The incidence, etiologies, outcomes, and predictors of mortality of acute liver failure in Thailand: a population-base study

Background Acute liver failure (ALF) is uncommon but progresses rapidly with high mortality. We investigated the incidence, etiologies, outcomes, and predictive factors for 30-day mortality in patients with ALF. Methods We conducted a population-based study of ALF patients hospitalized between 2009 and 2013 from the Thai Nationwide Hospital Admission database, which comprises 76% of all admissions from 858 hospitals across 77 provinces in Thailand. ALF was diagnosed using ICD-10 codes K72.0 and K71.11. Patients with liver cirrhosis were excluded. Results There were 20,589 patients diagnosed with ALF during the study period with 12,277 (59.6%) males and mean age of 46.6 ± 20.7 years. The incidence of ALF was 62.9 per million population per year. The most frequent causes of ALF were indeterminate (69.4%), non-acetaminophen drug-induced (26.1%), and viral hepatitis (2.5%). Acetaminophen was the presumptive cause in 1.7% of patients. There were 5502 patients (26.7%) who died within 30 days after admission. One patient (0.005%) underwent liver transplantation. The average hospital stay was 8.7 ± 13.9 days, and the total cost of management was 1075.2 ± 2718.9 USD per admission. The most prevalent complications were acute renal failure (ARF)(24.2%), septicemia (18.2%), and pneumonia (12.3%). The most influential predictive factors for 30-day mortality were ARF (HR = 3.64, 95% CI: 3.43–3.87, p < 0.001), malignant infiltration of the liver (HR = 3.37, 95% CI: 2.94–3.85, p < 0.001), and septicemia (HR = 1.96, 95%CI: 1.84–2.08, p < 0.001). Conclusions ALF patients have poor outcomes with 30-day mortality of 26.7% and high economic burden. Indeterminate etiology is the most frequent cause. ARF, malignant infiltration of the liver, and septicemia are main predictors of 30-day mortality.


Background
Acute liver failure (ALF) is a clinical condition characterized by an acute deterioration of liver function resulting in encephalopathy and coagulopathy (International Normalized Ratio: INR ≥ 1.5), which occurs within 8-26 weeks of the onset of symptoms in patients without preexisting liver disease [1][2][3]. However, the definitions have variations in time course and etiology [1][2][3][4]. Acetaminophen and drug toxicity, viral hepatitis, autoimmune hepatitis, and indeterminate ALF are the most frequent causes of ALF. There are different predominant causes of ALF for each geographical area and ethnicity [5][6][7]. In the United States and European countries, acetaminophen toxicity is the predominant cause, while acute viral hepatitis infection is common in South and East Asia [5][6][7][8][9][10].
ALF is an uncommon condition, and the average incidence is approximately 5.5-6.2 people per milllion population per year [11,12]. However, patients with ALF often rapidly progress to multiorgan failure and have a high mortality rate. Providing good supportive care in the intensive care unit, including cardiovascular monitoring, intracranial pressure (ICP) monitoring, and ventilator support, are important in the management of patients with ALF. Emergency liver transplantation remains the only definitive treatment for patients who do not achieve spontaneous recovery. However, a minority of ALF patients (7-18%) could receive emergency liver transplantation [13][14][15]. The in-hospital survival rates of patients with ALF treated with and without liver transplantation are 80-86% and 35-48%, respectively [9,15].
Identifying the predictive factors of mortality is important and may lead to early intensive care treatment and consideration for emergency liver transplantation. Thus, this study aimed to determine the predominant causes, clinical outcomes, and prognostic factors of ALF. We conducted a nationwide population-based cohort study of patients with ALF during a 5-year period to evaluate the incidence, etiologies, clinical outcomes, economic burden, and predictive factors of 30-day mortality.

Study population and data acquisition
Data were obtained from the discharge records of hospitalized patients whose medical treatment expenses were covered by the Universal Coverage Scheme (UCS). The data contained 76% of all admissions from primary, secondary, and tertiary hospitals across 77 provinces in Thailand. The National Health Security Office (NHSO) reviewed and improved the quality of the 28,294,685 admission summary records in the database starting from 2009 until 2013, which aimed to examine the epidemiology and outcomes of patients hospitalized with gastrointestinal or liver diseases in Thailand. In the current study, we investigated the demographic data, co-morbidities, causes of ALF, therapeutic procedures, clinical outcomes, and medical expense of ALF patients enrolled in the NHSO database during these 5-year period.
The study enrolled patients who were hospitalized for ALF and diagnosed for the first time using codes K72.0 (Acute and subacute hepatic failure) and K 71.11 (Toxic liver disease with hepatic necrosis, with coma) of the International Classification of Diseases, 10th revision (ICD-10) from January 1, 2009 to December 31, 2013 according to the NHSO database. Patients who had preexisting liver cirrhosis (K74) were excluded. The diagnosis of ALF, co-morbidities, and treatment considerations were performed by attending physicians based on clinical data and laboratory investigations in each hospital. Patients with ALF were diagnosed based on the compatible criteria as the onset of hepatic encephalopathy occurring within 26 weeks of the onset of jaundice in patients with no preexisting cirrhosis [3].
Co-morbidities, etiologies, and complications were identified using following codes: cerebrovascular disease (I60-69), ischemic heart disease (I20-25), diabetes mellitus (E08-13), chronic obstructive pulmonary disease (J44), chronic kidney disease ( The study protocol was approved by the Institutional Review Board (IRB 113/58) of the Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki under good clinical practice. The data were used by the Gastroenterological Association of Thailand (GAT) with an agreement from the NHSO.

Definitions
Most government hospitals in Thailand are operated by the Ministry of Public Health and are classified into three levels. Primary hospitals are community hospitals at the district level and have capacities of 10-200 beds. Secondary hospitals are general hospitals in major districts and have capacities of 200-500 beds. Tertiary hospitals are regional hospitals in provincial centers and have capacities of at least 500 beds and specialist physicians.

Statistical analysis
Categorical data are presented as numbers (percentages), and continuous data are shown as the mean ± standard deviation (SD). Comparison the difference between baseline characteristics, clinical outcomes and therapeutic procedures in patients who died and survived within 30 days is carried out using the chi-square test or Fisher's exact test for categorical variables and the unpaired t-test or Mann-Whitney U test for continuous variables. Two-sided P-values < 0.05 were considered statistically significant. Cox proportional hazards regression analysis was used to identify independent predictive factors for 30-day mortality, and the results are shown as the hazard ratio (HR) and 95% confidence interval. All analyses were performed using SPSS software version 22.0 (IBM, New York city, NY, USA).

Patient characteristics
Between January 2009 and December 2013, the incidence due to all-cause admissions was 5.6 million cases per year. Of these, a total of 159,061 admissions (or 21,165 admissions per year) were due to liver diseases, and 20,589 admissions (or 4117 admissions per year) were diagnosed with ALF and enrolled in this study. The annual admission rate for ALF was 0.07% of the total admissions per year and 19.4% of all admissions related to liver disease. According to the data from the National Statistical Office of Thailand (NSO), the estimated rate for the Thai population was 65.4 million people per year. The incidence of ALF in Thailand was 62.9 cases per million people per year in the present study. The mean age of hospitalized patients with ALF was 46.6 ± 20.7 years, and 12,277 patients were male (59.6%). Most of the patients were hospitalized in primary hospitals (n = 13,297, 64.6%) and in the Northeast part of Thailand (n = 8001, 38.9%). The patient's baseline characteristics and co-morbidities were shown in Table 1.
The mean cost per hospital admission for ALF was 1075.2 ± 2718.9 United States dollar (USD). The total medical expense of admission was 23 million USD per year. Of this amount, 4.2 million USD per year was spent but could not sustain the lives of patients. ALF patients who died within 30 days after admission had greater age (51.7 ± 20.6 vs. 44.7 ± 20.4 years, p < 0.001), a greater proportion of males (63.0% vs. 58.4%, p < 0.001), more co-morbidities (28.9% vs. 16.8%, p < 0.001), and more cases of indeterminate cause (73.6% vs. 67.9%) compared to patients who survived within 30 days (Table  1). In terms of clinical outcomes and interventions, the patients who died within 30 days after admission had shorter lengths of hospital stays, had significantly more complications (including ARF, septicemia, pneumonia, and urinary tract infections), and needed more blood transfusion, renal replacement therapy, plasma exchange and mechanical ventilator support (Table 2). No patient was treated with liver dialysis or monitored ICP in this database.

Predictors for 30-day mortality
Multivariate Cox regression analysis was performed to identify the predictor for 30-day mortality by including potential factors from univariate Cox regression analysis. Table 3 presents the independent predictive factors for 30-day mortality among hospitalized patients with ALF. The three most commons factors associated with 30-day mortality were ARF (HR = 3.64, 95% CI: 3.43-3.87, p < 0.001), malignant infiltration of the liver (HR = 3.37, 95% CI: 2.94-3.85, p < 0.001), and septicemia (HR = 1.96, 95%CI: 1.84-2.08, p < 0.001). The ARF, malignant infiltration of the liver, and septicemia were observed in 53.1, 4.9 and 37.1% of patients who died in 30 days after admission compared to 13.7, 0.6 and 11.3% of those who survived.

Discussion
The main finding of the current study was the incidence rate of ALF was 62.9 per million population per year in Thai population with a 30-day mortality rate after hospital admission of 26.7%. In addition, indeterminate etiology (69.4%) was the most common cause of ALF. Patients have higher risk of death within 30 days after admission if they have age ≥ 60 years, co-morbidities, malignant infiltration of the liver, Hepatitis B virus infection, Hepatitis C virus infection, indeterminate cause, and complications including ARF, septicemia and pneumonia.
The true incidence of ALF is unclear because there is scarce data from previous studies, which were mostly done in liver transplant units or tertiary referral centers. The estimated annual incidence of ALF is 5.5 cases per million population per year in the United States according to a population-based study and 6.2 cases per million population per year in Scotland in liver transplant centers [11,12]. Our study included data from primary, secondary, and tertiary hospitals and showed a higher incidence rate of ALF in comparison to these studies. However, the incidence rate from our study is close to that reported by Ho et al., who found an annual incidence rate of 80.2 cases per million person-years in Taiwanese hospitalized patients with ALF [7]. These data might indicate that ALF is not a rare condition in some countries, including Thailand.
The etiology of ALF is an important factor for prognosis and management. The main causes of ALF are different in each area of the world. The predominant cause of ALF is acetaminophen in the United States, the United Kingdom, and Australia (36-43%) [16][17][18]. In Asian countries, the predominant causes are acute hepatitis virus infection, specifically hepatitis E virus in India (43.9%) [19] and hepatitis B virus in Japan (32.7%) and Taiwan (33%) [7,20]. Indeterminate cause is the leading etiology of ALF in Scandinavian countries (43%), Chile (44%), and Sudan (38%) [21][22][23]. There were no previous data on etiology in adult patients with ALF in Thailand. However, data from a single tertiary care hospital reported that indeterminate cause (55%) was the main etiology among hospitalized Thai children with ALF [24]. Our study showed that indeterminate cause (69.4%) is the predominant etiology of ALF in Thailand. Of the established causes, non-acetaminophen drug toxicity (26.1%) was the most frequent. Thailand has a high burden of chronic hepatitis B infection with an estimated prevalence of 3.5% of the population [25]. However, our study showed that acute hepatitis B-induced ALF occurred in only 1.8% of patients, which might be underestimated because the diagnosis may be difficult by routine serology.    [17,26]. Acetaminophen overdose was the most common cause of ALF in both studies. Liver transplantation was done in 21.3 and 29% of ALF patients in the R Marudanayagam et al. study and the G Ostapowicz et al. study, respectively. Liver transplantation is considered a proven therapy for improving survival in patients with ALF. In our cohort, only one patient underwent liver transplantation due to the rapid progression of the disease and the limitations of the transplant system. From 2009 to 2013, Thailand had nine liver transplant centers, of which eight were located in the capital city and one was located in the northeastern region. According to the Thai Red Cross Organ Donation Center Database, there are approximately 87-158 donors and only 43-73 patients who received liver transplantation annually. Our information could be useful for improving health care policy for the management of ALF and promoting public awareness of organ donation.
Identifying patients who have poor survival outcomes is essential and will lead to prompt referral and consideration for emergency liver transplantation. Previous studies show that encephalopathy, extrahepatic organ failure (especially renal failure), and coagulopathy are poor prognostic factors for mortality [27]. In addition, short-term outcome varies according to etiology. The favorable prognosis was found with acetaminophen-and ischemic-induced ALF (60-80% short-term survival without transplantation) and poor prognosis with hepatitis B infection, indeterminate cause and drug other than acetaminophen-related ALF (17-25% short-term survival without transplantation) [17,26]. Consistent with our study, acetaminophen is inversely associated with the 30-day mortality in patients with ALF. Patients with acetaminophen-induced ALF had favorable prognosis possibly due to the availability of anti-dote (N-acetylcysteine: NAC) and prompt therapy from history of drug usage. Previous study showed 66 % of patients had spontaneous recovery with NAC and supportive care therapy [17,28]. Additionally, age ≥ 60 years, co-morbidities, malignant infiltration, Hepatitis B or C virus infection, indeterminate etiology, infectious complications including ARF, septicemia and pneumonia are independent predictors for 30-day mortality as well.
This study has some limitations. First, it was a retrospective study; there may have been some missing data and uncontrolled exposure. Furthermore, it relies on accurate data recordkeeping. Second, it might be interesting if we could analyze the difference of the model for end-stage liver disease (MELD) scores in patients with acute liver failure who died and survived within 30 days. However, some information about clinical findings and laboratory tests were lacking in the NHSO database, precluding us for estimating this score. Third, our study may not represent the true prevalence of indeterminate-ALF in Thailand; we hypothesized that several reasons led to a large number of indeterminate causes in our study, including (i) unavailability of further laboratory investigations in all hospitals especially in primary hospitals where the majority of ALF patients were hospitalized, (ii) co-existence of several potential etiologies possibly confounding the final diagnosis by attending physician, (iii) missing data and the inability to get a precise history due to encephalopathy. A recent publication shows that using diagnostic algorithms and expert review in combination with additional laboratory testing can verify etiologic diagnoses in 50% of patients initially considered as indeterminate-ALF [29].

Conclusions
The incidence rate of ALF in Thailand is 62.9 cases per million population per year. Indeterminate causes are the major cause, and non-acetaminophen drug-induced is the most common identifiable cause of ALF cases. Patients with ALF have high 30-day mortality and high economic burden. Age, etiologies and complications of ALF are predictive factors of 30-day mortality after hospital admission.