Heart failure (HF) and liver cirrhosis as separate disease entities contribute to significant morbidity and mortality in the United States and worldwide [1, 2]. Cirrhosis may influence HF outcomes as it promotes a hyperdynamic state, neurohormonal activation which can contribute to both systolic and diastolic myocardial dysfunction [3]. Furthermore, both liver cirrhosis and HF can independently lead to poor renal outcomes as reflected in the conditions of hepatorenal syndrome (HRS) and cardiorenal syndrome (CRS) [3, 4]. In hepatorenal syndrome, increased splanchnic blood flow with decreased central volume accompanied by neurohormonal activation leads to vasoconstriction decreasing renal blood flow and glomerular filtration rate [5]. However, more recently, this conventional liver-kidney or liver-heart crosstalk has been challenged by emerging evidence accentuating the role of the cardiac dysfunction as a potential mediator of renal impairment in liver cirrhosis [3, 4]. Interactions between the liver, heart, and kidney in the setting of CRS and HRS share several cardinal mechanistic pathways [5, 6], but clinical outcomes brought about by these interactions remain understudied.

We conducted a cross sectional analysis of nationally representative sample of hospitalizations for decompensated HF to assess the inpatient outcomes including all-cause mortality, acute kidney injury (AKI) and length of stay (LOS) duration among those with relative to without comorbid cirrhosis.

We conducted a cross sectional analysis of the 2010 thru 2014 years of the National Inpatient Sample (NIS). Briefly, the NIS of the Healthcare Cost and Utilization Project (HCUP), sponsored by the Agency for Healthcare Research and Quality (AHRQ), is a survey design-based database of discharge data for inpatient care from non-federal (excludes Veterans Hospitals and other federal facilities), non-rehabilitation, acute-care, short-term hospitals. The NIS is an annual sample of hospital discharges that provides national estimates of the characteristics of the patients, diagnoses, and hospital-based procedures performed in US acute-care hospitals. Annually NIS includes approximately 7 million hospital discharges. Detailed information on the NIS can be found on its website (https://www.hcup-us.ahrq.gov/databases.jsp).

In our study sample of interest, hospitalizations for decompensated HF, was identified using the hospitalization Diagnosis-Related Group (DRG) including 291, 292, and 293. Excluded were those aged younger than 18 years, and those with hemodialysis (ICD 9: 39.95, V45.1, V56.0, V56.1) or end stage renal disease (ICD-9: 585.6). HF subtype (diastolic, systolic, and combined systolic and diastolic) and comorbid conditions including cirrhosis (alcoholic and non-alcoholic), chronic kidney disease (CKD), cerebrovascular disease, chronic obstructive pulmonary disease, hypertension, diabetes mellitus, and cancer were extracted using their respective ICD-9 codes (see Supplemental Table 1). Target LOS (tLOS) was defined as 2.8 days (d) for hospitalization as determined by median time to discharge by Medicare level DRG (DRG 293, 4.0 d for DRG 292, and 5.2 d for DRG of 291).

The study outcomes included all-cause inpatient mortality, acute kidney injury (AKI), and length of stay (LOS) duration shorter than the DRG-level Medicare tLOS.

Descriptive and inferential statistics was calculated while incorporating the survey design features of the NIS data in order to provide population-based estimates. Descriptive statistics was calculated for categorical and continuous variables. Associations between categorical variables was determined using the design-adjusted Rao-Scott chi-square test. Survey logistic regression provided the odds of study outcome among those with related to without cirrhosis adjusting for age, gender, and additional comorbidities including CKD, cerebrovascular disease, chronic obstructive pulmonary disease, hypertension, diabetes mellitus, and cancer. Unadjusted and adjusted logistic regression by HF subtype provided the odds of study outcome among those with relative to without cirrhosis. A significance level of 0.05 with a 2-sided test was used for all hypotheses. All statistical analyses were conducted using SAS Statistical Software (SAS [9.4] Software, SAS Institute Inc., Cary, NC, USA).

This analysis is, to the best of our knowledge, the first observational report from a nationally representative database on clinical outcomes in patients with hospitalized HF with/without cirrhosis including AKI, mortality and impact on hospital LOS. The presence of underlying cirrhosis increases mortality risk in patients with HF. A recent analysis by Khalid et al. [7] showed an in-patient mortality rate of 3.4% among HF hospitalized patients with cirrhosis. Cirrhosis leads to a chronic hyperdynamic circulatory state coupled with low systemic vascular resistance, and provides a backdrop for independent cardiomyopathy from the risk factors for the etiology of cirrhosis in itself (such as NASH/amyloid/hemochromatosis) [8]. In the compensated state, the hemodynamic derangements predominantly in the form of arterial vasodilation are counterbalanced by the augmentation of cardiac output [9]. However, in the decompensated state, cardiac output may be inadequate to compensate for the much more pronounced systemic vascular resistance reduction that eventually leads to central hypovolemia and arterial hypotension, feeding into vicious cycle of sympathetic, renin-angiotensin-aldosterone, and vasopressin systems activation common to both cirrhosis as well as HF [10]. Furthermore, due to hepatic dysfunction, metabolism of drugs such as those used in HF may be impaired potentially limiting the use of guideline directed medical therapy in this specific patient population that may affect subsequent clinical outcomes [11]. This particular cohort of patients with “hepato-cardiac” disease represent a potentially high-risk group for poor outcomes when hospitalized for HF, given the close crosstalk between HF and cirrhosis, with several common aberrant, maladaptive pathways and possible limitations to therapy.

Cirrhosis and HF are independent drivers of worsening kidney function in their respective decompensated states vis a vis hepatorenal syndrome (HRS) and cardiorenal syndrome (CRS) respectively [4, 5, 6]. Interestingly, interactions between these organs share cardinal mechanistic pathways including neurohormonal activation and endothelial dysfunction [12]. In decompensated cirrhosis cardiomyopathy, the massive renal vasoconstriction coupled with reduced cardiac output adversely affects renal function [4, 13, 14]. A proof of concept study in 2003 examined cardiac output during and after resolution of spontaneous bacterial peritonitis in cirrhotic patients [15]. Eight of the patients who developed HRS were found to have lower cardiac output at baseline and declined further at infection resolution, compared with the fifteen who had normal function. This is the first clear evidence that cardiac dysfunction may play a collateral pathogenic role in HRS, with the use of the term “hepatocardiorenal syndrome” in the literature gaining recent traction [4]. Similarly, congestive hepatopathy in the setting of decompensated HF from elevated right atrial pressures adds the risk of further decompensation of liver function, with secondary effects on worsening kidney function [16, 17, 18]. Thus, the interplay of “hepato-cardiac” disease in those with HF and underlying cirrhosis is potentially a major driver for poor kidney outcomes as well as higher mortality and LOS with its attendant health care resource utilization, as shown in this analysis.

In this analysis, the impact of underlying cirrhosis and the development of worsening kidney function did not have a differential effect in those with HFrEF vs HFpEF. This is of clinical importance, as EF per se may not be the primary determinant of outcomes with the maladaptive “hepato-cardio-renal-axis” [4]. In fact, LV systolic dysfunction may be latent in cirrhosis, and may require provocative testing to identify the presence of cardiomyopathy [19]. Thus, the mere preservation of EF should not be a deterrent to a more nuanced approach to diagnosing cardiomyopathy in patients with cirrhosis, especially with the consideration to be able to modify the disease trajectory with appropriate guideline directed medical therapies for HF early in the course, given the high burden of mortality and resource utilization in the decompensated state, as shown in this cross-sectional analysis.

Our study has several limitations. The use of an administrative dataset and the cross-sectional nature of our study design limit the ability to capture patient level data. Other limitations include the lack of information in the database regarding the severity of the conditions described lack of an indicator for whether a condition was present on admission or not, and non-availability a disease state relevant to the outcome of interest, and potential residual confounding. Also, we relied on hospitalization DRG to identify the main indication for the hospitalization to have been for HF which allows for administrative error. We were not able to look at the influence of potential suboptimal use of guideline directed medical therapy given the more complex hemodynamic nature of the interplay of cirrhosis and heart failure with potentially lower blood pressure ranges. Nevertheless, these results describe the high burden of poor outcomes in patients with “cardio-hepatic” disease and its attendant costs on the health system, and are hypothesis generating towards design of future studies to alter the trajectory of this axis with the recognition of high-risk phenotypes and institution of appropriate guideline based medical therapies.

Patients hospitalized with HF and underlying cirrhosis may be at higher risk for poorer clinical outcomes. Prospective studies with control for potential confounders may better delineate this association.

All authors were involved with the conceptualization and the design of the study. AY was involved with data extraction and analysis. AY, MHM, JP, JS, EQ, KBL were involved with the first manuscript draft. ROM and JR served as the supervising senior authors who edited at every phase of the manuscript including multiple revisions. All authors reviewed and approved the final form of the manuscript submitted.

This research has been performed on the National Inpatient Sample as part of the databases and software tools developed for the Healthcare Cost and Utilization Project (HCUP). This research project was reviewed and approved by the Institutional IRB under expedited review due to its nature of a database study with no patient identifiers involved and no informed consent involved. IRB number: IRB-2021-602.

There are no acknowledgements to disclose.

This research received no external funding.

The authors declare no conflict of interest.