Mortality of patients with alcoholic liver disease admitted to critical care: a systematic review

Transcription

1 The Intensive Care Society 2012 Mortality of patients with alcoholic liver disease admitted to critical care: a systematic review S Flood, A Bodenham, P Jackson This review examined the burden of alcohol-related liver disease (ALD) in the intensive care unit, which is increasing, and whether scoring systems can assist in judging prognosis. Embase, Medline and internet databases were searched for relevant articles whose quality was then scored using the Centre for Evidence-Based Medicine s (CEBM) critical appraisal tool. Unit mortality of patients with ALD admitted to intensive care in these studies was between 40-50%. In comparison with liver-specific prognostic scoring, physiological scoring systems discriminated better between survivors and nonsurvivors. This is likely to be a reflection of the fact that patients with ALD in intensive care tend to die of multi-organ failure rather than isolated acute liver failure. Keywords: alcohol; cirrhosis; mortality; prognosis; intensive care; scoring systems Introduction The healthcare burden from alcohol-related illness has risen dramatically over the last decade. It is estimated that alcoholrelated disease costs the NHS in England 1.7 billion per year. 1 There were in excess of 800,000 alcohol-related admissions to hospital in the year , an increase of 69% from The number of alcohol-related deaths in the UK has increased from 4,023 (6.7 per 100,000 population) in 1992 to 9,031 (13.6 per 100,000) in The majority of these (4,764) were attributable to alcoholic liver disease (ALD). ALD encompasses a spectrum of liver injury, from steatosis or fatty liver through to acute alcoholic hepatitis and liver cirrhosis. The Intensive Care National Audit and Research Centre (ICNARC) database estimates that nationally, critical care admissions due to ALD increased from 550 in 1996 to 1,513 in 2005, with an estimated annual expenditure of 14.7 million (based on 2005 NHS costs). 4 The authors admit this is likely to be an underestimate. Patients with cirrhosis who become critically ill have traditionally been viewed as having a universally poor outcome. 5,6 This systematic review aims to objectively summarise the outcome data for patients with ALD and examine the prognostic value of liver-disease-specific versus physiology-based scoring systems. Methodology The medical literature databases Medline and Embase were searched, with a variety of thesaurus and non-thesaurus linked search terms. No date limits were applied. The searches were limited to articles published in English and were performed via the NHS Evidence Health Information Resources library at The search terms can be grouped into three categories according to the main aspects of the review s primary Mortality/Outcome Alcoholic liver disease Intensive care unit Prognosis Alcohol liver disease Intensive care Disease-free survival Alcohol liver cirrhosis Artificial ventilation Medical futility Alcoholic fatty liver Critical AND care Mortality alcohol* AND liver AND disease NOT non Intensive AND care Progno* alcohol* NOT non AND hepatitis Intensive AND therapy Survival alcohol* NOT non AND cirrho* ICU Outcome* alcohol* NOT non AND liver AND failure High AND dependency alcohol* NOT non AND hepatic AND failure Table 1 Search terms applied to Medline and Embase. Note* is a wildcard search strategy using unlimited truncation. eg alcohol* includes alcohol, alcoholic, alcoholics, 130 Volume 13, Number 2, April 2012 JICS

2 Inclusion criteria Adult patients only. The study should be based on patients with a diagnosis of alcoholic liver disease (ALD). In studies with a patient cohort of mixed aetiology liver disease, the largest group should be ALD. Patients should have been admitted to intensive care. Primary or main outcome measures should include mortality. Studies examining outcomes following liver transplantation or specialist intervention such as MARS were excluded. The study should be published in English. Table 2 Inclusion criteria applied to abstracts/full text articles. Total hits abstracts reviewed 51 abstracts 34 full text articles 21 articles critically appraised 110 rejected 38 transplant related 54 unrelated to ALD 15 mortality not outcome measured 2 non English 1 excluded ICU patients 17 duplicates removed 13 rejected 4 unrelated to ALD 5 non ICU patients 3 review articles 1? duplicate data set Questions 1. Was the defined representative sample of patients assembled at a common (usually early) point in the course of their disease? 2. Was patient follow-up sufficiently long and complete? 3. Were outcome criteria either objective or applied in a blind fashion? 4. If subgroups with different prognoses are identified, did adjustment for important prognostic factors take place? 5. How likely are the outcomes over time? 6. How precise are the prognostic estimates? 7. Can I apply this valid, important evidence about prognosis to my patient? Table 3 Centre for Evidence-Based Medicine s critical appraisal tool questions. question: What is the mortality of patients with alcoholic liver disease admitted to the intensive care unit? (Table 1). In addition to searching Medline and Embase, studies were obtained from the System for Information on Grey Literature in Europe (SIGLE) database, Google Scholar, reference lists of previously retrieved articles and personal communication with experts in the field. Abstracts were selected on the basis of six inclusion criteria (Table 2). These criteria ensured that the selected studies remained relevant to the review s title. Abstracts that met the inclusion criteria were obtained in full text. Full text articles were critically appraised using the Centre for Evidence-Based Medicine s (CEBM) scoring tool. The CEBM tool was one specifically designed to appraise prognostic studies. It posed seven questions (Table 3) concerning the methodological rigor with which the study had been conducted and the applicability of the results to everyday practice. The full CEBM tool can be accessed at Results From a total of 273 hits, 161 potentially relevant papers were obtained in abstract form. The full text article was sourced in 34 cases where the abstract suggested the study was likely to meet the set inclusion criteria. Following full text review, 13 studies were rejected for not fulfilling the inclusion criteria, Figure 1 Processing of search results. leaving a total of 21 articles to be critically appraised (Figure 1). Eight of these articles were published only in abstract or poster-presentation form. 7,9,14,15,17,19,20,22 The 21 studies contained data on a combined cohort of 20,476 patients. Four studies fulfilled all seven domains of the critical appraisal tool, while eight studies had areas of insufficient evidence and nine were deemed to have major flaws (Table 4). The majority of studies did not detail the grounds on which patients had acquired a diagnosis of alcohol-related cirrhosis or alcoholic liver disease. Seven studies 8,10,11,13,18,25,26 set out diagnostic inclusion criteria, requiring that enrolled patients had a diagnosis of ALD based on a combination of histology, imaging and clinical evidence of portal hypertension. Nineteen studies provided mortality data for patients with ALD admitted to intensive care. In four studies 19,20,21,25 it was not clear whether reported data referred to unit, hospital or alternative time-point mortality. The majority of studies recruited patients with a mixed aetiology of liver disease (with ALD as the largest group) but five studies 4,14,16,18,23 provided data on ALD patients exclusively. Four studies 12,16,21,25 examined the mortality of patients according to presenting pathology and six studies 12,16,18,19,20,25 assessed mortality according to treatment intervention. Six studies examined the relationship between mortality and number of organ system failures 10,11,16,19,23,24 while a total of nine articles analysed the ability of different scoring systems to discriminate between survivors and non-survivors. Discussion Mortality The primary objective of the review was to collate data estimating the mortality of patients with ALD who required intensive care. The majority of studies measured mortality at a minimum of two time points, ranging from unit mortality to five-year mortality. Unit mortality was the most frequently used outcome measure and rates ranged from 34 to 63%. The wide range of unit mortality probably arises as a result of studies recruiting patient cohorts with differing illness severity. Das 11 reported unit mortality to be 61%, but this study excluded ALD patients with gastrointestinal bleeding or encephalopathy, which as discussed below, have an arguably better prognosis. Mackle s 16 JICS Volume 13, Number 2, April

3 Author Year Level of n Patient population CEBM tool question* Limitations (citation) evidence* Abdel (7) c 100 Cirrhotic patients requiring mechanical Abstract only ventilation. Excluded elective surgery. Aggarwal (8) c 480 Consecutive patients admitted to a medical Non-UK patient ICU over a four-year period with cirrhosis. population Christensen (9) b 279 From 12,097 ICU admissions, 279 patients Abstract only had ALD. Cholongitis (10) c 312 Cirrhotic patients admitted to ICU, 65% Single centre study had alcohol as aetiology. Transplant patients excluded. Das (11) c 138 Patients with cirrhosis requiring medical Patients not requiring ICU in a regional liver transplant centre. extra-hepatic organ support were excluded Du Cheyron (12) b 186 Cirrhotic patients admitted to ICU; Single centre study majority had alcohol as aetiology. Gildea (13) c 422 Consecutive patients admitted to a medical Non UK centre ICU over a four-year period with diagnosis of cirrhosis. Transplant patients excluded Goutcher (14) c 63 All patients with ALD admitted to ICU Abstract only over a five-year period Mackay (15) c 500 Prospective case note review of Abstract only 500 admissions to ICU. Mackle (16) c 107 Patients with decompensated ALD admitted to general ICU. McPhail (17) c 486 Consecutive patients admitted to ICU Poster presentation with cirrhosis confirmed on histological, radiological or clinical criteria. Rabe (18) c 76 Paitents with liver cirrhosis requiring Only included intubation and ventilation over a patients requiring 10-year period. IPPV Rye (19) c 79 Retrospective review of patients with Abstract only alcoholic cirrhosis admitted to general ICU over six-year period. Saliba (20) c 308 Consecutive patients with decompensated Abstract only liver disease admitted to ICU over a three-year period. Singh (21) c 40 Patients with cirrhosis who required Specialist admission to ICU while awaiting transplant centre liver transplantation. Thomson (22) c 118 Data collected prospectively on patients Abstract only with cirrhosis admitted to general ICU. Wehler (23) c 143 Consecutive admissions of patients Non UK centre with hepatic cirrhosis who required ICU over a two-year period. Welch (4) c 4219 Secondary anaylsis of the Case Mix Program database (ICNARC) Multi centre, UK-based study. Zauner (24) b 208 Patients admitted to a general ICU with Non UK centre ALD Zauner (25) c 198 Patients with cirrhosis admitted to general Some data >30 medical ICU years old Zauner (26) c 196 Consecutive admissions of patients with liver Non UK centre cirrhosis to general ICU over a 10-year period Table 4 Summary of included studies. * see Oxford Centre for Evidence Based Medicine at Volume 13, Number 2, April 2012 JICS

4 study, which scored highly on critical appraisal, estimated overall unit mortality at 56%. This patient cohort was recruited from a university hospital with a regional tertiary referral centre for hepato-biliary disease and the Scottish Liver Transplant Unit. It is likely that patients were more severely ill than those managed in non-transplant centres. Comparing the average admission APACHE II scores from the studies supports this. Mackle s 16 cohort (unit mortality 56%) had a median APACHE II score of 25 (20-31) while Thomson s 22 patients (unit mortality 38%) scored a median APACHE II of 16 (13-22). Excluding studies that had methodological flaws on critical appraisal might make a more focussed estimate of mortality. Collating the four studies 8,13,22,23 that scored highest on critical appraisal, fulfilling all seven domains of the appraisal tool, unit mortality is estimated to be between 36 and 44%. Alternatively, data from studies that exclusively recruited ALD patients may be pooled. From the five studies that provide data on ALD patients exclusively, 50% of these patients died in intensive care. The predominant contribution to this data came from Welch et al, 4 who collated the ICNARC data for all admissions with ALD to UK intensive care units over a 10-year period from They document a slight reduction in mortality from 50% to 43% over the 10-year study period. Their data is based on a large cohort of patients (4,219) across a representative sample of UK units. It therefore seems reasonable to conclude that the unit mortality of patients with ALD admitted to intensive care is between 40 50%. This is higher than the 27% national average unit mortality for critically ill medical patients for the same time period. 27 Presenting features Patients with ALD commonly require intensive care for combinations of sepsis, hepatic encephalopathy, gastrointestinal bleeding, respiratory failure, spontaneous bacterial peritonitis or acute kidney failure. Four studies 12,16,21,25 estimated the mortality of patients admitted to ICU according to their presenting complication. ALD patients admitted to intensive care with sepsis reportedly have a mortality of 67-88%. The evidence base for this conclusion is very poor. Neither of the two studies involved clearly defined sepsis. Singh s 21 data is based on only three patients labelled as sepsis/hypotension and it is not clear whether data refers to unit or hospital mortality. Mackle 16 states that hospital mortality in the sepsis/multiorgan failure group was 88% but no further details of inclusion criteria are given in the text. A single study 16 provides mortality data for patients presenting with encephalopathy; Mackle estimates unit, hospital, 6-month and 12-months mortality at 14%, 33%, 60% and 80% respectively. This data suggest that, in the short term at least, patients presenting only with encephalopathy do relatively well and survive to hospital discharge. Their medium-to-long term prognosis appears more guarded, with a 12-month survival of only 20%. However, the data is based on only seven patients, and the authors provide no indication of encephalopathy severity, stating that the encephalopathy group were patients who presented with decompensated ALD in the absence of demonstrable sepsis or gastrointestinal haemorrhage. The Mackle 16 study is the only one to provide data on patients admitted with gastrointestinal haemorrhage, quoting a mortality of: 48%, 62%, 67% and 68% for unit, hospital, 6-month and 12-month (n=60). This suggests that gastrointestinal haemorrhage requiring intensive care has a significant initial risk of mortality (almost 50%) but that most patients surviving to hospital discharge are still alive at one year. No data was found on the outcome of patients admitted to intensive care with spontaneous bacterial peritonitis. Two studies have examined the prognosis of patients admitted with acute renal failure (ARF). Mackle defines ARF as a serum creatinine >120 μmol/l, quoting a unit mortality of 75% and a hospital mortality of 87% (n=60). Du Cheyron s 12 paper specifically addresses the attributable mortality of ARF in cirrhotic patients admitted to intensive care. He quotes a mortality of 46% (unit) and 51% (hospital) for mild and 78% (unit) and 84% (hospital) for severe renal failure (n=73). De Cheyron s definitions of mild and severe are as defined by the Acute Dialysis Quality Initiative group RIFLE classification 28 with mild equating to risk or injury and severe meaning failure. It is difficult to combine Mackle s and Du Cheyron s data sets based on differing definitions of ARF. The RIFLE definition is arguably more clinically relevant than an arbitrary creatinine level, which does not take into account the size/build of the patient or their baseline GFR. It is reasonable to speculate that a creatinine >120 μmol/l equates to Du Cheyron s mild ARF group and that the differing mortality between the two studies relates to admission illness severity (median APACHE II scores: vs ). There is some evidence 29 that the aetiology of acute kidney injury (sepsis, hypovolaemia, hepatorenal syndrome) in the cirrhotic patient influences outcome. This study is not formally included in this review as it did not recruit patients from critical care exclusively. Organ support Three papers provide mortality data according to organ system support. Patients may be categorised as requiring intermittent positive pressure ventilation (IPPV), vasoactive drugs and/or renal replacement therapy (RRT). Mackle 16 reports mortality figures for all three groups while Saliba 20 and Rye 19 discuss vasoactive infusions and RRT respectively. The Mackle data suggest that in general the outcome of patients with ALD who require invasive ventilation is poor, with 72% of all ALD patients on IPPV dying before hospital discharge. The authors also examined the subgroup of patients who required IPPV as a single system support and report a much lower mortality rate. Only 4% of patients requiring IPPV alone died in ICU and 69% survived to hospital discharge. The requirement for any additional organ support places patients in a much higher risk group (IPPV + vasoactive drugs: hospital mortality 80%, IPPV + RRT: hospital mortality 83%). 16 The outcome of patients with ALD requiring vasoactive drugs by continuous infusion is generally poor, with a unit mortality of 74-81% and a hospital mortality of 86%. 16,20 Studies did not clearly differentiate between inotropes and vasopressors or comment on the outcome of patients receiving only terlipressin. The need for renal replacement therapy is associated with the greatest risk of death, with Mackle and Saliba quoting a hospital mortality JICS Volume 13, Number 2, April

5 approaching 90%. Rye 19 found 100% mortality among patients requiring RRT. None of the three studies detailed the indications for RRT. Number of organ system failures The relationship between patient outcome and their number of organ system failures is relatively consistent between studies. A total of six papers 10,11,16,19,23,24 contributed to the data, three of which scored highly on critical appraisal. 16,19,23 Mortality rates ranged from 33-45% for single organ system failure, 65-75% for two organ system failures and % for three system failures. The data clearly suggest that a patient s chance of recovery decreases significantly with each additional system failure and that three system failures is associated with a very poor prognosis. Das found that with the exception of haematological failure, all other organ system failures including cardiovascular, respiratory, renal, hepatic and neurological failure were significant risk factors for in-hospital mortality. 11 Prognostic value of scoring systems The Child and Turcotte classification was first published in 1964 before being modified by Pugh in 1973 as the Child- (Turcotte)-Pugh Score (CPS). It was developed to assess the risk of death in patients with chronic liver disease undergoing porto-systemic shunting or oesophageal transection procedures. The modified score consists of five components: bilirubin, albumin, prothrombin time, severity of ascites, and encephalopathy. The Model for End-stage Liver Disease (MELD) is a more recently developed score used to predict short-term survival in cirrhotic patients and prioritise transplant lists. The Acute Physiology And Chronic Health Evaluation (APACHE) score and Sequential Organ Failure Assessment (SOFA) tools are based on acute physiological parameters and are familiar to those working in intensive care. Nine studies compared the prognostic value of liver-specific and physiology-based scoring systems in patients with ALD admitted to intensive care. The size of the study cohorts ranged from 76 to 486, with a total of 1,742 patients across all nine studies. Mean ages ranged from years. The commonest primary reason for admission was gastrointestinal haemorrhage, followed by hepatic encephalopathy. Other common reasons included sepsis and ARF. The ability of prognostic models to differentiate between survivors and nonsurvivors was tested in all nine studies by examining the area under the curve (AUC) of Receiver Operator Characteristic (ROC) curves. Seven out of the nine papers compared liverspecific scores against acute physiological scores and six found the physiology-based systems more discriminating. SOFA performed the best, with the majority of studies calculating an AUC >0.90. When Das compared modified (to exclude haematological failure) SOFA scores taken on days 1 and 3, he found the latter significantly more discriminating. With the exception of Rabe 18 who rated CPS above APACHE II, the remaining eight studies rated CPS the weakest discriminator, with AUC ranging from 0.61 to Many patients requiring admission to intensive care are CPS grade C and this may explain its inability to discriminate between survivors and nonsurvivors. Most studies scored MELD and APACHE II as moderately discriminating, with MELD achieving an AUC of 0.81 in three quarters of studies and APACHE II ranging from 0.66 to Two studies generated novel scoring systems. 17,26 The ability of the acute physiology scores to discriminate between survivors and non-survivors better is likely to be a reflection of the fact that patients with ALD in critical care tend to die of multi-organ failure rather than isolated acute liver failure. Das found that the severity of underlying liver disease was not a predictor of hospital mortality in patients with ALD admitted to intensive care. 11 The liver-specific scores, which focus on markers of hepatic function such as INR and albumin, are unsurprisingly insensitive to distant organ failures. MELD probably prognosticates better than CPS as it includes creatinine and will therefore detect acute kidney injury. Limitations of the analysis This review systematically searched two medical literature databases, with a variety of search terms and without date limitations. It strived to obtain other relevant publications from grey literature sources, reference lists of previously obtained articles and unpublished data via personal communication. The main limitations are that searching was restricted to the English language only, and that articles were selected and critically appraised by a single investigator (SF). Eight of the final twenty-one studies accepted for inclusion were published in abstract or poster form only. The inclusion criteria for the review included a requirement that alcohol was the single or dominant liver disease aetiology in the study population. This may have led to the exclusion of some mixed aetiology studies that selectively reported mortality data on ALD patients as a minority group. Conclusion There are no randomised trials examining the role of intensive care or invasive organ support in patients with ALD. The evidence for outcomes in this population of patients comes from a relatively small number of largely retrospective observational studies. The average unit mortality of patients with ALD admitted to intensive care is higher than the average non-surgical intensive care population, at 40-50%. Some subgroups of ALD patients have a better prognosis than others and mortality varies with presenting complication, treatment intervention required and number of organ system failures. Patients admitted with encephalopathy or those requiring single organ respiratory support have a unit mortality lower than the general non-surgical ICU population. Fewer than a third of patients presenting with acute kidney injury or requiring vasoactive infusions survive intensive care. Patients requiring renal replacement therapy have an over-80% average mortality, while those with three or more system failures consistently have a mortality rate greater than 90%. Acute physiology based scoring systems more closely predict mortality than traditional liver-specific scores; three-quarters of studies found that the SOFA score was the optimal tool for discriminating between survivors and non-survivors. Patients admitted to intensive care with ALD have a wide spectrum of outcomes. We suggest that the data summarised in this review supports a pragmatic approach to this patient cohort. In deciding whether a patient is likely to benefit from 134 Volume 13, Number 2, April 2012 JICS

6 invasive organ support, clinicians should avoid being unduly influenced by the severity of the underlying liver disease but be guided by the number, severity and apparent reversibility of non-haematological organ system failures. A trial of therapy with sequential SOFA scoring over two or three days allows those patients that are going to benefit to do so, while providing clarity to clinicians (and families) in those cases where the outlook is bleak that ongoing invasive support is not beneficial and a palliative approach more appropriate. This publication is an abbreviated version of a dissertation which was submitted and accepted as a component of the UK DICM in November Conflict of interest None. References 1. Institute of Alcohol Studies. Impact of Alcohol on the NHS.2009; Available from: Accessed 17th June The Information Centre for Health and Social Care. Statistics on Alcohol: England. 2009; Available from: alcoholeng2009/final%20format%20draft%202009%20v7.pdf. Accessed 17th June Office for National Statistics. Alcohol related deaths in the UK. 2010; Available from: Accessed 17th June Welch C, Harrison D, Short A, Rowan K. The increasing burden of alcoholic liver disease on United Kingdom critical care units: secondary analysis of a high quality clinical database. J Health Serv Res Policy 2008; 13:Suppl 2: Goldfarb G, Nouel O, Poynard T, Rueff B. Efficiency of respiratory assistance in cirrhotic patients with liver failure. Crit Care Med 1983;9: Shellman RG, Fulkerson W, DeLong E. Piantadosi CA. Prognosis of patients with cirrhosis and chronic liver disease admitted to the medical intensive care unit. Crit Care Med 1988;16: Abdelrazek W, Francoz C, Moreau R et al. Prognosis of critically ill cirrhotic patients undergoing mechanical ventilation: Implications for the management of acute-on-chronic liver failure. Liver Transplantation 2009;15(S7): Aggarwal A. Predictors of mortality and resource utilization in cirrhotic patients admitted to the medical ICU. Chest 2001;119: Christensen S, Johansen M, Jensen R et al. The impact of liver failure on 30-day mortality of alcoholic patients in intensive care. Acta Anaesthesiol Scand 2009;53: Cholongitas E, Senzolo M, Patch D et al. Risk factors, sequential organ failure assessment and model for end-stage liver disease scores for predicting short term mortality in cirrhotic patients admitted to intensive care unit. Aliment Pharmacol Ther ;23: Das V, Boelle P, Galbois A et al. Cirrhotic patients in the medical intensive care unit: early prognosis and long-term survival. Crit Care Med 2010;38: Du Cheyron D, Bouchet B, Parienti JJ et al. The attributable mortality of acute renal failure in critically ill patients with liver cirrhosis. Intensive Care Med 2005 ;31: Gildea TR, Cook W, Nelson DR, Aggarwal et al. Predictors of long-term mortality in patients with cirrhosis of the liver admitted to a medical ICU. Chest 2004;126: Goutcher C, Edwards J, Forrest E, Donaldson L. Mortality in patients with alcoholic liver disease admitted to intensive care: assessment of a new scoring system: A-734. European J Anaesthesiol 2006;23: Mackay A, Docking R, Kinsella J, Booth M. Co-morbidity, demographic and social factors as predictors of outcome from intensive care: An observational cohort study. Crit Care 2010;14 (Suppl 1):P Mackle IJ, Swann DG, Cook B. One year outcome of intensive care patients with decompensated alcoholic liver disease. Br J Anaesth 2006;97: McPhail MJ, Wendon J, Shawcross DL et al. Utility of organ failure prognostic scoring systems in a large cohort of critically ill patientswith cirrhosis: improved prediction of in-hospital mortality over MELD. Poster Presentation at the 46th annual meeting of the European association for the study of the liver Available from Accessed 29th December Rabe C, Schmitz V, Paashaus M et al. Does intubation really equal death in cirrhotic patients? Factors influencing outcome in patients with liver cirrhosis requiring mechanical ventilation. Intensive Care Med 2004;30: Rye K, Krishnamoorthy R, Skene H et al. Short-term prognosis and outcome predictors of alcohol-related cirrhosis admitted to the intensive care unit. Hepatol 2009;50:427A. 20.Saliba F, Levesque E, Ichai P et al. Outcome of cirrhotic patients admitted to the liver ICU and predictive factors of mortality: Results of a recent cohort of 308 patients. J Hepatol 2009;50 (Suppl 1): Singh N, Gayowski T, Wagener MM, Marino IR. Outcome of patients with cirrhosis requiring intensive care unit support: Prospective assessment of predictors of mortality. J Gastroenterol 1998;33: Thomson SJ, Moran C, Cowan ML et al. A study of patients with cirrhosis admitted to nontransplant general intensive care in the UK: Prevalence, case mix, outcomes and evaluation of critical illness and disease-specific scoring systems. Crit Care 2010;14 (Suppl 1):P Wehler M, Kokoska J, Reulbach U. Short-term prognosis in critically ill patients with cirrhosis assessed by prognostic scoring systems. Hepatology 2001;34: Zauner C, Schneeweiss B, Kranz A et al. Heavy chronic alcohol abuse has no additional adverse effect on the function of extrahepatic organs and ICU mortality in patients with liver cirrhosis. Wiener Klinische Wochenschrift 1999;111: Zauner CA, Kranz A, Kramer L et al. Outcome prediction for patients with cirrhosis of the liver in a medical ICU: a comparison of the APACHE scores and liver specific scoring systems. Intensive Care Med 1996;22: Zauner C, Schneeweiss B, Schneider B et al. Short term prognosis in critically ill patients with liver cirrhosis: an evaluation of a new scoring system. Euro J Gastroenterol Hepatol 2000;12: Intensive Care National Audit Centre. Case Mix Programme Data Available from: Summary%20statistics pdf Accessed 20th June Bellomo R, Kellum J, Ronco C. Defining and classifying acute renal failure: from advocacy to consensus and validation of the RIFLE criteria. Intensive Care Med 2007;33: Martín-LlahI M, Guevara M, Torre A et al. Prognostic importance of the cause of renal failure in patients with cirrhosis. Gastroenterol 2011;140: Simon Flood Specialty Registrar in Anaesthesia and ICM, Anaesthetic Department, St James s University Hospital, Leeds simonflood@doctors.org.uk Andrew Bodenham Consultant in Anaesthesia and Intensive Care Medicine, Leeds Teaching Hospitals NHS Trust Phil Jackson Consultant in Anaesthesia and Intensive Care Medicine, Leeds Teaching Hospitals NHS Trust JICS Volume 13, Number 2, April