Acute liver failure in Sweden: etiology and outcome

Transcription

1 Original Article doi: /j x : etiology and outcome G. Wei 1, A. Bergquist 2, U. Broomé 2, S. Lindgren 3, S. Wallerstedt 1, S. Almer 4, P. Sangfelt 5,Å. Danielsson 6, H. Sandberg-Gertzén 7,L.Lööf 8, H. Prytz 9 & E. Björnsson 1 1 From the Section of Gastroenterology and Hepatology, Department of Internal Medicine, Sahlgrenska University Hospital, Gothenburg; 2 Department of Gastroenterology and Hepatology, Karolinska University Hospital, Huddinge, Stockholm; 3 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital MAS, Malmö; 4 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital, Linköping; 5 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital, Uppsala; 6 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital, Umeå; 7 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital, Örebro; 8 Center for Clinical Research, Central Hospital, Västerås; and 9 Gastroenterology and Hepatology Division, Department of Medicine, University Hospital, Lund; Sweden Abstract. Wei G, Bergquist A, Broomé U, Lindgren S, Wallerstedt S, Almer S, Sangfelt P, Danielsson Å, Sandberg-Gertzén H, Lööf L, Prytz H, Björnsson E (Sahlgrenska University Hospital, Gothenburg; Karolinska University Hospital, Huddinge, Stockholm; University Hospital MAS, Malmö; University Hospital, Linköping; University Hospital, Uppsala; University Hospital, Umeå; University Hospital, Örebro; Central Hospital, Västerås; and University Hospital, Lund; Sweden). Acute liver failure in Sweden: etiology and outcome. J Intern Med 2007; 262: Objective. To determine the causes and outcome of all patients with acute liver failure (ALF) in Sweden and study the diagnostic accuracy of King s College Hospital (KCH) criteria and the model for end-stage liver disease (MELD) score with transplant-free deaths as a positive outcome. Research design and methods. Adult patients in Sweden with international normalized ratio (INR) of 1.5 due to severe liver injury with and without encephalopathy at admission between were included. Results. A total of 279 patients were identified. The most common cause of ALF were acetaminophen toxicity in 42% and other drugs in 15%. In 31 cases (11%) no definite etiology could be established. The KCH criteria had a positive-predictive value (PPV) of 67%, negative-predictive value (NPV) of 84% in the acetaminophen group. Positive-predictive value and negative-predictive value of KCH criteria in the nonacetaminophen group were 54% and 63% respectively. MELD score >30 had a positive-predictive value of 21%, negative-predictive value of 94% in the acetaminophen group. The corresponding figures for the nonacetaminophen group were 64% and 76% respectively. Conclusions. Acetaminophen toxicity was the most common cause in unselected patients with ALF in Sweden. KCH criteria had a high NPV in the acetaminophen group, and in combination with MELD score <30 predicts a good prognosis in acetaminophen patients without transplantation. Keywords: acute liver failure, etiology, outcome, King s College Hospital criteria, liver transplantation. Introduction Acute liver failure (ALF) is a complex medical condition defined as a rapid development of hepatic synthetic dysfunction associated with significant coagulopathy [1, 2]. The prognosis of patients with ALF is highly variable and has been shown to be dependent on the etiology, age of the patients, the course of the disease as well as the encephalopathy grade [1 3]. ALF is a heterogenous condition with a mortality rate without liver transplantation (LTX) in a range of 10 90% in different cohorts [1]. The ª 2007 Blackwell Publishing Ltd 393

2 etiology of ALF shows worldwide variation and prognosis depends greatly on the underlying cause, so it is an important to do studies in different cohorts. Because ALF is a rare disorder, the related clinical data are limited. A recently reported prospective study of the etiology and prognosis of patients with ALF in the United States (US) demonstrated that acetaminophen toxicity and idiosyncratic drug reactions were the most common causes of ALF [2]. Even higher prevalence of acetaminophen toxicity (72%) as the cause in patients with fulminant hepatic failure has been reported from the United Kingdom [4]. Ischemic hepatitis is being increasingly recognized as a cause of ALF, especially in elderly patients [5]. Information about the causes and outcome of patients with ALF in Scandinavia has been limited to patients listed for highly urgent liver transplantation (HULTX) [6]. Amongst the patients awaiting LTX for fulminant hepatic failure in Scandinavia a total of 43% of patients had indeterminate etiology [6]. The high proportion of patients with unknown etiology contrasts with the recent prospective US study with only 17% of cases [2]. In the current study, we aimed to determine the etiology of all patients fulfilling the criteria for ALF with and without encephalopathy at admission, not only those listed for HULTX. Thus, these patients represent the whole spectrum of patients with ALF. It is of major importance to identify those patients who will not survive the severe medical condition associated with ALF. The King s College Hospital (KCH) criteria for LTX in ALF are the most widely accepted and validated criteria used for prognostic assessment in these patients [7]. The model for endstage liver disease (MELD) score has been shown to be a good predictor of survival in patients with chronic liver disease but limited data exist in patients with ALF [8, 9]. In this study, we also wanted to explore the various conditions behind all cases of ALF in Sweden and to assess the potential of KCH criteria and MELD-score respectively to predict outcome in these patients. Research design and methods In this retrospective study, 279 patients 14 years of age or older during the 10 years period between 1994 and 2003, fulfilled the criteria for ALF in ten university hospitals in Sweden and were included in the analysis. LTX have been performed in Stockholm since 1984, in Gothenburg since 1985, and in Uppsala since The entry criteria for ALF were the presence of coagulopathy as defined by international normalized ratio (INR) 1.5 due to severe liver disease [2], without a previous underlying liver disorder. Hepatic encephalopathy was not a requirement for enrollment into the analysis, as information on the occurrence of the coma grade I and II can in many cases be difficult to retrieve from medical records. We made a search in computerized diagnoses hospital registry for the diagnosis of ALF, hepatitis and acetaminophen toxicity with the following diagnosis codes: Acute and subacute liver failure (K72.0), liver failure, unspecified (K72.9), toxic liver disease with liver necrosis (K71.1), toxic liver disease with acute hepatitis (K71.2), toxic liver disease with hepatitis(k71.6), toxic liver disease, unspecified (K71.9), acute hepatitis A (B15), acute hepatitis B (B16), acute hepatitis D,C,E (B17), viral hepatitis, unspecified (B19), intoxication with analgesic, non opium and antipyretic drugs (T39.8), intoxication with analgesic, non opium, antipyretic, and anti-rheumatic drugs (T39.9). Patients with acute decompensation of a chronic liver disease (acute-on-chronic) and patients with alcoholic hepatitis were excluded. The etiology of ALF was considered to be indeterminate if no diagnostic clues were obtained from the patient s history, imaging studies, serologic markers for viral hepatitis A, B, C, Cytomegalovirus and Ebstein-Barr virus, anti-smooth-muscle, antinuclear antibodies and ceruloplasmin (or other signs of Wilsons disease). In the Nordic organ allocation system, patients with ALF listed for HULTX are given a 72-h absolute priority for any ABO compatible donor organ [6]. The current study aimed to include all patients with ALF, both those who were not considered for transplantation and those listed for HULTX. A hepatologist from 394 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

3 each university hospital provided detailed demographic, clinical, laboratory and outcome data. The following variables were obtained from the medical records: age, gender, etiologic diagnosis, symptom duration before the presence of hepatic encephalopathy, coma grade at admission and maximum coma grade (I IV), ingestion of acetaminophen with or without suicidal intention and dose. INR, serum bilirubin, serum creatinine, serum alanine aminotransferase (ALT), serum aspartate aminotransferase (AST) on admission and maximum laboratory values were also recorded. The hepatologist from each university hospital determined the etiology and outcome analysis during the hospitalization.the KCH criteria [7] and MELD score [8, 9] are given in Table 1. Separate analyses were performed for acetaminophen and nonacetaminophen induced ALF. The static and dynamic variables assessed were age, period between onset of jaundice and encephalopathy, grade of Table 1 King s College Hospital criteria for the prognosis of nonsurvival amongst patients with acute liver failure. The model for end-stage liver disease score is based on the three biochemical parameters total serum bilirubin, INR and serum creatinine The KCH criteria: 1) For acetaminophen induced ALF: ph <7. 30 (irrespective of grade of encephalopathy) or INR >6.5 and serum creatinine >300 lmol L )1 in patients with grade III or IV encephalopathy 2) For ALF induced by etiologies other than acetaminophen INR >6.5; irrespective of grade of encephalopathy or any 3 of the following variables (irrespective of grade of encephalopathy) Age <10 or >40 years Cause: Non A Non B hepatitis, idiosyncratic drug reactions Duration of jaundice before onset of encephalopathy >7 days INR >3.5 Serum bilirubin >300 lmol L )1 MELD score: MELD ¼ 3.78 log e [bilirubin (mg dl )1 )] log e (INR) log e [creatinine (mg dl )1 )] encephalopathy, AST, ALT, serum bilirubin, serum creatinine, INR. Two values were analysed for each dynamic parameter the values obtained on admission (OA) and peak values (P) seen during the hospitalization. Mann Whitney test has been used for continuous variables and Fisher s Exact test for dichotomous variables for test between groups. All tests were two-tailed and conducted at 5% significance level. The variables having P-value <0.1 in univariate analyses were included as possible predictors in multivariate stepwise logistic regression models. We performed the analysis of sensitivity, specificity, positive-predictive value (PPV), negative-predictive value (NPV) and receiver operating characteristic (ROC)-curves as a comparison between spontaneous survivors and transplant-free deaths to assess the validity of the KCH criteria and MELD score >30. Transplant-free death was defined as a positive outcome. Separate analyses were performed in patients who died but who did not undergo LTX as the outcome for patients who underwent LTX could not be predicted with certainty. The study was approved by the ethical committee of the University of Gothenburg. Results During the 10 year period, , a total of 279 patients (61% women) fulfilled the criteria for ALF and were admitted to the different centres. The median age was 47 years, interquartile range (IQR; years). The different causes of ALF are demonstrated in fig. 1. Acetaminophen toxicity was the most common cause accounting for 118 out of 279 cases (42%), followed by idiosyncratic drug reactions whilst in 31 (11%) cases no cause could be determined (Fig. 1). The median dosage of acetaminophen ingested was 17 g d )1, (IQR g d )1 ). A total of 115 out of 118 patients (97%) had ingested more than 4 g d )1 which is the maximum package recommendation. There were 58 patients who were considered to have ingested acetaminophen with suicidal intention. In 31 cases the intoxication was unintentional, but in 29 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

4 Acetaminophen intoxication Other Drugs Indeterminate Ischemic Hepatitis Etiology Hepatitis B Autoimmune Hepatitis Hepatitis A Budd-Chiari Syndrome Cancer Mushroom poisoning Wilson's Disease Pregnancy Other causes Fig. 1 Etiologies of 279 patients with acute liver failure in Sweden during the 10 year s period, 1994 to Most common causes were acetaminophen toxicity accounting for 118 out of 279 cases (42%), followed by idiosyncratic drug reactions were the presumed cause of ALF in 42 cases (15%). Other drugs were Disulfiram (nine cases), Flucloxacillin (four cases), Diclofenac (three cases), Ibuprofen (two cases), and one case of the following drugs Phenoxymethylpenicilline, Enalapril, Hydroklortiazid, Simvastatin, Nitrofurantoin, Paroxetin, Moclobemid, Nefazodon, Ecstasy, Fluoxetin, Omeprazol, Amiodaron, Dextroproxiphen, Hydralazine, Cefadroxil, Ciprofloxacin, Amphetamine, Cocaine. In 31 cases (11%) the causes were unknown. cases the reason for overdose could not be determined by the medical records. A total of 95% of patients with acetaminopen toxicity were treated with acetylcystein. Idiosyncratic drug reactions were the presumed cause of ALF in 42 cases (15%). The most common drug amongst these patients was disulfiram in nine cases but a wide range of other drugs were also implicated (Fig. 1). The third most common known cause was ischemic hepatitis (shock liver) in 23 cases (8%), no patient had portal or arterial thrombosis. All these patients had underlying heart failure. Hepatitis B and hepatitis A were the causes in 11 cases (4%) and eight cases (3%), respectively (Fig. 1). The clinical and laboratory data of patients with ALF are shown in Table 2 according to the major etiology. In our cohort, 45% of patients were treated in the intensive care unit during the hospitalization. Patients with acetaminophen overdose had higher ALT and AST compared with other etiologies (Table 2). Figure 2 illustrates the outcome of the patients, showing that 170 out of 279 patients (61%) of the patients recovered spontaneously. Alltogether 52 patients were listed for HULTX and 39 patients received a liver graft. A total of 6 of 52 patients could be withdrawn from the urgent list because of improved condition and survived without transplantation, and three of them with acetaminopen toxicity. The median time from transplantation listing to actual transplantation were 2 days (IQR,1 to 4 days). Figure 3 shows the outcome according to the etiology amongst the major etiological groups of ALF. For the patients with acetaminophen toxicity, the survival without transplantation was 83%. Only 6 (5%) of these patients were transplanted. No patient with ischemic hepatitis was transplanted due to high age and/or bad general clinical condition. The mortality of patients with ischemic hepatitis was the highest (65%). For patients with indeterminate cause, the spontaneous survival was only 29% and LTX was performed in 52% of cases (Fig. 3). In the group caused by drugs other than acetaminophen, 21% of the patients were transplanted (Fig. 3). Table 3 shows the results of univariate analysis of static and dynamic variables in terms of difference between spontaneous survivors and transplant-free deaths as well as the patients who died and were transplanted put together. Stepwise logistic regression analysis in the acetaminophen group identified peak grade encephalopathy as the only independent predictor of mortality (Odds ratio ( ). In the group with ALF induced by etiologies other than acetaminophen, multivariate analysis identified age (Odds ratio (CI ) and peak serum creatinine (Odds ratio (CI ) as independent factors influencing outcome. Prognostic indicators used in KCH criteria were reevaluated in the current study in terms of their ability to predict a poor outcome (Table 4). Table 4 gives the 396 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

5 Table 2 The clinical and laboratory characteristics of patients with acute liver failure according to etiology All patients AM Other drugs Indeterminate Ischemic hepatitis n Women, n (%) 171 (61) 84 (71) 26 (62) 10 (32) 7 (30) Age, year 47 (14 91) 37 (14 78) 50 (21 91) 50 (14 86) 70 (21 90) No encephalopathy 39% 48% 36% 29% 17% Coma grade OA I/II 44% 30% 55% 52% 58% III/IV 17% 22% 10% 19% 25% Coma grade P I/II 30% 26% 38% 26% 46% III/IV 30% 25% 26% 45% 38% Symptom duration, day 7 (0 90) 1 (0 21) 7 (1 90) 7 (0 90) 7 (0 60) Intensive care, n (%) 125 (45%) 51 (45%) 17 (40%) 16 (52%) 11 (46%) Intensive care stay, days 3 (0 24) 3 (0 20) 3 (0 24) 3 (0 7) 4 (0 14) INR OA 1.9 ( ) 1.9 ( ) 1.8 ( ) 2.4 ( ) 2.0 ( ) INR P 2.7 ( ) 2.6 ( ) 2.6 ( ) 3.6 ( ) 2.4 ( ) Bilirubin OA, lmol L )1 71 ( ) 38 ( ) 179 (15 820) 241 ( ) 31 ( ) Bilirubin P, lmol L )1 154 ( ) 66 ( ) 333 (18 988) 521 (10 956) 56 (14 303) ALT OA, lkat L )1 17 ( ) 18 ( ) 16 ( ) 23 ( ) 15 (0.1 96) ALT P, lkat L )1 48 ( ) 100 ( ) 27 ( ) 28 ( ) 36 ( ) AST OA, lkat L )1 21 ( ) 24 ( ) 20 ( ) 21 ( ) 30 ( ) AST P, lkat L )1 45 ( ) 100 ( ) 27 (2 474) 25 ( ) 45 ( ) Creatinine OA, lmol L )1 95 (18 613) 87 (32 613) 102 (40 329) 75 (66 260) 173 (71 577) Creatinine P, lmol L )1 113 (37 979) 95 (42 963) 129 (46 623) 110 (61 979) 269 (71 625) ALT, alanine aminotransferase; AST, aspartate aminotransferase; INR, international normalized ratio; OA, on admission; P, peak; AM, acetaminophen overdose. Normal values, ALT (<0.7 lkat L )1 ), AST (<0.7 lkat L )1 ), bilirubin (<20 lmol L )1 ), creatinine (<100 lmol L )1 ), INR (<1.2). The results are shown as median and range. Fig. 2 The outcome for 279 patients with acute liver failure in Sweden during Four patients with acute liver failure but not listed for highly urgent liver transplantation were transplanted. sensitivity, specificity, PPV and NPV of the KCH criteria. KCH criteria had high NPV (84%) in acetaminophen patients. Regarding the acetaminophen cases, no patient fulfilled the alternate KCH criteria consisting of the combination of INR >6.5, serum creatinine >300 lmol L )1 and grade III or IV encephalopathy. ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

6 Table 4 Diagnostic accuracy of King s College Hospital (KCH) criteria and model for end-stage liver disease (MELD) score >30 in patients with acute liver failure in Sweden Specificity Sensitivity PPV NPV Predictive accuracy KCH criteria AM Non-AM MELD >30 AM Non-AM Fig. 3 The outcome according to etiology amongst the major etiological groups of acute liver failure. Figure 4 demonstrates the diagnostic performance of MELD >30 based on the sensitivity and specificity using receiver operating characteristic (ROC) analysis. MELD-score >30 was associated with a 64% PPV and 76% NPV in patients with other causes than acetaminophen toxicity. In acetaminophen patients MELD-score>30 the NPV was 94% (Table 4). PPV, positive-predictive value; NPV, negative-predictive value; AM, acetaminophen-induced acute liver failure; Non-AM, acute liver failure due to other etiologies than acetaminophen. The analysis was restricted to a comparison between spontaneous survivors and transplant-free deaths. We defined transplant-free deaths as a positive outcome and transplanted patients were excluded from the analysis. Discussion The current study is the first to characterize the presumed causes and outcome of patients with ALF in Sweden. An earlier study from the nordic countries, that included some patients from Sweden, was limited Table 3 The results of comparison of the values of static and dynamic parameters between transplant-free survivors, nonsurvivors without liver transplantation and LTX/death as assessed by Mann Whitney test AM Non-AM Spontaneous survivors Transplant-free deaths LTX/death Spontaneous survivors Transplant-free deaths LTX/death Age, year 37 (14 78) 50* (18 73) 43 (18 73) 50 (14 91) 68** (22 90) 66** (22 90) Duration of jaundice before the onset of encephalopathy, days 0 (0 14) 0.5 (0 7) 1 (0 7) 5 (0 42) 7 (0 34) 7 (0 34) Peak coma grade 1 (1 2) 4** (3 4) 4** (2 4) 1 (1 2) 3** (1 4) 3 (1 4) Bilirubin OA lmol/l 33 (3 568) 86 (15 335) 92 (15 335) 114 (4 447) 78 (8 820) 112 (8 820) Bilirubin P lmol/l 56 (3 930) 179** (53 442) 159 (53 442) 189 (10 628) 207 ( ) 283 ( ) Creatinine OA lmol/l 84 (32 419) 127 (48 613) 137* (48 400) 87 (18 538) 162** (40 577) 154* (40 577) Creatinine P lmol/l 91 (42 963) 173* (50 618) 164 (50 524) 94 (37 979) 264 ** (51 644) 262** (51 644) INR OA 1.8 ( ) 2.9* ( ) 3.1** ( ) 1.7 ( ) 2.0 ( ) 1.9 ( ) INR P 2.4 ( ) 3.8* ( ) 4.5 ( ) 2.1 ( ) 2.5* ( ) 2.8 ( ) The results are shown as median and range. AM, acetaminophen; Non-AM, other etiologies than acetaminophen; OA, on admission; P, peak; LTX, liver transplantation; INR, international normalized ratio. Normal values, Bilirubin (<20 l mol L )1 ), creatinine (<100 l mol L )1 ), INR (<1.2). *P-value < 0.05; **P-value < ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

7 Fig. 4 Receiver operating characteristics curves, for MELD score >30, showing the areas under the curves for acetaminophen (AM) group and the non-acetaminophen group (non-am). to patients listed for HULTX [6]. Our data does reflect the overall spectrum of patients with ALF and not only patients considered for transplantation. The presence of hepatic encephalopathy was not an absolute selection criteria in our study. The definition of ALF in the current study deviates somewhat from the one proposed by O Grady et al. [7]. There were several reasons for this. First of all it can be difficult to obtain reliable information about encephalopathy from medical records in a retrospective study. A total of 39% of the patients who were included in the current study did not have encephalopathy at the admission. However, encephalopathy grade I or II can be unrecognized and some patients might have developed mild encephalopathy not stated in the medical records whereas grade III or IV is usually obvious and information is usually present in the medical records. The development of encephalopathy is a late sign in the progression of ALF, and it is very important for the clinician to recognize this catastrophic illness before the onset of advanced grades of coma. Compared with the previously reported study on patients listed for HULTX the current study reports a smaller population of indeterminate etiology 11% vs. 43% [6]. Acetaminophen toxicity was found to be an uncommon reason for listing of patients in the Nordic countries for HULTX [6]. Acetaminophen toxicity was the cause of ALF in only 17% of these patients [6], whereas in unselected patients in the current study it was the cause in 42% of patients. The fact that only patients listed for HULTX were included in the previous study and in the present study patients with acetaminophen toxicity with milder liver failure were included making this a large group of patients might explain the major differences. It is conceivable that the major focus of the clinician taking care of a patient s life threatening illness, is to treat the condition and plan for transplantation, often with very tight time limits, rather than to rule out all possible causes. The proportion of our patients with an unknown cause (11%) is similar to what has been observed in both retrospective (15%) and prospective (17%) studies from the US [2, 10]. The patients with an unknown cause is a heterogenous group and some patients may have taken illegal narcotics, such as amphetamine, cocaine or ecstasy, which all have been associated with ALF [1]. Amongst patients with an unknown cause in the current study, only a few patients had undergone systematic toxicology screen in order to exclude these causes. Although acetaminophen toxicity is an uncommon cause of ALF in some parts of the world [1], it is according to our results the most common cause of ALF in Sweden, similar to experiences from Denmark, UK and the US [1]. Also in accordance with the experiences from the aforementioned prospective US study, acetaminophen toxicity was associated with the overall best transplant-free survival amongst the major etiological groups. The patients with acetaminophen toxicity had the best prognosis with 83% transplant free survival and only 5% of patients with ALF due to acetaminophen toxicity were transplanted. In the United Kingdom, most cases with acetaminophen toxicity were intentional self-poisoning whereas in the U.S., more than 50% of patients had unintentional acetaminophen toxicity [2, 11]. In the current study, 65% of patients with acetaminophen toxicity had suicidal intention amongst those in whom this could be determined from the medical records. The overall transplant-free survival of 61% in the current study is somewhat higher than previously ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

8 reports [2, 10, 11]. These results probably reflect the selection of patients, since encephalopathy was not a mandatory symptom and since we included some patients with mild to moderate progressive coagulopathy. The median waiting time between acceptance and transplantation was 2 days (IQR 1 to 4 days), similar to results recently reported [2, 6]. The short median time on waiting list for transplantation underlines the importance of a rapid diagnostic work up. Half of cases who could be withdrawn because of improved condition and ultimately survived were due to acetaminophen toxicity. Only 7 (13%) patients, listed for HULTX died on the waiting list, compared with 22% in the US [2]. The assesssment of prognosis in patients with ALF is of major importance in order to predict transplant-free survival. KCH criteria for prognosis in ALF reported by O Grady et al, in 1989 are still used clinically by many transplant centres [7, 12]. These criteria have been shown by other investigators to have a lower predictive accuracy in acetaminophen and non-acetaminophen patients than could be expected from the original study (85% and 94%) [7, 11 13]. However, when these KCH criteria were applied to our patients, we found a relatively low PPV in both acetaminophen and nonacetaminophen patients (67% and 54%) but high NPV in acetaminophen patients (84%). The low sensitivity in acetaminophen and nonacetaminophen patients (40% and 26%) indicates that other factors responsible for the patient s demise are not detected by KCH criteria. The reason for lower sensitivity observed in the current study is not clear. However, one factor that might explain the differences is the inclusion of a relatively large cohort of patients with ischemic hepatitis, which was not the case in the original KCH study [7]. In agreement with other studies analysing outcome in ALF, INR was significantly higher in the nonsurvivors versus those who survived in both the acetaminophen and the nonacetaminophen groups [7, 13]. However, only peak coma grade was independently associated with mortality in the acetminophen group in our study. Amongst the nonacetaminophen group, in line with the original report by O Grady et al. [7] age was found to be an independent predictor of bad outcome and peak serum creatinine was also an independent predictor of mortality in our patients. A potential draw-back of KCH criteria is the subjective assessment of the coma grade. Lethargy and drowsiness are common symptoms and signs associated with high fever and/or the occurrence of bacterial and infectious complications in these patients which may influence the assessment. Thus a reliable model based on objective measures would be valuable in this context. MELD score is used for allocation of organs for transplantation in patients with chronic liver disease [8, 9] and was recently shown to accurately predict mortality in patients with alcoholic hepatitis [14]. Recently, MELD score was reported to have a 100% NPV and a 90% PPV in nonacetaminophen induced ALF [15]. In our cohort, MELD score greater than 30, as previously suggested to be cut-off [15] was associated with high NPV (94%), but very low PPV of only 21% in the acetaminophen patients but associated with a relatively high NPV and PPV amongst the nonacetaminophen patients (Table 4). Thus, if MELD score is <30, this is a strong argument for a favourable outcome without transplantation in patients with acetaminophen toxicity if the patient does not fulfil KCH criteria for transplantation. Thus, clinical judgement is most important in the choice of conservative therapy versus transplantation in these patients. To conclude, in unselected patients with progressive coagulopathy in Sweden due to liver disease, acetaminophen toxicity was the most common cause with generally good prognosis. Patients with an unknown cause remain a challenge for the clinician although these patients constitute a much smaller proportion than in patients listed for HULTX. The KCH criteria were found to have a high NPV in the acetaminophen group which might be clinically important which together with a MELD score <30 predicts a spontaneous survival without transplantation in acetaminophen induced acute liver failure. Conflict of interest No conflict of interest was declared. 400 ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;

9 Acknowledgement This study was supported by the Faculty of Medicine, University of Gothenburg. References 1 O Grady JG. Acute liver failure. In: O Grady JG, Lake JR, Howdle PD, eds. Comprehensive Clinical Hepatology chapter 30. London: Mosby, 2000; Ostapowicz G, Fontana RJ, Schiodt FV et al. Results of a prospective study of acute liver failure at 17 tertiary care centers in the United States. Ann Intern Med 2002; 137: Lee WM. Acute liver failure in the United States. Semin Liver Dis 2003; 23: Williams R. Classification, etiology, and considerations of outcome in acute liver failure. Semin Liver Dis 1996; 16: O Grady JG. Acute liver failure. Postgrad Med J 2005; 81: Brandsaeter B, Höckerstedt K, Friman S et al. Fulminant Hepatic Failure: Outcome after listing for highly urgent liver transplantation-12 years experience in the nordic countries. Liver Transpl 2002; 8: O Grady JG, Alexander GJ, Hayllar KM, Williams R. Early indicators of prognosis in fulminant hepatic failure. Gastroenterology 1989; 97: Kamath PS, Wiesner RH, Malinchoc M et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001; 33: Kremers WK, van Ijperen M, Kim WR et al. MELD score as a predictor of pretansplant and posttransplant survival in OPTN/ UNOS status 1 patients. Hepatology 2004; 39: Schiodt FV, Atillasoy E, Shakil AO et al. Etiology and outcome for 295 patients with acute liver failure in the United States. Liver Transpl Surg 1999; 5: Shakil AO, Kramer D, Mazariegos GV, Fung JJ, Rakela J. Acute liver failure: Clinical features, outcome analysis, and applicability of prognostic criteria. Liver Transpl 2000; 6: Neuberger J. Prediction of survival for patients with fulminant hepatic failure. Hepatology 2005; 41: Anand AC, Neuberger JM, Nightingale P. Early indicators of prognosis in fulminant hepatic faliure: an assessment of the King s criteria. J Hepatol 1997; 26: Dunn W, Jamil LH, Brown LS et al. MELD accurately predicts mortality in patients with alcoholic hepatitis. Hepatology 2005; 41: Yantorno SET, Trentadue JJ, Ruf AE et al. The model for end stage liver disease (MELD): A usefull tool to assess prognosis in fulminant hepatic failure. Liver Transpl 2004; 10: C36. Correspondence: Einar Björnsson, Department of Internal Medicine, Sahlgrenska University Hospital, SE Gothenburg, Sweden. (fax: ; einar.bjornsson@medic.gu.se). ª 2007 Blackwell Publishing Ltd Journal of Internal Medicine 262;