Clinical review: The critical care management of the burn patient

Between 4 and 22% of burn patients presenting to the emergency department are admitted to critical care. Burn injury is characterised by a hypermetabolic response with physiologic, catabolic and immune effects. Burn care has seen renewed interest in colloid resuscitation, a change in transfusion practice and the development of anti-catabolic therapies. A literature search was conducted with priority given to review articles, meta-analyses and well-designed large trials; paediatric studies were included where adult studies were lacking with the aim to review the advances in adult intensive care burn management and place them in the general context of day-to-day practical burn management.


Pathophysiology of burn shock
Burn injury results in cardiogenic, hypovolaemic and distributive shock. Th e intravascular volume becomes depleted primarily due to increased capillary permeability and fl uid shifts. Above 30% total body surface area (TBSA), only partial compensation can be achieved by fl uid resuscitation due to a generalised reduction in sodium ATPase activity and disruption of the cellular transmembrane ionic gradient that persists for several days. Microvascular injury secondary to infl ammatory mediators such as histamine, bradykinin, prostaglandins, leukotrienes, vasoactive amines, platelet activation products and complement allows protein loss into the interstitium. Th e intravascular colloid osmotic pressure falls and fl uid escapes the vascular system. Th e result is a loss of intravascular fl uid, electrolytes and proteins with rapid equilibration with the interstitial compartments. Clinically, this is manifested by hypovolaemia, haemoconcentration, oedema, reduced urine output and cardiovascular dysfunction. Adequate resuscitation from burn shock is a critical therapeutic intervention in burn management.

Fluid resuscitation
Appropriate fl uid management is the foundation of acute burns management. Without early and eff ective treatment, burns involving greater than 15 to 20% TBSA will result in hypovolaemic shock [3]. Mortality is increased if resuscitation is delayed longer than 2 hours post burn injury [4]. Th e aim is to prevent the development of burn shock and to minimise disruption to physiologic parameters in the face of ongoing cellular and hormonal responses. Several formulae have been developed to optimise fl uid delivery whilst preventing over-resus citation and subsequent pulmonary oedema, and potentially compartment syndrome in unburned limbs or abdomen.
Th ere is limited evidence regarding burns resuscitation. Th e Parkland formula is most widely used and is the current consensus formula. Th e original Parkland formula included an element of colloid resuscitation but this has been omitted since 1979 due to fears over worsening acute respiratory distress syndrome (ARDS). Whilst formulae are a useful guide, the prescription should be adjusted to each patient. Initial fl uid resus citation is often inappropriate: Collis and colleagues reviewed burn size estimation and fl uid prescription in over 300 patients and found that, on average, patients received 150% of recommended fl uid based on the emergency department TBSA estimation, increasing to 200% after TBSA estimation by the burn unit [5]. Baxter identifi ed some patient groups who routinely required further fl uid in addition to that described by the Parkland formula. Th ese groups include patients with inhalation injuries, those with electrical burns and those receiving delayed resuscitation [6]. Both Holm and Csontos and colleagues have found evidence suggesting that the Parkland formula is not accurate for predicting fl uid requirements and instead suggest other methods and monitoring to guide resuscitation [7,8].
Base defi cit and lactate have been shown to correlate with mortality and fl uid resuscitation volumes [9,10]. Physiologic manipulation, however, does not change outcome; correction of acidosis and restoration of cardiac function takes between 24 and 48 hours irrespective of the resuscitation used [11,12]. It is important for clinicians to regularly review physiological parameters and resuscitation endpoints, particularly urine output. A urine output of 0.5 to 1 ml/kg/hour, as recommended by the American Burn Association, should be targeted in the adult patient whilst monitoring pulse, blood pressure and oxygen saturations [13]. Th ere is a paucity of evidence indicating the ideal urine output during resuscitation. Certain situations, including electrical or crush injury with associated rhabdomyolysis, merit additional monitoring and fl uid loading. A small study comparing permissive hypovolaemia using a haemodynamic-guided approach with retrospective controls who received the Parkland formula found a reduction of volume infusion (3.2 ± 0.7 vs. 4.6 ± 0.3 ml/kg/% burn, P <0.001), a reduction in positive fl uid balance (7.5 ± 5.4 vs. 12 ± 4.7 l/day, P <0.05) and a signifi cant reduction in Multiple Organ Dysfunction Score value (P = 0.003) with permissive hypovolaemia [14]. Currently, there is no reliable tool to guide burn shock resuscitation, which has implications for the detection of resuscitation endpoints. Reducing complications of over-resuscitation and under-resuscitation requires ongoing scrutiny from an experi enced burn team.
Pruitt has described the phenomenon of excess fl uid loading as fl uid creep; this is commonly seen. He states that 'adequate resuscitation has been succeeded by fl uid creep, producing excessive resuscitation in the apparent belief that if some fl uid is good, lots of fl uid will be even better' [15]. Fluid creep usually results from inaccuracies in calculating fl uid requirement, from clinician inattention to reducing unnecessary fl uid infusions, from the increased use of sedation and analgesic infusions, and from the excess administration of crystalloid in favour of colloid replacement -or a combination.
Fluid creep issues have led to re-evaluation of the use of colloid. In theory, colloid resuscitation may preserve plasma oncotic pressure, provide effi cient plasma expansion and reduce tissue and pulmonary oedema. An early meta-analysis of burn patients concluded the use of colloids to be deleterious, where the odds ratio for mortality with human albumin solution (HAS) usage was calculated to be as high as 2.40 (95% confi dence limits: 1.11, 5.19) [16]. Th is review evaluated only four trials involving burns, one of which showed an increase in mortality following the use of HAS. Goodwin and colleagues found colloid-resuscitated patients required less fl uid than those who received crystalloid alone (2.98 vs. 3.81 ml/kg/% TBSA) and their haemodynamic parameters were improved during the 12-hour to 24-hour period but patients developed progressive accumulation of lung water up to 7 days post burn [17]. Mortality was higher in the colloid group (11/40) than in the crystalloid group (3/39) although these patients eventually died of causes not obviously related to fl uid resuscitation.
A multicentre randomised control trial involving 42 burn patients found no increase in multiple organ failure rates or mortality following the administration of 5% HAS for burn shock resuscitation [18]. Cochran and colleagues completed a retrospective analysis of patients who received albumin resuscitation compared with a control group who did not receive albumin [19]. After controlling for age, TBSA of burn and inhalation injury, albumin resuscitation did not signifi cantly reduce mortality (odds ratio = 1.90, 95% confi dence interval = 0.85 to 4.22). HAS is available at 5% and 20% formu lations, and there is insuffi cient evidence to determine which might be most appropriate. Some consider the two formulations to be radically diff erent and maintain that a study on 20% albumin in burns resuscitation is sorely needed.
Th ere continues to be no consensus regarding the timing of colloid initiation. Fluid extravasation has been shown to stop by 8 to 12 hours, and Holm believes there is no evidence to indicate the need to delay colloid administration beyond this. Early colloid actually appears to have a pulmonary volume-sparing benefi t [7]. O'Mara and colleagues randomised patients to either a plasma or crystalloid-only resuscitation group, using the Parkland formula to calculate initial fl uid requirements and measured intra-abdominal pressures [20]. Although they did not fi nd an improvement in overall outcomes using plasma resuscitation, they did show a reduction in overall fl uid loading and signifi cantly lower intra-abdominal pressures (26.5 vs. 10.6 mmHg, P <0.0001) in the plasmatreated group.
Hypertonic saline is an alternative to colloid, potentially reducing fl uid shift to the interstitium. A doubling in mortality has been reported as well as a fourfold increase in renal failure, but a Cochrane Review was unable to ascertain whether there was any eff ect on patients with trauma, with burns or undergoing surgery due to lack of data [21,22]. Hypertonic saline is rarely used in our routine practice.
Currently, there is emerging evidence favouring routine use of colloid in managing burn shock. Colloid may reduce oedema-related complications and fl uid creep particularly in those with increasing fl uid requirements during resuscitation. Further studies are required to determine the most appropriate colloid, formulation and timing for its use.

Pharmacological resuscitation
Th ere are no pharmacological agents currently in widespread clinical use. Attempts to reduce the severity of burn shock by blocking some of the chemical mediators of acute infl ammation have been made with some success in clinical burns resuscitation. Such agents include hydralazine (a vasodilator), ketanserin (serotonin antagonist), hydrocortisone and ibuprofen (anti-infl ammatory agent). Interestingly, a randomised prospective study involving 37 patients where vitamin C was given in ultrahigh doses reduced fl uid requirements by 40% in burn shock and signifi cantly reduced the ventilation time (21.3 vs. 12.1 days, P <0.05) [23].

Transfusion, coagulation and blood products
Th e Transfusion Requirements in Critical Care study by the Canadian Clinical Trials Group prospectively randomised critically ill patients to a restrictive strategy of red cell transfusion (maintenance of haemoglobin at 7 to 9 g/dl) or a liberal strategy (maintenance of haemoglobin at 10 to 12 g/dl). Unfortunately this study excluded burns. Th e study found the restrictive strategy to be at least as eff ective as a liberal strategy with the possible exception of patients with acute myocardial infarction and unstable angina [24]. Following a multiple-centre cohort analysis highlighting an increased mortality associated with blood transfusion, there is increasing implementation of a restrictive transfusion strategy [25].
Implementation of a restrictive transfusion policy appears to have not adversely aff ected outcome in paediatric or adult burn patients [26,27]. Palmieri and colleagues undertook a study evaluating the eff ects of transfusion on outcome and revealed that the mean haemoglobin level in patients receiving their fi rst unit of blood was signifi cantly lower outside the operating theatre (8.9 vs. 10.2 g/dl, P <0.05), which is higher than that perceived in medical practice (8.1 g/dl) [25,28]. Nonsurvivors received signifi cantly more blood transfusions than survivors (17.9 vs. 13, P <0.05). Th is association was found after adjusting for multiple variables including burn size using multi-logistical regression. Patients with larger transfusions and poorer outcomes also had larger burns so increased transfusion requirements may simply be a surrogate marker for the severity of burn injury. In addition, the total number of units transfused correlated with the number of infectious episodes (Spearman rank correlation = 0.647; P <0.01) [25]. A prospective, randomised trial of restrictive (7 to 8 g/dl) versus liberal (9 to 10 g/dl) blood transfusion policy in burns >20% TBSA is currently being undertaken by the American Burn Association.
Th e burn patient will intermittently need surgery that may need large or massive transfusion, and sets the burn patient apart from the more general critical care patient. Red cells facilitate haemostasis through a rheological eff ect by pushing platelets to the periphery of the vessel lumen to better interface at the endothelium as well as through direct eff ects on platelet biochemistry. Haemato crit <30% can therefore lead to signifi cant extra blood loss [29].
Transfusion requirements also need to refl ect the surgical and rehabilitation plan. If a large burn excision requiring major transfusion is planned, the patient needs to be optimised for theatre and a low starting haemoglobin concentration seems unwise. If patients are entering the rehabilitation phase, anaemia compromising relevant activities should be avoided: the consequences of delayed or impaired rehabilitation are profound in this patient group.
Th ere has been considerable attention on the use of tranexamic acid to minimise perioperative blood loss and transfusion requirements. However, whilst the popularity of this strategy has increased -including anecdotally in burns -the randomised controlled trials to prove conclusive benefi t do not exist in the burns literature.

Inhalation injury
Airway management and ventilator support are often instigated early in patients with severe burns. Respiratory failure in these patients may be multifactorial. Primary injury to the lungs and upper airway occurs due to direct thermal inhalation injury. Secondary injury can occur early, following activation of the systemic infl ammatory response, or later, following the development of sepsis. Any pulmonary injury can be exacerbated by ventilatorassociated lung damage. Full thickness chest or abdominal wall burns can result in poor compliance and high inspiratory pressures.
Smoke inhalation injury occurs via several diff erent mechanisms not just direct thermal injury to the respiratory tract mucosa. Th e type of injury sustained depends upon factors including the patient's underlying respiratory function, the properties (water solubility) of the fumes inhaled and the extent of exposure. Th e exact pathophysiology of lung injury following smoke inhalation injury is unclear. Th e classic complement cascade is thought to be activated, followed by intrapulmonary leukocyte aggregation and oxygen free radical release resulting in pulmonary oedema [30]. Nitric oxide inhibits hypoxic vasoconstriction and is a key component of the infl ammatory cascade, giving rise to vasodilatation, ventilation/perfusion mismatch and diminished transpulmo nary oxygen transfer [31,32].
In addition to the ongoing infl ammatory cascades, casts formed from cellular debris, fi brin clots, polymorpho nuclear leucocytes, mucous, mucin B5 and airway oedema are likely to cause airway obstruction contributing to respiratory failure [31][32][33][34][35][36]. Partially obstructed airways can lead to air trapping and hyperinfl ation. Th is overdistention leads to release of a neutrophil chemoattractant, IL-8. Following activation, neutrophils produce superoxide that reacts with nitric oxide and causes endothelial damage and increased vascular permeability. Coagulation factors within the exuded plasma, tissue factor released from pulmonary epithelial cells and alveolar macrophages initiate the extrinsic coagulation cascade with resultant fi brin deposition in the alveolar space -a hallmark of smoke inhalation-induced acute lung injury. Fibrin also has an inhibitory eff ect on surfactant, further compounding alveolar collapse.
Systemic toxins such as hydrogen cyanide can be a signifi cant cause of mortality, characterised by signs of hypoxia despite adequate arterial oxygen tension. Various agents such as nitrates and hydroxycobalamin have been used, but some are themselves toxic, and there is little evidence from randomised trials to support their routine use. Carbon monoxide levels higher than 15 to 20% should be treated with 100% endotracheal oxygen. Th ere is little consensus regarding parameters or indications for hyperbaric oxygen, and availability is limited [37].

Ventilatory strategies
In the general intensive therapy unit, the incidence of ventilator-associated lung injury has been reduced following introduction of lung-protective ventilator strate gies using low tidal volumes and permitting a degree of hypercapnia [38]. Other adjunctive strategies, including maintenance of high positive end-expiratory pressure and prone positioning, have not shown any benefi cial eff ect on outcome [39,40].
High-frequency oscillatory ventilation (HFOV) is an unconventional form of mechanical ventilation that is increasingly being used in the management of oxygenation failure in adult ARDS. Th ere is limited information regarding the use of HFOV following inhalation injury [41][42][43]. Th e unique features of smoke inhalation-induced lung injury may limit the use of HFOV: distal alveolal recruitment may be limited by small airways obstructed with carbonaceous debris, oedema and sloughing mucosa; gas trapping and associated hypercapnia may be diffi cult to manage with HFOV; and management of secretions can be problematic. Th e use of HFOV also makes the use of nebulised therapies and therapeutic bronchoscopy impractical. Cartotto and colleagues retrospectively reviewed patients with ARDS after smoke inhalation and those with burns without an inhalational injury that received rescue HFOV for ARDS-related oxygen failure [42]. Patients with an inhalational injury did not achieve a signifi cant improvement in the partial pressure of oxygen/fractional inspired oxygen ratio until 72 hours of HFOV therapy, contrasting with those without an inhalational injury where there was a significant improvement in oxygenation within 8 hours. However, there was no predefi ned HFOV protocol and there were signifi cant baseline diff erences between the groups, including timing of HFOV initiation and pre-HFOV positive end-expira tory pressure.
Th ere are no studies to assess the use of HFOV as an early strategy in acute lung injury after thermal inhalation. Further prospective randomised control trials with well-defi ned strict protocols are required to guide usage of HFOV in ARDS following inhalation injury.

Tracheostomy
Evidence suggests percutaneous tracheostomy is safe in burn patients, but caution should be employed in those with severe head and neck burns or upper airway oedema because airway loss could result in serious complications [44]. Burn surgeons are experienced at these challenging tracheo stomies and coordination of tracheostomy (closed or open) with grafting of the neck requires close communi cation with the surgeon. A retrospective review of adult patients following severe burn injury found patients with tracheostomy had a signifi cantly shorter time to extu bation; however, there was no diff erence in ventilator support, pneumonia prevalence, length of stay or mor tality [45]. As with nonburn patients, there is no consen sus as to the optimal timing of tracheostomy. Saffl e and colleagues found no advantage in early tracheostomy [45], but there may be increased benefi t in children [46]. Early tracheostomy is likely to benefi t most those with major burns plus signifi cant inhalation injury because they will require numerous surgical procedures and prolonged ventilation, and respiratory sepsis is almost inevitable.

Pharmacologic ventilatory adjuncts
Studies in burned animals and humans have revealed multiple possible therapeutic eff ects of glycosaminoglycans for thermal injury. Heparin reduces airway obstruction by potentiating antithrombin-III-mediated inactivation of thrombin and possibly by acting as a free radical scavenger. Multiple studies have shown intravenous and nebulised heparin to be eff ective [47]. Tissue plasminogen activator and antithrombin have comparable eff ects to heparin [47,48]. A small, retrospective, single-centre study with 30 patients using historical controls receiving the same ventilatory strategy found a 38% reduction in mortality (number needed to treat 2.7) and reduced lung injury scores when nebulised unfractionated heparin, N-acetylcysteine and albuterol sulphate were administered in comparison with albuterol sulphate alone [49]. A paediatric case series with historical controls found a signifi cant reduction in mortality, incidence of atelectasis and reintubation rate in the group treated with an alternating regime of aero solised heparin alternating with 20% N-acetylcysteine solution [50]. A multicentre, prospective trial is needed to confi rm this fi nding.
Nitric oxide has been used in burn patients to treat hypoxic pulmonary vasoconstriction and to improve ventilation/perfusion mismatches and therefore tissue oxygenation [51,52]. However, Enkhbaatar and colleagues have shown there is an increase in nitric oxide levels in lung tissue secondary to inhalation injury and that the resultant loss of hypoxic vasoconstriction worsens ventilation perfusion mismatch [53]. Th ey examined the use of BBS-2 (intrinsic nitric oxide synthase inhibitor) on sheep and reported an improvement in pulmonary gas exchange and shunt fraction. Th e use of pharmacological adjuncts for the management of inhalation injury remains limited by the lack of commercial availability of some agents, a lack of human trials and confounding experimental results.

Sepsis
Following severe thermal injury, patients exhibit physiologic and metabolic responses with resultant persistent tachycardia, tachypnoea and rise in baseline core temperature. Virtually all patients therefore meet the criteria for the systemic infl ammatory response syndrome, resulting in little discriminative value in the burn patient. Th e American Burn Association has produced consensus guidelines suggesting modifi ed defi nitions for utilising the systemic infl ammatory response syndrome criteria in burn patients (Table 1) [54]. Other reliable markers of diag nosing sepsis are required. Th ose markers most commonly used include white cell count, C-reactive protein and erythrocyte sedimentation rate. More recently, procalcitonin has been described as a useful marker for sepsis in burn patients and in some studies has been found to be superior to C-reactive protein [55][56][57][58]. In the paediatric setting, however, procalcitonin lagged behind clinical diagnosis by 0.8 days and was less useful than C-reactive protein [59].
Sepsis prevention is critical in the management of the severely burned patient. In addition to standard infection control measures, early excision and skin grafting is now accepted practice. A meta-analysis comparing early excision and conservative management with late grafting identifi ed six studies that met their inclusion criteria. Two of these studies used sepsis as an outcome measure but used diff erent defi nitions of sepsis, making comparison diffi cult. One study showed the early excision group required a shorter duration of antibiotic therapy and had fewer positive wound cultures [60].
Split skin grafting may be limited by the availability of donor sites in extensive burns. In this case, wound care should be optimised with a dermal substitute or appropriate dressing. Silver is a broad-spectrum antimicrobial whose properties are utilised by a number of topical preparations and dressings. Meshing of autografts allows for greater wound cover whilst limiting the donor area. Cadaveric allograft, often in combination with autograft in a sandwich technique, is seen by many as the best available temporising measure and grafts retain immunological function. Th e huge expense of allograft and its preorder for surgery adds another dimension for communication between intensivists, anaesthetists and surgeons lest product is wasted.
In adults, there is no role for prophylactic antibiotics unless utilised for wound excision or manipulation procedures due to the risks of adverse events and antimicrobial resistance. A systematic review and metaanalysis found a potential benefi t of prophylactic antibiotics in burn patients but from data of poor methodological quality [61]. Selective decontamination of the digestive tract (SDD) in burns is generally supported by a small number of studies: a double-blind randomised control trial of 117 adult burn patients showed a signifi cant reduction in both mortality and pneumonia in the treatment group with respect to placebo without adverse eff ects [62]. A recent metaanalysis of the three published SDD randomised control trials in burn patients, recruiting 440 patients (289 SDD patients, 151 controls), confi rmed that SDD signifi cantly reduced mortality by 78% (P <0.001) [63]. Other studies have found a reduction in wound colonisation with the administration of SDD [64,65]. A small paediatric randomised control trial did not fi nd any signifi cant diff erences in rates of colonisation, pneumonia or sepsis [66].
Burns sepsis needs rapid recognition and treatment with appropriate antibiotics, guided where available by microbiological results. Large burns are associated with fungal colonisation and infection so additional investigation and antifungals may be required. Multi drugresistant organisms are becoming an increas ing problem in the burn care setting. Topical antibiotics may off er some effi cacy against these organisms and can be used in conjunction with systemic antimicrobials. Alternatively, drugs with less well tolerated profi les may be required, such as colistin for the treatment of multi-resistant acinetobacter species.
Th e pharmokinetics of many drugs can be altered due to the physiologic changes following major burns. Altered pharmokinetics of major antibiotic classes can result in subtherapeutic levels of the drug [67,68]. Consideration should be given to reduced dosing intervals or the use of continuous infusions guided by appropriate monitoring, to ensure therapeutic plasma levels.
Th ere is limited evidence to support routine use of corticosteroids in burn patients with sepsis [69,70]. Th e Surviving Sepsis Campaign recommends intravenous hydro cortisone should be used in patients with septic shock only if they are poorly responsive to both fl uid resusci tation and inotropic support because there is a paucity of evidence showing a reduction in mortality plus the possibility of a higher incidence of secondary infection [71].
Th ere remains controversy regarding the use of immune-enhancing dietary supplements. Studies have reported variable results following the use of glutamine supplementation. Current evidence from nine randomised control trials supports the use of enteral glutamine supplementation for patients with severe burn injuries [72]. Improvements were seen in wound healing, gut permeability, length of stay and mortality but there was heterogeneity between studies in both methodology and reporting. Further investigation is required to answer ques tions regarding dosage, timing, and length of supplemen tation. Research to date has provided insuffi cient evidence to support the use of other immunonutrients such as fi sh oil or arginine. Currently, there are no clinically approved topical immunomodulators available for the burn patient.

Nutrition and hyperglycaemia
Patients following a severe burn injury are hypermetabolic with resultant protein loss and reduction in lean body mass and hyperglycaemia through some of the same mechanisms as other critically ill trauma patients. Whilst the hypermetabolic state provides glucose to glucose-dependent tissues, it ultimately contributes to immune dysfunction, sepsis and organ failure. Holm and colleagues identifi ed increased adverse outcome in adults following hyperglycaemia in the fi rst 48 hours post injury [73]. Th e paediatric burn literature showed that hyperglycaemia is associated with increased catabolism, bacteraemia/fungaemia, skin graft loss and mortality whilst intensive insulin therapy was associated with survival [74][75][76][77][78].
Th e Normoglycaemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation Investigators assessed glucose targets in a mixed ICU population, using intensive insulin control versus conventional insulin control (4.5 to 6 vs. 8 to 10 mmol/l) in more than 6,000 patients and reported lower mortality in the conventional group (24.9 vs. 27.5%, P = 0.02) [78]. Following subgroup analysis, however, the trauma group (not stipulated whether this included patients with burns) benefi ted from tight glycaemic control. Control of hyperglycaemia can be diffi cult in burn patients and intensive insulin treatment can lead to a higher risk of hypoglycaemia, which can be compounded by frequent trips to theatre and breaks in enteral feed [79]. Cochrane and colleagues evaluated an intensive insulin protocol (aiming to maintain glucose <12 mmol/l) in 30 patients, 17 with burns and 13 with soft tissue infections [80]. Th ey experienced a 5%/day rate of hypo glycaemic episodes (glucose <6 mmol/l) but reported no neurological sequelae or deaths. Studies examining the eff ect of tight glycaemic control in burns are limited, but mortality benefi ts as evidenced by Van den Berghe and colleagues [81] and others has led to the control of hyper glycaemia becoming a standard of care.
Insulin has additional immunomodulatory eff ects beyond increasing cellular uptake of glucose. Insulin has a trophic eff ect on mucosal and skin barriers, reducing bacterial invasion and translocation, improves wound matrix formation and inhibits the production of proinfl am matory mediators.
Failure to satisfy the nutritional requirements of the hypermetabolic burn patient will lead to impaired wound healing, susceptibility to infection, organ failure and death [82][83][84][85]. A host of formulae to calculate calorifi c needs are used, and various feeding formulations are avail able. In general, the evidence greatly favours enteral to parenteral nutrition in relation to outcome measures and cost.
Nasogastric and nasojejunal feeding each have their proponents, with the advocates of duodenal feeding citing the ability to continue feed during surgical procedures without risk of aspiration, although reduced aspiration with intestinal feeding has not been proven [86,87]. Duodenal placement remains a challenging and inexact science requiring assistance with some imaging modality, but technological advancements have aided this at the bedside and many centres are moving towards jejunal feeding whether by fashion or evidence. Pro-motility agents such as erythromycin are often employed to assist maintenance of tubes in the correct location. Highquality evidence in many areas of nutrition remains very limited.
Th e combination of aggressive fl uid resuscitation, early enteral feeding and proton pump inhibitors has reduced the incidence of acute gastrointestinal ulceration (Curling's ulcer) from 15% to 3% with a sizeable decrease in mortality from former death rates of 70% [88].

Hypermetabolism
Burn-induced catabolism results in accelerated protein breakdown and reduced protein synthesis, culminating in a negative net nitrogen balance. Demling and DeSanti reported that oxandrolone (an oral anabolic agent) improved weight gain and was an unquantifi ed measure of muscle function [89], and in a later study showed a reduction in weight loss, urinary nitrogen loss and a shortened time to wound healing [90]. A prospective multicentre random ised double-blind trial of 81 patients with burns of 20 to 60% TBSA found that starting enteral oxandrolone 5 days post injury signifi cantly reduced the hospital stay (45.3 vs. 32 days, P = 0.035) without signifi cant adverse events [91].
Th e hypermetabolic response is in part mediated by endogenous catecholamines: shortly after severe burn or trauma, plasma catecholamine levels can increase up to 10-fold [92]. β-blockade following severe burns can reduce supraphysiologic thermogenesis, tachycardia, cardiac work and resting energy expenditure. Herndon and colleagues conducted a prospective randomised control trial in paediatric patients with severe burns to evaluate the eff ect of propanolol on muscle catabolism [93]. Th e net muscle-protein balance increased by 82% from baseline values in the propanolol group (P = 0.002) and was signifi cantly higher with respect to the control group (P = 0.001). Th e mechanism of action of propanolol remains unclear but appears to be a result of increased protein synthesis during a period of persistent protein breakdown and reduced peripheral lipolysis [94].

Acute kidney injury
Acute kidney injury (AKI) is common following a burn injury with an incidence as high as 30% but is associated with a reported mortality of between 80 and 100% [95][96][97]. When defi ned by the Risk, Injury, Failure, Loss and End-stage kidney classifi cation [98], AKI occurred in one-quarter of patients with severe burn injury (median mortality of 34.9%); and when defi ned by the need for renal replacement therapy, AKI occurred in 3% (median mortality of 80%) [99]. Mortality increases in a stepwise fashion with increasing Risk, Injury, Failure, Loss and End-stage class [95,100]. Steinvall and colleagues reported approximately one-half of patients, particularly those with a severe burn injury, developed AKI during the fi rst week post injury and the remaining patients during the second week [101]. In the majority of these patients, AKI was preceded by other organ dysfunction or sepsis. Th e burn shock resuscitation protocol used was successful at preventing early hypo volaemic AKI. Th is suggests that renal ischaemia, at least in the acute phase of burn shock, has a limited role in the development of AKI. Mosier and colleagues reported that both patients with or without an AKI averaged a urine output within the recommended 0.5 to 1.0 ml/kg/hour or greater; however, those who developed early AKI had a signifi cantly lower average urine output (0.6 vs. 1.16 ml/ kg/hour, P <0.01) [102].
A study looking at acute respiratory dysfunction in patients with major burns revealed that ARDS occurs early, usually within 3 days, and that renal dysfunction was more common amongst those with the most severe respiratory dysfunction [103]. In sepsis, there is increasing evidence to suggest that infl ammation and apoptosis play a key role in the aetiology of AKI [104] the literature in burns is limited. Late-onset AKI has been strongly associated with sepsis in the past [105]. A recent prospective cohort study found septic episodes were not always followed by renal dysfunction and two-thirds of late-onset AKI was not preceded by sepsis [101]. Th e development of intra-abdominal hypertension and abdominal compartment syndrome following overresuscitation may contribute to AKI of later onset. Exposure to nephrotoxic antibiotics and intravenous contrast may also have a role.
Aggressive utilisation of continuous renal replacement therapy (50.2 ± 13.2 ml/kg/hour) in severely burned casualties with AKI signifi cantly improved survival [106]. A recent study of patients with severe burn injury defi ned as TBSA >40% with AKI treated with continuous veno venous haemofi ltration with a maintenance dose of 20 to 35 ml/kg/hour compared with historical controls prior to the availability of continuous venovenous haemofi ltration found the use of continuous venovenous haemofi ltration to be associated with a reduction in 28-day stay (38 vs. 71%, P = 0.011) and hospital mortality (62 vs. 86%, P = 0.04) [107]. Significant clinical improvements were seen in those with respiratory dysfunction and shock. Th e American Burn Association's randomised controlled evaluation of haemofi ltration in adult burn patients with septic shock and acute renal failure (RESCUE) is currently recruiting patients as of May 2013.

Conclusion
Regional burn centres have developed to address resource requirements and the complexity of burn care. Th ere have been signifi cant improvements in outcome following burn injuries in the latter half of the 20th century, refl ecting advances in critical care -in parti cular, following the introduction of fl uid resuscitation protocols, early burn excision and closure, antimicrobials and infection control, nutritional support and modu lation of the metabolic response. Th ese improvements in care have resulted in the majority of patients outside extremes of age treated in a modern burn centre being expected to survive [108]. Th e US national burn repository with reports submitted from 91 hospitals showed that the length of stay has reduced in the past 10 years from approximately 11 days to 9 days, and mortality has fallen in females from 6.8% to 3.6% and in males from 4.5% to approximately 3% [109].
Recent evidence suggests the routine use of colloid resuscitation and permissive hypovolaemia to reduce fl uid creep and its attendant complications. Whilst many strategies including oxandralone have been utilised successfully to attenuate the hypermetabolic response, further studies are required to assess their safety and appropriate use. Currently, β-blockade using propanolol appears to be the most effi cacious anti-catabolic therapy. At the Mersey Regional Burn Centre we have adopted the routine use of oxandrolone and have a fl uid resuscitation protocol based on the Parkland formula but modifi ed to 2 to 3 ml/kg/% burn with albumin rescue [110]. Burns management is a rapidly evolving fi eld with 8,000 burnsrelated articles in the last 10 years as compared with 11,000 in the last 90 years, with numerous studies underway that will hopefully provide further guidance for the management of these critically ill patients.