key: cord-0976008-hr6ruijz
authors: Ge, Jin; Pletcher, Mark J.; Lai, Jennifer C.; Harper, Jeremy R.; Chute, Christopher G.; Haendel, Melissa A.
title: Outcomes of SARS-CoV-2 Infection in Patients with Chronic Liver Disease and Cirrhosis: a N3C Study
date: 2021-07-18
journal: Gastroenterology
DOI: 10.1053/j.gastro.2021.07.010
sha: def2db060459d640b427bd42b3607aafcb8557cd
doc_id: 976008
cord_uid: hr6ruijz

Background and Aims In chronic liver disease (CLD) patients with or without cirrhosis, existing studies on the outcomes with SARS-CoV-2 infection have limited generalizability. We used the National COVID Cohort Collaborative (N3C), a harmonized electronic health record (EHR) dataset of 6.4 million, to describe SARS-CoV-2 outcomes in patients with CLD and cirrhosis. Methods We identified all CLD patients with or without cirrhosis who had SARS-CoV-2 testing in the N3C Data Enclave as of 7/1/2021. We used survival analyses to associate SARS-CoV-2 infection, presence of cirrhosis, and clinical factors with the primary outcome of 30-day mortality. Results We isolated 220,727 patients with CLD and SARS-CoV-2 test status: 128,864 (58%) Non-Cirrhosis/Negative, 29,446 (13%) Non-Cirrhosis/Positive, 53,476 (24%) Cirrhosis/Negative, and 8,941 (4%) Cirrhosis/Positive patients. Thirty-day all-cause mortality rates were 3.9% in Cirrhosis/Negative and 8.9% in Cirrhosis/Positive patients. Compared to Cirrhosis/Negative, Cirrhosis/Positive had 2.38-times adjusted hazard of death at 30 days. Compared to Non-Cirrhosis/Positive, Cirrhosis/Positive had 3.31-times adjusted hazard of death at 30 days. In stratified analyses among patients with cirrhosis with increased age, obesity, and comorbid conditions (diabetes, heart failure, and pulmonary disease); SARS-CoV-2 infection was associated with increased adjusted hazards of death. Conclusions In this study of ∼221,000 nationally-representative, diverse, and gender-balanced CLD patients; we found SARS-CoV-2 infection in patients with cirrhosis was associated with 2.38-times mortality hazard, and the presence of cirrhosis among CLD patients infected with SARS-CoV-2 was associated with 3.31-times mortality hazard. These results provide an additional impetus for increasing vaccination uptake and further research regarding immune responses to vaccines in patients with severe liver disease.

Hepatic involvement is common in Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) infection with clinical manifestations ranging from liver function test elevations to acute hepatic decompensations. [2] [3] [4] [5] In patients with existing chronic liver diseases (CLD) and cirrhosis, the outcomes of SARS-CoV-2 infection have been mixed. [6] [7] [8] [9] [10] [11] Previous small-scale studies from tertiary referral centers have demonstrated mortality rates approaching 40% for patients with cirrhosis who were infected by SARS-CoV-2. 8, 11 Other studies, however, have shown that patients with cirrhosis who test positive for SARS-CoV-2 infection had similar mortality rates compared to those patients hospitalized with complications of cirrhosis without SARS-CoV-2 infection. 10 A study of patients with and without cirrhosis based on national data extracted from the United States Department of Veterans Affairs Clinical Data Warehouse demonstrated that patients with cirrhosis were less likely to test positive for SARS-CoV-2 but when positive were 3.5-times more likely to die from all-causes compared to those who tested negative. While this was one of the largest studies of outcomes of SARS-CoV-2 infection in patients with cirrhosis to date, the underlying patient population was 88% male, limiting generalization to other patient populations. 12 The National COVID Cohort Collaborative (N3C) was formed in April 2020 as a centralized resource of harmonized electronic health record (EHR) data from health systems around the United States. 1, 13 As of July 1, 2021, 214 clinical sites had signed data transfer agreements and 57 sites had harmonized data included in the N3C Data Enclave: a diverse and nationally representative central repository of harmonized EHR data and a new model for collaborative data sharing and analytics. Initial results up to December 2020 from the N3C main cohort has been previously characterized and described. 14 To address the conflicting results and gaps of previous studies, we used the N3C Data Enclave to answer three distinct questions regarding outcomes of SARS-CoV-2 infection in CLD patients:

1. What is the association between SARS-CoV-2 and mortality in CLD patients with cirrhosis? J o u r n a l P r e -p r o o f 2. What is the association between cirrhosis and mortality in CLD patients who tested positive for SARS-CoV-2?

3. What are the factors associated with mortality among CLD patients with cirrhosis who tested positive for SARS-CoV-2?

The National COVID Cohort Collaborative (N3C) The N3C is a centralized, curated, harmonized, secure, and nationally representative clinical data resource with embedded analytical capabilities. The N3C is comprised of members from the National The N3C Data Enclave includes EHR data of patients who were tested for SARS-CoV-2 or had related symptoms after January 1, 2020. For patients included in the N3C Data Enclave, encounters in the same source health system beginning on or after January 1, 2018 are also included to provide lookback data. N3C utilizes centrally maintained "shared logic sets" for common diagnostic and phenotype definitions. 1,14 All EHR data in the N3C Data Enclave are harmonized in the Observational Medical Outcomes Partnership (OMOP) common data model, version 5.3.1. 15, 16 In the OMOP common data model, classification vocabularies, such as International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), or Standard Nomenclature of Medicine (SNOMED); are mapped to J o u r n a l P r e -p r o o f standard OMOP concepts based on semantic and clinical relationships. 17 Vocabulary classification and mapping of various ontologies to the OMOP standard vocabulary is maintained by OHDSI and publicly available on ATHENA (http://athena.ohdsi.org/), which is a web-based vocabulary repository. 18 For all analyses, we utilized the de-identified version of the N3C Data Enclave, versioned as of July 1, 2021 and accessed on July 3, 2021. To protect patient privacy, all dates in the N3C Data Enclave are uniformly shifted up to plus/minus 180 days within each partner site in the de-identified database.

SARS-CoV-2 testing status was based on a modified version of the N3C shared logic set: specifically, OMOP concept identifiers signifying culture and nucleic acid amplification testing for SARS-CoV-2 (Supplemental Table 1 ) were queried among all patients included in the N3C Enclave. 1, 14 We did not query SARS-CoV-2 antibody testing as this may be a marker of remote infection or vaccination rather than active infection. The "index date" for all analyses was defined as the date of the earliest positive test (for SARS-CoV-2 positive patients) or earliest negative test (for SARS-CoV-2 negative patients). 12 Patients who underwent repetitive SARS-CoV-2 testing were classified based on the above definitions governing the earliest test. Patients who did not have SARS-CoV-2 testing by the above definitions (e.g., those who were clinically diagnosed with "suspected COVID-19" or those with antibody testing only) were excluded. To account for uniform date-shifting that occurs per partner site in the de-identified N3C Data Enclave, we calculated a "maximum data date" to reflect the last known date of records for each data partner and excluded patients who were tested <90 days of this "maximum data date."

CLD diagnoses were made based on documentation of at least one OMOP concept identifier corresponding to previously validated ICD-10-CM codes for liver diseases (Supplemental Table 2 ) at any J o u r n a l P r e -p r o o f time prior to the index date. [19] [20] [21] [22] As "steatosis of the liver" is a common finding in alcohol-associated liver disease (AALD) and non-alcoholic fatty liver disease (NAFLD), patients with OMOP concept identifier 4059290 (corresponding to ICD-10-CM code K76.0) and at least one OMOP concept identifier describing alcohol use (Supplemental Table 2 ) in accordance with definitions by the Centers for Disease Control and Prevention (CDC) and the National Institute on Alcohol Abuse and Alcoholism Alcohol Epidemiologic Data System, were categorized as those with AALD. [23] [24] [25] [26] Patients with OMOP concept identifier 4059290 without an alcohol use OMOP concept identifier were categorized as NAFLD.

Diagnoses were determined in a hierarchical manner such that NAFLD categorization was made only after exclusion of all other CLD causes. In those patients identified to have CLD, diagnoses of cirrhosis were made based on documentation of at least one OMOP concept identifier corresponding to previously validated ICD-10-CM codes for cirrhosis and its complications (Supplemental Table 2 ) at any time prior to the index date. 12, 27 Diagnoses of cirrhosis, therefore, can only take place in the setting of an existing CLD diagnosis. Patients who had undergone orthotopic liver transplantation (12, 170 patients) as signified by OMOP concept identifier 42537742 (corresponding to ICD-10-CM code Z94. 4) were excluded from all analyses.

Using the above definitions for SARS-CoV-2 testing and chronic liver disease/cirrhosis; we isolated our adult patients (with age ≥ 18 years documented) study population. We divided the study patients into four cohorts (Figure 1 

All patients were followed until their last recorded visit occurrence, procedure, measurement, observation, or condition occurrence in the N3C Data Enclave. The primary outcome was all-cause mortality at 30-days after the index SARS-CoV-2 test date. Secondary outcomes included hospitalization within 30-and 90-days after the index date, mechanical ventilation within 30-and 90-days, and allcause mortality at 90-days after the index date. The outcome of death was ascertained based on EHR data indicating in-hospital death, out-of-hospital death, or referral to hospice. The outcome of mechanical ventilation was ascertained by OMOP procedure or condition concepts. The outcome of hospitalization was ascertained based on recorded OMOP visits concepts. These outcomes were defined centrally based on concept sets in N3C shared logic and have been implemented on the full N3C cohort. 1, 14 To account for potential delays in data reporting/harmonization and outcome ascertainment J o u r n a l P r e -p r o o f from data partner sites, we had excluded all patients who had SARS-CoV-2 testing <90 days of the "maximum data date" as defined above.

Baseline demographic characteristics extracted from N3C Data Enclave included age, sex, race/ethnicity, height, weight, body mass index, and state of origin. States were classified into four geographic regions (Northeast, Midwest, South, and West) defined by the CDC's National Respiratory and Enteric Virus Surveillance System (NREVSS). 28 Patients were categorized as living in "Other/Unknown" region if they originated from territories not otherwise classified (e.g. Guam, Puerto Rico, United States Virgin Islands, or other dependencies) or if state of origin was censored to protect patient privacy in zip codes with few residents. We evaluated comorbid conditions based on the original Charlson Comorbidity Index (CCI), 29,30 consistent with central practices per the N3C consortium. 14 As per definitions established in N3C shared logic, CCI comorbid conditions were extracted centrally using the 'icd' R package, 14,31 which processes and categorizes diagnosis codes from raw data tables. To avoid double-counting liver related comorbidities in our analyses, we calculated a "Modified Charlson Index" based on the original assigned weights for comorbidities (Supplemental Table 3 ) excluding "mild liver disease" and "severe liver disease."

Components of common laboratory tests (basic metabolic panel, complete blood count, liver function tests, and serum albumin) were extracted based on N3C shared logic sets except for international normalized ratio (INR), which we custom-defined based on standard OMOP concept identifiers (Supplemental Table 4 ). We extracted the most complete values to calculate the model for end-stage liver disease-sodium (MELD-Na) score closest to or on the index date from within 30 days before to 7 days after the index date. 55% of patients had laboratory tests were performed within 2 days of the index date available. 17,653 patients, which represented 8% of the full analytical sample, J o u r n a l P r e -p r o o f had full laboratory data for calculation of MELD-Na scores. The time frame of 30 days before to 7 days after the index date was consistent with definitions used centrally by N3C to identify hospitalizations of interest in the main cohort. 14 

Clinical characteristics and laboratory data were summarized by medians and interquartile ranges (IQR) for continuous variables or numbers and percentages (%) for categorical variables.

Comparisons between groups were performed using chi-square and Kruskal-Wallis tests where appropriate. We used the Kaplan-Meier method to calculate 30-day and 90-day cumulative incidences of hospitalization, mechanical ventilation, and death. We used Cox proportional hazard models to evaluate the associations between SARS-CoV-2 and mortality among patients with cirrhosis, between cirrhosis and mortality among CLD patients who tested positive for SARS-CoV-2, and factors with mortality for Cirrhosis/Positive patients. In all multivariable analyses, we adjusted for age, sex, race/ethnicity, CLD etiology, CCI score, and region of origin.

We conducted stratified analyses based on categories of MELD-Na scores, categories of 

The baseline demographic and clinical characteristics of the four cohorts are presented in Table   1 . In general, the four cohorts differed significantly with regards to distributions of age, race/ethnicity, height, weight, body mass index (BMI), etiologies of chronic liver disease, Modified Charlson Index scores, NREVSS regions, and laboratory test values. Of note, patients with cirrhosis were less likely to be women: 53% of Non-Cirrhosis/Negative, 54% of Non-Cirrhosis/Positive, and 44% of Cirrhosis/Negative and 45% of Cirrhosis/Positive cohorts. Of CLD etiologies, there were notable differences in the distribution of patients with AALD: 34% and 28% of the Cirrhosis/Negative and Cirrhosis/Positive cohorts, respectively, compared to 6% and 7% of the Non-Cirrhosis/Negative and Non-Cirrhosis/Positive cohorts, respectively.

Full MELD-Na components were available with scores calculated in 17,653 patients, representing 8% of the total population. Among Non-Cirrhosis patients, full MELD-Na components were available for 6,866/158,310 (4%) patients, the median (IQR) MELD-Na were 9 (7-13) and 10 (8-13) in Non-Cirrhosis/Negative and Non-Cirrhosis/Positive patients, respectively. Among Cirrhosis patients, full MELD-Na components were available for 10,787/62,427 (17%) patients, the median (IQR) MELD-Na were 16 (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) and 17 (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) in Cirrhosis/Negative and Cirrhosis/Positive patients, respectively. For 121,703 (55%) CLD patients whose location data were available, every state and CDC NREVVS region was represented both in the full sample and among those with positive SARS-CoV-2 tests (Supplemental Table 2 . Thirty-day death rates increased progressively from 0.4% in Non-Cirrhosis/Negative patients to 1.7% in Non-Cirrhosis/Positive patients, and from 3.9%

in Cirrhosis/Negative patients to 8.9% in Cirrhosis/Positive patients. Ninety90-day death rates also increased progressively from 0.8% in Non-Cirrhosis/Negative patients to 2.3% in Non-Cirrhosis/Positive patients, and from 7.0% in Cirrhosis/Negative patients to 12.7% in Cirrhosis/Positive patients. Thirty30and 90-day mechanical ventilation rates also increased in a similar fashion based on SARS-CoV-2 status and presence of cirrhosis. Of note, 30-day and 90-day hospitalization rates were consistently higher among patients with cirrhosis compared to those patients without cirrhosis. Among both Non-Cirrhosis and Cirrhosis patients, those testing negative for SARS-CoV-2 had higher 30-and 90-day hospitalization rates. Kaplan-Meier curves for 30-day mortality among the four cohorts are presented in Figure 1 .

In univariate analyses, compared to Cirrhosis/Negative patients, SARS-CoV-2 positivity Table 3 .

In univariate analyses, compared to Non-Cirrhosis/Positive patients, the presence of cirrhosis Table 4 .

Demographic and clinical factors associated with all-cause 30-day mortality among Cirrhosis/Positive patients are presented in Table 5 

Stratified analyses of the contributions of various clinical factors and comorbidities to associations with 30-day mortality amongst Cirrhosis patients are presented in Table 6 . Among patients with compensated cirrhosis (defined as those without OMOP concept identifiers associated with variceal bleeding, ascites, spontaneous bacterial peritonitis, hepatic encephalopathy, hepatorenal syndrome, or hepatopulmonary syndrome), SARS-CoV-2 positivity (Cirrhosis/Positive) was associated with 5.00-times adjusted hazard of death within 30 days (aHR 5.00, 95%CI 3.92-6. 37 

In general, within stratified categories of age, the adjusted hazard ratios of death within 30 days for Cirrhosis/Positive compared to Cirrhosis/Negative patients increased from aHR 1.59 (age 30-49 years) to aHR 3.03 (age ≥65 years). Within stratified categories of BMI, the adjusted hazard ratios also increased from aHR 2.11 (BMI <25kg/m 2 ) to aHR 2.74 (BMI ≥35kg/m 2 ). Within stratified categories of MELD-Na scores, however, the adjusted hazard ratios deceased from aHR 3.49 (MELD-Na 6-15) to aHR 1.36 (MELD-Na ≥35). A similar trend was also seen within stratified categories of the Modified Charlson Index: the adjusted hazard ratios decreased from aHR 2.57 (Score 1-2) to aHR 2.37 (Score ≥5).

When stratified based on comorbid conditions, the adjusted hazard ratios of death within 30 days for Cirrhosis/Positive compared to Cirrhosis/Negative patients increased in the presence of diabetes (aHR 2.58 versus aHR 2.28 for no diabetes), heart failure (aHR 2.45 versus aHR 2.34 for no heart failure), and pulmonary disease (aHR 2.63 versus aHR 2.27 for no pulmonary disease). When stratified based on chronic renal disease, however, the adjusted hazard ratios were lower for those with renal disease (aHR 2.34 versus aHR 2.30 for no renal disease).

As calculated MELD-Na scores were available for only 17,653 (8%) patients and serum albumin values were available for 75,267 (34%) patients, we conducted sensitivity analyses to determine the influence of these variables on the above multivariate models comparing patients with cirrhosis (Supplemental Table 5 ). For the multivariate model evaluating the association of SARS-CoV-2 infection with death in patients with cirrhosis, further adjustments for MELD-Na and serum albumin did not change the significance of the association (aHR ranged from 1.66 to 2.38). For the multivariate model evaluating factors associated with death among Cirrhosis/Positive patients, further adjustments for MELD-Na and serum albumin did not change the significance of the association for age and death (aHR ranged from 1.02 to 1.05). These adjustments for MELD-Na and serum albumin did, however, eliminate 

In this study of nearly 221,000 patients with chronic liver disease in the National COVID Cohort Collaborative, we found that SARS-CoV-2 infection was associated with 2.38-times hazard of all-cause mortality within 30 days among patients with cirrhosis. Among all CLD patients (with and without cirrhosis) who tested SARS-CoV-2 positive, the presence of cirrhosis was associated with 3.31-times hazard of all-cause mortality within 30 days. Our results are consistent with previous studies of CLD patients with and without cirrhosis, but our use of the N3C Data Enclave has several unique features that enhance the generalizability of our results and advance our understanding of SARS-CoV-2 infection in CLD patients. The number of clinical sites included in this study (harmonized data from 57 as of July 1, 2021) confers a major strength to this study in terms of the number of patients, national scope, and demographic representation. Notably, 51% of the participants were women and 32% were racial/ethnic minorities: 16% identified as Black/African-American, 13% Hispanic, and 3% Asian.

In addition, compared to previous studies which only included data in the early phases of the COVID-19 pandemic, this study covers a longer duration up to July 2021 and reflects changes in treatment and therapy advances. For example, we found a lower cumulative incidence of all-cause 30day mortality at 8.9% for Cirrhosis/Positive patients compared to previous studies with estimates up to 17%. 12 Consistent with previous studies, we also found comparatively higher hospitalization rates in SARS-CoV-2 negative groups (Non-Cirrhosis/Negative and Cirrhosis/Negative) likely due to changes in J o u r n a l P r e -p r o o f healthcare delivery during the COVID-19 pandemic as standardized testing prior to hospital admissions and procedures became widespread. 33, 34 With regards to demographic and clinical factors associated with adverse outcomes in SARS-CoV-2 infection, our findings were also consistent with existing literature. We found female gender was associated with a lower hazard of death (aHR 0.84, 95%CI 0.74-0.95, p<0.01) among all CLD patients with SARS-CoV-2 infection (Table 4 ). This gender association, however, did not remain once we stratified to only Cirrhosis/Positive. Consistent with extensive racial/ethnic disparities described, 35, 36 we found an increased hazard of mortality for those who identified as Hispanic (aHR 1.20, 95%CI 1.02-1.42, p=0.03) and those who identified as Other/Unknown (aHR 1.25, 95%CI 1.01-1.55, p=0.04) among CLD patients with positive SARS-CoV-2 test (Table 4 ). When we stratified to only Cirrhosis/Positive patients, we found that this association between Hispanic ethnicity and mortality was no longer significant. The reasons for this are likely multifactorial and reflect present disparities in differential rates of SARS-CoV-2 infection, 35, 36 and longstanding disparities in access to treatment for liver diseases in the United States. [37] [38] [39] The broader questions regarding gender and racial/ethnic disparities during the COVID-19 pandemic are active areas of exploration amongst several N3C teams. 1, 13, 14 To further understand risk factors and patterns associated with adverse outcomes in SARS-CoV-2 infection, we conducted stratified analyses comparing mortality between Cirrhosis/Positive and Cirrhosis/Negative patients (Table 6 ). Consistent with previous literature, [40] [41] [42] [43] we found that age, obesity, and comorbid conditions (diabetes, heart failure, and pulmonary disease) were significant cofactors in increasing the mortality risk for patients with cirrhosis when infected with SARS-CoV-2. For instance, among patients with cirrhosis between the ages of 30-49 years, the adjusted hazard of 30-day mortality associated with SARS-CoV-2 infection was 1.59 -this adjusted hazard increased to 3.03 among those who were greater than 65 years old. Similarly, among patients with cirrhosis with BMIs <25kg/m 2 , the adjusted hazard of 30-day mortality associated SARS-CoV-2 infection was 2.11 -this adjusted hazard J o u r n a l P r e -p r o o f increased to 2.74 among patients with cirrhosis with BMIs ≥35kg/m 2 . Interestingly, we found that the adjusted hazard ratios of 30-day mortality decreased when we stratified by MELD-Na score categories. This is likely due to high baseline mortality rates seen among patients with more severe liver disease regardless of SARS-CoV-2 infection, e.g. cumulative incidence of death at 30 days of 41.8% among Cirrhosis/Negative patients with MELD-Na score ≥35. A similar phenomenon was seen with increasing Modified Charlson Index scores in which the adjusted hazard ratios decreased when we stratified by higher score categories. This is also likely due to a higher baseline mortality rate amongst Cirrhosis/Negative patients with higher comorbidity scores. We did not include smoking status in our stratified analyses as there have been data ascertainment issues (as missingness was only suggestive of non-smoking status) per discussions with central N3C teams.

Due to the methodology by which we derived our SARS-CoV-2 negative comparison populations (Non-Cirrhosis/Negative and Cirrhosis/Negative), we likely introduced selection bias as these cohorts were more likely to undergo procedures or be hospitalized. These comparison populations, therefore, do not reflect baseline populations of CLD patients with and without cirrhosis. As such, the adjusted hazard ratios for 30-day mortality associated with SARS-CoV-2 infection among patients with cirrhosis and various comorbidity categories calculated in this study may be an underestimate of the true hazard ratio as our comparison populations were more clinically ill.

We acknowledge the following limitations. First, N3C is a collaboration amongst multiple NCATS-supported CTSA program hubs and, therefore, has an overrepresentation of tertiary academic medical centers as data partners, which limits the generalizability of the study. Moreover, there is substantial oversampling of data from certain states -notably North Carolina, New York, Illinois, and Colorado. The national scope and gender/demographic characteristics of our study population, however, are unique strengths of this study compared to previous research. Second, as data were aggregated from many sites, there is systematic missingness of certain variables. In our study, this is J o u r n a l P r e -p r o o f most apparent in that we were only able to calculate the MELD-Na scores for 17,653 patients. We accounted for this by conducting sensitivity analyses that showed our main findings did not change. In addition, our sensitivity analyses revealed that certain geographic and CLD etiology associations with mortality were eliminated once adjustments were made in Cirrhosis/Positive patients. This most likely reflected not-at-random data missingness in N3C. Third, although N3C has standardized protocols for data curation and harmonization, there likely remains variations in terminology and ontology between sites. The use of the OMOP common data model, however, decreases such differences and enforces a degree of standardization. 15, 16, 44 Fourth, due to date-shifting employed in the process of de-identification in the N3C Data Enclave and differences in data harmonization times between data partner sites, there may be a delay in ascertainment of outcomes. There may be misclassification of outcomes if the date of SARS-CoV-2 testing was close to the latest known date of records ("maximum data date") for that site. To account for these issues, we employed two methods: 1. We attempted to maximize follow up for each patient by defining last follow up as any encounters or records (visit occurrence, procedure, measurement, observation, or condition occurrence) in the OMOP data model. 2. We excluded patients whose date of SARS-CoV-2 testing was within 90 days of the maximum data date -this exclusion criterion only affected 1% of potential patients to be included in the analytical sample.

Fifth, we utilized the de-identified version of the N3C Data Enclave to conduct our analyses. To protect patient privacy, date shifting was uniformly applied. This means that our analyses could not investigate temporal trends with each COVID-19 surge in the United States. Lastly, there is likely misclassification between patients with AALD and NAFLD given the non-specific nature of OMOP concept identifier 4059290 "steatosis of the liver" (corresponding to ICD-10 code K76.0). This is most apparent in only 6% of CLD patients without cirrhosis who were classified to have AALD while 33% of J o u r n a l P r e -p r o o f patients with cirrhosis were classified with AALD (due to more specific ICD-10 codes for cirrhosis due to AALD).

Despite these limitations, our study is one of the largest studies of outcomes of SARS-CoV-2 infection in patients with chronic liver diseases with and without cirrhosis to date. Our results are consistent with those from previous studies and show that SARS-CoV-2 infection is associated with an increased risk of all-cause mortality among patients with cirrhosis. This study provides an additional impetus for increasing vaccine uptake among patients with cirrhosis. 45 Moreover, as patients with advanced liver diseases have well recognized immune dysfunction with attenuated immune responses to other vaccines, [46] [47] [48] [49] further research is urgently needed regarding immune responses to COVID-19 vaccines in CLD patients to guide public health recommendations. Given the continued expansion of N3C and ongoing acquisition of longitudinal data, our study in the N3C Data Enclave lays the foundation for studying future potential clinical questions, such as clinical responses to vaccinations, which affect liver disease patients as the COVID-19 pandemic continues to evolve. 50 J o u r n a l P r e -p r o o f †Modified Charlson Index was calculated based on the original Charlson Comorbidity Score excluding weights for "mild liver disease" and "severe liver disease." ǂMELD-Na scores were calculated for 17,653 patients, which represent 8% of the total sample. Abbreviations: AALD, alcohol-associated liver disease; AST, aspartate transaminase; ALT, alanine transaminase; BMI, body mass index; INR, international normalized ratio; MELD-Na, Model for End-Stage Liver Disease-Sodium; NAFLD, nonalcoholic fatty liver disease. .01 †Modified Charlson Index was calculated based on the original Charlson Comorbidity Score excluding weights for "mild liver disease" and "severe liver disease." Abbreviations: AALD, alcohol-associated liver disease; aHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; NAFLD, non-alcoholic fatty liver disease. <0.01 †Modified Charlson Index was calculated based on the original Charlson Comorbidity Score excluding weights for "mild liver disease" and "severe liver disease." Abbreviations: AALD, alcohol-associated liver disease; aHR, adjusted hazard ratio; CI, confidence interval; HR, hazard ratio; NAFLD, non-alcoholic fatty liver disease. 

The National COVID Cohort Collaborative (N3C): Rationale, design, infrastructure, and deployment

SARS-CoV-2 infection of the liver directly contributes to hepatic impairment in patients with COVID-19

Interrelationship between coronavirus infection and liver disease

Clinical characteristics of coronavirus disease 2019 in china

Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China

Outcomes following SARS-CoV-2 infection in patients with chronic liver disease: An international registry study

Predictors of Outcomes of COVID-19 in Patients with Chronic Liver Disease: US Multi-center Study

High rates of 30-day mortality in patients with cirrhosis and COVID-19

Clinical characteristics and outcomes of coronavirus disease 2019 among patients with preexisting liver disease in the united states: A multicenter research network study

Comparison of mortality risk in patients with cirrhosis and COVID-19 compared with patients with cirrhosis alone and COVID-19 alone: multicentre matched cohort

High mortality rates for SARS-CoV-2 infection in patients with pre-existing chronic liver disease and cirrhosis: Preliminary results from an international registry

Cirrhosis and SARS-CoV-2 infection in US Veterans: risk of infection, hospitalization, ventilation and mortality

The national COVID cohort collaborative: clinical characterization and early severity prediction

COVID-19 Clinical Data Warehouse Data Dictionary Based on OMOP Common Data Model Specifications Version 5.3.1. 17. Anon. documentation:vocabulary:mapping [Observational Health Data Sciences and Informatics

Changing trends in etiology-based annual mortality from chronic liver disease

Trends in mortality from chronic liver disease

Mortality attributable to cholestatic liver disease in the United States

prevalence and mortality of autoimmune hepatitis in England 1997-2015. A population-based cohort study

Alcohol-related and viral hepatitis C-related cirrhosis mortality among Hispanic subgroups in the United States

Publications | National Institute on Alcohol Abuse and Alcoholism | Surveillance Report #93

Application of Machine Learning Methods to Predict Non-Alcoholic Steatohepatitis (NASH) in Non-Alcoholic Fatty Liver (NAFL) Patients

Patterns of inpatient opioid use and related adverse events among patients with cirrhosis: A propensity-matched analysis

A new method of classifying prognostic comorbidity in longitudinal studies: development and validation

Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries

Anon. icd: Comorbidity Calculations and Tools for ICD-9 and ICD-10 Codes version 4.0.9 from CRAN. Available at

R: A language and environment for statistical computing

COVID-19 screening of asymptomatic patients admitted through emergency departments in Alberta: a prospective quality-improvement study

Utility of asymptomatic inpatient testing for COVID-19 in a lowprevalence setting: A multicenter point-prevalence study

Racial and Ethnic Disparities in COVID-19-Related Infections, Hospitalizations, and Deaths : A Systematic Review

COVID-19 and racial/ethnic disparities

Racial and ethnic disparities in diagnosis of chronic medical conditions in the USA

Prevalence of chronic liver disease and cirrhosis by underlying cause in understudied ethnic groups: The multiethnic cohort

Disparities in End-Organ Care for Hispanic Patients with Kidney and Liver Disease: Implications for Access to Transplantation

Factors associated with death in critically ill patients with coronavirus disease 2019 in the US

Cardiovascular Disease, Drug Therapy, and Mortality in Covid-19

COVID-19 in People With Diabetes: Urgently Needed Lessons From Early Reports

Prevalence and 30-Day Mortality in Hospitalized Patients With Covid-19 and Prior Lung Diseases

Transforming and evaluating electronic health record disease phenotyping algorithms using the OMOP common data model: a case study in heart failure

AASLD Expert Panel Consensus Statement: Vaccines to Prevent COVID-19 Infection in Patients with Liver Disease

Pneumococcal vaccine response in cirrhosis and liver transplantation

Effectiveness of influenza vaccines in adults with chronic liver disease: a systematic review and meta-analysis

Response to hepatitis B vaccination in patients with liver cirrhosis

SARS-CoV-2 vaccination in patients with liver disease: responding to the next big question

The COVID-19 pandemic will have a long-lasting impact on the quality of cirrhosis care

J o u r n a l P r e -p r o o f In this study of 220,727 patients with liver disease: 30-day mortality was 8.9% for Cirrhosis/SARS-CoV-2+ patients and SARS-CoV-2 infection was associated with a 2.38-times hazard of death.What You Need to Know: BACKGROUND AND CONTEXT We used the National COVID Cohort Collaborative (N3C), a harmonized electronic health record (EHR) dataset of 6.4 million, to describe SARS-CoV-2 outcomes in patients with CLD and cirrhosis.

In this study of 220,727 patients with liver disease: 30-day mortality was 8.9% for Cirrhosis/SARS-CoV-2+ patients and SARS-CoV-2 infection was associated with a 2.38-times hazard of death.

Comparison population of Cirrhosis/SARS-CoV-2-is likely sicker than the general cirrhosis population. There is substantial not-at-random missingness of multiple covariates.

This study corroborates previous research on the increased risk of adverse outcomes in Cirrhosis/SARS-CoV-2+ patients. This study provides additional impetus for increasing vaccine uptake among this vulnerable population.