in agreement with Duane et al.6 that the complex nature of the operative management for liver trauma renders the mortality high regardless of the choice of procedure. Furthermore any operative procedure used when all else has failed and the patient is extremely unstable, is likely to result in a poor outcome.6 The surgeon must therefore be familiar with all the techniques including packing, and must tailor the surgical approach to the individual injury.6, 7 A familiarity with the indications, limitations and applications of the various approaches should also be developed.7

In conclusion therefore shock on admission, high grade of injury, high Injury Severity Score (ISS) and presence of associated organ injury are associated with increased mortality rate and can be regarded as predictors of outcome. Further studies with large numbers are necessary to address the role of injury mechanisms as a predictor of outcome.