Alimentary Pharmacology and Therapeutics SUMMARY

Transcription

1 Alimentary Pharmacology and Therapeutics The sequential organ failure assessment (SOFA) score is prognostically superior to the model for end-stage liver disease (MELD) and MELD variants following paracetamol (acetaminophen) overdose D. G. N. Craig, T. W. D. J. Reid, E. C. Wright, K. G. Martin, J. S. Davidson, P. C. Hayes & K. J. Simpson Scottish Liver Transplantation Unit, Royal Infirmary of Edinburgh, Little France, Edinburgh, UK. Division of Clinical and Surgical Sciences, University of Edinburgh, Edinburgh, UK. SUMMARY Background The prognostic value of the model for end-stage liver disease (MELD) and sodium-based MELD variants in predicting survival following paracetamol overdose remains unclear. Correspondence to: Dr K. J. Simpson, Scottish Liver Transplantation Unit, Royal Infirmary of Edinburgh, Little France, Edinburgh, EH16 4SA, UK. Publication data Submitted 5 November 211 First decision 24 November 211 Resubmitted 18 December 211 Accepted 2 January 212 EV Pub Online 2 January 212 Aim To examine the prognostic accuracy of sodium-based MELD variants in paracetamol-induced acute liver injury compared with the sequential organ failure assessment (SOFA) score. Methods Retrospective analysis of 138 single time point paracetamol overdoses admitted to a tertiary liver centre. Individual laboratory samples were correlated with the corresponding clinical parameters in relation to time post-overdose, and the daily MELD, MELD-Na, MELDNa, MESO, imeld, UKELD, updated MELD and SOFA scores were calculated. Results Sixty-six (47.8%) patients developed hepatic encephalopathy, of whom 7 were transplanted and 21 died without liver transplantation. SOFA had a significantly greater area under the receiver operator characteristic for the prediction of spontaneous survival compared with MELD at both 72 (P =.24) and 96 (P =.17) h post-overdose. None of the sodium-based MELD variants improved the prognostic accuracy of MELD. A SOFA score >6 by72hor>7 by 96 h, post-overdose predicted death/transplantation with a negative predictive value of 96.9 (95% CI ) and 98.8 (95% CI ) respectively. SOFA and MELD had similar accuracy for predicting the development of hepatic encephalopathy (P =.493). Conclusions The SOFA score is superior to MELD in predicting spontaneous survival following paracetamol-induced acute liver injury. Modification of the MELD score to include serum sodium does not improve prognostic accuracy in this setting. SOFA may have potential as a quantitative triage marker following paracetamol overdose. Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd 75 doi:1.1111/j x

2 D. G. N. Craig et al. INTRODUCTION Accurate prognostication following paracetamol (acetaminophen)-induced acute liver failure (ALF) is vital to effectively utilise critical care resources including emergency orthotopic liver transplantation (OLT). 1 The Kings College Hospital poor prognostic criteria (KCC), 2 modified in 26 by the addition of arterial lactate, 3 are used throughout the UK to determine those patients who will most likely die without OLT. The modified KCC have been criticised by a number of authors for their low negative predictive value, 4 and a plethora of alternative prognostic variables to the KCC have subsequently been proposed. 5 Several proposed markers, including the sequential organ failure assessment (SOFA) and systemic inflammatory response syndrome scores, utilise features of extrahepatic organ injury in an attempt to predict outcome following paracetamol hepatotoxicity. 6 8 The model for end-stage liver disease (MELD) score, which was initially developed to predict survival in patients undergoing a transhepatic portosystemic shunt, incorporates three laboratory indices: international normalised ratio (INR), serum bilirubin and serum creatinine. 9 MELD is now in widespread use as a donor allocation tool for chronic liver disease patients, but has also received much attention as a potential prognostic tool in ALF In particular, MELD appears to accurately predict survival in patients with non-paracetamolinduced ALF. 13 The prognostic accuracy of the MELD score in chronic liver disease patients awaiting OLT has been improved by the incorporation of serum sodium, namely in the MELD-sodium (MELD-Na and MELD- 14, 15 Na), United Kingdom model for End-stage Liver Disease (UKELD), 16 integrated MELD (imeld), 17 and the MELD to serum sodium ratio (MESO) scores. 18 Other models have updated the original MELD formula by assigning a higher weight to bilirubin and lower weight to creatinine and INR (UpMELD). 19 However, to our knowledge, these markers have not been assessed in the setting of paracetamol-induced acute liver injury. The aims of this study were to examine the temporal changes of MELD and the sodium-based MELD variants in a cohort of patients who required tertiary level care after taking a single intentional paracetamol overdose, and to compare the prognostic accuracy of these scores with SOFA and the KCC. To minimise temporal confounding effects, analysis was performed in relation to time from overdose, rather than in relation to the time from hospital admission, as many patients present to hospital following paracetamol overdose due to psychological, rather than physical morbidity. PATIENTS, METHODS, AND DEFINITIONS The cohort retrospectively analysed were 138 consecutive single time point intentional paracetamol overdoses admitted to the Scottish Liver Transplant Unit (SLTU) between April 23 and July 21. Single time point overdose was defined as a paracetamol overdose (>4g) taken at a single defined time point with the objective of self-harm. We specifically excluded patients with staggered overdoses, or overdoses taken accidentally in an attempt to relieve pain, due to the confounding temporal and diagnostic problems associated with these overdoses. 2 Ethical approval for the study was obtained from the Scotland A Research and Ethics Committee. Paracetamol-induced acute liver injury was defined as at least 2/3 of a history of ingestion of potentially toxic amounts of paracetamol (>4 g/day); detection of paracetamol in the serum >1 mg/l; or a serum alanine aminotransferase (ALT) level >1 IU/L within 7 days of a history of paracetamol ingestion irrespective of the serum paracetamol concentration. 21 All 138 patients fulfilled all of these three criteria. Severe acute liver injury was defined as sudden deterioration in liver function with associated coagulopathy in the absence of a history of chronic liver disease, whereas, the term ALF (i.e. fulminant liver failure) was restricted to those patients developing HE. 22 Guidelines for accepting patients from referring hospitals were based on previously published criteria from the British Society of Gastroenterology and remained unchanged over the time course of the study. 23 Patients admitted to the SLTU are managed using a standard protocol as previously described, 24 the main goals of which have remained similar over the duration of this cohort study. Decisions to list patients for OLT were made on the basis of the modified KCC after 26. For the purposes of this study, death and OLT were considered equivalent when undertaking survival analysis; post hoc reanalysis with transplanted patients excluded did not significantly change the results (data not shown). Laboratory parameters Retrospective analysis of all hospital electronic laboratory records was performed to obtain all standard haematological, biochemical and coagulation samples obtained for each patient during the first 7 days of admission. Each sampling time point was noted from either the clinical notes and/or the laboratory receipt, and the corresponding haemodynamic and clinical parameters for each sampling time point were retrieved from the clinical notes. These data were then individually compared against the reported time from overdose for each patient, 76 Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

3 SOFA and MELD scores following paracetamol overdose and the corresponding laboratory and clinical values were recorded in a dedicated database. SOFA score Organ failure was assessed using the SOFA score, which assesses six organ systems: hepatic, renal, coagulation, cardiovascular, respiratory and central nervous; and provides a graded score from to 4 points for each organ system, supplementary Table S1. 25 The SOFA score was calculated for each 24-h period following overdose, using the most deranged values for each parameter in each 24- h period. Due to potential confounding, SOFA scores were not calculated following the administration of platelet transfusions. MELD and MELD variants The seven separate MELD-based scores were obtained from the Bologna Liver Oncology Group online calculator ( 26 The MELD score was calculated from the formula by Kamath et al. 9 as follows: MELD ¼ 9:57 ln [creatinine (mg/dl)] þ 3:78 ln [bilirubin (mg/dl)] þ 11:2 lnðinrþ þ6:43 with the resulting score rounded to the nearest integer, and with higher values indicating more severe disease. The MELD variants were calculated from the following formulae: MELD-Na ¼ MELD þ 1:59 ½135 Na (mmol/l)š; with a Na range of12 135mmol/L MELDNa ¼ MELD - Na (mmol/l) f:25 MELD ½14 Na (mmol/l)š þ 14; with a Na range of125 14mmol/Lg MESO ¼ [MELD/Na (mmol/l)] 1 imeld ¼ MELD + [age (years) :3Š ½:7Na (mmol/l)] þ 1 Table 1 Demographics of study population and clinical outcomes Variable Dead/OLT (n = 28) Spontaneous survival (n = 11) P Sex (male/female) 1/18 (35.7%/64.3%) 51/59 (46.4%/53.6%).395 Age (years) 41 (31 47) 33 (22 43).115 Ingested paracetamol dose (g) 27.5 ( ) 25. ( ).226 Admission paracetamol level 52 (32 136) 73 (24 148).798 (lmol/l) Mixed overdose 11 (39.2%) 51 (46.3%).115 Time from overdose to NAC (h) 15.5 ( ) 23.5 ( ).791 Time from overdose to SLTU admission (h) 41. ( ) 53. ( ).45 Admission laboratory parameters (SLTU) Platelets (91 9 /L) 57 (34 14) 126 (74 187).27 Sodium (mmol/l) 135 ( ) 136 ( ).555 Creatinine, lmol/l 174 ( ) 91 (71 179).4 (mg/dl) 2. ( ) 1. (.8 2.) ALT (IU/L) 6697 ( ) 8292 ( ).148 Bilirubin, lmol/l 8 (69 17) 81 (45 11).475 (mg/dl) 4.4 ( ) 4.7 ( ) Albumin (g/l) 31 (23 35) 34 (31 37).15 PT (s) 76 (52 96) 5 (36 71).4 Developed encephalopathy 28 (1%) 36 (32.7%) <.1 Met King s College Criteria 24 (85.7%) 7 (6.4%) <.1 Renal replacement therapy 24 (85.7%) 18 (16.4%) <.1 Inotropic support 24 (85.7%) 9 (8.2%) <.1 Mechanical ventilation 24 (85.7%) 21 (19.1%) <.1 Transplanted 7 (25.%) (%) <.1 Data are presented as median (±interquartile range) or percentages as appropriate. NAC, N-acetyl cysteine; PT, prothrombin time; SLTU, Scottish Liver Transplant Unit. Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

4 D. G. N. Craig et al. UKELD ¼ 5 f1:5lnðinrþþ:3 ln½creatinineðlmol/l)šþ:6 lnðbilirubinðlmol/l)þ 13 ln (Na (mmol/l) þ 7g UpMELD ¼ 1:27 lnð1 þ creatinine (mg/dl)š þ :94 ln½1 þ bilirubin (mg/dl)š þ 1:66 lnð1 þ INRÞ To minimise confounding, no MELD or MELD-variant scores were calculated after the administration of fresh frozen plasma (FFP) or after the initiation of renal replacement therapy (RRT). Statistical analysis Statistical analysis was performed using MedCalc (Med- Calc Software, Mariakerke, Belgium), SPSS version 16. (SPSS, Chicago, IL, USA) and Graphpad Prism (Graph- Pad Software Inc., La Jolla, CA, USA). Data values are presented as median ± interquartile range (IQR) or percentages, unless otherwise stated. When undertaking survival analysis, death and OLT were considered equivalent. Receiver operating characteristic (ROC) analysis was used to identify optimum threshold values to discriminate non-survivors. The area under the curve (AUC) and standard error for each ROC and the differences between AUCs, were calculated according to the method of Delong et al. 27 Continuous data were compared using either analysis of variance or the Kruskal Wallis test if intergroup variances were unequal. Categorical data were analysed using Chi-squared tests or Fishers exact test. Bonferroni corrections were undertaken for repeated measures. Missing data were treated in a last observation carried forward manner. Results were considered statistically significant when two-sided P <.5. RESULTS Patients and details of overdose A total of 138 patients (61 male, 77 female) admitted to the SLTU between April 23 and February 21 were included in the study (Table 1). During this period, a total of 415 patients had been admitted, of whom 125 were classified as non-paracetamol cases, and 123 patients had taken a staggered paracetamol overdose or overdosed on paracetamol accidently and in the remaining 29 cases; the details of the paracetamol overdose remain unclear. The median patient age of the 138 included patients was 34 (IQR 24 43) years. A total of 18 patients were transferred to the SLTU from outlying health boards, at a median time of 5 (36 67) h following overdose. The remaining patients were admitted to the SLTU from wards within the Royal Infirmary of Edinburgh or from local hospitals. The median ingested paracetamol dose was 25 (16 42) g. All 138 patients received NAC treatment at their local hospital, at a median time from overdose of 23 (9.5 42) h. A total of 62 (45%) of patients had taken a mixed overdose. A total of 34 (24.6%) of paracetamol-induced acute liver injury patients were encephalopathic on admission to the SLTU, and a further 32/14 (3.8%) went on to develop HE during admission. A total of 66 (47.8%) patients therefore developed HE, and thus ALF, at some point during their illness. Of these patients, 31 (46.9%) subsequently met the modified KCC, at a median time from overdose of 52. ( ) h from overdose. Of these 31 patients, seven were transplanted, 17 died without OLT and seven recovered spontaneously. The seven transplanted patients fulfilled the original KCC at the time of OLT. Four patients died without meeting the modified KCC. MELD and temporal relationship with overdose Time course analysis of the seven MELD-based scores (MELD, MELD-Na, MELDNa, MESO, imeld, UKELD and UpMELD) and the SOFA score was performed for each 24-h period following paracetamol overdose according to clinical outcome (spontaneous survival vs. death/ OLT), with the results shown in Figure 1. The MELD, MELDNa and SOFA scores significantly discriminated between survivors and nonsurvivors by 48 h, with outcome based upon the remaining MELD-based scores diverging significantly by 72 h post-overdose. ROC analysis of maximum SOFA and MELD-based scores We performed pairwise comparisons of the AUC for SOFA against these seven MELD-based parameters at 48, 72 and 96 h post-overdose (Table 2). SOFA outperformed MELD-Na, imeld and UKELD by 48 h postoverdose, and displayed significantly greater prognostic accuracy than all seven MELD parameters at both 72 and 96 h. None of the MELD variants outperformed MELD at any time point. The ROC analyses identified a SOFA score of >6 and >7, and a MELD score of >37 and >42, as the most discriminatory cut-off values at 72 and 96 h, respectively, and comparative analysis of these thresholds with the KCC at these time points is shown at Table 3. Analysis of the 66 patients with HE, and 78 Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

5 SOFA and MELD scores following paracetamol overdose (a) SOFA SOFA score 5 N (b) MELD N MELD score (c) 6 MELD-Na score (e) 5 MESO score (g) 8 UKELD score MELD-Na MESO UKELD N OD N (d) MELDNa Updated MELD score imeld score MELDNa score (f) imeld (h) UpMELD N N N N Figure 1 Sequential SOFA (panel A), MELD (panel B) and sodium-based MELD (panels C H) scores (mean and 95% CI) in patients who survived (hashed line, n = 11) or died/underwent emergency orthotopic liver transplantation (solid line, n = 28) relative to the time of paracetamol overdose. P <.5. OLT, orthotopic liver transplantation; SOFA, sequential organ failure assessment; MELD, model for end-stage liver disease; MELD-Na, MELD-sodium; MELDNa, MELD-sodium; UKELD, United Kingdom End-stage Liver Disease; MESO, MELD to serum sodium ratio; imeld, integrated MELD; UpMELD, Updated MELD. Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

6 D. G. N. Craig et al. Table 2 Pairwise comparison of area under the receiver operator characteristic (AUC) for prediction of death/olt using maximum SOFA and MELD-based scores obtained by 48, 72 and 96 h following overdose Prognostic test SOFA MELD MELD-Na MELDNa MESO imeld UKELD UpMELD 48 h AUC SE % CI (vs. SOFA) (vs. MELD) 72 h AUC SE % CI (vs. SOFA) (vs. MELD) 96 h AUC SE % CI <.1.4 (vs. SOFA) (vs. MELD) therefore ALF, revealed a trend towards greater AUCs for SOFA compared with MELD at both 72 (SOFA.84 (95% CI.672.9), MELD.697 (95% CI ), P =.134) and 96 (SOFA.795 (95% CI ), MELD.698 (95% CI ), P =.23) h post-overdose. Analysis of the 42 patients who developed grade III/IV HE during their illness revealed similar AUCs for SOFA compared with MELD at both 72 (SOFA.733 (95% CI ), MELD.611 (95% CI ), P =.264) and 96 (SOFA.674 (95% CI ), MELD.631 (95% CI ), P =.696) h post-overdose. SOFA and MELD scores as predictors of the development of FHF Of the 66 (47.8%) patients who developed HE, and therefore ALF, a total of 34 (51.5%) had developed HE prior to SLTU admission. Analysis of the remaining non-encephalopathic patients was performed to identify whether SOFA or MELD could predict the subsequent development of HE. At 72 h following overdose, the AUCs for HE development were.814 (95% CI ) for SOFA and.781 (95% CI ) for MELD (P =.493). The AUC of the two scores for the development of grade III/IV HE were also similar [SOFA,.786 (95% CI ); MELD.749 (95% CI ), P =.582]. DISCUSSION This detailed retrospective temporal analysis demonstrates that the SOFA score is superior to MELD in predicting spontaneous survival following single time point paracetamol overdose. Modification of MELD by the addition of serum sodium does not improve the predictive accuracy of MELD in the setting of paracetamolinduced acute liver injury. Compared with the KCC, both SOFA and MELD have high negative predictive values, but lack specificity, suggesting that neither MELD nor SOFA should be used as definitive transplant listing criteria. Instead, future studies should prospectively explore the value of SOFA as a gatekeeper for access to tertiary care following paracetamol overdose, and exam- 71 Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

7 SOFA and MELD scores following paracetamol overdose Table 3 Predictive accuracy of the maximum MELD and SOFA scores obtained during the first 72 and 96 h following single time point paracetamol overdose Time postoverdose (h) Prognostic marker Threshold score N/ deaths >Threshold/ deaths Sensitivity (95% CI) Specificity (95% CI) Negative predictive value (95% CI) Positive predictive value (95% CI) 72 SOFA >6 111/22 47/2 9.9 ( ) MELD >37 66/22 1. ( ) Modified KCC +ve 23/ KCC ( ) 96 SOFA >7 138/28 57/ ( ) MELD >42 56/ ( ) Modified KCC +ve 3/ KCC ( ) 69.7 ( ) 5. ( ) 94.4 ( ) 72.7 ( ) 7.9 ( ) 93.6 ( ) 96.9 ( ) 1. ( ) 95.5 ( ) 98.8 ( ) 95.1 ( ) 94.5 ( ) 42.6 ( ) 33.8 ( ) 78.3 ( ) 47.4 (4. 49.) 42.9 ( ) 75.9 ( ) Optimal thresholds for predicting death or liver transplantation were derived from receiver operator characteristic analysis. The accuracy of the MELD and SOFA scores are compared with the modified King s College Hospital poor prognostic criteria (KCC), including patients without hepatic encephalopathy. ine its prognostic accuracy following staggered paracetamol overdose. This study is the first to quantitatively examine the prognostic accuracy of sodium-based MELD scores in the setting of paracetamol-induced acute liver injury. The strengths of the study include the analysis of these prognostic indicators in relation to time from overdose, rather than at the time of hospital admission or the onset of HE, both of which introduce considerable clinical heterogeneity and potential confounding. 28 Heterogeneity is further reduced by the single centre nature of this study, where both criteria for patient admission and clinical management protocols have remained unchanged during the time course of the study. However, we recognise that this study has several important limitations. Clearly, the results of this retrospective study are not directly applicable to patients who have taken a staggered overdose or in cases where accurate overdose timings are unavailable. We also recognise the possibility of selection bias, and that some patients with severe organ failure secondary to paracetamol overdose may not have been transferred to the SLTU. 29 An important caveat to these data is that not all the patients in this study developed HE, and therefore ALF. However, the mere absence of HE following severe acute liver injury does not preclude the development of other severe systemic complications and waiting until HE has developed before considering the potential need for OLT is unjustifiable. 3 Conversely, we also recognise that some patients had already developed HE by the time of admission, limiting the conclusions that can be drawn regarding the accuracy of SOFA or MELD in predicting the development of HE. In keeping with previous studies of MELD in paracetamol-induced liver injury, we have demonstrated that MELD has high sensitivity, but unacceptably low specificity, when applied as a marker of spontaneous survival. 5, 11 MELD has several important limitations as a prognostic marker in the setting of ALF; it is not widely utilised outside liver centres, is difficult to calculate at the bedside without handheld computer devices and is susceptible to artificial manipulation of its three components by treatment modalities such as RRT, artificial liver support devices and reversal of coagulopathy prior to invasive line insertion or intracranial pressure monitoring. 31 Another potentially confounding issue is the lack of international standardisation of laboratory variables, such as prothrombin time and creatinine, and changes to assay reagents over time. 32 The bilirubin component of MELD does not discriminate acute from chronic hepatic injury, and a recent study suggested that the prognostic accuracy of MELD in ALF could be improved by the replacement of bilirubin by total cytokeratin This study included only nine paracetamol cases; our recent study examining cytokeratin-18 and other cell death markers suggest that these immunoassays add little Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

8 D. G. N. Craig et al. prognostic benefit in patients with paracetamol hepatotoxicity. 34 Furthermore, such immunoassays are neither routinely available nor have been validated in clinical practice, and therefore are unlikely to gain widespread acceptance, unless a clear prognostic benefit can be demonstrated. An interesting issue is why sodium-based modifications to the MELD score fail to improve the prognostic accuracy of MELD in this setting. Hyponatraemia is a well-recognised adverse prognostic marker in patients with chronic liver disease, 14 and in ALF, and hyponatraemia is thought to worsen cerebral oedema, in itself an adverse prognostic marker. One explanation for this apparent discrepancy may lie in the use of prolonged infusions of NAC, leading to iatrogenic hyponatraemia, and therefore erroneously increased MELDs. The greatest benefit of the MELD adjustments is in identifying those patients with low MELDs on OLT waiting lists who are at unexpectedly high risk, 15 and these benefits may be outweighed by the very high MELD scores seen during ALF. It also remains unclear as to why the incorporation of patient age, an independent risk factor in ALF, into the imeld equation does not improve prognostic accuracy, but again this may reflect the simultaneous incorporation of serum sodium into the imeld equation. This study further supports the use of the SOFA score as a triage marker in the context of acute liver injury. SOFA is a dynamic test, which can be rapidly recalculated throughout admission, and is widely used in intensive care units. It incorporates measures of respiratory, cardiovascular and neurological failure not included in the MELD calculation, all of which are independently predictive of outcome. Furthermore, SOFA is an ordinal rather than dichotomous variable, and therefore a rising SOFA score could help to identify deteriorating patients at risk of death at an earlier stage and expedite transfer to liver centres. Despite these benefits, SOFA has several limitations in its use due to the difficulties in calculating the neurological component in intubated patients and the inability of the SOFA score to differentiate acute 35, 36 from chronic hepatic dysfunction. Although the SOFA score was not originally developed to predict outcome, its high sensitivity makes it attractive as a means of identifying all those high-risk paracetamol overdose patients who might die without transplant, although its relatively lower specificity suggests that SOFA should not replace the KCC as definitive listing criteria. In conclusion, this retrospective analysis of 138 single time point paracetamol overdoses has demonstrated that sodium-based modification of the MELD score is unhelpful in predicting spontaneous survival. The SOFA score is superior to MELD at predicting spontaneous survival, and equally accurate at predicting future HE, suggesting that SOFA, rather than MELD, should be prospectively evaluated as a triage marker in paracetamol-induced acute liver injury. ACKNOWLEDGEMENT Declaration of personal and funding interests: None. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Table S1. The sequential organ failure assessment (SOFA) score. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. REFERENCES 1. Craig DG, Lee A, Hayes PC, Simpson KJ. Review article: the current management of acute liver failure. Aliment Pharmacol Ther 21; 31: O Grady JG, Alexander GJ, Hayllar KM, Williams R. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 1989; 97: Bernal W, Donaldson N, Wyncoll D, Wendon J. Blood lactate as an early predictor of outcome in paracetamolinduced acute liver failure: a cohort study. Lancet 22; 359: Bailey B, Amre DK, Gaudreault P. Fulminant hepatic failure secondary to acetaminophen poisoning: a systematic review and meta-analysis of prognostic criteria determining the need for liver transplantation. Crit Care Med 23; 31: Craig DG, Ford AC, Hayes PC, Simpson KJ. Systematic review: prognostic tests of paracetamol-induced acute liver failure. Aliment Pharmacol Ther 21; 31: Schmidt LE, Larsen FS. Prognostic implications of hyperlactatemia, multiple organ failure, and systemic inflammatory response syndrome in patients with acetaminophen-induced acute liver failure. Crit Care Med 26; 34: Cholongitas EB, Betrossian A, Leandro G, Shaw S, Patch D, Burroughs AK. King s criteria, APACHE II, and SOFA scores in acute liver failure. Hepatology 26; 43: Craig DG, Reid TW, Martin KG, Davidson JS, Hayes PC, Simpson KJ. The systemic inflammatory response syndrome and sequential organ failure 712 Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd

9 SOFA and MELD scores following paracetamol overdose assessment scores are effective triage markers following paracetamol (acetaminophen) overdose. Aliment Pharmacol Ther 211; 34: Kamath PS, Wiesner RH, Malinchoc M, Kremers W, Therneau TM, Kosberg CL, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 21; 33: Schmidt LE, Larsen FS. MELD score as a predictor of liver failure and death in patients with acetaminophen-induced liver injury. Hepatology 27; 45: Zaman MB, Hoti E, Qasim A, Maguire D, McCormick PA, Hegarty JE, et al. MELD score as a prognostic model for listing acute liver failure patients for liver transplantation. Transplant Proc 26; 38: Yantorno SE, Kremers WK, Ruf AE, Trentadue JJ, Podestá LG, Villamil FG. MELD is superior to King s college and Clichy s criteria to assess prognosis in fulminant hepatic failure. Liver Transpl 27; 13: Kremers WK, Van Ijperen M, Kim WR, Freeman RB, Harper AM, Kamath PS, et al. MELD score as a predictor of pretransplant and posttransplant survival in OPTN/UNOS status 1 patients. Hepatology 24; 39: Biggins SW, Kim WR, Terrault NA, Saab S, Balan V, Schiano T, et al. Evidence-based incorporation of serum sodium concentration into MELD. Gastroenterology 26; 13: Kim WR, Biggins SW, Kremers WK, Wiesner RH, Kamath PS, Benson JT, et al. Hyponatremia and mortality among patients on the liver-transplant waiting list. N Engl J Med 28; 359: Barber KM, Pioli S, Blackwell JE, Collett D, Neuberger JM, Gimson AE. Development of a UK score for patients with end-stage liver disease. Hepatology 27; 46: 51A. 17. Luca A, Angermayr B, Bertolini G, Koenig F, Vizzini G, Ploner M, et al. An integrated MELD model including serum sodium and age improves the prediction of early mortality in patients with cirrhosis. Liver Transpl 27; 13: Huo TI, Wang YW, Yang YY, Lin HC, Lee PC, Hou MC, et al. Model for endstage liver disease score to serum sodium ratio index as a prognostic predictor and its correlation with portal pressure in patients with liver cirrhosis. Liver Int 27; 27: Sharma P, Schaubel DE, Sima CS, Merion RM, Lok AS. Re-weighting the model for end-stage liver disease score components. Gastroenterology 28; 135: Gyamlani GG, Parikh CR. Acetaminophen toxicity: suicidal vs. accidental. Crit Care 22; 6: Larson AM, Polson J, Fontana RJ, Davern TJ, Lalani E, Hynan LS, et al. Acetaminophen-induced acute liver failure: results of a United States multicenter, prospective study. Hepatology 25; 42: O Grady JG, Schalm SW, Williams R. Acute liver failure: redefining the syndromes. Lancet 1993; 342: Devlin J, O Grady J. Indications for referral and assessment in adult liver transplantation: a clinical guideline. British Society of Gastroenterology. Gut 1999; 45(Suppl. 6): Simpson KJ, Bates CM, Henderson NC, Wigmore SJ, Garden OJ, Lee A, et al. The utilization of liver transplantation in the management of acute liver failure: comparison between acetaminophen and non-acetaminophen etiologies. Liver Transpl 29; 15: Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996; 22: Biselli M, Gitto S, Gramenzi A, Di Donato R, Brodosi L, Ravaioli M, et al. Six score systems to evaluate candidates with advanced cirrhosis for orthotopic liver transplant: which is the winner? Liver Transpl 21; 16: DeLong ER, DeLong DM, Clarke- Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: O Grady JG. Prognostication in acute liver failure: a tool or an anchor? Liver Transpl 27; 13: Blair CS, Simpson KJ, Jones AL, Squires T, Masterton G, Gorman DR, et al. Deaths from paracetamol poisoning in Scotland, impact of the Scottish Liver Transplantation Unit. Gut 1998; 42 (Suppl. 1): Pakravan N, Simpson KJ, Waring WS, Bates CM, Bateman DN. Renal injury at first presentation as a predictor for poor outcome in severe paracetamol poisoning referred to a liver transplant unit. Eur J Clin Pharmacol 29; 65: Warrillow SJ. Predictions and outcomes for the critically ill patient with cirrhosis: is it time to settle on the SOFA and let jaundiced views on outcome MELD away? Crit Care Med 21; 38: Newsome PN, Henderson NC, Germain L, Ludlam CA, Simpson KJ. Prothrombin time to assess fulminant hepatic failure. Lancet 21; 358: Bechmann LP, Jochum C, Kocabayoglu P, Sowa JP, Kassalik M, Gieseler RK, et al. Cytokeratin 18-based modification of the MELD score improves prediction of spontaneous survival after acute liver injury. J Hepatol 21; 53: Craig DG, Lee P, Pryde EA, Masterton GS, Hayes PC, Simpson KJ. Circulating apoptotic and necrotic cell death markers in patients with acute liver injury. Liver Int 211; 31: Moreno R, Vincent JL, Matos R, et al. The use of maximum SOFA score to quantify organ dysfunction/failure in intensive care. Results of a prospective, multicentre study. Working Group on Sepsis related Problems of the ESICM. Intensive Care Med 1999; 25: Janssens U, Graf C, Graf J, et al. Evaluation of the SOFA score: a singlecenter experience of a medical intensive care unit in 33 consecutive patients with predominantly cardiovascular disorders. Sequential Organ Failure Assessment. Intensive Care Med 2; 26: Aliment Pharmacol Ther 212; 35: ª 212 Blackwell Publishing Ltd