Multiple bacterial infections increase the risk of hepatic encephalopathy in patients with cirrhosis

PLOS ONE, May 2018

Objective Patients with liver cirrhosis (LC) are at increased risk for bacterial infections. It is not fully understood how exposure to infections induces further development of hepatic encephalopathy (HE). This study estimated risks of infection associated with HE among patients with LC. Methods A nested case-control study of 14,428 adult patients with LC was performed using the population-based Longitudinal Health Insurance Database 2000 in Taiwan. Cases were cirrhotic patients who developed HE during follow-up. Controls were matched to each case by age at LC diagnosis (±2 years), sex, Charlson Comorbid index score, year of LC, and follow-up time with a 1:1 ratio. A multivariate logistic regression model was used to determine and compare the odds of developing HE based on exposure to various risk factors, including site of infection, cirrhosis-related complications, Helicobacter pylori eradication therapy, and peptic ulcer bleeding. Patient survival was evaluated using the time-dependent Cox regression model. Results Cirrhotic patients with HE (n = 714) and without HE (n = 714) were matched to compare risks. Infections and more frequent yearly infections were significantly associated with increased risk of HE. Independent predictors of HE included spontaneous bacterial peritonitis (aOR, 5.13; 95% CI, 3.03–8.69), sepsis (aOR, 2.54; 95% CI, 1.82–-3.53), and biliary tract infection (aOR, 2.03; 95% CI, 1.2–3.46), controlling for confounders. Conclusion Frequent infections are associated with increased risk of HE in cirrhotic patients. More frequent exposure to infection increases the risk of HE and mortality rates. Appropriate prevention of infection and the use of antibiotics for cirrhotic patients at risk for HE are needed.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

http://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0197127&type=printable

Multiple bacterial infections increase the risk of hepatic encephalopathy in patients with cirrhosis

May Multiple bacterial infections increase the risk of hepatic encephalopathy in patients with cirrhosis Lan-Ting Yuan 0 1 2 Seng-Kee Chuah 0 2 Shih-Cheng Yang 0 2 Chih-Ming Liang 0 2 Cheng- Kun Wu 0 2 Wei-Chen Tai 0 2 Tsung-Hsing Hung 0 2 Seng-Howe Nguang 0 2 Jiunn-Wei Wang 0 2 Kuo-Lun Tseng 0 2 Ming-Kun Ku 0 2 Pin-I Hsu 0 2 Deng-Chyang Wu 0 2 Chien-Ning Hsu 0 2 0 Abbreviations: CI , Confidence interval; H. pylori, Helicobacter pylori; HE, Hepatic encephalopathy; ICD-9-CM , International Classification of Diseases 1 Division of Gastroenterology, Yuan General Hospital , Kaohsiung, Taiwan , 2 Division of Hepatogastroenterology, Department of Internal Medicine, Kaohsiung Chang Gung Memorial Hospital , Kaohsiung, Taiwan , 3 Chang Gung University, College of Medicine, Kaohsiung, Taiwan, 4 Division of Hepatogastroenterology, Department of Internal Medicine, Buddist Tzu Chi General Hospital , Dalin Branch, Chia-Yi, Taiwan , 5 Division of Gastroenterology, Pin-Tung Christian Hospital , Pin-Tung, Taiwan , 6 Division of Gastroenterology, Department of Internal Medicine, Kaohsiung Medical University Hospital and Kaohsiung Medical University , Kaohsiung, Taiwan , 7 Division of Gastroenterology, Kaohsiung Municipal TaTung Hospital , Kaohsiung, Taiwan , 8 Division of Gastroenterology, Cishan Hospital , Kaohsiung, Taiwan , 9 Division of Gastroenterology, FooYin University Hospital , Pin-Tung, Taiwan , 10 Division of Gastroenterology, Department of Internal Medicine, Kaohsiung Veterans General Hospital, National Yang-Ming University , Kaohsiung, Taiwan , 11 Department of Pharmacy, Kaohsiung Chang Gung Memorial Hospital , Kaohsiung, Taiwan , 12 School of Pharmacy, Kaohsiung Medical University , Kaohsiung , Taiwan 2 Editor: Gianfranco D. Alpini, Texas A&M University , UNITED STATES - Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Objective LC. Methods Patients with liver cirrhosis (LC) are at increased risk for bacterial infections. It is not fully understood how exposure to infections induces further development of hepatic encephalopathy (HE). This study estimated risks of infection associated with HE among patients with A nested case-control study of 14,428 adult patients with LC was performed using the population-based Longitudinal Health Insurance Database 2000 in Taiwan. Cases were cirrhotic patients who developed HE during follow-up. Controls were matched to each case by age at LC diagnosis (±2 years), sex, Charlson Comorbid index score, year of LC, and follow-up time with a 1:1 ratio. A multivariate logistic regression model was used to determine and compare the odds of developing HE based on exposure to various risk factors, including site of infection, cirrhosis-related complications, Helicobacter pylori eradication therapy, and peptic ulcer bleeding. Patient survival was evaluated using the time-dependent Cox regression model. Results Cirrhotic patients with HE (n = 714) and without HE (n = 714) were matched to compare risks. Infections and more frequent yearly infections were significantly associated with Ninth Revision, Clinical Modification; LC, liver cirrhosis; LHID 2000, Longitudinal Health Insurance Database 2000; NHI, National Health Insurance; OR, Odds ratio; PPI, Proton pump inhibitor; SBP, Spontaneous bacterial peritonitis; UTI, urinary tract infections. increased risk of HE. Independent predictors of HE included spontaneous bacterial peritonitis (aOR, 5.13; 95% CI, 3.03±8.69), sepsis (aOR, 2.54; 95% CI, 1.82±-3.53), and biliary tract infection (aOR, 2.03; 95% CI, 1.2±3.46), controlling for confounders. Conclusion Frequent infections are associated with increased risk of HE in cirrhotic patients. More frequent exposure to infection increases the risk of HE and mortality rates. Appropriate prevention of infection and the use of antibiotics for cirrhotic patients at risk for HE are needed. Introduction Hepatic encephalopathy (HE) is a commonly encountered complication in cirrhotic patients with advanced liver disease or portosystemic shunts. The incidence of HE ranges from 2% to 20% per year in patients with liver cirrhosis.[1±3] HE is associated with increased morbidity and mortality as well as significant utilization of health care resources.[4±6] Identifying risk factors for HE would be paramount for implementing preventive measures to improve overall outcomes for cirrhotic patients. When HE is diagnosed, underlying precipitating factors should be sought and treated first. Common culprits include gastrointestinal bleeding, infection, constipation, excessive dietary protein, hypovolemia, shock, hypokalemia, alkalosis, surgical portosystemic shunts or transjugular intrahepatic portosystemic shunts, hyponatremia, and medications such as opiates and benzodiazepines.[ 7,8 ] HE is a reversible neuropsychiatric condition, and elevated ammonia level in the serum has been considered the primary pathophysiologic cause. It is widely accepted that ammonia is derived primarily from enteric bacterial flora.[ 7 ] Bacterial infection is present at admission (community-acquired infections) or develops during hospitalization in patients with liver cirrhosis (nosocomial and health care±related infections), and it occurs in more than 50% of hospitalized cirrhotic patients.[ 9 ] Spontaneous bacterial peritonitis (SBP), urinary tract infections (UTI), pneumonia, and cellulitis are the most frequent infections among cirrhotic patients in different settings.[ 7,9 ] Furthermore, ammonia toxicity is greatly attributed to fecal bacteria. Helicobacter pylori (H. pylori) in the stomach hydrolyzes urea and converts it to ammonia, which can be rapidly absorbed and increases blood ammonia concentrations in H. pylori– infected patients with cirrhosis.[10±12] However, ammonia levels are not related to severity of HE.[13] Whether eradication therapy for H. pylori–infected patients can lower ammonia levels and the risk of HE development remains inconclusive.[ 14 ] Infections in cirrhotic patients are detrimental to hepatic functions and increase mortality fourfold.[15±17] To date, there is not enough empirical evidence of risk factors for HE in cirrhotic patients to support a prevention strategy for HE. To determine the independent effect of infections on the development of HE, we used a nested case-control design to avoid incorrect sampling of cases and controls and compared individual infections according to system and intensity. Materials and methods The study was reviewed and approved by the Institutional Review Board and Ethics Committee of Chang Gung Medical Foundation in Taoyuan, Taiwan (IRB #201601548B1). All 2 / 14 personal identifying information for patients was anonymized; therefore, the need for informed consent was waived for the study. Data sources Data were extracted from the Longitudinal Health Insurance Database 2000 (LHID 2000) of 1 million individuals who were randomly sampled from the year 2000 Registry for Beneficences of 23.75 million individuals involved in Taiwan's National Health Insurance (NHI) program. [ 18 ] The Taiwan NHI is a single-payer health insurance program that covers 99.9% of Taiwan's population.[19] The LHID 2000 contains demographic information, diagnostics, medical treatments, prescriptions, and total costs from January 1, 1997 to December 31, 2012. Data analysts were staff members of the Center for Medical Informatics and Statistics at Kaohsiung Medical University, a site of the Application of Health and Welfare Informatics of the Ministry of Health and Welfare in Taiwan. Study design and population The population-based case study involved a cohort of patients aged older than 18 years with a hospital discharge diagnosis of liver cirrhosis (LC) (International Classification of Diseases, 9th edition [ICD-9], codes 571.5, 571.2, 571.6) recorded in the LHID 2000 between 1997 and 2012 (Fig 1). The study index date was defined as the date of the first ICD-9 code for LC was assigned. Patients were excluded if they had prior peptic ulcer disease (ICD-9 531, 532, 533, 534), malignancy (140±239), HE (572.2), ever underwent a transjugular intrahepatic portosystemic shunt procedure (33113A, 3313B), or had infections including antimicrobial combination therapy or a diagnosis of H. pylori (041.86), pneumonia (481±487), spontaneous bacterial peritonitis (SBP) (567.2, 567.8, 567.9), sepsis (038, 020.0, 790.7, 112.81), UTI (590.1, 595.0, 595.9, 599.0), biliary tract infection (574.00, 574.01, 574.1, 574.30, 574.31, 574.4, 574.60, 574.61, 574.80, 574.81, 576.1, 575.0), cellulitis (681, 682, 728.86), inflammatory disease of the central nervous system (324, 320), septic arthritis (711), endocarditis (421), perianal abscess (566), or liver abscess (572.0) recorded as both inpatient and outpatient claims within 365 days before the index date. H. pylori infection and eradication triple or quadruple therapy was defined as proton-pump inhibitor (PPI) or histamine type 2 receptor antagonists (H2RA) plus clarithromycin or metronidazole plus amoxicillin or tetracycline, with or without bismuth for 7±14 days.[ 20,21 ] All patients were followed from the first date of LC until death, withdrawal from the NHI program, or the last date used for the dataset (December 31, 2012). Patients in the case group who developed overt HE with more severe symptoms (Grade III-IV) requiring hospital admission for treatment during the follow-up period were defined by an ICD-9 code (572.2) for HE at hospital discharge.[ 22,23 ] For controls, the LC group without HE during follow-up was identified and subsequently matched to each HE case. The 1:1 exact matching for age at the index date within a 2-year difference, sex, and propensity score (including age at the index date, sex, Charlson comorbidity index [CCI] score, year of the index date, and length of follow-up in months) was performed for cases and controls.[24] Covariates Patients' demographic information, time at HE diagnosis, length of follow-up, comorbidities, liver diseases, and liver transplantation (LT) (ICD9, V427) were identified within 365 days prior to the index date. Comorbidities were assessed by the CCI using ICD-9 codes with at least two records from inpatient, emergency, and outpatient claims. Prevalent liver diseases 3 / 14 Fig 1. All-cause mortality among cirrhotic patients by frequency of infection episodes. were hepatitis B virus (HBV) infection (070.2, 070.3, V0261), hepatitis C virus (HCV) infection (070.41, 070.44, 070.51, 070.54, 070.70, 070.71, V0262), other viral hepatitis (V0269), alcoholrelated liver disease (ALD) (571.0±571.3), decompensated cirrhosis with esophageal variceal bleeding (456.0, 456.20, 530.82), ascites (789.5, 567.2,567.8, 567.9, ICD-9 procedure 54.91), and others such as jaundice (782.4), portal hypertension (572.3), hepatorenal syndrome (572.4), other sequelae of chronic liver disease (572.8), and hepatic cellular carcinoma (HCC) (1550) within 1 year before the index date. During follow-up, ICD-9 codes for infections recorded on inpatient claims were obtained. To investigate the dose relationship between infection and HE risk, the number of infections during hospitalization was categorized as 0, 1±3, and more than 3. Undergoing endoscopy for upper gastroenterology bleeding (billing code 47043B with the exclusion of ICD-9 44.43) and developing HCC and decompensated cirrhosis were accounted for by all-cause mortality analysis. 4 / 14 Statistical analysis Descriptive statistics and counts with proportions were presented as the mean (standard deviation [SD]) for categorical data and as the median (25th-75th percentile) for continuous data. Categorical variables were compared between groups using the χ2 test. Multiple logistic regression models were used to calculate adjusted odds ratios (aOR) with 95% confidence intervals (CI) for development of HE among cirrhotic patients. Model 1 included exposure to any infection, model 2 included frequency of infection, and model 3 included individual infections. All three models included patient demographic characteristics, CCI scores, H. pylori therapy initiation, HCC, and decompensated cirrhosis development that may predispose to HE. The timedependent Cox proportional hazards regression model was used to calculate the adjusted hazard ratio (aHR) with 95% CI for all-cause mortality and accounted for potential time-varying covariate effects.[ 25 ] The independent effect of HE in the model was adjusted according to infection, patient demographic characteristics, CCI score, H. pylori therapy, HCC, and decompensated cirrhosis development. The level of statistical significance was 5%, and a two-sided P<0.05 was considered significant. All statistical analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC). Results Baseline patient characteristics In the study cohort, 6078 cirrhotic patients (42%) who ever had an infection before development of HE were excluded (S1 Fig). The incidence of HE among patients with liver cirrhosis was 11.3% (2136 of 18824 cirrhotic patients). The demographic characteristics of the LC patients with and without HE are shown in S1 Table. Using the predefined matching criteria, 714 patients with HE during hospitalization (cases) and 714 randomly selected patients without HE (controls) were analyzed. In the LC-matched cohort, the mean age was 55.74±13.35 years and 548 (76.75%) patients were male. The demographic characteristics of the LCmatched cohort are shown in Table 1. Ascertainment of risk factors The potential risk factors for HE were categorized from the study index date to the first event of HE during hospitalization. As shown in Table 2, a higher proportion of patients in the case group (58.82%) than in the control group (30.81%) was exposed to any type of infection. H. pylori infection (13.31% vs. 8.68%; p<0.01), pneumonia (14.99% vs. 10.50%; p = 0.01), spontaneous bacterial peritonitis (SBP) or peritonitis (14.29% vs. 2.52%; p 0.001), sepsis (25.63% vs. 9.52%; p<0.001), UTI (18.77% vs. 11.20%; p<0.0001), biliary tract infection (7.14% vs. 3.22%; p<0.001), and cellulitis (11.62% vs. 3.98%; p = 0.02) increased the risk of HE. Hospitalization due to infection occurred for 6.44±42.26 of the case group and 0.87±1.27 of the control group. Risk factors associated with hepatic encephalopathy The proportion of the case group ever exposed to infection was higher than that of the control group (69.19% vs. 41.18%; p<0.001). Tables 3 and 4 show risk factors associated with HE. Infections were associated with hepatic encephalopathy development (aOR, 3.04; 95% CI, 2.44±3.78; p<0.0001). Compared to patients without any infection, patients with 1 to 3 episodes of any type of infection had approximately threefold odds of developing HE (aOR, 2.68; 95% CI, 2.13±3.37; p<0.001), and the odds increased for patients with more frequent infections (aOR, 10.27; 95% CI, 5.17±20.4; p<0.001) when controlling for age at LC diagnosis, gender, CCI score, and severity of liver diseases. 5 / 14 LC, liver cirrhosis; CCI, Charlson comorbid index; SD, standard deviation; HCC, hepatic cellular carcinoma; PPI = proton pump inhibitors, Case: patients with HE, control: patient without HE, Patient's comorbid conditions were identified within 1 year prior to the date of LC diagnosis In model 3 (Table 4), an independent effect associated with HE development was observed among patients who ever had SBP (aOR, 5.13; 95% CI, 3.03±8.69; p<0.0001), sepsis (aOR, 2.54; 95% CI, 1.82±3.53; p<0.0001), or biliary tract infection (aOR, 2.03; 95% CI, 1.20±3.46; p = 0.009). Although UTI was associated with an additional 38% risk for HE, its effect did not reach statistical significance in this study cohort. Risk factors associated with all-cause mortality The overall mortality rate was 43.1% (n = 615) for the matched study cohort and 76% for those who ever had an infection before HE development. The overall survival rate was not statistically different between patients with and without development of HE (48.6% vs. 37.54%; P = 0.081) (S2 Fig) 6 / 14 SBP, spontaneous bacterial peritonitis; LC, liver cirrhosis; SD, standard deviation; PUB, peptic ulcer bleeding, EV, esophageal varices; The exposure of risk factors for hepatic encephalopathy was categorized from the date of LC diagnosis to the first event of hospitalized HE. Discussion HE is one of the most debilitating complications of cirrhosis because it greatly affects society and individual patients and their caregivers.[ 5,6 ] Bacterial infections are common in patients with LC; however, the epidemiological features of infection associated with HE have not been clearly demonstrated. In the present study, we found the following. First, infection was an independent risk for HE, requiring hospitalization among cirrhotic patients. Second, more frequent infections significantly increased the dose-response risk of HE development and 7 / 14 H. Pylori therapy Pneumonia SBP and unspecified peritonitis Sepsis without infection focus Urinary tract infection Biliary tract infection Cellulitis Perianal abscess Liver abscess Age at LC, years 50±59 vs 49 60±69 vs 49 70 vs 49 Male vs Female CCI score Prior to HE development HCC PUB Decompensated cirrhosis 8 / 14 mortality. In addition, prevalent infections such as SBP and unspecified peritonitis were strong risk predictors of HE among cirrhotic patients. Intestinal bacterial overgrowth is associated with higher ammonia production and higher serum ammonia concentrations, resulting in the induction of encephalopathy through the promotion of cerebral edema, modulation of the blood±brain barrier, and neuro-inhibition. [ 26 ] The present population-based epidemiological study confirmed that prior to the development of HE, cirrhotic patients with HE were more likely to have been exposed to infection than those without HE (58.81 vs. 30.81%; p<0.0001). The significant predictive factors of HE development for cirrhotic patients were SBP and unspecified peritonitis, sepsis, and biliary tract infection. UTI (30%) and pneumonia (25%) were the common sites of infection among cirrhotic patients, which is consistent with previous cirrhotic cohorts.[ 7 ] However, the individual predictive value of HE was not confirmed in the present study cohort. When the intensity of exposure to infection associated with HE development was examined, our data showed that a subgroup of cirrhotic patients with 3 infectious episodes per year is at higher risk for HE than those with fewer exposures. More frequent exposure to infection ( 3 episodes per year) was a strong predictive factor for all-cause mortality in cirrhotic patients. The adjusted morality (aHR, 1.03; 95% CI, 0.87± 1.22; p = 0.711) revealed no statistical difference between patients with and without HE. Although the therapeutic effectiveness of cirrhotic encephalopathy needs further investigations, these findings suggest that HE occurrence was not independent of survival outcome. In the present study, effect of HE occurrence on overall mortality among cirrhotic patients was assessed in time-dependent Cox model to justify cirrhosis complications that occurred before end of the study period. Future applications in survival research, it is important to consider the presence of competing events during follow-up, i.e. receiving liver transplantation, death on 9 / 14 first HE event that might influence the results and interpretations in conventional Cox proportional hazard model. As previously reported, a history of infection in the past 12 months, advanced liver disease (model of end-stage liver disease score 15), and diagnosis of malnutrition were independent predictors of infection and sepsis.[ 9 ] The present study results support that routine screening and identifying who is at risk for bacterial infections among hospitalized cirrhotic patients would facilitate compliance with empirical antibiotic therapy and may prevent resistance and worsening infection.[ 7 ] The results of the present study support that H. pylori is more prevalent in cirrhotic patients with HE than in those without HE (13.31% vs. 8.68%; p = 0.005); however, the administration of standard combinations of H. pylori eradication regimens (22%) did not benefit HE prevention (aOR, 1.34; 95% CI, 0.88±2.04; p = 0.178) after adjusting for other sites of infection and deterioration of liver cirrhosis. It is worth noting that eradication of H. pylori was significantly associated with a 28% lower risk of mortality for patients with liver cirrhosis. Similarly, Dasani et al observed that symptoms in infected encephalopathic patients improved following H. pylori eradication therapy.[ 27 ] Based on these study results and previous evidence,[ 14 ] the beneficial effects of eradication therapy for H. pylori are insufficient for recommending the use of this therapy in clinical practice. Because there are geographic differences in the incidence and prevalence of H. pylori infection and antimicrobial resistance and in the availability of medications and endoscopy, cirrhotic patients with a history of peptic ulcer, gastric mucosa±associated lymphoid tissue lymphoma, or history of endoscopic resection of early gastric cancer should be tested for H. pylori infection and administered the most appropriate combination of antibiotic regimens recommended in the clinical setting.[ 28,29 ] Bajaj et al report the recent results of an open-label, randomized, standard of care (SOC) controlled study using fecal microbiota transplantation (FMT) in patients with recurrent HE. Twenty patients received a single FMT enema in combination with SOC and a 5-day antibiotic pre-treatment or SOC-only were followed for 150 days. During the follow-up, none of the HE episodes developed in the FMT group, compared with 5 (50%) in the SOC-only group.[ 30 ] The FMT group had cognitive function improvement and a lower rate of serious adverse event (20% vs 80%) than the SOC-only group. [ 30 ] Although FMT has shown promising results in the treatment of overt HE, stronger evidence on efficacy, long-term safety and which patients with HE should be considered are needed. A registry assessing short- and long-term patient outcomes after FMT has been developed in this regard. [ 31 ] A systematic review with meta-analysis synthesized 21 trials, 1420 participants and suggest that probiotics has a beneficial effect on HE symptoms recovery (10 trials, 574 participants, RR 0.67, 95% CI 0.56±0.79) and development of overt HE (10 trials, 585 participants, RR 0.29, 95% CI 0.16±0.51) comparing to no intervention. [ 32 ] Serum ammonia level was lower (10 trials, 705 participants, mean difference -8.29 umol/L, 95% CI -13.7± -3.41) and slightly imporved quality of life for pariticipants with probiotcs. There were no reliable difference between probiotics and lactulose.[ 32 ] Considerable efforts of research in the field of gut microbiota are needed for the future therapeutic use in overt HE. The overall mortality rate was high among cirrhotic patients with HE (48.6% of matched cases of HE and 51.26% of all HE cases) in the present study cohort, and 76% of them were exposed to infection before HE development. The present findings support other studies that have shown infections in cirrhotic patients with HE present additional risks of worsening outcomes.[33±35] For example, acquisition of infection during the mild stage of HE was significantly associated with progression to the advanced stage.[ 33 ] Moreover, recent reports have 10 / 14 shown that bacterial infections were significantly associated with increased overall mortality among patients with HE.[ 34,35 ] Most previous studies regarding the influence of the microbiota on LC were highly diverse in design and involved small numbers of patients in single-center settings.[ 14,36 ] One of the strengths of the present study was its relatively large sample size derived from a populationbased cohort, thus allowing for the investigation of independent associations between individual sites of infection and HE development. The advantages of longitudinal data have enabled us to further investigate the dose-response association between infection and risk of HE development. To date, the mechanism that could further induce HE development and deterioration of patient outcomes among those with LC remains unclear. Our data focusing on exposure to infection and the probability of the first episode of HE have relevant clinical implications for the detection and prevention of cirrhosis for high-risk patients. First, a synergistic effect of infections is likely to be important in developing HE. Prior exposure to infection and medication history is easy to obtain and should be identified during routine practice for those cirrhotic patients at higher risk for HE. In addition, recommended empirical therapy for bacterial infections should be closely monitored and justified according to the prevalence of resistant pathogens.[ 7 ] Few prophylactic therapy strategies are recommended for patients with previous episodes of SBP and variceal bleeding.[ 37 ] Studies focusing on who will benefit from prevention of the first HE episode and what should be included in the standard of care therapy are needed. Despite the diverse range of bacterial pathogens, it has been suggested that interactions between gut microbiota, etiology of chronic LC, and host-related precipitating factors (e.g., malnutrition, low protein ascites <1.5 g/dL, and alcohol consumption)[ 38 ] should be incorporated into an algorithm for standard of care therapy.[ 8 ] It is worthy to noting that acute kidney injury (AKI) is a recognized mortality predictor in cirrhotic patients with infection. The 30-day mortality was 10-fold higher in cirrhotic patients admitted with infection who developed AKI than those who did not. Advancing stages of AKI in cirrhotic patients were associated with a higher incidence of bacteremia, pneumonia and UTI, and cirrhosis-related encephalopathy and spontaneous bacterial peritonitis (SBP). [ 39 ] For caring cirrhotic patients with infection, factors associated with worsening HE should be identified and closely monitored, including diuretic therapy and hypokalemia which may facilitate the conversion of ammonium (NH4) to ammonia (+NH3) and lead to acute mental dysfunction.[ 26 ] Furthermore, due to kidney involved both excretion and production of ammonia, early detection of hypovolemia and AKI, and prevention of patients affected by this disease (e.g., infection, diuretic therapy) are imperative to better survival outcome.[ 26,39 ] This study was subject to certain limitations common to studies using claims data. First, laboratory results regarding the severity of LC and index HE are not available in the NHI dataset. Using the liver disease diagnosis as a proxy for the presence of decompensated cirrhosis (esophageal varices bleeding, ascites) at baseline and during years of follow-up can minimize the potential bias when estimating risk factors for the development of HE. In addition, there is a lack of bacterial culture results to determine microbial diversity and the effects of antimicrobial therapy for HE. We used the advanced stage HE diagnosis at hospital discharge, which could underestimate the incidence of mild infections without hospitalization; however, it ensured diagnostic infection specificity. Furthermore, other unmeasured or unknown health factors for HE development, including actual alcohol use, dietary protein intake, and medication use, among cirrhotic patients may result in a biased estimate of exposure to infection, thus limiting the generalizability of the current study results to different cirrhotic populations. 11 / 14 In conclusion, bacterial infections, especially SBP and unspecified peritonitis, sepsis, and biliary tract infection, are common and important predictors of the development of HE. Furthermore, a dose-response effect associated with an increased risk of HE highlights the need to promote appropriate infection prevention strategies and the use of antibiotic therapy for cirrhotic patients with different precipitating risks for HE. Additional research is needed regarding the associations between antibiotic therapy and the occurrence of HE or changes in its severity. Supporting information S1 Fig. Schematic flowchart of study design. (TIF) S2 Fig. All-cause mortality among cirrhotic patients by occurrence of hepatic encephalopathy. (TIF) S1 Table. Baseline characteristics of cirrhotic patient cohort before matching. (DOCX) Acknowledgments The authors thank Professor Yi-Hsin Yang and the staff of the Center for Medical Informatics and Statistics at Kaohsiung Medical University, a site of the Application of Health and Welfare Informatics, Ministry of Health and Welfare, in Taiwan for their suggestions and help with data analysis. Author Contributions Conceptualization: Lan-Ting Yuan, Seng-Kee Chuah, Chien-Ning Hsu. Data curation: Seng-Kee Chuah, Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, WeiChen Tai, Tsung-Hsing Hung, Seng-Howe Nguang, Jiunn-Wei Wang, Kuo-Lun Tseng, Ming-Kun Ku, Pin-I Hsu, Deng-Chyang Wu, Chien-Ning Hsu. Formal analysis: Chien-Ning Hsu. Investigation: Seng-Kee Chuah, Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, Wei Chen Tai. Methodology: Chien-Ning Hsu. Project administration: Seng-Kee Chuah, Pin-I Hsu, Deng-Chyang Wu. Resources: Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, Wei-Chen Tai, TsungHsing Hung, Seng-Howe Nguang, Jiunn-Wei Wang, Kuo-Lun Tseng, Ming-Kun Ku, Pin-I Hsu, Deng-Chyang Wu. Supervision: Chien-Ning Hsu. Validation: Seng-Kee Chuah, Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, WeiChen Tai, Tsung-Hsing Hung, Seng-Howe Nguang, Jiunn-Wei Wang, Kuo-Lun Tseng, Ming-Kun Ku, Pin-I Hsu, Deng-Chyang Wu. Visualization: Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, Wei-Chen Tai, Tsung Hsing Hung, Seng-Howe Nguang, Jiunn-Wei Wang, Kuo-Lun Tseng, Ming-Kun Ku. 12 / 14 Writing ± original draft: Lan-Ting Yuan. Writing ± review & editing: Seng-Kee Chuah, Pin-I Hsu, Deng-Chyang Wu, Chien-Ning Hsu. 13 / 14 1. Li YY , Nie YQ , Sha WH , Zeng Z , Yang FY , Ping L , et al. ( 2004 ) Prevalence of subclinical hepatic encephalopathy in cirrhotic patients in China . World J Gastroenterol 10 : 2397 ± 2401 . https://doi.org/10. 3748/wjg. v10.i16.2397 PMID: 15285027 2. Michitaka K , Tokumoto Y , Uesugi K , Kisaka Y , Hirooka M , Konishi I , et al. ( 2008 ) Neuropsychiatric dysfunction in patients with chronic hepatitis and liver cirrhosis . Hepatol Res 38 : 1069 ± 1075 . https://doi. org/10.1111/j.1872- 034X . 2008 . 00374 . x PMID : 19000057 3. Muto Y , Sato S , Watanabe A , Moriwaki H , Suzuki K , Kato A , et al. ( 2005 ) Effects of oral branched-chain amino acid granules on event-free survival in patients with liver cirrhosis . Clin Gastroenterol Hepatol 3 : 705 ± 713 . PMID: 16206505 4. Stepanova M , Mishra A , Venkatesan C , Younossi ZM ( 2012 ) In-hospital mortality and economic burden associated with hepatic encephalopathy in the United States from 2005 to 2009 . Clin Gastroenterol Hepatol 10 : 1034 ± 1041 e1031. https://doi.org/10.1016/j.cgh. 2012 . 05 .016 PMID: 22642955 5. Poordad FF ( 2007 ) Review article: the burden of hepatic encephalopathy . Aliment Pharmacol Ther 25 Suppl 1 : 3 ± 9 . 6. Bajaj JS , Wade JB , Gibson DP , Heuman DM , Thacker LR , Sterling RK , et al. ( 2011 ) The multi-dimensional burden of cirrhosis and hepatic encephalopathy on patients and caregivers . Am J Gastroenterol 106 : 1646 ± 1653 . https://doi.org/10.1038/ajg. 2011 .157 PMID: 21556040 7. Fernandez J , Gustot T ( 2012 ) Management of bacterial infections in cirrhosis . J Hepatol 56 Suppl 1 : S1 ± 12 . 8. Hepatic encephalopathy in chronic liver disease: 2014 practice guideline by the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases . J Hepatol 61 : 642 ± 659 . https://doi.org/10.1016/j.jhep. 2014 . 05 .042 PMID: 25015420 9. Merli M , Lucidi C , Giannelli V , Giusto M , Riggio O , Falcone M , et al. ( 2010 ) Cirrhotic patients are at risk for health care-associated bacterial infections . Clin Gastroenterol Hepatol 8 : 979 ± 985 . https://doi.org/ 10.1016/j.cgh. 2010 . 06 .024 PMID: 20621200 10. Neithercut WD , Rowe PA , el Nujumi AM , Dahill S , McColl KE ( 1993 ) Effect of Helicobacter pylori infection on intragastric urea and ammonium concentrations in patients with chronic renal failure . J Clin Pathol 46 : 544 ± 547 . PMID: 8331178 11. Ito S , Kohli Y , Kato T , Abe Y , Ueda T ( 1994 ) Significance of ammonia produced by Helicobacter pylori . European journal of gastroenterology & hepatology 6 : 167± 174 . 12. Mokuolu AO , Sigal SH , Lieber CS ( 1997 ) Gastric juice urease activity as a diagnostic test for Helicobacter pylori infection . Am J Gastroenterol 92 : 644 ± 648 . PMID: 9128315 13. Albrecht J ( 2005 ) Astrocytes in ammonia neurotoxicity: A target, a mediator and a shield . The Role of Glia in Neurotoxicity: 329 ± 342 . 14. Schulz C , Schutte K , Malfertheiner P ( 2014 ) Does H. pylori eradication therapy benefit patients with hepatic encephalopathy?: systematic review . J Clin Gastroenterol 48 : 491 ± 499 . https://doi.org/10. 1097/ MCG .0000000000000108 PMID: 24583758 15. Assy N , Rosser BG , Grahame GR , Minuk GY ( 1999 ) Risk of sedation for upper GI endoscopy exacerbating subclinical hepatic encephalopathy in patients with cirrhosis . Gastrointest Endosc 49 : 690 ± 694 . PMID: 10343210 16. Arroyo V , Jimenez W ( 2000 ) Complications of cirrhosis. II. Renal and circulatory dysfunction. Lights and shadows in an important clinical problem . J Hepatol 32 : 157 ± 170 . PMID: 10728802 17. Gustot T , Durand F , Lebrec D , Vincent JL , Moreau R ( 2009 ) Severe sepsis in cirrhosis . Hepatology 50 : 2022 ± 2033 . https://doi.org/10.1002/hep.23264 PMID: 19885876 18. National Health Insurance Research Database. LHID2000.Taiwan. 20. Wen CP , Tsai SP , Chung WS ( 2008 ) A 10-year experience with universal health insurance in Taiwan: measuring changes in health and health disparity . Ann Intern Med 148 : 258 ± 267 . PMID: 18283203 Wang JW , Hsu CN , Tai WC , Ku MK , Hung TH , Tseng KL , et al. ( 2016 ) The Association of Helicobacter pylori Eradication with the Occurrences of Chronic Kidney Diseases in Patients with Peptic Ulcer Diseases . PLoS One 11 : e0164824 . 21. Hsu PI , Lai KH , Lin CK , Chen WC , Yu HC , Cheng JS , et al. ( 2005 ) A prospective randomized trial of esomeprazole- versus pantoprazole-based triple therapy for Helicobacter pylori eradication . Am J Gastroenterol 100 : 2387 ± 2392 . https://doi.org/10.1111/j.1572- 0241 . 2005 . 00264 . x PMID : 16279889 22. Vilstrup H , Amodio P , Bajaj J , Cordoba J , Ferenci P , Mullen KD , et al. ( 2014 ) Hepatic encephalopathy in chronic liver disease: 2014 Practice Guideline by the American Association for the Study of Liver Diseases and the European Association for the Study of the Liver . Hepatology 60 : 715 ± 735 . https://doi. org/10.1002/hep.27210 PMID: 25042402 23. Lo Re V 3rd, Carbonari DM , Forde KA , Goldberg D , Lewis JD , Haynes K , et al. ( 2015 ) Validity of diagnostic codes and laboratory tests of liver dysfunction to identify acute liver failure events . Pharmacoepidemiol Drug Saf 24 : 676 ± 683 . https://doi.org/10.1002/pds.3774 PMID: 25866286 24. GMATCH Algorithm SAS Code ReferenceÐSAS paper 214 ± 26 Reducing Bias in a Propensity Score Matched-Pair Sample Using Greedy Matching Techniques . 25. SAS/STAT(R) 9.3 User's Guide. The PHREG Procedure . Model using time-dependent explanatory variable . 26. Prakash R , Mullen KD ( 2010 ) Mechanisms, diagnosis and management of hepatic encephalopathy . Nat Rev Gastroenterol Hepatol 7 : 515 ± 525 . https://doi.org/10.1038/nrgastro. 2010 .116 PMID: 20703237 27. Dasani BM , Sigal SH , Lieber CS ( 1998 ) Analysis of risk factors for chronic hepatic encephalopathy: the role of Helicobacter pylori infection . Am J Gastroenterol 93 : 726 ± 731 . https://doi.org/10.1111/j.1572- 0241 . 1998 . 214_a . x PMID: 9625117 28. Chey WD , Leontiadis GI , Howden CW , Moss SF ( 2017 ) ACG Clinical Guideline: Treatment of Helicobacter pylori Infection . Am J Gastroenterol 112 : 212 ± 239 . https://doi.org/10.1038/ajg. 2016 .563 PMID: 28071659 29. Sheu BS , Wu MS , Chiu CT , Lo JC , Wu DC , Liou JM , et al. ( 2017 ) Consensus on the clinical management, screening-to-treat, and surveillance of Helicobacter pylori infection to improve gastric cancer control on a nationwide scale . Helicobacter 22 . 30. Bajaj JS , Kassam Z , Fagan A , Gavis EA , Liu E , Cox IJ , et al. ( 2017 ) Fecal microbiota transplant from a rational stool donor improves hepatic encephalopathy: A randomized clinical trial . Hepatology 66 : 1727 ± 1738 . https://doi.org/10.1002/hep.29306 PMID: 28586116 31. Kelly CR , Kim AM , Laine L , Wu GD ( 2017 ) The AGA's Fecal Microbiota Transplantation National Registry: An Important Step Toward Understanding Risks and Benefits of Microbiota Therapeutics . Gastroenterology 152 : 681 ± 684 . https://doi.org/10.1053/j.gastro. 2017 . 01 .028 PMID: 28143773 32. Dalal R , McGee RG , Riordan SM , Webster AC ( 2017 ) Probiotics for people with hepatic encephalopathy . Cochrane Database Syst Rev 2 : CD008716 . 33. Vaquero J , Polson J , Chung C , Helenowski I , Schiodt FV , Reisch J , et al. ( 2003 ) Infection and the progression of hepatic encephalopathy in acute liver failure . Gastroenterology 125 : 755 ± 764 . PMID: 12949721 34. Dionigi E , Garcovich M , Borzio M , Leandro G , Majumdar A , Tsami A , et al. ( 2017 ) Bacterial Infections Change Natural History of Cirrhosis Irrespective of Liver Disease Severity . Am J Gastroenterol 112 : 588 ± 596 . https://doi.org/10.1038/ajg. 2017 .19 PMID: 28220780 35. Hung TH , Lay CJ , Chang CM , Tsai JJ , Tsai CC , Tsai CC ( 2013 ) The effect of infections on the mortality of cirrhotic patients with hepatic encephalopathy . Epidemiol Infect 141 : 2671 ± 2678 . https://doi.org/10. 1017/S0950268813000186 PMID: 23433300 36. Piotrowski D , Boron-Kaczmarska A ( 2017 ) Bacterial infections and hepatic encephalopathy in liver cirrhosis-prophylaxis and treatment . Adv Med Sci 62 : 345 ± 356 . https://doi.org/10.1016/j.advms. 2016 . 11 . 009 PMID: 28514703 37. Fernandez J , Ruiz del Arbol L , Gomez C , Durandez R , Serradilla R , Guarner C , et al. ( 2006 ) Norfloxacin vs ceftriaxone in the prophylaxis of infections in patients with advanced cirrhosis and hemorrhage . Gastroenterology 131 : 1049 ±1056; quiz 1285. https://doi.org/10.1053/j.gastro. 2006 . 07 .010 PMID: 17030175 38. Cordoba J , Ventura-Cots M , Simon-Talero M , Amoros A , Pavesi M , Vilstrup H , et al. ( 2014 ) Characteristics, risk factors, and mortality of cirrhotic patients hospitalized for hepatic encephalopathy with and without acute-on-chronic liver failure (ACLF) . J Hepatol 60 : 275 ± 281 . https://doi.org/10.1016/j.jhep. 2013 . 10 .004 PMID: 24128414 39. Wong F , O'Leary JG , Reddy KR , Patton H , Kamath PS , Fallon MB , et al. ( 2013 ) New consensus definition of acute kidney injury accurately predicts 30-day mortality in patients with cirrhosis and infection . Gastroenterology 145 : 1280 ± 1288 e1281. https://doi.org/10.1053/j.gastro. 2013 . 08 .051 PMID: 23999172


This is a preview of a remote PDF: http://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0197127&type=printable

Lan-Ting Yuan, Seng-Kee Chuah, Shih-Cheng Yang, Chih-Ming Liang, Cheng-Kun Wu, Wei-Chen Tai, Tsung-Hsing Hung, Seng-Howe Nguang, Jiunn-Wei Wang, Kuo-Lun Tseng, Ming-Kun Ku, Pin-I Hsu, Deng-Chyang Wu, Chien-Ning Hsu. Multiple bacterial infections increase the risk of hepatic encephalopathy in patients with cirrhosis, PLOS ONE, 2018, DOI: 10.1371/journal.pone.0197127