Predictors of inpatient outcomes of COVID‐19 infection in patients with cirrhosis in the early pandemic phase: A nationwide survey

Abstract Background and Aim Previous studies conducted at single centers have suggested that patients with cirrhosis are at a greater risk for worse outcomes with COVID‐19. However, there is limited data on a national level in the United States. We aimed to study hospital‐related outcomes and identify the predictors of poor outcomes in patients with cirrhosis and concurrent COVID‐19. Methods We queried 2020 National Inpatient and Readmission databases to identify all hospitalizations due to cirrhosis in adults with a diagnosis of COVID‐19. Primary outcomes included inpatient mortality, mechanical ventilation (MV), and intensive care unit (ICU) utilization. Secondary outcomes included mean length of stay (LOS) and mean hospitalization costs. We classified cirrhosis into compensated (CC) and decompensated (DC) groups. Results We identified 25194 hospitalizations of adult patients due to cirrhosis with a concurrent diagnosis of COVID‐19. These patients had higher mortality (19.50% vs 6.19%, P ≤ 0.01), MV (11.7% vs 2.8%, P ≤ 0.01), ICU utilization (17.3% vs 8.1%, P ≤ 0.01), LOS (8.89 days vs 6.16 days, P ≤ 0.01), and total hospitalization costs ($24 817 vs $18 505, P ≤ 0.01) than those without COVID‐19. On subgroup analysis, patients in the DC group had higher mortality, LOS, and hospitalization costs compared to those in the CC group. On multivariate analysis, we also found that COVID‐19 infection, age, Charlson Comorbidity Index ≥3, acute kidney injury, end‐stage renal disease, septic shock, acute respiratory failure, MV, and ICU status were independent predictors for mortality. Conclusion Our study suggests that COVID‐19 infection is an independent predictor of mortality in patients with cirrhosis, with threefold higher mortality and increased resource utilization. Early intervention through immunizations and advanced COVID‐19 therapies can help improve these outcomes.


Introduction
The coronavirus disease 2019 (COVID-19) pandemic is an ongoing global health challenge that has triggered unprecedented direct and indirect medium-to long-term effects on mortality and morbidity.According to the World Health Organization, there were more than 640 million confirmed cases as of December 2022, and more than 6 million people died because of the pandemic. 1Patients with chronic health conditions or already reduced life expectancies are considered the most vulnerable individuals.Although preventive health measures and immunizations have partially offset the risk of contraction and poor outcomes, patients with chronic diseases are still at an increased risk for poor outcomes. 2Among these patients, chronic liver disease (CLD), especially cirrhosis, is considered a high-risk comorbid condition for severe COVID-19 disease due to inherent immune dysregulation. 3etween 1999 and 2010, cirrhosis had a prevalence of 0.27% in the United States, with an all-cause mortality rate of up to 26.4% per 2-year interval in cirrhosis patients compared with 8.4% in propensity-matched controls. 4Earlier studies during the pandemic showed that COVID-19 is associated with significantly higher mortality and morbidity in patients with cirrhosis compared to the general population.Numerous studies have indicated that cirrhotic patients with COVID-19 have a dire prognosis, with overall mortality rates as high as 35%. 5,6In addition to the adverse inpatient outcomes, in some patients COVID-19 symptoms tend to linger, causing repeated hospitalizations.Patients with cirrhosis, especially decompensated cirrhosis (DC), are a high-risk population at baseline for readmissions, mainly due to the complications of portal hypertension.
Most previous studies on COVID-19 in cirrhotic patients were from single centers or different healthcare systems, or had a small sample size.In addition, there is no nationwide study regarding the readmission rate and causes of hospital readmissions in patients with cirrhosis with concomitant COVID-19 infection.A better understanding of the national data for inpatient outcomes and readmissions in patients with cirrhosis with COVID-19 can aid in identifying areas for improvement and planning measures to improve patient care and minimize costs.We used the two largest databases to determine the inpatient outcomes, resource utilization, and 30-day readmission rate in cirrhotic patients with COVID-19.

Methods
Data source.We conducted a retrospective cohort study using the recently released National Inpatient Sample (NIS) 2020 and National Readmission Database (NRD) 2020 to identify all patients with cirrhosis and concomitant COVID-19 infection.NIS and NRD are a part of the Healthcare Cost Utilization Project (HCUP), sponsored by the Agency of Healthcare Research and Quality (AHRQ). 7NIS and NRD are the largest publicly accessible databases in the United States, covering all payer healthcare information.The NIS 2020 contains data from 48 states covering 97% of the U.S. population. 8Similarly, the NRD 2020 includes data from 31 states covering 60.8% of all U.S. hospitalizations. 9The NIS database provides estimates of in-hospital mortality and resource utilization associated with various conditions to improve the quality and outcomes of inhospital care.In comparison, the NRD provides national-level estimates of in-hospital readmissions in the United States.Both are designed in a stratified probability sample representing 20% of all non-federal acute care inpatient hospitalizations.The stratification follows the American Hospital Association criteria and is based on the U.S. census region, division, hospital control, location, teaching status, and hospital bed size. 7Both databases contain patient-level and hospital-level information.Hospital-level data include control/ownership, teaching/non-teaching status, number of beds, hospital region (not in NRD), and rural/urban location.Patient-level data contain demographics including age, gender, race (not in NRD), median income by zip quartile, principal and secondary diagnoses at discharge, the total number of patients who died, length of stay (LOS), procedures performed, and total hospitalization cost and charges.The discharge diagnosis and procedure codes are identified using the International Classification of Diseases, Tenth Revision, and Clinical Modification (ICD10-CM).NIS is de-identified data, but NRD assigns a unique database identification number for each patient. 8This unique number can be used to identify all admissions for each patient.A readmission was defined as any non-traumatic admission within 30 days of the index admission.If the patient had multiple readmissions, we considered only the first readmission.All patients who died during the index hospitalization were excluded from the readmission study.In our study, the NIS data were used to estimate inpatient mortality and resource utilization while NRD was used to assess the 30-day readmission rate.
Ethical statement.Both NIS and NRD datasets are deidentified and do not directly involve "human subjects," as defined by federal regulations and guidance (https://www.hhs.gov/ohrp/regulations-and-policy/regulations/common-rule/index.html).Therefore, NIS/NRD do not require institutional review board (IRB) approval.All procedures performed were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Study population.
We identified all hospitalization related to cirrhosis and concomitant COVID-19 infection using ICD-10 codes (provided in Supplementary Data).We also performed subgroup analysis to categorize cirrhosis into compensated (CC) and decompensated cirrhosis (DC) groups based on the presence of one or more diagnosis codes for clinically significant portal hypertension-related complications.We excluded patients under 18 years of age and those with elective admissions.To arrive at an accurate readmission rate, we excluded all the index hospitalizations in December because NRD keeps track of hospitalizations on a calendar year basis (i.e.January 1 to December 31) without linkage to the previous or following year.
Study outcome.The primary outcomes of our study were inpatient mortality, mechanical ventilation (MV), acute respiratory failure (ARF), intensive care utilization (ICU) requirement, and the all-cause 30-day readmission rate, with the most common reason leading to these readmissions.The secondary outcomes included healthcare utilization, including the LOS and the total hospitalization charges in both groups.We also conducted univariate and multivariate regression analyses to identify the independent predictors of mortality, LOS, and hospitalization cost.
Patient and hospital characteristics.We used NIS and NRD variables to identify the patient's age, gender, median household income for the patient's zip code, primary insurance payer (Medicare, Medicaid, private insurance, self-pay, or uninsured), patient residential area (a large metropolitan area with 1 million residents, a small metropolitan area with less than 1 million residents, micropolitan areas, and not metropolitan/ micropolitan).Hospital characteristics, including size (large, medium, or small), location (urban or rural), and teaching status (teaching or non-teaching) were also assessed.The ICD-10 codes were used to identify comorbid conditions, including hypertension, acute kidney injury (AKI), chronic kidney disease (CKD), end-stage renal disease (ESRD), diabetes mellitus (DM), acute respiratory failure (ARF), chronic pulmonary disease (CPD), and MV (Supplemental Files).NIS also provides information about the total hospitalization charges the hospital billed for each hospitalization.The Charlson Comorbidity Index (CCI) was used to assess the burden of comorbidities in both groups of patients.In studies utilizing administrative databases, the CCI score predicts 1-year survival in patients based on the presence or absence of multiple comorbidities. 10atistical analysis.All statistical analyses were conducted using STATA, version 16.0 (Stata Corp, College Station, Texas, USA).To analyze descriptive statistics for patients and hospital characteristics, Pearson's chi-square test was used for categorical variables and Student's t-test for continuous variables.
Continuous variables are described as means with standard deviation (SD), whereas categorical variables are reported as proportions or percentages.We used case-by-case deletion to remove missing data, and less than 5% of the data for LOS, hospitalization cost, and mortality were missing.An α-value of less than 0.05 was chosen as the level of significance.To assess the effect of COVID-19 on secondary outcomes, we used bivariate linear and logistic regression analyses, followed by multivariate regression models to account for confounding factors.We chose variables for the multivariate models with a significance level of 0.2 in the bivariate analysis.All observations with missing values were excluded from the study.The total hospitalization cost variable is not available in NIS as a default; hence we used the HCUP cost-to-charge ratio. 11The survival analysis was performed, with the time from the discharge to readmission considered a time variable and death as a failure.Patients were censored if they were alive on their 30th day of discharge.

Results
Baseline characteristics.We identified a total of 733 354 cirrhosis-related hospitalizations in adult patients, out of which 25 194 had a concurrent diagnosis of COVID-19 infection (Fig. 1).In the cirrhosis + COVID group, 12 745 (50.59%) met the definition of DC compared to 12 449 (49.41%) for CC.The cirrhosis + COVID group patients were older, with a mean age of 62.20 years versus 60.27 years for those with cirrhosis but without COVID-19 (P < 0.01).A significantly larger proportion of the patients in the cirrhosis + COVID group were older than 65 years compared to those with cirrhosis without COVID-19 infection (49.2% with COVID vs 36% without COVID-19, P < 0.01).In both groups, the majority of patients were females (52.1% with COVID-19 vs 55.2% without COVID-19; P < 0.01) and White (73.6% with COVID-19 vs 77.8% without COVID-19; P < 0.01).More than half of the patients with both cirrhosis and COVID-19 (55.4%) had Medicare as their primary payer; this proportion was smaller among the patients with cirrhosis without COVID-19 (46%; P < 0.01).In terms of individual comorbidities, more patients had a CCI > 3 (29.3%with COVID-19 vs 26.7% without COVID-19; P < 0.01) and CPD (24.7% with COVID-19 vs 22.4% without COVID-19; P < 0.01) in the COVID group compared to the non-COVID group; however, ESRD was significantly higher in the non-COVID group (6.9% with COVID-19 vs 9.1% without COVID-19; P < 0.01).A larger number of the patients in the cirrhosis + COVID group were discharged to a skilled nursing facility compared to the non-COVID group (18.7% with COVID-19 vs 11.2% without COVID-19, P < 0.01).There were no statistically significant differences in median income between the two groups of patients based on zip code.The remaining patient and hospital characteristics were comparable between the two groups (Table 1).

Discussion
To our knowledge, this is the first study in the United States to present hospital-related outcomes of COVID-19 infection in patients with cirrhosis at the national level.Our study findings suggest that the inpatient mortality rate of patients with cirrhosis and COVID-19 infection, particularly those with clinically significant portal hypertension, is 3 times higher than patients with cirrhosis without COVID-19.In addition, COVID-19 infection is an independent predictor of mortality in these patients after controlling for confounding factors.Moreover, advanced age, CCI > 3, AKI, ESRD, ICU care or MV requirement, and development of septic shock and respiratory failure were independent predictors of mortality (Table 3).Patients with cirrhosis are also more likely to develop severe disease with an increased requirement for MV and intensive care utilization with COVID-19 infection, which may contribute to increased LOS and hospitalization charges.However, patients with cirrhosis and concurrent COVID-19 infection have a lower all-cause 30-day readmission rate than those without COVID-19, which may be explained by the higher inpatient mortality in patients with COVID-19 infection during index hospitalizations.
Although the exact mechanism is still unclear, excess systemic inflammation, intestinal dysbiosis, cirrhosis-induced immunological dysfunction, and coagulopathies may play a role in the etiology of cirrhosis patients with a greater risk of mortality and severity following COVID-19 infection. 124][15] However, most of these studies were conducted in single centers, had a small sample size from various healthcare systems, and did not distinguish the differences in outcomes between CC and DC. 13,14Our study period corresponds to the earlier phase of the pandemic (2020) and reports relatively low inpatient mortality compared to some earlier studies.Moreover, on subgroup analysis, patients with DC had higher mortality with COVID-19 than those with CC.Our study design has several unique features that improve the generalizability of our findings and advance our understanding of COVID-19 disease in cirrhosis.These features include a subgroup analysis of cirrhosis, a broad representation of demographics, and a large nationwide sample size.On multivariate analysis among patients with cirrhosis, COVID-19 infection was associated with a 2.63 times higher risk of in-hospital mortality.A recent meta-analysis analyzing 40 studies by Nagarajan et al. found that patients with cirrhosis and concurrent COVID-19 had 3 times higher odds of mortality than those without cirrhosis, which is consistent with our study findings. 15However, most of the studies included in that meta-analysis were conducted outside the United States, representing different patient populations with varying access to various healthcare systems.In contrast, our study's results represent the U.S. population at a national level regardless of hospital size, academic status, rural/urban location, and geographical region.
To gain a better understanding of the risk factors and patterns associated with adverse COVID-19 infection outcomes, we conducted a multivariate analysis to find the predictors of inpatient mortality.Earlier, some studies had also reported that certain demographic factors were associated with poor outcomes in cirrhosis and COVID-19.These included increased age, male gender, and ethnicity, primarily Black and Hispanic populations. 16,17owever, we found that COVID-19 infection, advanced age, and CCI ≥ 3 were major cofactors in raising the mortality risk of patients with cirrhosis.Consistent with previous research, we found that female gender was associated with a lower risk of death in patients with cirrhosis and COVID-19. 17Furthermore, in the COVID group, other comorbid conditions and inpatient complications, such as AKI, ESRD, ARF, MV, and ICU status, were independent predictors of mortality.
To date, our study is the only nationwide estimate of the 30-day readmission rate and its causes among patients with cirrhosis and COVID-19 infection.Using NRD, the 30-day readmission rate among patients with cirrhosis and COVID-19 infection was 8.28%, which is lower than the rate of 17.54% among patients conducted a retrospective analysis to identify the most common causes of readmissions in patients with cirrhosis alone. 18They reported that the three most common causes of readmissions among these patients were liver diseases-not otherwise specified, substance abuse, and hepatitis, with only a small percentage of  shock, and some studies report mortality rates as high as 75%. 19,20ur study results should encourage clinicians to implement infection control protocols and closely monitor these patients for signs of infection after discharge.Additionally, further research is needed to understand the underlying factors that increase the susceptibility of these patients to infections.In terms of resource utilization, when compared to the non-COVID group, patients in the COVID group, especially those with DC, had higher mean LOS and mean total hospitalization costs.Moreover, a more significant proportion of the patients in the COVID group were discharged to a skilled nursing facility compared to the non-COVID group (Table 1).From 2001 to 2011, the annual cost of cirrhosis-related hospitalizations in the United States rose more than twofold, from $4.8 billion to $9.8 billion. 21Inpatient hospitalization and readmissions significantly increase healthcare resource utilization among patients with cirrhosis. 22Hirode et al. found that from 2012 to 2016, the average inflation-adjusted cost of cirrhosis-related hospitalizations in the United States increased from $16 285 to $19 185 per admission. 23Our results show a significantly higher mean hospitalization cost for patients with cirrhosis and COVID-19 ($24 817), which could be explained by the increased utilization of MV, ICU, and increased LOS. 24To the best of our knowledge, our study is unique in providing data on the resource utilization of patients with cirrhosis and COVID-19 infection in the United States.
We acknowledge certain limitations to our study.Most of these limitations stem from using a claims-based administrative database.These databases utilize ICD-10 codes for diagnosis and procedure documentation, which are prone to under-or overreporting erroneously.Our study includes only the inpatient data and does not include information on the outcomes of outpatients with cirrhosis who had non-severe COVID-19.Since this database does not have laboratory data, we were unable to assess the severity of illness based on the Model End-Stage Liver Disease-Sodium scores or Child-Turcotte-Pugh Class.Similarly, the database lacks the necessary granularity to perform certain detailed analyses.We could not determine the patient's liver transplant status.This database also lacks pharmacological information, limiting our ability to assess the efficacy of specific therapies.Finally, we could not capture out-of-state readmissions because NRD is derived from state-based databases and does not use the exact linkage identification between states.We assume, however, that a limited number of out-of-state patients were readmitted.
Despite its limitations, our study is one of the first to describe hospital-related COVID-19 outcomes, readmission rates with causes, and resource utilization in patients with cirrhosis in the U.S. population at a national level.Our findings indicate that COVID-19 infection is associated with an increased risk of allcause inpatient mortality and resource utilization in patients with cirrhosis.Moreover, a considerable number of cirrhosis patients get readmitted with sepsis post COVID infection for reasons that are currently not understood.This leads to increased resource utilization, adding a burden to the healthcare costs at the national level.Our study results should encourage future studies to address the causes of these readmissions to improve morbidity and mortality in these patients.This study also adds to the growing body of evidence for encouraging mandatory immunization and booster doses in patients with cirrhosis.Because of their high risk of developing severe disease, early initiation of advanced COVID-19 therapies during hospitalization could improve the outcomes for these patients.Although this data predates the availability of vaccines and advanced treatments for COVID-19, it would be interesting to see the results of future studies to see whether these trends persist for subsequent years of the pandemic.

Table 1
The 30-day readmission rate in patients with cirrhosis was 8.28% in COVID-19 versus 17.54% without COVID-19.Sepsis was the most common cause of readmissions in the cirrhosis + COVID group, followed by alcoholic cirrhosis of the liver and hepatic failure, unspecified without coma.Secondary outcomes and resource utilization.The patients in the cirrhosis + COVID group had higher mean LOS (8.89 days with COVID-19 vs 6.16 days without COVID-19, P < 0.01) and mean total hospitalization cost ($24 817 with COVID-19 vs $18 505 without COVID-19, P < 0.01) compared to patients with cirrhosis but without COVID-19.On subgroup analysis, the DC group with COVID-19 had even higher mean LOS (9.86 days vs 7.93 days; P < 0.01) and total hospitalization cost ($28 476 vs $21 218; P < 0.01) than the CC Group with COVID-19.COVID-19 infection was also independently associated with Patient selection flow diagram.*Non-adult patients, elective admissions, patients with missing data related to outcomes.Baseline patient and hospital characteristics for hospitalized cases of cirrhosis with and without COVID-19 infection

Table 2
Outcomes for hospitalized cases of cirrhosis with and without COVID-19 infection with cirrhosis without COVID-19 infection.The higher death rate during the initial hospital stay may have contributed to a lower readmission rate among patients with COVID-19.Although there is no previous data on the readmission rate and causes in patients with cirrhosis and concurrent COVID-19 infection, Garg et al.

Table 3
Bivariate and multivariate logistic regression showing predictors for mortality in patients with cirrhosis hospitalized with COVID-19 infection readmitted patients being diagnosed with sepsis.In contrast, our findings suggest that sepsis was the most common reason for readmissions among patients with cirrhosis and COVID-19 infection, indicating that these patients may be more susceptible to infections after discharge, resulting in repeat hospitalizations.Patients with cirrhosis have worse outcomes with sepsis and septic

Table 4
Bivariate and multivariate logistic regression showing predictors for length of stay (LOS) in patients with cirrhosis hospitalized with COVID-19 infection

Table 5
Bivariate and multivariate logistic regression showing predictors for hospitalization costs in patients with cirrhosis hospitalized with COVID-19 infection