Early reports suggest that alcohol misuse increased in 2020 as a result of the COVID-19 pandemic. Using retrospective data from Henry Ford Health System in Detroit MI—an area that experienced an early and severe COVID-19 outbreak—we investigated the impact of the pandemic on alcohol-related liver disease (ARLD) in the summer of 2020 compared with the same period in 2016–2019. Both the number of ARLD admissions and the proportion of total admissions represented by ARLD patients increased significantly in 2020 compared with previous years. The number of ARLD admissions as a proportion of all hospitalizations was 50% higher in 2020 than in 2016–2019 (0.31% vs 0.21%; P = .0013); by September 2020, the number of admissions was 66% higher than previous years. Despite racial and geographical disparities in direct and indirect COVID-related stressors across the Detroit metropolitan area, the demographic profile of ARLD patients did not change compared with previous years.
In March 2020, the state of Michigan experienced an early and severe surge in COVID-19 hospitalizations, primarily concentrated in the Detroit area. A number of statewide public health measures were enacted; hospitals cancelled elective procedures and paused transfers from outside hospitals, and both schools and workplaces were closed. These types of COVID-19-related stressors have been associated with increased alcohol consumption and binge drinking.[1] Alcohol sales increased markedly during the first months of quarantine measures in the USA. Nielsen et al. reported a 54% increase in national sales of alcohol as early as March 2020 and a 262% increase in online sales compared with 2019.[2] Likewise, another study documented self-reported increases in daily alcohol use.[3] Preliminary reports suggest an increase in hospital admissions for alcohol-related liver disease (ARLD),[4,5] but comparisons with prior years have not yet been documented. We investigated whether new hospital admissions resulting from ARLD increased in the summer of 2020 compared with previous years.
We collected retrospective data for the months of May through September of the years 2016 through 2020 at Henry Ford Health System (HFHS), a vertically integrated health system with four acute care hospitals serving the Detroit metropolitan area. The summer period was selected because the initial wave of COVID-19 infections and hospitalizations in the Detroit area had subsided, and elective transfers had resumed at most area hospitals. All study procedures were approved by the HFHS Institutional Review Board; requirements for informed consent were waived because of the observational and de-identified nature of the data. Discharge data, diagnosis and procedure codes, and laboratory data from the 12 months preceding the admission were extracted from electronic medical records.
Patients with probable hospitalizations for ARLD were identified using principal and secondary ICD-10-CM discharge diagnosis codes, classified into three lists: (a) Alcoholic hepatitis with/without ascites (K70.10/K70.11), and Alcoholic hepatic failure with/without coma (K70.40/K70.41); (b) Acute and subacute hepatic failure with/without coma (K72.00/K72.01), and Hepatic failure, unspecified, with/without coma (K72.90/K72.91); and (c) Alcoholic cirrhosis with/without ascites (K70.30/K70.31) and Alcoholic liver disease, unspecified (K70.9). Those with a principal discharge code from List A were automatically classified as ARLD admissions, as were those with a principal discharge code from List B and at least one secondary discharge code from List A or C. Patients with any other combination of the discharge codes of interest were identified for case confirmation by chart review. Criteria included: history of ARLD or heavy alcohol use (≥40 g/day and ≥50 g/day for women and men respectively); laboratory evidence of hepatitis (serum bilirubin ≥3 mg/dL, aspartate aminotransferase [AST] 50–500 IU/mL and AST/aminotransferase [ALT] ratio > 1.5); recent onset or worsening of jaundice with <60 days of abstinence before the onset of the jaundice; and investigation excluded non-alcohol causes of liver disease or hepatitis. All other hospitalizations during the study period were classified as non-cases.
Inverse probability of treatment weight (IPTW) was estimated through propensity score modelling to account for selection bias by age, sex and race. Weighted multivariable logistic regression was used to investigate the association between ARLD admissions and era using SAS 9.4. Chi-squared tests were used to compare the proportion of all hospital admissions resulting from ARLD in the pre-COVID era (2016–2019) versus 2020.
During the study period, there were 116 438 total admissions during May-September 2016–2019 (avg 29 110/5-month period; 5822/month) and 30 060 during May-September 2020 (avg 6012/month). A total of 337 ARLD hospitalizations were identified; 193 that were automatically classified based on discharge diagnosis code(s) in addition to 144 that were confirmed via review of medical records. There were 244 ARLD admissions during the pre-COVID era (avg 12.2/month) and 93 in 2020 (avg 18.6/month). Patient characteristics are detailed in Table 1. In 2020, the mean age of patients was slightly but not significantly higher (47.33 years vs 45.74 years, P = .257). Likewise, the proportion of ARLD admissions who were female (45% versus 41%) was higher than in 2016–2019, as was the proportion of Black/African American admissions (19% vs 11%), but these differences were not significant (P = .445 and .131 respectively).
As shown in Figure 1, there was an average of fewer than 15 monthly ARLD admissions in 2016–2019; in 2020, admissions were slightly lower in May but increased rapidly across the rest of the observation period, with 25 ARLD admissions in September 2020. In multivariable analysis, the proportion of ARLD admissions increased by more than 50% in 2020 compared with 2016–2019, representing 0.31% vs 0.21% of total hospital admissions (P = .0013). Similar to the univariate analysis, we did not observe a significant interaction between sex, race or age and the increased rate of admissions in 2020. A sensitivity analysis showed that the proportion of admissions for non-ARLD decompensated cirrhosis significantly decreased in 2020 (P < .0001, data not shown), suggesting that the observed increase in ARLD admissions was a cause-specific phenomenon and not simply a reflection of an overall increase in liver-related admissions.
Figure 1.
Alcohol-related liver disease admissions to Henry Ford Health System hospitals for May through September for 2016–2019 (monthly average) compared with 2020
We observed a significant increase in the number of ARLD admissions within a Detroit-area health system from May to September 2020. Our findings augment observations made early in 2020 that persons with alcohol use disorder would likely be particularly vulnerable to the impacts of the pandemic,[6] as well as recent work documenting a rise in liver transplants resulting from ARLD during the last year.[7] Given the magnitude of the initial outbreak in south-eastern Michigan, social and economic disruptions persisted throughout most of 2020—we speculate that the consistent increase in admissions during our observation period reflects the cumulative effects of sustained increases in heavy alcohol use among susceptible persons during this timeframe. Likewise, our observation that the demographic profile of ARLD patients admitted in 2020 was similar to that of previous years suggests that these were likely individuals with existing alcohol use disorder, exacerbated by COVID-related stressors. Health systems and community mental health organizations will need to prioritize outreach and resources for these vulnerable patients and populations to reduce the downstream impact of the pandemic.
Funding Information
This manuscript was funded by Henry Ford Health System.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Liver International. 2022;42(4):762-764. © 2022 Blackwell Publishing
