In-hospital mortality of hepatorenal syndrome in the United States: Nationwide inpatient sample

Wisit Kaewput, Charat Thongprayoon, Carissa Y Dumancas,Swetha R Kanduri, Karthik Kovvuru, ChalermratKaewput, Pattharawin Pattharanitima, Tananchai Petnak, Ploypin Lertjitbanjong, Boonphiphop Boonpheng,Karn Wijarnpreecha, Jose L Zabala Genovez, Saraschandra Vallabhajosyula, Caroline C Jadlowiec, FawadQureshi,Wisit Cheungpasitporn

Abstract

Key Words: Hepatorenal syndrome; Liver transplantation; Mortality; Nationwide; Big data;Hospitalization; Outcomes; Predictors

INTRODUCTION

Hepatorenal syndrome (HRS) is a serious complication of cirrhosis with an incidence as high as 32 % among patients with advanced liver disease[1 -7 ]. Previous studies have consistently demonstrated high morbidity, mortality, and resource utilizations[1 ,8 -17 ].Several factors have been associated with poor outcomes, including high model for end stage liver disease (MELD) score[18 ], degree of acute kidney injury (AKI)[11 ,19 ],extrahepatic organ failure[20 ], and sepsis[18 ,21 ].

In recent decades, there have been significant advances in knowledge, treatment,and optimal management of patients with HRS[1 ,7 -17 ,22 -24 ]. While terlipressin, a synthetic vasopressin analog with predominant vasopressin 1 A receptor effect[25 ], has been used to treat HRS in many Asian and European countries, it is currently not yet available in the United States for the treatment of HRS[1 ,26 ]. Thus, currently available treatment options for HRS in the United States include albumin volume expansion,octreotide with or without midodrine, and intravenous cardiovascular medications like vasopressin and norepinephrine[1 ]. Nevertheless, there have been improvements in the overall care for patients with HRS, including liver transplantation and renal replacement therapy. In addition, several studies have suggested the use of transjugular intrahepatic portosystemic shunt (TIPS) for patients with HRS[27 -29 ]. However, data specific to HRS, hospital mortality trends, and hospital admission resource utilization remain limited.

In this study, we aimed to evaluate patient characteristics, in-hospital treatments,mortality, and resource utilization during hospital admissions for HRS in the United States. We also assessed the temporal trend in mortality and identified the predictors for mortality.

MATERIALS AND METHODS

Study population

We conducted a retrospective cohort study of hospital admissions for HRS from 2005 to 2014 in the national inpatient sample (NIS) database. The detail of the NIS database was previously described[30 ]. We identified hospital admission with a primary discharge diagnosis using the international classification of disease-9 (ICD-9 ) diagnosis code of 572 .4 . The Mayo Clinic institutional review board approved this study (IRB number 21 -007353 and date of approval; July 27 , 2021 ) and exempted the need for informed consent because the data in NIS database was publicly available and deidentified.

Data collection

We abstracted patient and hospital characteristics, procedures, outcomes, and resource utilization from the database (Supplementary Table 1 ). Patient characteristics included age, sex, race, etiology of liver disease, medical comorbidity based on Elixhauser index[31 ], and admission day. Hospital characteristics included hospital size, ownership,location, teaching status, and region. Procedures included renal replacement therapy,liver transplantation, TIPS, abdominal paracentesis, and mechanical ventilation.Outcomes included hospital mortality, resource utilization, including length of hospital stay, and hospitalization cost. Since this study used data over 10 different calendar years, we adjusted hospitalization costs for inflation using the consumer price index and converted them to 2014 United States dollar equivalents.

Statistical analysis

The NIS database contains hospitalization data from a stratified sample of 20 % of hospitals in the United States. As such, we used discharge weight provided by the Healthcare Cost and Utilization (HCUP) to estimate the total number of hospital admissions for HRS. We used descriptive statistics to summarize patient and hospital characteristics, procedures, outcomes, and resource use of HRS admission. We fitted logistic regression model for hospital mortality and liver transplantation, and standard least square linear regression for length of hospital stay, and hospitalization cost, using calendar years as the independent variable to assess the annual trend from 2005 through 2014 . We estimated adjusted odds ratio (OR) for hospital mortality from multivariable multi-level mixed effect logistic regression, employing hospital identification number as random effect with patients-level characteristics clustered within hospital-level characteristics. We performed all statistical analyses using STATA,version 15 (StataCorp LP, College Station, TX, United States).

RESULTS

Patient characteristics, in-hospital treatments, outcomes, and resource use in hospital admission for HRS

There were 4938 hospital admissions with HRS as the primary diagnosis in the unweighted sample and 23973 admissions in the weighted sample. Table 1 shows patient and hospital characteristics of hospital admissions for HRS. The mean age was 58 .8 ± 12 .3 years, and the majority of patients were males (63 %). Alcohol-related liver disease (46 %) and viral hepatitis (25 %) were the most common liver disease etiologies.Most patients were admitted to large urban teaching hospitals. Of those patients admitted for HRS, 21 % received renal replacement therapy and 2 % underwent liver transplant during their hospitalization. During this 10 -year period, there was a 32 %mortality observed for HRS admissions. The mean length of hospital stay was 8 .8 d and the mean hospitalization cost was 73731 United States dollars.

Trends in hospital mortality, liver transplantation, length of stay, hospitalization cost in hospital admission for HRS

Table 2 showed the annual trend in hospital mortality, liver transplantation, length of hospital stay, and hospitalization cost in HRS admissions from 2005 to 2014 .

There was a decreasing trend in hospital mortality from 44 % in 2005 to 24 % in 2014 among hospital admissions for HRS in the United States (OR: 0 .92 , 95 %CI: 0 .90 -0 .94peryear;P＜ 0 .001 ) (Figure 1 A).

Meanwhile, there was an increase in the rate of liver transplantation (OR: 1 .11 ,95 %CI: 1 .02 -1 .20 per year; P = 0 .02 ) (Figure 1 B) and renal replacement therapy (OR:1 .05 , 95 %CI: 1 .02 -1 .08 per year; P ＜ 0 .001 ) (Figure 1 C) performed in hospitalization for HRS.

There was an increasing trend in mean length of hospital stay (coefficient estimate 0 .2 d per year; P ＜ 0 .001 ) (Figure 1 D) and hospitalization cost (coefficient estimate 5778 United States dollarsperyear;P＜ 0 .001 ) (Figure 1 E) among hospitalization for HRS during 10 -year period from 2005 to 2014 .

Predictors for hospital mortality

In multivariable analysis (Table 3 ), older age (OR: 1 .45 for 40 -59 years, 1 .77 for 60 -79 years, 2 .12 for ≥ 80 years, compared to 18 -39 years; all P ＜ 0 .001 ), alcohol use (OR: 1 .35 ;P＜ 0 .001 ), coagulopathy (OR: 1 .15 ; P = 0 .001 ), and presence of a neurological disorder(OR: 1 .38 ; P ＜ 0 .001 ) predicted higher hospital mortality.

春夏之交是采摘柳蒿芽的最隹之際，婦女們成群結(jié)隊(duì)來到江河邊采摘柳蒿芽，休息之余還要邊歌邊舞，抒發(fā)對美好生活和大自然的愛戀情懷。采摘柳蒿芽除了嘗鮮之外，其余大部分要曬干或用開水焯了以后，放入冰箱儲存起來，待日后食用。過去災(zāi)荒年青黃不接時，是柳蒿芽采添飽了達(dá)斡爾人饑腸轆轆的腸胃，拯救了達(dá)斡爾人的生命，從此，達(dá)斡爾人同柳蒿芽—庫木勒結(jié)下了難以割舍的情緣。遠(yuǎn)居他鄉(xiāng)的達(dá)斡爾人，總是把家鄉(xiāng)帶回或其它方式捎來的柳蒿芽視為珍品隹肴，每每品嘗都會增添思念家鄉(xiāng)的情愫，拉近同家鄉(xiāng)的距離。

Need for mechanical ventilation (OR: 9 .24 ; P ＜ 0 .001 ) was associated with higher mortality, whereas liver transplantation (OR: 0 .15 ; P ＜ 0 .001 ) and TIPS (OR: 0 .23 ;P＜0 .001 ), and abdominal paracentesis (OR: 0 .48 ; P ＜ 0 .001 ) were associated with lower hospital mortality. Renal replacement therapy was not significantly associated with mortality risk.

DISCUSSION

In this study based on a large United States database of hospitalizations, the mortality rate for hospitalized patients with HRS decreased by approximately 50 % during the 10 -year study period. During the same period, there was a 2 -fold increase in the incidence of HRS patients receiving a liver transplant and the incidence of in-hospital renal replacement therapy increased by 60 %. Notably, there were also increase in length of hospital stay and a 2 -fold increase in the estimated hospital cost, which is likely related to higher utilization of healthcare resources. This highlights the high economic burden of chronic liver disease in the United States[32 ,33 ].

The marked improvement in the in-hospital mortality rate for HRS is likely reflective of changes in both medical and surgical management during the study period. Our study shows that there was an apparent increase in the number of liver transplants and renal replacement therapy around 2007 to 2008 . This trend coincided with overall changes in clinical practice over the preceding years[34 ]. Although the unique pathophysiology of HRS has long been recognized as a functional renal failure occurring as a result of advanced liver disease[35 ], its treatment, including the initiation of in-hospital dialysis, and the role for liver transplantation have significantly evolved[36 ]. Historically, the initiation of renal replacement therapy in patients with HRS was felt to be controversial and futile. Increasing experience with liver transplantation in the setting of HRS as well as improved access to continuous renalreplacement have resulted in a change in practice and a decrease in mortality[37 ].

Table 1 Patient characteristics, in-hospital treatments, outcomes, and resource use in hospital admission for hepatorenal syndrome(mean ± SD)

SE: Standard error; HIV/AIDS: Human immunodeficiency virus/acquired immunodeficiency syndrome; TIPS: Transjugular intrahepatic portosystemic shunt; LTA: Liver transplant alone; SLKT: Simultaneous liver-kidney transplantation.

In 2007 -2008 , multiple randomized control trials on terlipressin were published and have influenced the medical management HRS as well as patient outcomes[9 -11 ,14 ,15 ]. Studies have shown potential beneficial effects of terlipressin, a potent selective splanchnic and extrarenal vasoconstrictor, on kidney function among patients with HRS[10 ,38 ,39 ]. Additionally, non-response to vasoconstrictors can also predict HRS mortality[40 ,41 ]. Unfortunately, as of 2020 , the FDA has not yet approved the use of terlipressin for HRS in the United States. Results from the phase 3 trial terlipressin did not show any significant survival benefit and its use was associated with adverse events, such as respiratory failure[42 ,43 ]. Although terlipressin is currently not yet available in the United States[1 ,26 ], the observed findings of decreasing mortality trends for HRS in the Unites States are likely due to improvements in healthcare,increased access and acceptance of chronic intermittent hemodialysis for patients with liver disease as well as increased acceptance of liver transplantation for patients with acute decompensation[44 ].

In addition to liver transplantation, our study interestingly showed that TIPS and abdominal paracentesis were associated with lower hospital mortality among patients with HRS. Possible mechanisms underlying reduced mortality among patients who received paracentesis were that those who had abdominal paracentesis received moreaggressive treatments such as albumin and vasopressors, TIPS, and liver transplantation than those who received palliative care. Furthermore, abdominal paracentesis may have led to the diagnosis and treatment for spontaneous bacterial peritonitis[45 ].The use of TIPS in patients with HRS remains controversial, although there is increasing data suggesting there may be benefit[24 ,29 ]. According to current best practice recommendations, the presence of HRS is not an absolute contraindication for TIPS and the presence of other indications, such as ascites, should guide decision making[29 ]. Specific to this topic, there is a clear need for additional randomized controlled trials, however, in the interim, there are an increasing number of small studies demonstrating positive outcomes in select HRS patients receiving TIPS[24 ,46 ,47 ]. Since mortality in patients with HRS undergoing TIPS is driven mainly by poor liver function it may be possible that there was a population selection bias and these patients had initially better liver function resulting in better survival.

Table 2 The annual trend in hospital mortality, liver transplantation, renal replacement therapy, length of hospital stay, hospitalization cost in hepatorenal syndrome admission from 2005 to 2014 (mean ± SD)

Our study also showed several risk factors associated with in-patient mortality for HRS. These factors include advanced age, history of alcohol use, coagulopathy and presence of a neurological disorder. It is well known that older age, coagulopathy, and neurological disorder are associated with poor outcomes in patients with HRS[11 ,18 -21 ]. Hepatic encephalopathy is known to be associated with mortality[48 ], and thus this could be the underlying reason for association between neurological disorder and increased in-patient mortality for HRS. Although specific knowledge regarding the duration and timing of alcohol use prior to hospitalization is a limitation of this dataset, active alcohol use is a known decompensating event that can result in AKI and HRS. It is also possible that recent alcohol use prevented certain patients from being suitable for liver transplantation. In this foreseeable scenario, initiation of renal replacement therapy has increasingly been used as a bridge to liver transplant eligibility and liver compensation.

There are several limitations in our study. The NIS is a hospitalized database. Thus,we did not evaluate the long-term outcomes of HRS following hospitalization.Although our study showed a decreasing trend of in-hospital mortality rates, it should not be generalized to the overall survival of patients with HRS. Estimates of inhospital mortality do not include deaths that occur after discharge. The database did not contain MELD score, which predicted mortality in HRS patients[48 ]. In addition,treatment of HRS was not assessed in this study[40 ,41 ]. Data on medications including midodrine, octreotide, vasopressor, albumin infusion were not available in the database. Thus, we could not assess the effects of these agents and the response totreatments on the outcomes of HRS. Lastly, HRS was identified by ICD-9 diagnosis code. Given definition of the HRS has changed over the years, these changes in definition may have affected the incidence of HRS in our study overtime.

Table 3 Clinical characteristics associated with in-hospital mortality

HIV/AIDS: Human immunodeficiency virus/acquired immunodeficiency syndrome; TIPS: Transjugular intrahepatic portosystemic shunt; OR: Odds ratio;95 %CI: 95 % confidence interval.

Figure 1 Data on admissions in the United States due to hepatorenal syndrome. A: Decreasing trend in hospital mortality among hospital admissions;B: Increase in the rate of liver transplantation among hospital admissions; C: Trend of renal replacement therapy among hospital admissions; D: Trend of mean length of hospital stay among hospital admissions; E: Trend of hospitalization cost among hospital admissions. NIS: National inpatient sample.

CONCLUSION

In summary, our study showed a decreasing trend of in-hospital mortality rates in patients with HRS. These trends were likely related to advances in medicine, increased access and acceptance of renal replacement therapy, and increased utilization of liver transplantation which is the definitive treatment for HRS. Future studies are needed to understand if these trends are impacted by other factors such as facility performance,patient care teams, health insurance reimbursement policies, or other factors.

ARTICLE HIGHLIGHTS