NAFLD and NASH are obesity‐independent risk factors in COVID‐19: Matched real‐world results from the large PINC AI™ Healthcare Database

Non‐alcoholic fatty liver disease (NAFLD) and non‐alcoholic steatohepatitis (NASH) are potential risk factors for severe pneumonia and other infections. Available data on the role of NAFLD/NASH in worsening outcomes for COVID‐19 are controversial and might be confounded by comorbidities.


| INTRODUC TI ON
Infections with new genetic variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pose an ongoing threat to both infected patients and healthcare systems.Representing the biggest global healthcare crisis since the 1918 influenza pandemic, about 766 million confirmed infections with COVID-19 have resulted in a staggering 6.9 million deaths worldwide. 1In recent months, the severity of individual cases has subdued due to the spread of the highly contagious, but less clinically consequential omicron variant, as well as rising percentages of (repeated) immunisation. 2The need to analyse both the dynamics of the pandemic and individual risk profiles to protect potential high-risk patients remains essential.
Numerous studies have identified advanced age, underlying respiratory diseases, impaired renal function, cardiovascular disease, obesity, type 2 diabetes mellitus, and immunosuppression as risk factors for COVID-19 disease progression, hospitalisation, and deaths.Unfortunately, the COVID-19 pandemic intersects with a steady increase in the prevalence of metabolic syndrome (i.e., obesity, diabetes, hypertension, dyslipidaemia, or hyperlipidaemia) and non-alcoholic fatty liver disease (NAFLD). 3,4In fact, the new nomenclature of 'metabolic dysfunction-associated steatotic liver disease' (MASLD) requires the presence of ≥ 1 characteristic cardiometabolic comorbidity in individuals with a fatty liver in order to establish the diagnosis. 5patic fat accumulation gains major importance in the event of necroinflammation of fatty liver tissue such as non-alcoholic steatohepatitis (NASH), possibly resulting in cirrhosis and/or hepatocellular carcinoma.Recently, the incidence of both NAFLD and NASH has experienced a meteoric rise.Combined, they have become the leading cause of liver disease in Western countries, affecting 20 to 40% of the population. 6,7NAFLD is now the fastest-growing contributor to liver-related morbidity and mortality, 6 while NASH-related cirrhosis has already established itself as one of the leading causes for liver transplantation in adults. 8,91][12][13] The most common causes of death in patients with COVID-19 and chronic liver disease include lung failure and liverrelated mortality 14 ; cohesively, end-stage liver disease/cirrhosis drives mortality through comorbidities. 15terestingly, the available data about NAFLD/NASH as causal risk for COVID-19 severity are polyvocal.7][28][29][30] On a wider note, NAFLD/NASH in pre-cirrhosis stages have been linked to more severe pneumonia and other respiratory infections and thus generally seem to pose a risk for infection in affected individuals. 31,32As the issue remains challenging, a recent position paper by the European Association for the Study of the Liver (EASL) has opted to-at least epidemiologically-consider patients with NAFLD at '…increased overall risk of developing severe COVID-19 which may be attributed to the presence of other high-risk comorbidities'. 33e primary objective of this study was thus to characterise and dissect the impact of NAFLD/NASH on hospitalisation-related outcomes for patients with COVID-19 after controlling for other risk factors such as obesity, diabetes, hypertension, and cardiovascular disease in a large cohort of patients using real-world data (RWD).

| Study population
The PINC AI™ Healthcare Database is an administrative database with patient-level hospital chargemaster data from over 1000 hospitals in the United States (US) that together represent approximately 25% of all inpatient admissions. 34This study used the PINC AI™ Special Release COVID-19 database, which included approximately 900 hospitals.

| Ethical approval
Since this retrospective analysis used commercially available deidentified data, it was deemed by the authors to be exempt from review by an institutional or independent ethics committee.
This study was conducted in accordance with legal and regulatory requirements, as well as scientific purpose, value, and rigour, following the reporting guideline of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. 35

| Study aims
The primary aim of the analysis was to compare hospitalisationrelated outcomes among hospitalised patients with COVID-19 with and without NAFLD/NASH in a matched case-control design to address comorbidities, including those related to metabolic syndrome.Outcomes included overall hospital LOS, admission to ICU, ICU LOS, use of IMV, mortality, and samehospital readmission within 30 and 90 days of the index event (discharge date).

| Statistical analysis
Controls without NAFLD/NASH were matched 1:1 to patients with NAFLD/NASH based on age (within one year), gender, and race/ ethnicity.Using logistic regression, propensity scores were then calculated based on hospital census division, primary insurance type, point of origin, admission type, and the presence of diabetes, obesity, and asthma based on relevant ICD-10 diagnosis codes.A standardised difference of >10% (>.1) between groups was the threshold to determine whether additional baseline variables should be controlled in a regression analysis.Inverse probability of treatment weighting (IPTW) was used to control for differences between cases and controls after matching, such as cirrhosis and malignancies.Data are expressed as frequencies or means with standard deviation unless otherwise indicated.Differences between groups were analysed by two-sided t-test for continuous variables and the chi-square test for categorical or binary outcomes.Linear regression models with IPTW were applied for least-squares mean (LSM) adjusted differences in LOS.Binary logistic regression analysis with IPTW was performed to calculate odds ratios (ORs) for hospitalisation-related outcomes.Statistical analysis was executed using SAS version 9.4 (SAS Institute, Cary, NC, USA).Visualisation of data was performed using GraphPad Prism version 9 (GraphPad Software, San Diego, CA, USA).All p values reported are two-sided and considered significant at a value of <.05.

| Patients with COVID-19 and NAFLD/NASH have higher FIB-4 scores
To correlate chronic liver disease codes to laboratory values and illuminate an influence of fibrosis levels, patients were stratified alongside cut-off values for the non-invasive Fibrosis-4 index for liver fibrosis (FIB-4). 38The FIB-4 score comprises of age, liver function tests    reported in a study using RWD in Germany that reported the prevalence of NAFLD/NASH as 4.7% (215 655 out of 4 580 434 patients in the database). 41It is possible that patients with an ICD-10 diagnosis code for NAFLD/NASH represent those with more advanced cases than in the broader population, which would also explain the higher than expected prevalence of ACLD/cirrhosis noted in our study. 42is study is bound by the limitations of electronic healthcare databases as well as ICD-10 coding and its published underutilisation for subclinical diseases, such as NAFLD. 43,44It is thus probable that the control cohort includes a relevant number of patients with NAFLD/NASH.Moreover, both a timeframe-dependent and regional variability of ICD-10 coding remains unaccounted for.To tackle such a variability bias, the PINC AI™ database uses data from non-profit, non-governmental, community, and teaching hospitals from both rural and urban areas.

| NAFLD/NASH not associated with higher in-hospital mortality or readmission at 30 and 90 days
We have tried to counteract these biases through expanding the timeframe of inclusion and rigorous matching of concomitant diseases.Limitations imposed by a reduced availability of laboratory, histologic, or imaging data pertain, even though our analysis demonstrated that the RWD cohort with coded COVID-19 and NAFLD/ NASH has greater non-invasive FIB-4 scores.
Patients without NAFLD/NASH combined with undiagnosed cases of NAFLD/NASH within the control cohort might have displayed significantly lower levels of fibrosis, leading to the relevance witnessed.In fact, fibrosis might be the underlying factor leading to increased susceptibility. 45,46ven that vaccinations in the US started in December 2020 and slightly overlapped with the observation period of our study (

Conclusions:
In this large real-world cohort of patients hospitalised for COVID-19 in the United States, NAFLD/NASH were obesity-independent risk factors for complicated disease courses.K E Y W O R D S chronic liver disease, COVID-19, fatty liver, MASH, MASLD, metabolic syndrome, NAFLD, NASH, SARS-CoV-2 Key points The available data on the role of NAFLD/NASH in COVID-19 are controversial and might be confounded by concomitant aspects of the metabolic syndrome.This study demonstrates NAFLD/NASH as independent risk factors in a large real-world cohort through exact matching of comorbidities.The study examined adults (≥18 years) diagnosed with COVID-19 (International Classification of Diseases [ICD]-10 diagnosis code U07.1) hospitalised for the first time (index event) and discharged between 1 April 2020 and 31 March 2021.Exclusion criteria are comprised of pregnancy, human immunodeficiency virus (HIV), diagnosed alcohol dependency or abuse, and underlying liver-related diseases (viral hepatitis, biliary cirrhosis [PBC], and sclerosing cholangitis [PSC]).Cases were further defined as those diagnosed with NAFLD (ICD-10 diagnosis code K75.81) or NASH (ICD-10 diagnosis code K76.0) during the index hospitalisation or three months prior.A control cohort was established using identical inclusion criteria and no concomitant coding of either K75.81 and/or K76.0 in primary or secondary position either during index hospitalisation or three months prior to the index event.Patient characteristics included demographics (e.g., age, gender, race/ethnicity) and comorbidities (e.g., chronic liver disease [CLD], liver cirrhosis, alcohol-related disorders, asthma, chronic obstructive pulmonary disease [COPD], hypertension, diabetes, obesity, cardiovascular diseases, cerebrovascular disease, neurological diseases, chronic kidney disease [CKD], malignancies, HIV, prior solid organ or stem cell transplant, rheumatic diseases, sickle cell disease, cystic fibrosis, pulmonary fibrosis); obesity was defined as a body mass index of 30 kg/m 2 or more, irrespective of body composition.Hospitalisation characteristics included dates (e.g., date of hospitalisation, discharge, and readmissions (if applicable), which were used to calculate length of stay (LOS)) and use of healthcare services within the hospital (e.g., ICU, invasive mechanical ventilation (IMV), and use of extracorporeal membrane oxygenation (ECMO)).The study period was prior to the general availability of COVID-19 vaccines, so patients were assumed to be unvaccinated.The predominant COVID-19 variants in the US at the time were B.1.427(epsilon), B.1.429(epsilon), B.1.526(iota), and the alpha lineage B.1.1.7.

(
transaminases), and platelet count.Patients were stratified according TA B L E 1 Characteristics of COVID-19 patients with and without NAFLD/NASH.

4 |
DISCUSS IONUsing a large source of hospital RWD in the US, we tried to dissect the influence of NAFLD/NASH from important comorbidities by carefully matching and adjusting the groups compared.We demonstrated that patients hospitalised with COVID-19 and diagnosed with NAFLD/NASH have a higher FIB-4 score, longer hospital LOS, F I G U R E 1 (A) Patient disposition and matched cohort design.a Eligible patients were aged ≥18 years and diagnosed with confirmed COVID-19 (ICD-10 U07.1 in the primary or secondary position) with index event and discharge between 1 April 2020 and 31 March 2021, excluding pregnant patients and those with human immunodeficiency virus, alcohol dependence/abuse, viral hepatitis, biliary cirrhosis, or sclerosing cholangitis.IPTW, inverse probability of treatment weighting; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; PHD-SR, PINC AI™ Healthcare database, w/o, without.(B) Length of stay in hospitalised patients with COVID-19 with and without NAFLD/NASH.Data refer to index hospitalisation due to COVID-19.Least-square mean adjusted differences are based on linear regression analysis.Regression analysis with Inverse Probability of Treatment Weighting (IPTW), adjusted for cirrhosis and malignancy.CI, confidence interval; LOS, length of stay; LSM, least-square mean.(C) Odds ratios for hospitalisation-related outcomes in COVID-19 patients with and without NAFLD/NASH.Data refer to index hospitalisation due to COVID-19.Odds ratios are based on logistic regression analysis.Regression analysis with Inverse Probability of Treatment Weighting (IPTW), adjusted for cirrhosis and malignancy.CI, confidence interval; COVID-19, coronavirus disease 2019; ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; IMV, invasive mechanical ventilation, OR, odds ratio.longer ICU LOS, and increased use of IMV.Based on data between April 2020 and March 2021, no difference was identified in hospital mortality or readmission rates within 30 or 90 days.The main strengths of our study are the large sample size and rich clinical information that was used to select matched controls from the same time period and adjust for differences remaining after matching, as well as the rigorous statistical methods used.Key limitations to acknowledge are that data from the hospitals that contributed to the PINC AI™ database may not be representative of all hospitals in the US, that outcomes were limited to the hospital inpatient setting, with no ability for follow-up after discharge other than readmissions, use of hospital chargemaster data to indicate healthcare resource utilisation, and inaccuracies or incompletes that may exist in the ICD-10 diagnosis codes reported in hospital chargemaster data.The prevalence of NAFLD/NASH based on ICD-10 diagnosis codes was lower than the estimated prevalence of about 25%-30% in the broader US population, suggesting an underreporting in real-world clinical settings.These findings were similar to those

Standard difference a p value
Baseline characteristics are presented as absolute numbers with percentages or means with standard deviation (SD).P values are based on either chi-square for categorical variables or two-sided t-test for continuous variables.

with COVID-19 and NAFLD/NASH are hospitalised longer, stay in the ICU longer, and are more likely to require mechanical ventilation
17.5%, adjusted OR 1.07 [95% CI 1.01-1.14]);adjustedORswere only calculated after balancing the cohorts for differences in cirrhosis and malignancies by applying IPTW; see Figure1C.
26,27could cause severe COVID-19 outcomes.26,27Similarly,Liet al. used the UK Biobank database to analyse the relationships between ICD-10 diagnosis codes U071 (COVID-19) and K760 (NAFLD) in patients with COVID-19. 26he prevalence of NAFLD was 4.31% in 556 patients with severe COVID-19 > 70 years of age, with no association between NAFLD and severe COVID-19 in multivariate regression.It is unclear if underlying differences in methods or populations can explain these discrepancies.Furthermore, studies are needed to investigate the full clinical and healthcare economic burden of patients with NAFLD/NASH, including in hospital outpatient and other care settings.NAFLD poses a relevant contribution to the public health burden, and societal, political, and medical measures need to be implemented to tackle this challenge. 47ndeed, a recently published study by Shang et al.Stratification and non-invasive estimate of relevant hepatic fibrosis present through Fibrosis-4 Index (FIB-4) in COVID-19 patients with and without NAFLD/NASH.Note: FIB-4 data are stratified along cut-off points 1.3 (absence of advanced fibrosis), 1.3-2.67(indeterminate),and >2.67 (presence of advanced fibrosis) and presented as absolute numbers and percentages.pvalues are based on two-sided t-test for continuous variables.Abbreviations: COVID-19, coronavirus disease 2019; FIB-4, Fibrosis-4 index for liver fibrosis; N, total analysis population; n, subgroup analysis population; NAFLD, non-alcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis.exact matching prior to inverse probability of treatment weighting to avoid balancing out advanced cases of liver fibrosis.
has reported increased overall susceptibility towards incidence and TA B L E 2 a 1:1